THESE DE DOCTORAT DE L’UNIVERSITE PARIS 6 Ecole doctorale Diversité du Vivant

Spécialité

ECOLOGIE

Présentée et Soutenue publiquement par

MICKAËL HENRY

Le 04 Novembre 2005

Pour obtenir le grade de DOCTEUR DE L’UNIVERSITE PARIS 6

LE DECLIN DES POPULATIONS DE CHAUVES-SOURIS FRUGIVORES EN FORET NEOTROPICALE FRAGMENTEE

– CONSEQUENCES SUR LA DISPERSION DES GRAINES.

Devant le jury composé de :

Pierre CHARLES-DOMINIQUE, DR CNRS, Cayenne ………………. Directeur de thèse

Elisabeth KALKO, Professeur, Université de Ulm, Allemagne ……. Codirectrice de thèse

Robert BARBAULT, Professeur, Université Paris 6 ………………………………. Président

Theodore FLEMING, Professeur, Université de Miami, USA …………………. Rapporteur

Jean-Louis MARTIN, DR CNRS, CEFE, Montpellier ………………………. Rapporteur

Jean-François COSSON, CR INRA, Montpellier …..……………………….…. Examinateur

Thèse effectuée au Département d’Ecologie et Gestion de la Biodiversité

UMR 5176, CNRS-MNHN – 4, avenue du petit château ; 91800 Brunoy ; France.

2

A ma compagne Solenne

et notre fils Elouann.

3

REMERCIEMENTS

Je tiens à introduire ce travail de thèse en remerciant les personnes qui m’ont assisté dans ces trois années de labeurs, tant sur le plan scientifique que technique, administratif, moral ou matériel. A tous, membres de mon comité et de mon jury de thèse, collègues du laboratoire d’Ecologie Générale de Brunoy, compagnons des missions de terrain, parents, amis, et ceux que j’aurais malencontreusement oubliés dans les lignes suivantes, je leur adresse mes plus sincères remerciements.

Je remercie vivement Robert Barbault, président du jury, et Ted Fleming et Jean-Louis Martin qui ont accepté de se constituer rapporteurs de cette thèse. Je suis conscient de l’investissement de temps que représente l’évaluation d’une thèse, et leur en suis d’autant plus reconnaissant que leurs fonctions respectives ne leur laisse que peu de flexibilité pour un tel supplément de travail.

J’exprime une reconnaissance toute particulière envers mon directeur de thèse Pierre Charles-Dominique pour m’avoir offert cette opportunité d’étude et pour s’y être investi avec autant d’intérêt. J’ai apprécié sa patience, son optimisme, et son goût à partager sa connaissance de la biologie tropicale.

Je remercie chaleureusement ma codirectrice Elisabeth Kalko pour sa confiance, sa patience, ses conseils avisés et ses encouragements rassurants. Elle m’a fait profiter de son sens du pragmatisme et appris à structurer mes écrits autour d’un fil rouge plutôt que dans un sac de nœuds.

Je remercie Pierre-Michel Forget pour m’avoir offert les moyens de réaliser mon projet d’étude sur les chauves-souris de Guyane, et aussi d’en promouvoir les résultats au cours de divers congrès internationaux. J’espère que j’aurais su me montrer à la hauteur de sa confiance et de ses encouragements.

Je dois exprimer ma profonde gratitude aux deux autres membres de mon comité de thèse, Jean-François Cosson et Jean-Marc Pons, également les deux compagnons de terrain qui m’ont introduit pour la première fois à la « jungle » Guyanaise en 1999. Sources inépuisables de bons conseils, ils m’ont aidé à progresser avant et pendant la thèse. J’espère que ces interactions enrichissantes se poursuivront aussi après.

Martine Perret m’a aimablement accueilli au sein du laboratoire d’Ecologie Générale de Brunoy. Outre ses conseils avisés sur le plan scientifique, j’ai bénéficié à plusieurs reprises de son habileté à dénouer les situations administratives complexes.

Je remercie Ted Fleming, Tatyana Lobova, Scott Mori et Jérôme Chave qui ont généreusement accepté de commenter les chapitres de cette étude et d’en corriger l’anglais.

Ce travail aurai probablement duré une année supplémentaire sans l’assistance, l’expertise, l’efficacité, l’acharnement, en un mot l’abnégation de Sylvie Jouard. Je lui dois des heures à arpenter la forêt, de jour comme de nuit et par tous les temps, et des journées entières à trier et identifier des graines à la lueur de sa loupe binoculaire au laboratoire.

Les diverses interactions que j’ai pu avoir avec les chercheurs et étudiants du laboratoire d’Ecologie Générale de Brunoy m’ont été profitables. Certains d’entre eux ont contribué à améliorer cette étude de façon substantielle par leurs commentaires et suggestions, parfois sans même le savoir par quelque remarque anodine en apparence mais déterminante en réalité. Je pense en particulier à Natalia Norden, Marguerite Delaval, François Feer, Jean-

4

François Ponge, Sandra Ratiarison, Bernard Riéra, Fabienne Aujard, Christian Erard, Marc Théry, Françoise Bayart, Florence Cayetanot, Maud Séguy, Florence Némoz, Sandrine Salmon, Nicolas Bernier, Florence Moyen et Sylvie Peronny. Je garde une pensée émue et respectueuse de mes quelques discussions avec André Brosset.

J’ai bénéficié des aimables conseils des étudiants et chercheurs du groupe de recherche d’Elisabeth Kalko au Département d’Ecologie Expérimentale de l’Université de Ulm, Allemagne. Moritz Weinbeer et Njikoha Ebigbo m’ont familiarisé avec respectivement les analyses de domaines vitaux et les analyses de patron spatial de dispersion des graines. Je remercie également Marco Tschapka, Christoph Meyer et Stefan Klose pour nos discussions fructueuses.

Ce travail de thèse est largement teinté de l’influence de mes précédents professeurs, Marc Colyn (univ. Rennes I) et Don Thomas (univ. Sherbrooke, Québec), les premiers « maîtres spirituels » qui m’ont patiemment initié à la science chiroptérologique.

Les séjours de terrains, souvent éprouvants, seraient moins exaltants sans la présence réconfortante et la bonne humeur des collègues. Ainsi le camp des Nouragues doit-il beaucoup à Patrick Châtelet. Tout se passe tellement mieux avec sa sérénité. Il restera indissociable des meilleurs souvenirs que je garderai de ce petit coin de forêt primaire. Cyrielle Bobiller, Alexandre Cartier et Jean-Louis Filiol ont aussi contribué au bon déroulement des collectes de données aux Nouragues, et notamment pour les captures de chauves-souris et les suivis télémétriques. Le travail de terrain, ce fut également une multitude de souvenirs ineffables partagés avec Natalia Norden, François Feer, Holger Teichert, Heiko Hentrich, Wemo Betian, Sophie Bentz, François Catzeflis, Olivier Cleassens, Gilles Cheylan, Arnaud Lyet, Jean-Christophe De Massary, Amadou, Koffi, Van Raymond et la Belle Cabresse®.

Pour le dépouillement de mes échantillons, le traitement des données et la gestion des crises informatiques, j’ai bénéficié de l’aide de Laurent Dhennin, Adeline Caubert, Pierre Belbenoit, Roger Botalla et Isabelle Hardy.

Je remercie Françoise Bertay, Mireille Charles-Dominique, Marielle Peroz, Nadine Desplanques, Manuela Da Fonseca et Marie-Ange Delamare, pour leur disponibilité, leur bienveillance et leur précieuse assistance tout au long du combat bureaucratique que chaque thésard doit traverser.

Les missions en Guyane ont été financées par l’UMR CNRS/MNHN 5176 et, pour les travaux au barrage de Petit-Saut, par la convention MNHN/EDF n° CQZH 1294. Je remercie également la Fondation des Treilles pour la bourse qu’elle m’a attribué.

A l’issus de ces longues années d’étude, mes pensées les plus affectueuses vont vers ma famille et mes amis, qui ont tous souffert de mes absences prolongées, mais sans jamais manquer de m’encourager et de m’épauler. Merci à mes parents, mes grands-parents et ma sœur Klervi pour leurs sacrifices, leur dévouement et leur soutien en toute circonstance. Merci à Elisabeth, Sylvie et Denis pour leur simplicité et tous leurs efforts proposés spontanément pour soulager mon emploi du temps. Merci à mes oncles, tantes et cousins, et à Nicolas, Virginie, Florent et Christèle pour leur sympathie et leur soutien. Merci à Olivier pour ces deux décennies d’amitié.

Par dessus tout, merci Solenne pour tout ce que tu m’as offert sans compter, ta patience, ton équilibre, les nombreux renoncements, ton amour et Elouann…

5

« Petits camps de bois rond plantés à la diable.

Baraques mal bâties par des gens qui tenaient une hache

pour la première fois. L’argile et la mousse entre les billes pour

affronter l’hiver furibond. Hommes épuisés venus de partout

réunis là par une même folie. Certains assez déments pour avoir

traîné jusqu’en ces lieux perdus leur femme et leurs enfants.

Ainsi naissaient des villages partout où l’on flairait le métal

jaune.

La terre s’ouvrait pour cracher son or, les forêts s’ouvraient

pour que naissent les villages. […] Il semblait que le monde

entier repoussait ses limites vers le nord, comme si la terre,

soudain, eût manqué sous les pas des hommes. »

Bernard Clavel

(« L’or de la terre » ; l’épopée de la ruée vers l’or dans le nord

canadien).

6

TABLE DES MATIERES INTRODUCTION GENERALE.............................................................................................9

FORAGING STRATEGY AND BREEDING CONSTRAINTS OF RHINOPHYLLA PUMILIO, A FRAGMENTATION-SENSITIVE PHYLLOSTOMID FRUIT BAT. ......22

ABSTRACT .............................................................................................................................22 INTRODUCTION ......................................................................................................................23 METHODS ..............................................................................................................................26

Study site...........................................................................................................................26 Bat captures and radio-tracking.......................................................................................27 Data collection .................................................................................................................29 Range size .........................................................................................................................29 Foraging strategy .............................................................................................................30 Effect of reproductive status on movement pattern ..........................................................31

RESULTS ................................................................................................................................32 Range size .........................................................................................................................32 Foraging strategy .............................................................................................................36 Effect of reproductive status on movement pattern ..........................................................36

DISCUSSION ...........................................................................................................................41 The movement pattern of R. pumilio.................................................................................41 Evidences for food intake increase in lactating females ..................................................43 Effect of reproductive status on movement pattern ..........................................................45 Conclusions on fragmentation sensitivity of R. pumilio ...................................................47

REFERENCES..........................................................................................................................48

THE ROLE OF HABITAT FRAGMENTATION AND FOOD AVAILABILITY IN LIMITING POPULATIONS OF UNDERSTORY FRUIT BATS IN FRENCH GUIANA. .................................................................................................................................59

ABSTRACT .............................................................................................................................59 INTRODUCTION ......................................................................................................................60 METHODS ..............................................................................................................................61

Study area and time periods .............................................................................................61 Bat surveys........................................................................................................................62 Landscape descriptors ......................................................................................................64 Movement pattern of R. pumilio .......................................................................................66 Food availability...............................................................................................................67 Determinants of bat abundances ......................................................................................68

RESULTS ................................................................................................................................69 Bat surveys........................................................................................................................69 Movement pattern and landscape descriptors..................................................................69 Food availability...............................................................................................................72 Determinants of bat abundances ......................................................................................72

DISCUSSION ...........................................................................................................................75 Food availability hypothesis vs. habitat connectivity hypothesis.....................................75 Further decomposing fragmentation-sensitivity: the role of foraging strategies.............76 Fragmentation age and sustainability of populations ......................................................78 Implications for conservation...........................................................................................79

REFERENCES..........................................................................................................................80 APPENDIX ..............................................................................................................................87

7

CONSEQUENCES OF AN EXPERIMENTAL DISTURBANCE OF BAT ACTIVITY ON THE SEED RAIN PATTERN OF SOME KEYSTONE BAT PLANTS IN A NEOTROPICAL RAIN FOREST. .......................................................................................89

ABSTRACT .............................................................................................................................89 INTRODUCTION ......................................................................................................................90 METHODS ..............................................................................................................................92

Study area .........................................................................................................................92 Study species.....................................................................................................................93 Design of the experiment ..................................................................................................94 Seed rain sampling ...........................................................................................................95 Bat disturbance.................................................................................................................96 Measures of seed species diversity ...................................................................................97 Measures of seed limitation..............................................................................................97 The effects of experimental disturbance on seed limitation .............................................98 Seed species grouping ......................................................................................................99

RESULTS ..............................................................................................................................100 General seed rain description ........................................................................................100 Bat disturbance...............................................................................................................102 Seed source and dispersal limitation..............................................................................107

DISCUSSION .........................................................................................................................109 General seed rain description ........................................................................................110 The effect of bat disturbance on seed diversity and seed limitation...............................112 The origin of seed limitation: restricted seed source or seed dispersal? .......................113 Are bats effective dispersers? .........................................................................................114 Conclusions on seed dispersal in a fragmentation context ............................................115

REFERENCES........................................................................................................................115 APPENDIX ............................................................................................................................122

DISCUSSION GENERALE ................................................................................................125 LE PATRON D’ACTIVITE DE RHINOPHYLLA PUMILIO ..............................................................125

La classification des stratégies de quête alimentaire. ....................................................125 La stratégie de quête alimentaire et les capacités mnésiques. .......................................126

LE DECLIN DES CHAUVES-SOURIS FRUGIVORES EN MILIEU FRAGMENTE...............................127 Fragmentation et compétition interspécifique................................................................127 Mesurer la sensibilité à la fragmentation.......................................................................131

CONSEQUENCES DU DECLIN DES CHAUVES-SOURIS FRUGIVORES SUR LA PLUIE DE GRAINES 131 CONCLUSIONS SUR LES IMPLICATIONS EN CONSERVATION...................................................133 REFERENCES........................................................................................................................136

ANNEXES .............................................................................................................................139 ANNEXE 1 : PUBLICATION DELAVAL ET AL. (2005). ...........................................................140 ANNEXE 2 :COMMUNICATION AFFICHEE PRESENTEE A L’ATBC 2003. ...............................156 ANNEXE 3 : LISTE DES COMMUNICATIONS SCIENTIFIQUES EN CONGRES. .............................160

ABSTRACT ..........................................................................................................................161

RESUME ...............................................................................................................................162

8

Introduction générale

Philodendron grandifolium dont les fruits sont occasionnellement consommés et les graines dispersées par la petite chauve-souris de sous-bois Rhinophylla pumilio (dessin S. Jouard).

9

INTRODUCTION GENERALE Les chauves-souris (ordre des chiroptères) sont largement répandues à travers le monde, à

l’exception des régions polaires, et représentent au sein de la classe des mammifères les

richesses génériques et spécifiques les plus élevées après les rongeurs. Environ 1100 espèces

appartenant à 19 familles sont actuellement recensées, soit le quart des mammifères actuels

(Simmons et Conway 2003, Wilson et Reeder 1993). Les chiroptères se subdivisent en deux

sous-ordres, les microchiroptères (18 familles, environ 917 espèces) et les mégachiroptères

(une seule famille, environ 188 espèces). Tandis que les microchiroptères sont présents sur

tous les continents et ont adopté une variété de spécialisations alimentaires (insectivores,

frugivores, carnivores, nectarivores, piscivores, hématophages ; Findley 1993), le groupe des

mégachiroptères se limite aux espèces frugivores et nectarivores de l’ancien monde. Les

microchiroptères se distinguent des mégachiroptères essentiellement par la présence d’un

système d’écholocalisation sophistiqué. En générant des ultrasons depuis le larynx et en

percevant les échos renvoyés par leur environnement, les chauves-souris localisent proies et

obstacles dans l’obscurité. Les mégachiroptères auraient perdu leur système

d’écholocalisation au cours de leur radiation évolutive à partir d’un ancêtre commun aux

microchiroptères (Simmons et Conway 2003, Simmons et Geisler 1998). Ils utilisent donc

uniquement la vision et l’olfaction pour localiser leurs items alimentaires.

Au niveau mondial, le principal centre de diversité des microchiroptères s’étend de

l’Amérique Centrale au nord de l’Amazonie (Hutson et al. 2001). La région du Nicaragua est

plus particulièrement désignée comme le « hotspot » mondial de la diversité chiroptérienne

avec 58 genres recensés. Dans ces régions d’Amérique centrale et d’Amazonie, plus de 70

espèces de chauves-souris peuvent vivre en sympatrie (Arita 1997, Bernard et Fenton 2002,

Bonaccorso 1979, Delaval et al. 2005, Fleming et al. 1972, Handley 1967, Kalko et al. 1996,

Lim et Engstrom 2001, Reis et Peracchi 1987, Simmons et Voss 1998), un chiffre inégalé par

les autres groupes de mammifères. Une telle diversité ne peut être assurée que par un partage

complexe des niches écologiques. Outre la variété des régimes alimentaires, les différences de

sélection d’habitat et de mode d’alimentation peuvent promouvoir la coexistence de tant

d’espèces.

La morphologie de leurs ailes et la structure de leurs signaux d’écholocalisation prédisposent

les chauves-souris à préférer certains habitats et modes d’alimentation plutôt que d’autres

(Aldridge et Rautenbach 1987, Norberg et Rayner 1987). Plus les ailes sont courtes et larges

10

(faible rapport d’aspect) et la masse corporelle faible par rapport à la surface alaire (faible

charge alaire), plus le vol sera manoeuvrable et adapté au sous-bois de fort encombrement

végétal. Les espèces insectivores possédant une telle morphologie, par exemple, produisent

des signaux à haute fréquence et plutôt longs (ou à fréquence modulée) pour une meilleure

capacité à discerner les insectes volants des obstacles. Les espèces insectivores adaptées à la

chasse en milieu ouvert, au contraire, ont des ailes plus longues et étroites leur conférant un

vol rapide, et produisent des signaux à basse fréquence et plutôt courts. Les chauves-souris

peuvent ainsi être regroupées en une dizaine de guildes, i.e. groupes d’espèces exploitant les

mêmes ressources et de façon similaire, telles que les insectivores de milieu ouvert, les

insectivores de milieu encombré, les frugivores préférant les fruits de canopée, les frugivores

préférant les fruits de sous-bois, etc. (Kalko 1998, Kalko et al. 1996).

La diversité taxonomique et écologique des chauves-souris en milieu tropical est associée à

une diversité des interactions avec d’autres organismes. Parmi les plus étudiées et les plus

révélatrices de l’utilité des chauves-souris dans le fonctionnement des écosystèmes, la

pollinisation et la dispersion des graines sont régulièrement citées en exemple (voir les revues

de Dumont [2003] et Helversen et Winter [2003]). Ces deux interactions sont de type

mutualiste. Les plantes offrent aux chauves-souris de l’énergie sous forme de nectar ou de

pulpe de fruit, et en contrepartie, les chauves-souris disséminent le matériel génétique des

plantes en transportant des grains de pollen vers des fleurs à féconder ou les graines vers des

sites propices à la germination. Geiselman et al. (2002) ont recensé en littérature pas moins de

5100 interactions impliquant d’une part 109 espèces de chauves-souris néotropicales et

d’autre part 476 espèces de graines et 384 espèces de pollen représentant 97 familles

végétales, soient autant de situations où les chauves-souris sont susceptibles de favoriser les

flux de gènes de plantes.

Or, la déforestation progresse à un rythme sans précédent dans les régions tropicales et

menace la diversité des chauves-souris (Brosset et al. 1996, Cosson et al. 1999, Estrada and

Coates-Estrada 2001, 2002, Estrada et al. 1993, Gorresen et Willig 2004, Kalko 1998, Pons et

Cosson 2002, Schulze et al. 2000) ainsi que l’équilibre de leurs interactions avec les plantes

(e.g. Medellín et Gaona 1999, Quesada et al. 2003). Entre 1990 et 2000, les pays d’Amérique

du sud ont perdu en moyenne 0,4% de leur surface boisée chaque année, soit au total près de

37000 km²/an dont 63% sont imputables au seul Brésil (FAO 2005). Ces forêts subissent des

altérations variées et voient leur surface se réduire face au développement des plantations

(hévéas, canne à sucre, café, cacao), des pâtures, des abatis et cultures sur brûlis,

particulièrement le long des récents axes de circulation trans-amazoniens (FAO 2005, Skole et

11

Tucker 1993). La déforestation ne se résume pas à une simple perte d’habitat pour les

animaux forestiers. La déforestation fragmente leur habitat en isolant des patchs de forêt

résiduelle par des étendues inhospitalières. La fragmentation de l’habitat telle qu’elle a été

définie par Lovejoy et al. (1986) se réfère à « tout processus conduisant à la réduction de la

taille d’un habitat et aboutissant à la création d’une ou plusieurs parcelles d’habitat de plus

petite taille ». En outre, la fragmentation de la forêt est associée à de multiples perturbations

indirectes telles que la modification des conditions microclimatiques entraînant une mortalité

accrue des arbres le long des bordures, l’invasion d’espèces végétales et animales adaptées

aux perturbations, le développement de la chasse et du braconnage, etc. (Tabarelli et al. 2004).

La déforestation et la fragmentation de la forêt sont considérées comme les principales

menaces pour la diversité des régions tropicales (Bierregaard et al. 2001, Turner 1996).

Parmi les 288 espèces de chauves-souris recensées dans les régions néotropicales, 57 (19,8%)

sont déclarées menacées et 60 autres (20,8%) sont susceptibles de le devenir (Hutson et al.

2001). Les chauves-souris sont d’autant plus sensibles aux perturbations et instabilités de leur

environnement qu’elles présentent une faible capacité de résilience démographique du fait de

leur stratégie de reproduction de type « K » (sensu Pianka 1970) : une longue espérance de vie

(>20 ans pour certaines espèces) et un taux de reproduction limité (le plus souvent un seul

jeune par portée, et une à deux portées par an ; Hayssen et Kunz 1996). Ces traits d’histoire de

vie font des chiroptères une exception parmi les petits mammifères de cette taille, favorisant

généralement une stratégie de type « r » (de nombreux jeunes et une espérance de vie courte ;

e.g. rongeurs).

La protection des chauves-souris est entre autres limitée par le manque d’informations

biologiques et écologiques les concernant. Dans son plan d’action pour les microchiroptères

(Hutson et al. 2001 : p 56), l’IUCN recommande donc d’« initialiser des recherches sur divers

aspects de la biologie et de l’écologie des chauves-souris, y compris l’étude de leur

comportement de quête alimentaire, des suivis de populations pour en évaluer les fluctuations

et les limites de leur viabilité, l’étude de leur rôle dans les processus de maintien des

écosystèmes […] ». L’objectif général de cette étude est d’examiner ces trois aspects clés

pour la conservation des chauves-souris, en focalisant plus spécifiquement sur les chauves-

souris de la guilde des frugivores de sous-bois, un groupe d’espèces sensibles à la

fragmentation de l’habitat forestier (Cosson et al. 1999). Dans un premier temps, le patron

d’activité et de quête alimentaire d’une espèce modèle en particulier a été décrit. Ensuite, en

considérant davantage d’espèces de cette guilde, les fluctuations de populations sur une

12

période de 10 ans en milieu fragmenté ont été étudiées en relation avec la structure du

paysage. Finalement, les conséquences du déclin de ces espèces sur la dispersion des graines

ont été examinées. Les travaux de terrains ont été réalisés en Guyane Française entre 2002 et

2004. La Guyane a une superficie de 88000 km² dont 90% sont occupés par la forêt primaire.

Elle constitue l’un des derniers blocs de forêt intacte de la région Amazonienne (Turner et

Corlett 1996). Avec le Surinam et l’Uruguay, la Guyane est l’un des rares territoires

d’Amérique du Sud n’ayant pas subi de réduction significative du couvert forestier entre 1990

et 2000 (FAO 2005). Le plateau des Guyanes figure également parmi les quelques 19 zones

jugées importantes pour la conservation du plus grand nombre possible de genres de

microchiroptères à travers le monde (Hutson et al. 2001).

En Guyane Française, les chauves-souris frugivores de sous-bois, toutes des phyllostomidés,

rassemblent d’une part les espèces de la sous-famille des carollinés (Carollia perspicillata, C.

brevicauda, Rhinophylla pumilio) et d’autre part deux espèces de la sous-famille des

stenodermatinés, soient Sturnira tildae et S. lilium. Ces cinq espèces consomment

essentiellement les fruits des plantes arbustives des genres Piper (Piperaceae), Solanum

(Solanaceae) et Vismia (Clusiaceae), ainsi que les fruits des épiphytes du genres Philodendron

(Araceae) et de la famille des cyclanthacées, qui se développent sur les troncs en sous-bois

(en général à moins de 8 m du sol). Au sein des chauves-souris frugivores de sous-bois

(« understory fruit bats »), on distinguera la spécialiste des épiphytes R. pumilio (« epiphyte

specialist ») des quatre autres espèces qualifiées de frugivores d’arbustes (« shrub-

frugivores »). En effet, R. pumilio est la seule espèce dont le régime alimentaire soit

principalement composé d’infructescences d’épiphytes.

Dans le premier chapitre, le patron d’activité de R. pumilio sera étudié, avec une attention

particulière aux contraintes de l’allaitement pour les femelles. Le patron d’activité se réfère

non seulement à la mobilité des individus dans l’espace (taille du domaine vital et des aires

d’activité), mais aussi à leur stratégie de quête alimentaire, i.e. la façon dont ils se déplacent

dans le paysage et dont ils utilisent leur budget temporel pour s’alimenter. Ces

caractéristiques sont importantes car elles peuvent déterminer la capacité des individus à

s’accommoder aux discontinuités de leur habitat. En outre le patron d’activité des femelles

allaitantes sera comparé à celui des non reproductrices, car la théorie laisse présager une

réorganisation post-partum des budgets énergétiques et temporels des femelles. En effet, Le

coût énergétique de l’allaitement augmente avec la diminution de la taille corporelle. Chez les

petits mammifères, le rendement quotidien de lait relativement à leur masse corporelle est si

13

important que les femelles doivent augmenter largement, voire doubler, leur consommation

journalière de nourriture. Les petits mammifères sont alors appelés « reproducteurs à

revenus » (« income breeders »), par opposition aux grands mammifères dits « reproducteurs

à capital » (« capital breeders ») dont l’accumulation de réserves graisseuses tend à

compenser la recrudescence des dépenses énergétiques imposées par la production de lait

(Jonsson 1997). En tant que reproductrices à revenus et stratèges K à la fois, les chauves-

souris doivent en période d’allaitement intensifier leur quête alimentaire tout en prodiguant

des soins de qualité à leur jeune (Kunz et Stern 1995, Kurta et al. 1989, Wilde et al. 1995).

Pour concilier ces deux contraintes d’ordres énergétique et temporel, les femelles sont

susceptibles de réorganiser leur patron d’activité nocturne et de présenter une moindre

flexibilité face à la fragmentation de l’habitat. Pour dresser un portrait du patron d’activité

typique de R. pumilio, des suivis télémétriques ont été entrepris dans la forêt primaire de la

réserve des Nouragues.

Le deuxième Chapitre a pour objectif d’examiner les rôles respectifs du degré de connectivité

de l’habitat et de la disponibilité alimentaire dans le maintien en milieu fragmenté des

populations de chauves-souris frugivores de sous-bois. Il existe plusieurs approches pour

appréhender l’effet de la fragmentation sur les populations et communautés animales.

L’approche la plus simple s’inspire de la relation espèces-aire (« species-area » relationship,

Gleason 1922, 1925, Williams 1943). De la même façon que la richesse spécifique observée

lors d’un échantillonnage augmente avec la taille de la surface d’échantillonnage, la richesse

spécifique abritée par un fragment d’habitat devrait augmenter en fonction de sa taille. Ainsi,

l’hypothèse de l’échantillonnage aléatoire (« random sampling hypothesis ») prédit que les

communautés habitant des fragments devraient s’apparenter à autant de sous-échantillonnages

aléatoires de la communauté d’origine en milieu continu (Andrén 1996). Cette approche

rattache l’effet de la fragmentation à la simple perte d’habitat. Cependant, la diminution de la

richesse spécifique en milieu fragmenté est empiriquement plus marquée que ce que prédirait

un simple échantillonnage aléatoire (Andrén 1996), en partie parce que les probabilités

d’extinction sont plus grandes pour les petites populations habitant les fragments.

Une approche alternative consiste à appliquer la théorie de la biogéographie insulaire

(MacArthur et Wilson 1963, 1967) qui présuppose que le nombre d’espèces habitant une île

résulte d’un équilibre dynamique entre les taux d’immigration et d’extinction. Ainsi, par

analogie aux îles, la richesse spécifique d’un fragment d’habitat sera d’autant plus faible à

l’équilibre que l’île sera isolée et de petite taille (As et al. 1997). Cette « métaphore insulaire »

14

a cependant des limites car il y a une discordance spatio-temporelle entre les processus

régissant la richesse spécifique des îles véritables (échelle continentale, long terme) et celle

des fragments forestiers (échelle locale ou régionale, court à moyen terme). Ainsi, selon les

organismes étudiés, la taille des domaines vitaux peut dépasser celle des fragments. La notion

d’« isolation », qui peut se mesurer par la simple distance au continent dans le cas d’une île

océanique, prend alors un caractère subjectif dans un contexte de fragmentation où les

fragments d’habitat sont suffisamment nombreux et proches les uns des autres pour jouer le

rôle de passage à gué (« stepping stones ») pour les organismes étudiés. Dans ce cas, la

présence d’une espèce donnée dans un milieu fragmenté peut dépendre largement de la forme,

de la configuration spatiale et de la distribution relative des fragments et de l’habitat continu

(Taylor 1987ab).

La discipline récente de l’écologie du paysage utilise des outils modernes pour contourner ces

problèmes conceptuels. A partir d’images satellitaires, les systèmes d’information

géographique permettent de calculer des indices de paysage complexes intégrant à la fois les

notions de taille et d’isolation des fragments d’habitat. Le deuxième Chapitre utilise de tels

indices, parallèlement à des données d’inventaires de ressources alimentaires, pour mieux

cerner les caractéristiques du paysages favorisant ou limitant le maintien en milieu fragmenté

des populations de chauves-souris frugivores de sous-bois. Ces indices sont construits d’après

le patron d’activité de R. pumilio (Chapitre 1) considérée ici comme espèce modèle. Cette

étude a été réalisée à Saint-Eugène, dont la forêt primaire a été fragmentée par la mise en eau

d’un barrage hydroélectrique en 1994. La matrice étant un lac de retenue, elle n’abrite pas de

végétation secondaire susceptible d’influencer ou de confondre les effets de la fragmentation

en soit (Leigh et al. 2002). Les inventaires de chauves-souris s’y sont échelonnés sur 10 ans.

Le troisième Chapitre s’attachera à étudier les conséquences du déclin des populations de

chauves-souris frugivores sur le patron de pluie de graines des plantes qu’elles consomment

habituellement. En effet, attendu que les chauves-souris sont affectées par la fragmentation de

l’habitat (Chapitre 2), des répercussions négatives sont à prévoir sur la dispersion des graines.

Or, la dispersion des graines est un mécanisme crucial dans le fonctionnement des

écosystèmes car elle constitue le lien unique entre les plantes reproductrices et l’établissement

des plantules, et influence ainsi en grande partie le recrutement reproductif et le patron

démographique des plantes (Herrera et al. 1994, Nathan et Muller-Landau 2000, Schupp

1995).

15

Selon l’hypothèse de l’évasion (« escape hypothesis », Howe et Smallwood 1982), une

dispersion efficace permet aux graines d’échapper au taux élevé de mortalité des graines

autour des plantes parent résultant de l’action accrue des prédateurs, des pathogènes et de la

compétition intraspécifique (Connell 1971, Janzen 1970). En outre, la dispersion à longue

distance favorise la colonisation d’habitats éloignés ou isolés (« colonisation hypothesis »,

Howe et Smallwood 1982). La colonisation par la dispersion est particulièrement importante,

car c’est précisément par ce biais que la diversité des plantes peut être maintenue ou rétablie

dans les forêt fragmentées (Hanski 1994). Dans le Chapitre 3, l’accent sera donc mis sur les

petites graines endozoochores (<5 mm) car elles sont susceptibles de parcourir de plus

grandes distances que les graines synzoochores avant d’être rejetées. En effet, les premières

sont avalées avec la pulpe et retenues dans le tractus digestif pendant une quinzaine de

minutes à plusieurs heures avant d’être déféquées, ce qui leur donne l’opportunité d’être

dispersées à des centaines, voire des milliers de mètres de leur origine. Les secondes, plus

grosses, sont au contraire rejetées immédiatement lors de la consommation du fruit, dans un

rayon d’autant plus restreint autour de la plante parent que ce fruit est lourd et

énergétiquement coûteux à transporter (quelques dizaines ou centaines de mètres).

Le patron de pluie de graines (diversité, densité et homogénéité) peut être étudié au moyen de

collecteurs de graines, de simples bâches ou tissus tendus horizontalement au dessus du sol de

façon à intercepter les graines dans leur chute et les préserver de l’action des animaux se

déplaçant au sol. La problématique posée dans ce chapitre peut être abordée par une approche

descriptive étudiant les variations naturelles de la pluie de graines à travers de nombreux sites

d’un milieu fragmenté. Cependant, quantifier convenablement la pluie de graines dans un site

nécessite a priori une grande quantité de collecteurs. Il a donc été choisi de privilégier une

approche expérimentale, plus économe en logistique, impliquant une perturbation artificielle

de l’activité des chauves-souris. Cette perturbation a été réalisée au moyen de sessions de

captures au filet localement intensives autour des collecteurs.

A l’issus de ces chapitres, une conclusion dressera un bilan des résultats obtenus, en dégagera

la portée et les implications en conservation, et fournira des éléments de discussion pour

orienter de futures études. Ce travail s’appuie en partie sur les informations synthétisées dans

Delaval et al. (2005) sur la communautés des chauves-souris de la Réserve Naturelle des

Nouragues, en Guyane (Annexe 1). Les résultats présentés à travers les trois chapitres

suivants ont fait l’objet de plusieurs communications en congrès internationaux (Annexes 2 et

3).

16

RÉFÉRENCES Aldridge, H., et I. L. Rautenbach. 1987. Morphology, echolocation and resource partitioning

in insectivorous bats. Journal of Animal Ecology 56: 763–778.

Andrén, H. 1996. Population Responses to Habitat Fragmentation - Statistical Power and the

Random Sample Hypothesis. Oikos 76: 235–242.

Arita, H. T. 1997. Species composition and morphological structure of the bat fauna of

Yucatan, Mexico. Journal of Animal Ecology 66: 83–97.

As, S., J. Bengtsson, et T. Ebenhard. 1997. Archipelagoes and theories of insularity.

Ecological Bulletin 46: 88–116.

Bernard, E., et M. B. Fenton. 2002. Species diversity of bats (Mammalia: Chiroptera) in forest

fragments, primary forests, and savannas in central Amazonia, Brasil. Canadian

Journal of Zoology 80: 1124–1140.

Bierregaard, R. O., Jr., C. Gascon, T. E. Lovejoy, et R. Mesquita (éds.). 2001. Lessons from

amazonia: the ecology and conservation of a fragmented forest. Yale University

Press, New Haven.

Bonaccorso, F. J. 1979. Foraging and reproductive ecology in a Panamanian bat community.

Bulletin of the Florida State Museum, Biological Sciences 24: 359–408.

Brosset, A., P. Charles-Dominique, A. Cockle, J.-F. Cosson, et D. Masson. 1996. Bat

communities and deforestation in French Guiana. Canadian Journal of Zoology 74:

1974–1982.

Connell, J. H. 1971. On the role of natural enemies in preventing competitive exclusion in

some marine mammals and in rain forest trees. Pp. 298–312 dans Dynamics of

populations (P. J. D. Boer, et G. Gradwell, éds.). PUDOC, Wageningen, The

Netherlands.

Cosson, J.-F., J.-M. Pons, et D. Masson. 1999. Effects of forest fragmentation on frugivorous

and nectarivorous bats in French Guiana. Journal of Tropical Ecology 15: 515–534.

Delaval, M., M. Henry, et P. Charles-Dominique. 2005. Interspecific competition and niche

partitioning: example of a Neotropical rainforest bat community. Revue d'Ecologie

(Terre et Vie) 60 : 149–166.

Dumont, E. R. 2003. Bats and fruits: an ecomorphological approach. Pp 398–429 dans Bat

Ecology (T. H. Kunz, et M. B. Fenton, éds.). The University of Chicago Press.

Estrada, A., et R. Coates-Estrada. 2001. Bat species richness in live fences and in corridors of

residual rain forest vegetation at Los Tuxtlas, Mexico. Ecography 24: 94–102.

17

Estrada, A., et R. Coates-Estrada. 2002. Bats in continuous forest, forest fragments and in an

agricultural mosaic habitat-island at Los Tuxtlas, Mexico. Biological Conservation

103: 237–245.

Estrada, A., R. Coates-Estrada, et D. Merrit. 1993. Bat species richness and abundance in

tropical rain forest fragments and in agricultural habitats at Los Tuxtlas, Mexico.

Ecography 16: 309–318.

FAO. 2005. State of the world’s forests. Forestry Department, Food and Agricultural

Organization of the United Nations. Rome.

Findley, J. S. 1993. Bats: a community perspective. Cambridge University Press, Cambridge,

UK.

Fleming, T. H., E. T. Hooper, et D. E. Wilson. 1972. Three central American bat

communities: structure, reproductive cycles, and movement patterns. Ecology 53:

555–569.

Geiselman, C. K., S. A. Mori, et F. Blanchard.( 2002 onwards). Database of Neotropical

Bat/Plant Interactions. http://www.nybg.org/ botany/tlobova/ mori/batsplants/

database/ dbase_frameset.htm

Gleason, H. A. 1922. On the relation between species and area. Ecology 3: 158–162.

Gleason, H. A. 1925. Species and area. Ecology 4: 66–74.

Gorresen, P. M., et M. R. Willig. 2004. Landscape responses of bats to habitat fragmentation

in Atlantic forest of Paraguay. Journal of Mammalogy 85: 688–697.

Handley, C. O. 1967. Bats of the canopy of an Amazonian forest. Atlas do Simposio sôbre a

biota Amazônica 5: 211–215.

Hanski, I. 1994. Patch-occupancy dynamics in fragmented landscapes. Trends in Ecology and

Evolution 9: 131–135.

Hayssen, V., et T. H. Kunz. 1996. Allometry of litter mass in bats: comparisons with respect

to maternal size, wing morphology, and phylogeny. Journal of Mammalogy 77:

476–490.

Helversen, O., von, et Y. Winter. 2003. Glossophagine bats and their flowers: costs and

benefits for plants and pollinators. Pp 346–397 dans Bat Ecology (T. H. Kunz, et M.

B. Fenton, éds.). The University of Chicago Press.

Herrera, C., P. Jordano, L. López-Soria, et J. Amat. 1994 Recruitment of a mastfruiting, bird-

dispersed tree: bridging frugivore activity and seedling establishment. Ecological

Monographs 64: 315–344.

18

Howe, H. F., et J. Smallwood. 1982 Ecology of seed dispersal. Annual Review of Ecology

and Systematics 13: 201–228.

Hutson, A. M., S. P. Mickleburgh, et P. A. Racey (comp.). 2001. Microchiropteran bats:

global status survey and conservation action plan. IUCN/SSC Chiroptera Specialist

Group. IUCN, Gland, Switzerland and Cambridge, UK.

Janzen, D. 1970. Herbivores and the number of tree species in tropical forests. American

Naturalist 104: 501–528.

Jonsson, K. I. 1997. Capital and income breeding as alternative tactics of resource use in

reproduction. Oikos 78: 57–66.

Kalko, E. K. V. 1998. Organisation and diversity of tropical bat communities through space

and time. Zoology 101: 281–297.

Kalko, E.K.V., C. O. Handley, Jr., et D. Handley. 1996. Organization, diversity, and long-

term dynamics of a neotropical bat community. Pp. 503–553 dans Long-term

studies of vertebrate communities (S. M. Cody, et J. Smallwood, éds.). Academic

Press, New York.

Kunz, T. H., et A. A. Stern. 1995. Maternal investment and post-natal growth in bats.

Symposia of the Zoological Society of London 67: 123–138.

Kurta, A., G. P. Bell, K. A. Nagy, et T. H. Kunz. 1989. Energetics of pregnancy and lactation

in free-ranging little brown bat (Myotis lucifugus). Physiolical Zoology 62: 804–

818.

Leigh, E. J., J.-F. Cosson, J.-M. Pons, et P.-M. Forget. 2002. En quoi l’étude des îlots

forestiers permet-elle de mieux connaître le fonctionnement de la forêt tropicale ?

Revue d’Ecologie (Terre et Vie) 57 : 181–194.

Lim, B. K., et M. D. Engstrom. 2001. Bat community structure at Iwokrama forest, Guyana.

Journal of Tropical Ecology 17: 647–665.

Lovejoy, T. E., R. O. Bierregaard, Jr., A. B. Rylands, J. R. Malcolm, C. E. Quintela, L. H.

Harper, K. S. Brown, Jr., A. H. Powel, G. V. N. Powell, H. O. Schubart, et M. B.

Hays. 1986. Edge and other effects of isolation on Amazon forest fragments. Pp

257–285 dans Conservation biology: the science of scarcity and diversity (M.E.

Soulé, éd.). Sinauer, Sunderland, Massachusetts.

MacArthur, R. H., et E. O. Wilson. 1963. An equilibrium theory of insular zoogeography.

Evolution 17: 373–387.

MacArthur, R. H., et E. O. Wilson. 1967. The theory of island biogeography, Princeton

University Press.

19

Medellín, R. A., et O. Gaona. 1999. Seed dispersal by bats and birds in forest and disturbed

habitats of Chiapas, México. Biotropica 31: 478–486.

Nathan, R., et H. C. Muller-Landau. 2000. Spatial patterns of seed dispersal, their

determinants and consequences for recruitment . Trends in Ecology and Evolution

15: 278–285.

Norberg, U. M., et J. M. V. Rayner. 1987. Ecological morphology and flight in bats

(Mammalia; Chiroptera): wing adaptations, flight performance, foraging strategy

and echolocation. Philosophical Transactions of the Royal Society London, B.

Biological Sciences 316: 335–427.

Pianka, E. R. 1970. On r- and K-selection. American Naturalist 104: 592–597.

Pons, J.-M., et J.-F. Cosson. 2002. Effect of forest fragmentation on animalivorous bats in

French Guiana. Revue d’Ecologie (Terre et Vie) 57: 117–130.

Quesada, M., K. E. Stoner, V. Rosas-Guerrero, C. Palacios-Guevara, et J. A. Lobo. 2003.

Effects of habitat disruption on the activity of nectarivorous bats (Chiroptera:

Phyllostomidae) in a dry tropical forest: implications for the reproductive success of

the neotropical tree Ceiba grandiflora. Oecologia 135: 400–406.

Reis, N. R., et A. L. Peracchi. 1987. Quiropteros da regiao de Manaus, Amazonias, Brasil

(Mammalia, Chiroptera). Boletim Museu Paraense Emilio Goeldi, serie Zoologia 3:

161–182.

Schulze, M. D., N. E. Seavy, et F. F. Whitacre. 2000. A comparison of the Phyllostomid bat

assemblages in undisturbed neotopical forest and in forest fragments of a slash-and-

burn farming mosaic in Petén, Guatemala. Biotropica 32: 174–184.

Schupp, E. W. 1995. Seed-seedling conflicts, habitat choice, and patterns of plant recruitment.

American Journal of Botany 82: 399–409.

Simmons, N. B., et T. M. Conway. 2003. Evolution of ecological diversity in bats. Pp 493–

535 dans Bat Ecology (T. H. Kunz, et M. B. Fenton, éds.). The University of

Chicago Press.

Simmons, N. B., et J. H. Geisler. 1998. Phylogenetic relationships of Icaronycteris,

Archaeonycteris, Hassianycteris, and Palaeochiropteryx to extant bat lineages, with

comments on the evolution of echolocation and foraging strategies dans

Microchiroptera. Bulletin of the American Museum of Natural History 235: 1–182.

Simmons, N. B., et R. Voss. 1998. The mammals of Paracou, French Guyana: a neotropical

lowland rainforest fauna. Part I. Bats. Bulletin of the American Museum of Natural

History 237: 0–219.

20

Skole, D., et C. Tucker. 1993. Tropical deforestation and habitat fragmentation in the

Amazon: satellite data from 1978–1988. Science 260: 1905–1910.

Tabarelli, M., J. M. Cardoso da Silva, et C. Gascon. 2004. Forest fragmentation, synergisms

and the impoverishment of neotropical forests. Biodiversity and Conservation

13:1419–1425.

Taylor, R. J. 1987a. The geometry of colonization: 2. Islands. Oikos 48:225–231.

Taylor, R. J. 1987b. The geometry of colonization: 2. Peninsulas. Oikos 48:232–237.

Turner, I. M. 1996. Species loss in fragments of tropical rain forest: a review of the evidence.

Journal of Applied Ecology 33: 200–219.

Turner, M., et R. T. Corlett. 1996. The conservation value of small, isolated fragments of

lowland tropical rain forest. Ecology and Evolution 11: 330–333.

Wilde, C. J., M. A. Keer, C. H. Knight, et P. A. Racey. 1995. Lactation in verspertilionid bats.

Symposia of the Zoological Society of London 67: 139–149.

Williams, C. B. 1943. Area and number of species. Nature 152: 264–267.

Wilson, D. E., et D. M. Reeder (éds.). 1993. Mammal species of the world: a taxonomic and

geographic reference. Smithsonian Institution Press, Washington, D.C.

21

Chapitre 1

Foraging strategy and breeding constraints of Rhinophylla pumilio, a fragmentation-sensitive

phyllostomid fruit bat. Rhinophylla pumilio équipée d’un émetteur télémétrique (dessin S. Jouard).

22

FORAGING STRATEGY AND BREEDING CONSTRAINTS OF RHINOPHYLLA PUMILIO, A FRAGMENTATION-SENSITIVE

PHYLLOSTOMID FRUIT BAT.

ABSTRACT

The understanding of processes leading to population declines and diversity loss in

fragmented tropical forests is critical because deforestation is occurring at an unprecedented

rate. This requires a thorough knowledge of the ecology and behavior of fragmentation-

sensitive species. In that respect, we studied the movement pattern, including range size and

foraging strategy, of the fruit-bat Rhinophylla pumilio (Phyllostomidae), with a particular

emphasis on the constraints females have to deal with when rearing a young. Judging from the

well scattered distribution of its main food resource that consists of epiphyte infructescences,

we hypothesized that R. pumilio would spend most of its flight time searching for food within

few foraging areas rather close from each other, while commuting flights between foraging

areas would be infrequent, resulting in small home ranges relatively to other frugivorous bats

with different food preferences. Furthermore, we predicted that lactating females that are

challenged to both increase their food intake and feed their young during nighttime would

perform more search flights, less commuting flights, and reduce the size of their home range

and foraging areas. We radio-tracked nine females (four non-reproductive, four lactating, one

juvenile) and two males along a total of 49 nights at the Nouragues primary rain forest,

French Guiana, in rainy seasons 2003 and 2004. Supporting our predictions, their foraging

strategy was mostly restricted to short (40-120 m) search flights in a single small foraging

area (3.5-14.1 ha). Lactating females most probably transported their young and nursed it in

their foraging area at night. This was associated with a decrease in flight distances and size of

foraging area, and an increase in total flight time throughout the night. We propose that the

foraging strategy of R. pumilio that is mostly restricted to short-distance search flights, is

incompatible with the need to regularly cross expanses of inhospitable matrix in fragmented

forests. Fragmentation may also decrease the breeding success because lactating females

apparently cannot afford flights as long as non-reproductive females while foraging.

23

INTRODUCTION

Bats (Chiroptera) radiated into a great variety of ecological niches throughout their

exceptional evolutionary diversification. The diversity of their life history traits is particularly

prominent in the tropics where many bat species play important roles in complex ecological

processes including pollination, seed dispersal, and regulation of insect populations (Charles-

Dominique 1986, Findley 1993, Helversen and Winter 2003, Kalka and Kalko 2005, Marshall

1983, Medellín and Gaona 1999). Bats can be grouped into distinct functional groups or

guilds (Kalko et al. 1996) according to their main diet (e.g., nectarivores, frugivores,

insectivores), foraging habits (aerial insectivores, gleaners) and habitat preferences ranging

from unobstructed, open spaces to highly cluttered space within forest. Species forming a

particular guild generally belong to the same family or subfamily and often co-exist locally in

species-rich ensembles (sensu Fauth et al. 1996). Local neotropical bat communities may

contain more than 70 species (Arita 1997, Bernard and Fenton 2002, Bonaccorso 1979,

Delaval et al. 2005, Fleming et al. 1972, Handley 1967, Kalko et al. 1996, Lim and Engstrom

2001, Reis and Peracchi 1987, Simmons and Voss 1998) and are striking illustrations of

intricate resource partitioning systems that promote such diversity. In the highly diverse

neotropical lowland forests, bats represent half of the mammalian fauna.

However, like for many other neotropical vertebrate and invertebrate communities

(Bierregaard et al. 2001, Laurance and Bierregaard 1997, Laurance et al. 2002, Turner 1996),

bat diversity is increasingly threatened by habitat disturbances and fragmentation resulting

from intensive deforestation and changes in land use practices (e.g. Chapter 2; Brosset et al.

1996, Cosson et al. 1999, Estrada and Coates-Estrada 2001, 2002, Estrada et al. 1993,

Gorresen and Willig 2004, Kalko 1998, Schulze et al. 2000). Habitat fragmentation is a loss

of habitat connectivity with the consequence that habitat fragments become isolated from

each other by expanses of inhospitable matrix.

An important issue in tropical ecology is to identify life history traits that make some species

more prone to decline or go extinct in fragmented habitats than others. This requires a

thorough knowledge of the bats’ use of space and movement patterns within landscapes, in

particular their range size and foraging strategy, i.e. the manner in which bats move across the

landscape to search for and exploit food resources. Acquiring such knowledge on these highly

mobile animals is limited by technical difficulties. Radio-tracking surveys remain the most

efficient method to monitor bats’ nocturnal movements. Until recently, when the number of

24

studies increased including a wide range of species (e.g., Bernard and Fenton 2003, Gannon

and Willig 1997, Kalko et al. 1999, Meyer et al. 2005, Thies 1998, Weinbeer and Kalko

2004), detailed radio-tracking surveys on neotropical bats had been mostly restricted to two

common frugivorous species, namely Artibeus jamaicensis (Handley and Morrison 1991,

Morisson 1978ab) and Carollia perspicillata (Charles-Dominique 1991, Heithaus and

Fleming 1978). These two species became reference models to illustrate the foraging

strategies of two groups of fruit-bats, the large bats of the genus Artibeus, mostly specialized

on figs, Ficus spp., Moraceae, and the medium-sized bats of the genus Carollia mostly

specialized on fruits of Piper, Solanum and Vismia.

Foraging behavior of fruit-eating bats is mainly composed of search flights, devoted to locate

food sources, and commuting flights devoted to straightforward moves between day roost and

foraging areas as well as among foraging areas. The large fig-eating A. jamaicensis (56 g) and

the smaller shrub-frugivorous C. perspicillata (17 g) have adopted different foraging

strategies with different use of search and commuting flights because of the spatiotemporal

distribution of their foods. A. jamaicensis frequently feeds on the fruits of a range of Ficus

trees that produce big-bang crops thus providing huge amounts of figs for short periods of

time, usually only for a few days (Korine et al. 2000). At the tree, A. jamaicensis conducts

shuttle flights to pick up ripe figs and consumes them at temporary feeding roosts located 25-

400m away from fruiting trees. Because Ficus trees are distributed over large areas and fruit

asynchronously over the year, it is assumed that a given foraging area is associated with a

single tree that is fruiting at that moment. A. jamaicensis regularly conducts rather long

commuting flights, not only between day roosts and foraging areas, but also among foraging

areas with 2 to 5 shifts per night (Morisson 1978a). Commuting flights may exceed 2 km and

can reach up to 10 km (Morrison 1978ab). Although they start foraging at night by

commuting directly to a fruiting Ficus that they had already visited the previous 2-5 nights,

they are obliged to constantly search for new fruiting trees. Therefore, foraging areas last only

a few days and shift as soon as the bat has found another fruiting tree. Occasional long flying

bouts of 10-45 min after the main feeding period at the beginning of the night might be

performed for that purpose (Morrison 1978a).

In comparison, C. perspicillata feeds mainly on fruits of shrubs, namely Piper, Solanum and

Vismia that offer a steady state fruit resource characterized by low but continuous fruit

production extending over longer periods of time, usually spanning several weeks to several

months (Fleming 1981, 1985, Fleming et al. 1977, Thies and Kalko 2004). To cope with the

characteristics of this food resource that is not concentrated in space and time, in contrast to

25

individual fig trees with huge fruit crops, but shows a spatially more scattered distribution, the

shrub-frugivorous C. perspicillata has to spend more time flying from shrub to shrub to

search and collect enough food within its foraging areas. However, as it mainly exploits

patches of Piper that are typically composed of a relatively high number of individuals often

growing rather closely together, many of its search flights are less than 50 m. In Heithaus and

Fleming (1978), individuals typically used two foraging areas per night and a total of only 2-6

foraging areas over mean tracking periods of 12-days (range 3-19 days). Commuting flights

among foraging areas were mostly much shorter than in A. jamaicensis, rarely exceeding1.5

km.

To summarize, within the time frame of a radio-tracking session on an individual bat (1-2

weeks) those species that feed on the highly patchily-distributed fig resource with

asynchronous big-bang crops are expected to frequently shift foraging areas within and

among nights by using long commuting flights. On the contrary, those species specialized on

the more scattered shrub plants with small steady-state fruit crops are expected to spend more

time engaged in search flights within each foraging area, and to use fewer foraging areas

among which they commute less often and over shorter distances. As a consequence, the

former species may typically use larger home ranges than the latter species.

In this study, we focused our attention on Rhinophylla pumilio (Phyllostomidae), a fruit-bat

widespread across Amazonian forests and the Guiana shield, that ranks among the most

fragmentation-sensitive species (Cosson et al. 1999). This study aims at documenting

movement patterns, in particular range size and foraging strategy in an undisturbed rainforest

to better understand possible causes for its decline in fragmented habitats. Special emphasis is

given to physiological constraints faced by females when rearing a young.

The small 9-g R. pumilio is specialized on infructescences produced by epiphytic Araceae and

Philodendron spp. (Cyclanthaceae) (Cokcle 1997, Cosson 1994, Delaval 2004, Delaval et al.

2005). These epiphytes are widely scattered in the forest, mostly growing on tree trunks

between 1 and 5-8 m above ground (Cockle 1997, Cockle 2001). They produce a single or

only a few infructescences at a time. The spatiotemporal distribution of the epiphyte fruit

resource shares more characteristics with shrubs (steady state crops with spatially scattered

fruits) than with fig trees (big bang crops with locally abundant fruits). Therefore, we

hypothesized that similarly to the shrub frugivore C. perspicillata, the foraging strategy of R.

pumilio would rely on frequent search flights and less on commuting flights between foraging

areas. Accordingly, they might use few foraging areas and these should be close to each other

and form a relatively small home range whose size and location should be more stable over

26

time than in A. jamaicensis. The peculiar roosting behavior of R. pumilio may also influence

range size. Individuals roost under leaves from epiphytes or large palm fronts that are mostly

modified into tents (Charles-Dominique 1993, Simmons and Voss 1998, Zortéa 1995).

Rhinophylla pumilio forms small polygynous groups with one or two adult males and up to 4

females (Rinehart 2003, Simmons and Voss 1998). The low roost-fidelity usually observed in

tent-roosting bats compared to species roosting in caves or hollow-trees may be associated

with frequent shifts between roosts to achieve low commuting distances between day roosts

and foraging areas, leading to smaller home range size (Lewis 1995).

Finally, we argue that physiological constraints of rearing a young may force females R.

pumilio to modify their movement pattern, as has been previously suggested for C.

perspicillata (Charles-Dominique 1991) and also for a wide range of insectivorous bats (de

Jong 1994, Henry et al. 2002, Kurta et al. 1989, Racey and Swift 1985, Reynolds and Kunz

2000, Swift 1980). Indeed, as typical “income breeders”, female bats rely on current food

intake to support costs of reproduction and require high food intake to meet the demands of

milk production. For instance, energy requirements increased by 1.5 to 2 times for lactating C.

perspicillata (Fleming 1988). A higher energy intake can be achieved by conducting more

search flights in order to harvest more fruits in a given foraging area. At the same time, the

need of feeding the young during the night may force females to decrease the length and/or

frequency of their commuting flights among foraging areas. The presumed behavioral shifts

accompanying the onset of lactation may reduce fitness of females in fragmented areas where

large expanses of inhospitable matrix may disrupt once continuous foraging areas, forcing

individuals to increase the length and frequency of commuting flights. This, in turn, exerts

pressure on their time and energy budget.

METHODS

Study site

The study was carried out at the Nouragues research station, in the centre of the Réserve

Naturelle des Nouragues, northern French Guiana (4°50’ N, 52°42’ W). Les Nouragues is part

of a large block of continuous tropical lowland rainforest where human influence has been

absent or negligible over the past two centuries (Charles-Dominique 2001). Annual

temperature averages 26.3°C and average annual rainfall ranges from 2500 to 3200 mm, with

27

a marked dry season from August to November. Trees of the families Caesalpinaceae,

Sapotaceae and Lecythidaceae dominate the canopy (Poncy et al. 2001).

Rhinophylla pumilio is the fourth most commonly captured species in the understory of the

study site (Delaval et al. 2005).

Bat captures and radio-tracking

All R. pumilio were captured with 12-m mist-nets (height 2.5 m, four shelves) set within a

300×400 m area transected by a 5-m wide creek, and located in the middle of the 100-ha

Nouragues research station quadrat which is completely covered by a 100-m spacing grid of

forest trails. Three nets were arranged in a T-pattern, and moved after each capture night.

Captured bats were kept in cloth bags before examination. Reproductive status of females

(pregnant, lactating or non-reproductive) was assessed by checking for the presence of a

palpable fetus or of prominent hairless nipples (Racey 1988). Juveniles and subadults were

distinguished from adults based on the degree of fusion of metacarpal epiphyses (Anthony

1988). Individuals were weighed with an electronic scale (Okaus Corporation, USA) to the

nearest 0.1 g. Only bats >8.5g were used for radio-tracking. The transmitters (0.70 g,

Biotrack, UK) accounted for 6.9±0.9% of the bats’ body mass, which slightly exceeded the

recommended 5% rule (Aldridge and Brigham 1988), but remains below the critical 10%

limit above which transmitter mass may negatively affect the animals’ foraging behavior

(Brander and Cochran 1969). We attached the transmitters with cyanoacrylate glue (Henkel

France SA, Boulogne Billancourt) in the interscapular region of the bats after trimming the

fur. No lesions were found on recaptured bats, neither in short term (4 to 45 days, n=4) nor in

medium term (5 and 12 months, n=2) after tagging. Bats fitted with a transmitter were fed

with sugar water and released at the capture site within one hour after capture. We radio-

tracked one bat at a time, using a FT-290R receiver (YAESU electronics) and a 4-element

Yagi antenna (Tonna electronics). A total of eleven bats were tracked, including 4 non-

reproductive and 4 lactating females (F1-F4 and F5-F8 respectively), 1 subadult female (F9),

and 2 non-reproductive males (M1 and M2). All tracking sessions were conducted in the

middle of the wet season between February and May in 2003 and 2004, except for M1 that

was tracked in October 2003 (dry season). Contrary to others individuals, M2 was fitted with

a 0.84 g position-sensitive transmitter (9.7% of body mass; model LB-2B, Holohil Systems

Ltd., Carp, Ontario, Canada) where a mercury switch modulates pulse rhythm according to its

inclination with a high repetition rate when the bat was flying and a slower rate when the bat

was hanging. The lactating female F5 was caught together with a volant juvenile that must

28

have been her own as she fed it in the capture bag. They were released together. Bats were

tracked for 5 consecutive nights unless the transmitter fell off before this time (n=4). Bats

were not tracked during the night of release because they may have been stressed after

handling. No tracking session was undertaken during the 6-days period encompassing full

moon.

During tracking sessions, we determined temporary night hanging locations using

triangulation with two bearings. Bearings were taken following the direction of maximal

signal intensity according to the receiver intensity gauge. For that purpose, we constantly

moved along trails, either alone or with two persons who stayed in radio-contact, to follow the

bat as closely as possible. Bats were considered hanging when the transmitter signal was

judged constant in direction and intensity during at least 1 min, and were considered flying

otherwise. We occasionally failed to locate <2-min hanging phases because we could not

reach adequate bearing positions in this short period of time.

The precision of triangulation depends on the distance between the transmitter and the

receivers and on the angle between bearings. To assess the range of our equipment, we took

bearings in the forest understory from 65 positions ranging from 15 to 195 m from a

transmitter placed 2 m above ground level. The error of the mean angle (absolute angle

between expected and measured bearings) reached minimum values (3-10°) between 45 and

120 m from the transmitter. It peaked at 14-17° between 15 and 45 m because signal direction

was subject to strong reflections from the surrounding vegetation, mainly trees, and the

ground. Therefore, we determined the signal intensity level corresponding to this range of

high reflection risks and afterward preferred whenever possible to triangulate from positions

with lower signal intensity. Furthermore, the rather flat terrain and the accurate 100-m spacing

grid of straight forest trails marked every ten meters allowed us to easily ensure that each pair

of bearings formed an adequate angle for valuable triangulation (20 to 160°) as tracking

proceeded. We estimated the triangulation resolution (distance error between bearing

intersection and real location of transmitter) to rarely exceed ±15 m.

Coordinates of the bearing positions, mostly at trail intersections, were obtained from a

custom-made map of the study site, based on a 1/50000 map (Institut Géographique National,

France) and computed by P. Charles-Dominique on the software Designer 4.1 (2)

(MicrographX corporation). The fixes, i.e. bat hanging locations as given by bearing

intersections, were then determined and analyzed with the software Tracker 1.1 (Camponotus

AB, Solna, Sweden, 1994). All fixes exceeding maximum detection range around bearing

29

positions (>350 m) were discarded before analyses. These accounted for only 1.5% of total

fixes.

Data collection

To avoid any influence of day roost localizations on estimations of FAs, we did not take data

during a 20-30 min buffering period both after emergence from and before return to the day

roosts. Thus, analyses of foraging strategy were restricted to the time interval 19:15 - 06:15.

During this whole time interval, we continuously monitored the bats’ activity by noting time

(rounded to the nearest min) for each transition between flight and hanging phases, and

between each signal loss and retrieval. Bearings were taken for each hanging phase until

midnight.

We also located day roosts every day for each tracked bat. To find tent roosts without

disturbing the bats, we followed the direction of the radio signal until signal intensity

indicated to us that we were very close (<20 m) to the roost. We then carefully observed the

surrounding vegetation until we found a tent. We counted the number of congeners that were

roosting with the tracked bat (pups excluded) in the tent, sometimes aided by binoculars,

when the tents were high (1.5 to 4 m above ground) and permitted disturbance-free

observation. In case the tents were low (<1.5 m) above ground or within dense vegetation,

bats were counted at dusk when leaving the roost, and a mirror was placed on the ground to

facilitate further counting in the same roost. When the bats roosted in tight clusters, we

visually estimated the minimum number of individuals taking into account the number of ears

whose color slightly contrasted with the bats’ fur. Accuracy of this method was estimated at

±2 individuals for groups of >5 bats.

Based on these data, two main aspects of movement patterns were characterized: range size,

in particular home range, foraging areas and core areas and foraging strategy, namely use of

search and commuting flights. Furthermore, activity rhythm across the night was studied in

relation to female reproductive status.

Range size

Range size parameters used herein, i.e. size of home range (HR), foraging area (FA) and core

area (CA), aim at characterizing the spatial extent of the habitat that the bats used in the forest

with regard to their activities devoted to roosting and foraging including food search,

consumption and digestion during the night. Type and duration of activities were estimated

for each individual from all of its tracking localizations gathered during day and night within

30

5-day sessions. HR comprises both roosting and foraging activity, which are distinct

components of the time budget. We used the 100% minimum convex polygon (MCP) method

to define HR. MCP is a non-parametric method connecting the outermost fixes so as to

delineate a single area enclosing all fixes. Its size is independent of the position and density of

fixes inside.

In contrast, FAs and CAs are restricted to a single type of activity (foraging activity) and were

analyzed using the probabilistic Kernel method (adaptive Gauss method, density CV=0.15;

Worton 1989). This method lays out isopleths that delineate zones of equal probability of

presence of tracked individuals. While FAs refer to the whole foraging activity, CAs were

defined as preferred zones within FAs where the bulk of the activity is concentrated. To

permit comparisons with other studies on bats (Meyer et al. 2005, Weinbeer and Kalko 2004),

we defined FAs and CAs as the surfaces enclosing 95% and 50%, respectively, of probability

of presence of the bats.

Probabilistic range size estimations should be performed on series of independent (i.e.

randomly sampled) localizations (Worton 1989). However, in many studies, successive

animal localizations may suffer from temporal autocorrelation. To minimize this problem, it is

suggested to choose longer time intervals between localizations than the time the animal

would require to cross its HR. In our study, we localized bats only during hanging phases so

that time intervals were not standardized but depended on duration of flights between hanging

phases. However, these intervals were rarely less than 3-4 min, which was enough for bats to

commute across their HR (see results).

The fixes used to estimate FAs and CAs represent the localization of temporary night roosts

and of fruiting plants used by bats. Indeed, bearings were taken only during hanging phases

that corresponded to either resting phases in temporary night roosts or to feeding phases on

epiphyte infructescences. The infructescences on which R. pumilio is specialized are too large

to be removed by bats and cannot be transported and consumed into temporary feeding roosts

in contrast to figs, fruiting spikes of Piper or berries of Solanum (e.g. Fleming 1981, Morrison

1978a). Instead, bats land on the infructescences and consume in situ part of the pulp of the

ripe fruits, each containing up to several hundreds of tiny seeds that are dispersed by

endozoochory (Cockle 1997).

Foraging strategy

To distinguish between search and commuting flights and to assess their relative frequency

during foraging activity, we used two definitions. First, we refer to flights between two FAs

31

as commuting flights, and to flights within a single FA as search flights. Second, we

established a quantitative definition distinguishing between commuting and search flights

based on average flight speed determined from flight distance and duration. Flight distance is

the linear distance between the two hanging locations that were successively visited by the

bats. Average flight speed was obtained by dividing flight distance by flight duration. We

assumed that flight speed of a bat commuting between foraging areas or from the day roosts

to the foraging area will be higher than average flight speed of a bat searching for food and

thereby flying back and forth within a feeding patch. As average flight speed of a bat flying

straight in the understory from its roost to a foraging area is about 6 m/s (Heithaus and

Fleming 1978) we arbitrarily set 2 m/s as the limit between search and commuting flights.

Effect of reproductive status on movement pattern

Some bats are known to transport their young into their FAs (see reviews by Jones [2000] and

Kunz and Hood [2000]) where they probably also feed them during the night. In a first step,

we wanted to know whether this also applies to lactating R. pumilio, and in a second step,

how this may in turn modify their general movement pattern. To find out whether females

took their young with them during foraging, we checked, whenever possible, for the presence

of a young remaining in a day roost after the emergence of a tracked, lactating bats, using a

low-intensity diffuse white light (Tikka LED headlamp, Petzl®, France). As a complementary

method, we also assessed from our tracking data whether lactating bats frequently returned to

a specific hanging location within their FAs. This would possibly indicate shuttling flights

between foraging areas and a night roost where lactating females may deposit their young and

regularly feed it. Assuming that two hanging locations were potentially the same when they

were situated <15 m apart (i.e. estimated upper limit of the error occurring for fixes), we

defined an index of revisitation rate of hanging locations, calculated as the ratio [(total nb of

hanging phases – nb of distinct hanging locations) / (total nb of hanging phases – 1)]. These

values ranged from 0% (all hanging phases were at different locations) to 100% (all hanging

phases were at the same location) and were computed for each night using all available fixes.

They were then arcsine-transformed and compared between reproductive groups (non-

reproductive vs. lactating females) using a nested ANOVA.

Afterwards, we analyzed the effect of reproductive status on components of the bats’

movement pattern: range size, including HR size, size and number of FAs and CAs, distances

between day roosts and nearest FAs, flight distances, foraging strategy, namely proportions of

search and commuting flights and activity rhythm across the night encompassing cumulative

32

flight time per night, duration and frequency of flights and hanging phases. When an

individual used more than one FA or CA, areas were pooled. To compare differences in flight

time over the night, we first calculated the time each bat spent flying in 30-min intervals for

the whole tracking nights (19:15 – 06:15). We then performed a two-way repeated measures

ANOVA to assess possible fluctuations in time spent in flight across the night intervals and

between reproductive groups. This analysis was performed separately for the two parts of the

night (before and after midnight). For the other parameters of movement pattern, we used

either t-tests or nested ANOVAS, depending on whether few or many values were assigned to

each bat. In the latter case, values were, whenever necessary, log- or square-root-transformed

to re-establish normality before comparison. Departures from normality were detected using

Kolmogorov-Smirnov tests. All statistical tests were performed with Systat 9.0 (SPSS, Inc.,

Chicago, Illinois).

RESULTS

Range size

We obtained a total of 529 fixes in 42 full tracking nights (11 hr) for the nine females and 7

half tracking nights (5h) for the two males. Periods where we lost the signal and other

interruptions such as bad weather were short (rarely more than 20 min) and accounted for

only 2.7% of total tracking time. We successfully localized all day roosts of the tagged

animals that were always situated within the 100-ha quadrat of the research station. The bats

used 23 roosts, only three of which could not be visually identified. Thirteen of the 20

identified roosts consisted of large leaves of epiphytes (Philodendron fragrantissimum and P.

ornatum, Araceae) and young fronts of juvenile palms Astrocaryum sciophilum (Arecaceae).

Philodendron leaves were all modified into tents with chewed veins along the central vein

whereas most A. sciophilum leaves were unmodified. Bats were also found four times under

large unmodified mature leaves of young Jessenia bataua (Arecaceae) and twice under

unmodified dry leaves of Cecropia sciadophylla (Cecropiaceae) that formed an umbrella-like

shelter. All identified day roosts were situated 1 to 4 m above ground.

HR size varied from 2.5 to 16.9 ha and FAs from 3.5 to 14.1 ha. Each bat used a single FA,

except F5 that used two FAs that were separated by 15 m and that merged at a slightly

reduced isopleth smoothing parameter (density CV=0.135 instead of 0.150). Flights that

33

Figure 1. Examples of spatial use by six individuals of R. pumilio at the Nouragues study area: home range (100% inclusion convex polygon), foraging area (95% kernel, thin line), core area (50% kernel, bold line) and day roosts (black squares). Individual number and reproductive status (NR for non-reproductive and L for lactating females) are indicated in upper right corners. Variation in height between isoclines (grey curves) is 20m.

0 200 400 m

Nou

ragu

es c

reek

Camp

0 200 400 m

Nou

ragu

es c

reek

Camp

0 200 400 m

Nou

ragu

es c

reek

Camp

0 200 400 m

Nou

ragu

es c

reek

Camp

0 200 400 m

Nou

ragu

es c

reek

Camp

0 200 400 m

Nou

ragu

es c

reek

Camp

N N

N N

N N

F3 (NR) F1 (NR)

F8 (L) F7 (L)

M2 M1

Table 1: Results of tracking sessions conducted on 11 R. pumilio (females F1-F9 and males M1-M2). Time budget parameters are given as means ±±±±SD per individual and part of night (first part 19:15 – 23:59 and second part 00:00 – 06:15).

Individual F1 F2 F3 F4 F5 F6 F7 F8 F9 M1 M2 Reproductive status a NR NR NR NR L L L L SA NR NR Nb of tracking nights 5 3 5 5 5 5 5 4 5 3 4 Total contact time (%) 98.2 98.4 95.7 96.3 98 97.2 97.2 93.6 100 98.9 96.1 Spatial use Nb of bearings 58 34 55 63 52 37 44 35 63 28 52 Home range (ha) 12.3 5.8 16.9 9.2 10.7 4.5 5.9 7.5 5.8 9.0 2.5 (day roosts excluded) (9.7) (5.3) (8.6) (6.6) (3.6) (4.3) (2.0) (2.2) (3.8) (8.4) (2.1) Foraging area (95% kernel) (ha) (Nb of foraging areas)

12.5 (1)

11.7 (1)

13.4 (1)

8.0 (1)

5.3 (2)

7.8 (1)

3.7 (1)

5.1 (1)

4.3 (1)

14.1 (1)

3.5 (1)

Core area (50% kernel) (ha) (Nb of core areas)

2.4 (1)

0.6 (2)

1.6 (3)

1.2 (1)

0.7 (2)

0.7 (1)

0.5 (2)

0.6 (2)

0.2 (1)

2.1 (1)

0.5 (1)

Mean flight distances (m) 94±71 91±75 115±111 96±53 75±67 85±61 97±55 41±28 70±55 118±78 71±42 Hanging location revisitation (%) 15±9 27±1 19±12 13±15 29±19 24±13 20±14 41±8 25±15 0 14±12 Day roosts Nb of distinct day roosts 2 1 5 2 5 2 3 2 1 2 2 Nb of roosts out of foraging areas 2 1 1 1 3 0 3 1 1 2 0 Range of distances to the nearest foraging area (m)

75-130 5 280 150 165-365 - 100-430 475 120 25-35 -

Nb of congeners in visited roosts 0-4 ? 0-2 4-5 2-6 2 3-4 3-5 4-5 1-4 0-1 Time Budget Mean flight duration (min ± SD) First part of night 9±5 8±5 9±4 12±7 10±6 15±6 14±8 13±5 5±2 12±5 7±5 Second part of night 11±6 11±4 11±5 13±6 11±5 13±6 17±7 14±7 9±6 Mean cumulative flight time (min ± SD) First part of night 104±25 89±7 110±22 121±23 79±29 106±16 128±18 117±7 60±9 110±16 79±29 All night long 196±38 160±17 213±51 255±50 199±42 234±34 316±37 252±31 135±31 Mean duration of hanging phases (min ± SD) First part of night 15±17 15±09 15±10 16±11 25±16 22±13 16±13 19±06 21±16 18±15 17±20 Second part of night 35±24 50±33 30±28 21±17 22±17 27±13 17±8 21±14 37±27 Mean nb of hanging phases per hour First part of night 2.28 1.58 2.80 1.94 1.56 1.44 1.73 1.80 2.20 1.76 2.27 Second part of night 1.25 0.76 1.68 1.63 1.76 1.44 1.66 1.32 1.28 a NR=non-reproductive; L=lactating; SA=non-reproductive subadult

35

extended far beyond the limits of the FAs where we mostly lost signal contact were rare

given the fairly high contact time (99.2%). FAs enclosed 1-3 small CAs totaling 0.5 to 2.4 ha

(Fig. 1, Table 1). Although we obtained relatively few fixes per bat (47.3±12.3), estimations

of FA of most individuals varied little as number of fixes exceeded 30 (Fig. 2), indicating that

the actual size of their FA had been covered well. F6 and M1 were exceptions, because the

size of their FA still varied when we added fixes. Figure 2. Estimation of foraging area (95% kernel) with increasing number of fixes for 4 non-reproductive and 4 lactating females R. pumilio (full and open symbols respectively).

Contrary to our prediction, bats mostly flew to day roosts that were located outside of their

FA (15 out of 27 day roosts). Accordingly, HRs as calculated by the MCP method were all

larger (by 81% on average, range 5% to 242%) with the day roosts included (Table 1). The

distance between day roost and FA averaged 101±140 m when attributing a nil distance to

roosts situated inside FAs, and 212±146 m when considering only roosts outside of FAs.

Group size in the day roosts generally encompassed 3 to 7 individuals (n=40 observations),

independently from roost type. Large commuting distances between day roosts and FA were

associated with larger group size of the respective roosts. Although weak, this correlation was

significant (Fig. 3).

0

5

10

15

20

20 30 40 50 60 70

Estim

atio

n of

fora

ging

are

a (h

a)

Number of fixes

36

Figure 3. Correlation between number of congeners in day roosts and distance that tracked R. pumilio had to cover to get to their day roosts from their respective foraging area. Overlapping dots are indicated by larger dot size.

Foraging strategy

As all tracked individuals used a single FA each, none of the 420 recorded flights matched

our definition of commuting flights as flights between two FAs. Similarly, only 2 (0.5%) of

them matched our quantitative definition of commuting flight with an average flight speed of

>2 m/s. These were long 400-m flights performed in 2 min (ca. 3.3 m/s). Only five flights

(1.2%) were close to our definition of commuting flight with an average speed of 1.0 to 1.8

m/s at flight distances of 120 and 220 m, respectively. Therefore, we regarded almost all

flights as search flights. As much as 98.3% of the flights had an average flight speed of less

than 1 m/s.

Effect of reproductive status on movement pattern

When observing emergence of lactating females from day roosts at dusk (n=8), pups were

either taken by the females right away or after a maximum of only 10 to 12 min (n=2).

Furthermore, during the night, lactating females exhibited a higher hanging location

revisitation rate than non-reproductive females (28% and 18%, respectively, Table 1 and 2),

indicating possible regular returns to particular night roost(s) for nursing.

Compared to non-reproductive females, lactating R. pumilio displayed significantly smaller

FAs (-42%, t=4.081, df=5.4, p=0.008) and shorter flight distances (-25%, Table 2). However,

Distance between day roost and foraging area (m)

Num

ber o

f con

gene

rs in

the

day

roos

t

0

1

2

3

4

5

6

7

8

0 100 200 300 400 500

Pearson coefficient=0.466;p=0.002; n=40

37

Table 2: Results of nested ANOVAs regarding factors shaping nocturnal activity pattern of 4 non-reproductive and 4 lactating females R. pumilio. Analyses test the effects of reproductive status (non-reproductive vs. lactating), inter-individual variability (nested in reproductive status) and part of night (first part 19:15 – 23:59 vs. second part 00:00 – 06:15). Probability values p are indicated in bold for significant effects.

Dependant variable and factors a df F p Flight distances (first part of night) (n=318, R2=0.076)

reprod. 1 8.957 0.003 individuals (reprod.) 6 2.840 0.010

Hanging location revisitation rate (n=37, R2=0.325) reprod. 1 5.237 0.030 individuals (reprod.) 6 1.393 0.251

Cumulative flight duration (first part of night) (n=37, R2=0.407) reprod. 1 0.060 0.808 individuals (reprod.) 6 3.319 0.013

Cumulative flight duration (all night long) (n=37, R2=0.617) reprod. 1 11.331 0.002 individuals (reprod.) 6 6.066 <0.001

Duration of flights (n=648, R2=0.100) reprod. 1 26.980 <0.001 part of night 1 6.391 0.012 reprod. × part of night 1 1.445 0.230 individuals (reprod.) 6 5.244 <0.001

Duration of hanging phases (n=665, R2=0.118) reprod. 1 0.355 0.552 part of night 1 35.756 <0.001 reprod. × part of night 1 24.833 <0.001 individuals (reprod.) 6 4.682 <0.001

Number of hanging phases per hour (n=74, R2=0.553) reprod. 1 3.881 0.053 part of night 1 39.244 <0.001 reprod. × part of night 1 29.060 <0.001 individuals (reprod.) 6 1.247 0.295

neither the size of CAs (t=2.235, df=3.1, p=0.108) nor the size of HR (t=1.440, df=4.4,

p=0.213) was affected by reproductive status.

We did not find a strong link between reproductive status of the females and roosting

behavior. As the other females, lactating females roosted in a wide variety of plant species,

they displayed a low roost fidelity and changed roosts on average every second day (Table 1).

They did not shorten distances between day roosts and FA (t=1.516, df=20, p=0.145).

Furthermore, there was a slight trend that lactating females joined larger groups compared to

non-reproductive females (t=1.915, df=28, p=0.066). They roosted always with at least 2

congeners (n=16) whereas non-reproductive females roosted significantly more often alone

(n=4 out of 14, χ²=5.27, df=1, p=0.022) or with less than 2 congeners (n=5 out of 14, χ²=6.86,

df=1, p=0.009).

38

The typical activity rhythm of non-reproductive adult males and females was characterized

by alternations of medium duration flights (10-15 min) and somewhat longer hanging phases

(15-20 min, Table 1). Long hanging phases (1-2 hours) were occasionally observed, mostly

during the second part of night. Two-way repeated measures ANOVAs revealed significant

differences in the pattern of nightly activity rhythm (flight time per 30-min interval, Fig. 4a)

between reproductive groups, depending on the part of night and the reproductive status of

bats. Before midnight, there was a significant variation of flight time among the 10 time

intervals (F=7.60, p<0.001) and a significant interaction of the time intervals with

Figure 4. Mean flight time (a) and mean cumulative flight time (b) per 30-min interval between 19:15 and 06:15 for 4 non-reproductive (bold line) and 4 lactating females R. pumilio (thin line).

0

3

6

9

12

15

19:00 20:00 21:00 22:0023:0000:0001:0002:0003:0004:0005:00 06:00

0

50

100

150

200

250

300

19:00

20:00

21:00

22:00

23:00

00:00

01:00

02:00

03:00

04:00

05:00

06:00

Mea

n fli

ght t

ime

/ 30-

min

inte

rval

±

1 SE

(min

) M

ean

cum

ulat

ive

fligh

t tim

e ±

1 SE

(min

)

Time of the night

(a)

(b)

39

reproductive status (F=3.13, p=0.032), but no effect of reproductive status alone (F=1.27,

p=0.270). After midnight, there was a significant effect of reproductive status on flight time

(F=8.10, p=0.009), but no significant variation among the 12 intervals nor any interaction

between both (F=1.11, p=0.41 and F=1.29, p=0.31, respectively).

From these results, we conclude that activity rhythm of females varied significantly but

asynchronously between both groups before midnight, and that activity was rather even after

midnight. However, lactating females spent significantly more time in flight than non-

reproductive females (Fig. 4a). This resulted in a fairly linear cumulative flight time for

lactating females that remained steeper than that of non-reproductive females after midnight

(Fig. 4b). At the end of night, cumulative flight time of lactating females differed

significantly from non-reproductive females and was on average 42 min (20.0%) longer

(Table 2).

Neither duration nor number of flights and hanging phases varied from the first to the second

part of night for lactating females (Fig. 5). In contrast, the general decline in flight time of

non-reproductive females during the second part of night was associated with a marked

increase of the mean duration of hanging phases (Fig. 5bc). In non-reproductive females,

13.9% of hanging phases lasted more than 1 hour, in contrast to only 1.1% for lactating

females. Among the outputs of statistical tests (Table 2), the two significant interactions

between reproduction status and part of night were the most informative. They support the

previous observations that non-reproductive females increased their hanging time during the

second half of night with several long hanging phases (>1 hr) whereas lactating females

maintained a high activity level all night long by alternating short flights and hanging phases

(ca. 14 and 22 min, respectively).

A second trend was the significant increase of mean duration of flights for lactating females

(Fig. 5a and Table 2) compared to non-reproductive females. They flew 2.2 to 4.3 min longer

(18 to 46%) than non-reproductive females between two consecutive hanging phases. Finally,

the effect of reproductive status on the duration of hanging phases was non significant (Table

2), but while it increased for non-reproductive females after midnight, it remained unchanged

for lactating females, as indicated by the significant interaction of hanging phase duration

with the part of night. Limiting the analysis to the first two hours of the night, where activity

peaked simultaneously in both reproductive groups (Fig. 4a) and was probably mostly

devoted to food intake, actually revealed significant differences regarding flight time and

hanging phases. Flight time and hanging phases of lactating females were 5-6-min longer

than those of non-lactating females. Non-reproductive and lactating females flew 8.7±5.3 and

40

Figure 5. Mean duration of flights and hanging phases (a and b) and mean number of hanging phases per hour (c) of 4 non-reproductive and 4 lactating females R. pumilio during the first and the second part of night. Sample sizes are indicated above bars (see Table 2 for statistics).

0

5

10

15

20

25

non-reproductive lactating

0

10

20

30

40

50

60

non-reproductive lactating

0

1

2

3

non-reproductive lactating

183 146

151

185

185

143

139

181

Mea

n fli

ght d

urat

ion

±1SD

(min

utes

)

Part of night: ■ 19:15 – 23:59 □ 00:00 – 06:15

(a)

(b)

Mea

n nu

mbe

r of h

angi

ng

phas

es p

er h

our ±

1SD

M

ean

dura

tion

of h

angi

ng

phas

es ±

1SD

(min

utes

)

18 19

18 19

(c)

41

13.7±8.1 min (SD) respectively (nested ANOVA; n=141; effect of reproductive status:

F=19.121, df=1, p<0.001; effect of individuals: p>0.05; R2=0.170) and spent 11.4±6.4 and

17.7±8.9 min (SD) hanging, respectively (nested ANOVA; n=136; effect of reproductive

status: F=16.941, df=1, p<0.001; effect of individuals: p>0.05; R2=0.182).

DISCUSSION

In this study, we aimed at describing the movement pattern of the small tent-using R. pumilio

whose diet is specialized on epiphyte fruits. According to our predictions, its foraging

strategy consisted mainly of search flights. Longer commuting flights were almost non-

existent and individuals tended to use a single, rather small FA. The overall HR remained

smaller than in other fruit bats, although day roosts were often located away from the FA of

individuals, probably because these bats mainly roost in groups. Lactating females seem to

increase their food intake by consuming more pulp at each feeding phase on epiphyte

infructescences as they increase hanging phase durations during the foraging peak following

dusk. They apparently transport their young and nurse it in their FA at night. This was

associated with a decrease in flight distances and size of FA, and an increase in total flight

time over the night.

The movement pattern of R. pumilio

The foraging strategy of R. pumilio as depicted by radio-tracking fits predictions based on the

spatiotemporal distribution of its main diet. Given the steady-state production of

infructescences and the well scattered distribution of epiphytes, we expected individuals to

spend most of their flight time in search flights, to use few FAs, and then to conduct fewer

and shorter commuting flights than the fig-eating A. jamaicensis. Rhinophylla pumilio used a

single FA and virtually never conducted commuting flights while foraging, which is

consistent with these theoretical predictions. This places the shrub-frugivorous C.

perspicillata at an intermediate place between R. pumilio and A. jamaicensis regarding

number of FAs as well as length and frequency of commuting flights between FAs. Female

C. castanea, another small shrub-frugivorous bat, tended to use a single small FA (2.6 to 8.6

ha) like R. pumilio, but males foraged in up to 3 FAs (Thies 1998). As only two males were

tracked here, we cannot fully compare the data sets.

42

An alternative interpretation of these results would be that shrub-frugivorous bats actually

combine search and commuting flights, while fig-eating bats separate commuting and

searching flights. Fleming et al. (1977) investigated this hypothesis by means of a fruit

removal experiment at different distances from food patches. They proposed that small shrub

frugivores are constantly “on the alert” while commuting to increase chances of encountering

food sources, while fig-eating bats appear less efficient in finding individual ripe figs apart

from fruiting Ficus. Because epiphytes are abundant and ubiquitous, it may pay bats to be

constantly searching food while commuting. Cockle (1997) estimated that adults or subadults

of epiphyte Evodianthus funifer, one of the commonest food items of R. pumilio, are

established on about 140 trees per ha (i.e. on 10 trees per 30-m diameter circular plots) in the

part of our study site that encloses the HRs. Additional botanical surveys on a 0.4-ha plot

within the mist-netting area (M.H., unpublished data) revealed that >80% of all trees bore

epiphytes whose fruits are known to be part of the diet of R. pumilio.

Partly owing to its foraging strategy mostly restricted to search flights in a single FA, R.

pumilio is probably one of the species with the smallest HR (mean 8.2 ha, range 2.5 – 16.9

ha) among the well-studied phyllostomid fruit-bats. Although differences in methodological

approaches preclude meaningful comparisons, both C. perspicillata and A. jamaicensis

undoubtedly have larger HRs. In Costa Rica, C. perspicillata females regularly commuted

between and among day roosts and FAs that were 1 to 2 km apart (Heithaus and Fleming

1978). This represents twice to five times the maximum flight distance measured for R.

pumilio in this study. Distances are even greater for A. jamaicensis that fed up to 10 km from

its day roost in Mexico (Morrison 1978b). Only females of the shrub-frugivorous C. castanea

had HRs in Panama that were similar in size to those of R. pumilio (mean 13.7 ha, range 5.5 –

34.9 ha, Thies 1998), while those of males were three times larger.

Small body size per se cannot explain the small HR of R. pumilio. Discrepancies in HR size

remained marked when comparing R. pumilio to other phyllostomid bats of similar size but

from insectivorous guilds. In particular, the small trawling insectivorous Macrophyllum

macrophyllum (Meyer et al. 2005) and the small gleaning insectivorous Lampronycteris

brachyotis (Weinbeer and Kalko 2004) both radio-tracked in Panama displayed mean HRs of

43 and 46 ha respectively, reaching up to >150 ha in both cases and even more for L.

brachyotis if its distant day roost had been taken into account. The gleaning insectivorous

Lophostoma silvicolum has an intermediate HR size that still remains on average larger than

those reported here for R. pumilio (mean 17 ha, range 11 – 31 ha, Kalko et al. 1999,

unpublished data).

43

The small HR size of R. pumilio compared to shrub-frugivorous bats may also partly result

from a greater proximity of the day roosts with regard to the FAs (<475 m for R. pumilio, vs.

1.2 to 1.6 km for C. castanea and C. perspicillata respectively; Heithaus and Fleming 1978,

Thies 1998). These differences in distance seem to be associated with different roosting

behaviors between the “nomadic” R. pumilio on one hand, that forms small groups and

regularly shifts day roost, and the Carollia species on the other hand, considered as refuging

species (sensu Hamilton and Watt 1970) that form larger colonies in permanent day roosts to

which they are faithful (Heithaus and Fleming 1978, Thies 1998). Lower roost fidelity is

expected to be related to higher roost availability (Lewis 1995). Accordingly, as much as

35% of identified day roosts used by R. pumilio in our study were actually unmodified leaves

of juvenile palms ranking among the dominant species of the local palm community (de

Granville 2001), suggesting a high availability of potential roosting sites compared to caves

and hollow-trees used by Carollia species.

Yet, we found evidence that R. pumilio individuals regularly roosted outside the immediate

vicinity of their respective FA, probably to take advantage in roosting together with

conspecifics. The significant positive correlation between the distance individuals cover to

join a particular day roost in relation to their respective FA and the number of congeners with

whom they share this day roost suggests that some advantages of roosting in a group make it

worthy to cover long distances. We propose at least two advantages of roosting in a group for

R. pumilio including predator avoidance and thermoregulatory advantages. More individuals

increase the mean vigilance level of the group and thus the chances of detecting predators.

Furthermore, hanging in a cluster contributes to reducing heat losses (Kurta 1985, Tuttle

1976). The latter point may be crucial because temperature in tents used by R. pumilio is not

buffered against fluctuations and also not warmer than ambient temperature (M.H.,

unpublished data). Yet, even in tropical lowland forests, ambient temperature at understory

level remains below the minimal thermal neutrality temperature of small bats during most of

the daytime (ca. 31-32°C for <15g species, Speakman and Thomas 2003). The formation of

clusters may compensate for low ambient temperatures.

Evidences for food intake increase in lactating females

We expected the physiological constraints of lactation to force females to substantially

modify their movement pattern. As a proximal constraint of lactating, female bats need to

increase their food intake for producing milk (Anthony and Kunz 1977, McLean and

Speakman 1999) like in other small mammals (Hammond and Diamond 1994, Perrigo 1987,

44

Rogowitz and McClure 1995). Food intake can be roughly quantified by the number of

flights per unit of time as has been done for C. perspicillata that carries and consumes food

items in temporary night roosts (Charles-Dominique 1991). In our study, the number of

flights and hanging phases per night remained unchanged among reproductive groups,

suggesting that lactating R. pumilio did not visit more epiphyte infructescences than non-

reproductive ones.

Probably, increased food intake is achieved through increased duration of visits to epiphytes

with ripe fruits. This is suggested by the duration of the hanging phases during peak foraging

activity following dusk that exceeds more than half (55%) of those in non-reproductive

females. Indeed, assuming that hanging phases in the first two hours of the night mostly

correspond to feeding phases on infructescences, we suggest that duration of hanging phases

at this time mostly indicates food intake. Thus, the time that it takes to process pulp can be

estimated to average 11.4±6.4 for non-reproductive and 17.7±8.9 min lactating females

respectively. This is similar to or shorter than fig-eating bats that may need 7 to 31 min to

process a fig, but much longer than the small shrub-frugivore Carollia spp. that handle and

consume Piper spikes in less than 1 min (Bonaccorso and Gush 1987, Dumont 2003).

However, ripe fruits of epiphytes offer more pulp than a bat requires to reach gut capacity

(Cockle 1997, Cosson 1994), while Carollia spp. needs to rapidly consume several Piper

spikes fruits within 5-15 min to reach gut capacity (Bonaccorso and Gush 1987). It takes up

to two nights until a large chiropterochorous epiphyte infructescences is entirely consumed

(Cockle 1997, Cosson 1994). Piper spikes that offer only a small amount of pulp compared to

epiphyte infructescences are removed by the bats within hours after ripening (e.g. Thies and

Kalko 2004).

The hypothesis that lactating R. pumilio females increase their food intake by ingesting more

pulp at a time on individual epiphyte infructescences might hold true if they display a greater

gut capacity or a faster food intake or assimilation process than non-reproductive females.

Small animals have to deal with an alimentary bottleneck that limits the quantity of food they

can ingest at a time. It constrains their rate of food intake and may oblige them to take food in

small portions over the day (Koteja 1996a, Król and Speakman 2003, Speakman et al. 2001).

Rhinophylla pumilio fits well into this scenario because as in many fruits, a great proportion

of the ingested pulp volume in Araceae and Cyclanthaceae infructescences is accounted for

by water alone (80-90% wt/wt; Dinerstein 1986, Worthington 1989). Because the mass-

relative daily milk production in lactating mammals increases exponentially with decreasing

body mass (Hanwell and Peaker 1977), it places a huge demand on energy and water reserves

45

on a small mammal like bats (e.g. Kurta et al. 1989, 1990). The allometric relation proposed

by Hanwell and Peaker (1977) predicts for the small (9 g) R. pumilio that daily milk

production may reach 25% of mother’s body mass. Empirical values of 36 to 44% have been

reported for vespertilionid bats of similar size (see Table 10.4 in Kunz and Hood, 2000). Milk

production combined with the suckling of the young require mobilization of both water and

nutrients at mammary glands, and thus may partly reduce the alimentary bottleneck by

expelling ingested water at a greater rate, which in turn may allow lactating R. pumilio to

ingest more pulp at each hanging phase on an epiphyte infructescence.

Alternatively, by extending the mean duration of hanging phases beyond 15-20 min, lactating

bats may have time to cover most of the first digestive cycle, partly emptying their gut after a

rapid passage of food through the digestive tract (5 to 30 min for R. pumilio, Cockle 1997),

and initiating a second feeding cycle on the same infructescence. It is also possible that gut

capacity of females increases during lactation, as it was observed for small mammals at peak

lactation (Koteja 1996b). In any case, the mammary glands must be regularly emptied within

the foraging time to permit fast milk production, which might constrain females’ movement

pattern.

Effect of reproductive status on movement pattern

Compared to non-reproductive females, lactating females R. pumilio did not use smaller HRs.

Furthermore, they did not select day roosts closer to their respective FA nor did they display

higher site fidelity to day roosts. On the contrary, they often roosted well outside of their FA,

which means that nocturnal nursing imposes them to either regularly commute back to their

day roost at nighttime or to transport their young up to their FA. We found indirect evidence

that females transport their young into their FA during the night, leave it in temporary night

roosts and regularly fly back to feed it. This scenario stems from the observations that

lactating females (i) do not appear to leave their young in the day roosts during the night, (ii)

never returned to day roosts situated outside of their FA during the night and (iii) revisited

significantly more often (+56% according to our estimations) hanging locations that they had

already visited earlier in the night compared to non-reproductive females. Accordingly,

lactating R. pumilio mist-netted at the beginning of the night were more likely to carry a

young than lactating females caught later at night (Comm. Pers. J.-F. Cosson and M.

Delaval). This result conforms well with observations on some fruit-bats (C. perspicillata,

Pine 1972, Cosson 1994; Uroderma bilobatum, Lewis 1992) and nectar bats (e.g.

Baumgarten and Vieira 1994) that also transport their young. Transportation of young from

46

the day roost to different night roosts may reduce predation pressure at day roosts during

nighttime, or may be necessary if young of fruit-bats eat fruits with their mother before they

become volant (Fenton 1969). Our results contrast, however, with studies on most

insectivorous bats that form maternity colonies and leave their young in the day roost while

foraging. Here, lactating females have to regularly come back for nursing during the night

(e.g. Grinevitch et al. 1995, Henry et al. 2002, Swift 1980, but see also Marimuthu 1988 and

Radhamani et al. 1990 for Hipposideros speoris).

Although transporting young into their FA imposes additional energy expenditure on females

(Hughes and Rayner 1993), it may also allow substantial energy saving because they do not

need to regularly commute back to their day roost for nursing during the night. This may

allow females to spend more time foraging in total. Indeed, modifications of movement

patterns accompanying lactation in this study suggest that breeding females may be under

time pressure when foraging. On one hand, they maintained high flight activity all night long

and on the other hand, they reduced the size of their FA and flight distances by 42% and 25%

respectively compared to non-reproductive females. Temperate insectivorous also tend to

forage longer when lactating (up to >100%, Barclay 1989, Rydell 1993) and to reduce their

HR by up to 51% (Henry et al. 2002, Racey and Swift 1985, Swift 1980). In these species,

births are generally synchronized with peak insect abundance in summer, making it

impossible to distinguish between the respective effects of lactation and food availability on

activity rhythm (Henry et al. 2002).

Other mechanisms of energy compensation during lactation have been suggested for

temperate bats, including reduction of grooming activity or use of torpor (McLean and

Speakman 1999). Torpor is widely used by temperate bats to maintain a positive energy

balance in case of low ambient temperatures or food deprivation. However, it considerably

slows the rate of milk production and the growth of the young (Audet and Fenton 1988,

Hickey and Fenton 1996, Racey and Swift 1981, Wilde et al. 1999). Similar milk yield

slowdown may occur in phyllostomid bats that can use torpor as last resort in case of food

shortage (Audet and Thomas 1997, Studier and Wilson 1970). Sharing day roosts with

congeners may reduce costs of thermoregulation and risks of such physiological stress. Our

group size data indicate a significant trend of lactating R. pumilio to roost together with

conspecifics with their pup in contrast to non-reproductive females, that tend to roost more

singularly. However, this remains to be supported with larger sample sizes.

Finally, we have no satisfying explanation for the 18 to 46% increase in flight duration for

lactating R. pumilio. Tentatively, we propose that lactating females may spend more time in

47

search flight to locate food sources as close as possible to temporary night roosts where they

left their young. In contrast, flight durations in lactating C. perspicillata were estimated to be

35% shorter than in non-reproductive ones (Charles-Dominique 1991).

Conclusions on fragmentation sensitivity of R. pumilio

Traits that favor maintenance of populations in fragmented habitats include a small HR and

the ability to efficiently exploit the matrix that surrounds the fragments (Gascon et al. 1999,

Laurance et al. 2002). Accordingly, small shrub-frugivore bats Carollia and Sturnira are

more abundant in fragmented habitats compared to large fig-eating bats (Brosset et al. 1996,

Estrada and Coates-Estrada 2002, Estrada et al. 1993, Kalko 1998, Schulze et al. 2000),

probably because of high concentrations of pioneer plants, particularly Piper and Solanum

shrubs along forest edges and in areas of second growth vegetation surrounding the survey

sites of all these studies, while Ficus trees are less abundant in fragmented areas due to forest

loss. However, removing the matrix effect may produce very different pattern, as illustrated

by bat surveys in the fragmented forest of Saint-Eugène, French-Guiana, were the matrix is a

flooded area devoid of second growth vegetation (Chapter 2; Cosson et al. 1999). In this area,

the capture rate of R. pumilio and of the shrub-frugivorous bats C. brevicauda, C.

perspicillata and Sturnira tildae rapidly declined in fragments (1 to 7 ha) while that of large

fig-eating Artibeus spp. remained less affected.

As outlined by our study, search flights are an important component of the foraging strategy

of frugivorous bats feeding on a spatially scattered food resource. This foraging strategy

would be inefficient within an area where inhospitable matrix devoid of fruiting plants is too

large and requires long commuting flights. This is particularly true for R. pumilio that is

specialized on epiphytes that do not establish in young secondary growth vegetation.

Eventually, small shrub-frugivore bats may not be able to efficiently exploit a highly

fragmented habitat in which they are forced to repeatedly conduct long commuting flights.

As an alternative but not mutually exclusive explanation, small shrub-frugivore bats may not

be able to afford long commuting distances like some fig-eating bats that are up to five times

larger. This is likely to limit their ability to move between remote forest fragments. In that

respect, Cosson et al. (1999) found a significant negative correlation between fragmentation

sensitivity and body size of fruit-bats. Larger bats conduct faster flights due to higher wing

loading, which is energetically beneficial for commuting (Norberg and Rayner 1987).

Conversely, smaller bats are known to fly less efficiently (Speakman and Thomas 2003) as a

greater proportion of energy expenditure during flight is lost as heat dissipation (96% against

48

75-80%). As a consequence, when covering similar flight distances, small fruit-eating bats

expend a greater proportion of their daily energy budget than large fruit-eating bats. In

support of the link between body size and energetics, within genus Carollia, smaller species

tend to spend more time foraging and eat smaller fruits but of higher nutritional quality than

larger species (Fleming 1991, Thies 1998).

Finally, we believe that constraints of rearing a young represent a critical factor for females

R. pumilio in fragmented habitats, but also probably for females of other species whose

foraging strategy consists largely of search flights. Lactating females cover shorter distances

and spend more time foraging, while forest fragmentation implies longer flight distances

because animals have to fly back and forth between fragments and thus additional time loss.

Although some studies report evidences of breeding activity in fragmented habitats (Estrada

and Coates-Estrada 2002), this does not indicate that breeding success equals that of

populations established in continuous forest. Interestingly, the relative abundance of the

shrub-frugivorous bats in the fragmented forest of Saint-Eugène, French-Guiana,

continuously decreased over the 10 years following fragmentation (Chapter 2; Cosson 1999),

suggesting that local recruitment is possibly too low to ensure a self-sustaining population.

REFERENCES

Aldridge, H. D. J. N., and R. M. Brigham. 1988. Load carrying and maneuverability in an

insectivorous bat: a test of the 5% “rule” of radio-telemetry. Journal of Mammalogy

69:379–382.

Anthony, E. L. P. 1988. Age determination in bats. Pp. 47–58 in Ecological and behavioral

methods for the study of bats (T. H. Kunz, ed.). Smithsonian Institution Press,

Washington, D. C.

Anthony, E. L. P., and T. H. Kunz. 1977. Feeding strategies of the little brown bat, Myotis

lucifugus, in southern New Hampshire. Ecology 58: 775–786.

Arita, H. T. 1997. Species composition and morphological structure of the bat fauna of

Yucatan, Mexico. Journal of Animal Ecology 66: 83–97.

Audet, D., and M. B. Fenton. 1988. Heterothermy and the use of torpor by the bat Eptesicus

fuscus (Chiroptera: Vespertilionidae): a field study. Physiological Zoology 61:

197–204.

49

Audet, D., and D. W. Thomas. 1997. Facultative hypothermia as a thermoregulatory strategy

in the phyllostomid bats, Carollia perspicilliata and Sturnia lilium. Journal of

Comparative Physiology (B) 167: 146–152.

Barclay, R. M. R. 1989. The effect of reproductive condition on the foraging behavior of

female hoary bats, Lasiurus cinereus. Behavioral Ecology and Sociobiology 24:

31–37.

Baumgarten, J. E., and E. M. Vieira. 1994. Reproductive seasonality and development of

Anoura geoffroyi (Chiroptera: Phyllostomidae) in central Brazil. Mammalia 58:

415–422.

Bernard, E., and M. B. Fenton. 2002. Species diversity of bats (Mammalia: Chiroptera) in

forest fragments, primary forests, and savannas in central Amazonia, Brasil.

Canadian Journal of Zoology 80: 1124–1140.

Bernard, E., and M. B. Fenton. 2003. Bat mobility and roosts in a fragmented landscape in

central Amazonia, Brazil. Biotropica 35: 262–277.

Bierregaard, R. O., Jr., C. Gascon, T. E. Lovejoy, and R. Mesquita (eds.). 2001. Lessons from

amazonia: the ecology and conservation of a fragmented forest. Yale University

Press, New Haven.

Bonaccorso, F. J. 1979. Foraging and reproductive ecology in a Panamanian bat community.

Bulletin of the Florida State Museum, Biological Sciences 24: 359–408.

Bonaccorso, F. J., and T. J. Gush. 1987. Feeding behaviour and foraging strategies of captive

phyllostomid fruit bats: an experimental study. Journal of Animal Ecology 59: 907–

920.

Brander, R. B., and W. W. Cochran. 1969. Radio-location telemetry. Pp. 95–103 in wildlife

management techniques manual (R. H. Giles, ed.). Wildlife Society, Washington D.

C.

Brosset A., P. Charles-Dominique, A. Cockle, J.-F. Cosson, and D. Masson. 1996. Bat

communities and deforestation in French Guiana. Canadian Journal of Zoology 74:

1974–1982.

Charles-Dominique, P. 1986. Inter-relations between frugivorous vertebrates and pioneer

plants: Cecropia, birds and bats in French Guyana. Pp. 119–135 in Frugivores and

seed dispersal (A. Estrada, and T. H. Fleming, eds.). Dr W. Junk Publishers,

Dordrecht, The Netherlands.

50

Charles-Dominique, P. 1991. Feeding strategy and activity budget of the frugivorous bat

Carollia perspicillata (Chiroptera : Phyllostomidae) in French Guiana. Journal of

Tropical Ecology 7: 243–256.

Charles-Dominique, P. 1993. Tent-use by the bat Rhinophylla pumilio (Phyllostomidae:

Carolliinae) in French Guiana. Biotropica 25: 111–116.

Charles-Dominique, P. 2001. The filed station. Pp 1–7 in Nouragues: Dynamics and plant-

animal interactions in a Neotropical rainforest (F. Bongers, P. Charles-Dominique,

P.-M. Forget, and M. Théry, eds.). Kluwer Academic Publishers, Dordrecht, The

Netherlands.

Cockle, A. 1997. Modalités de dissémination et d'établissement de lianes de sous-bois

(Cyclanthaceae et Philodendron) en forêt guyanaise. PhD Thesis, Université Paris

VI, France.

Cockle, A. 2001. The dispersal and recruitment of Cyclanthaceae and Philodendron

(Araceae) understorey root-climbing vines. Pp 251–263 in Nouragues: Dynamics

and plant-animal interactions in a Neotropical rainforest (F. Bongers, P. Charles-

Dominique, P.-M. Forget, and M. Théry, eds.). Kluwer Academic Publishers,

Dordrecht, The Netherlands.

Cosson, J.-F. 1994. Dynamique de population et dispersion de la chauve-souris frugivore

Carollia perspicillata en Guyane Française. PhD Thesis, Université Paris XI,

France.

Cosson, J.-F., J.-M. Pons, and D. Masson. 1999. Effects of forest fragmentation on

frugivorous and nectarivorous bats in French Guiana. Journal of Tropical Ecology

15: 515–534.

de Granville, J.-J. 2001. Appendix 2: practical guide to the palms. Pp 343–350 in Nouragues:

Dynamics and plant-animal interactions in a Neotropical rainforest (F. Bongers, P.

Charles-Dominique, P.-M. Forget, and M. Théry, eds.). Kluwer Academic

Publishers, Dordrecht, The Netherlands.

de Jong, J. 1994. Habitat use, home-range and activity pattern of the northern bat, Eptesicus

nilssoni, in a hemiboreal coniferous forest. Mammalia 58: 535–548.

Delaval, M. 2004. Impacts des perturbations d’origine anthropique sur les peuplements de

chauves-souris en Guyane Française. PhD Thesis, Université Paris VI, France.

51

Delaval, M., M. Henry, and P. Charles-Dominique. 2005. Interspecific competition and niche

partitioning: example of a Neotropical rainforest bat community. Revue d'Ecologie

(Terre et Vie) 60 : 149–166.

Dinerstein, E. 1986. Reproductive ecology of fruit bats and the seasonality of fruit production

in a Costa Rican cloud forest. Biotropica 18: 307–318.

Dumont, E. R. 2003. Bats and fruits: an ecomorphological approach. Pp 398–429 in Bat

Ecology (T. H. Kunz, and M. B. Fenton, eds.). The University of Chicago Press.

Estrada, A., and R. Coates-Estrada. 2001. Bat species richness in live fences and in corridors

of residual rain forest vegetation at Los Tuxtlas, Mexico. Ecography 24: 94–102.

Estrada, A., and R. Coates-Estrada. 2002. Bats in continuous forest, forest fragments and in

an agricultural mosaic habitat-island at Los Tuxtlas, Mexico. Biological

Conservation 103: 237–245.

Estrada, A., R. Coates-Estrada, and D. Merrit, Jr. 1993. Bat species richness and abundance

in tropical rain forest fragments and in agricultural habitats at Los Tuxtlas, Mexico.

Ecography 16: 309–318.

Fauth, J. E., J. Bernardo, M. Camara, W. J. Resetarits, J. Van Buskirk, and S. A. McCollum.

1996. Simplifying the jargon of community ecology: a conceptual approach.

American Naturalist 147: 282–286.

Fenton, M. B. 1969. The carrying of young by females of three species of bats. Canadian

Journal of Zoology 47: 158–159.

Findley, J.S. 1993. Bats: a community perspective. Cambridge University Press.

Fleming, T. H. 1981. Fecundity, fruiting pattern, and seed dispersal in Piper amalago

(Piperaceae), a bat-dispersed tropical shrub. Oecologia 51: 42–46.

Fleming, T. H. 1985. Coexistence of five sympatric Piper (Piperaceae) species in a tropical

dry forest. Ecology 66: 688–700.

Fleming, T. H. 1988. The short-tailed fruit bat – A study in plant-animal interactions. The

University of Chicago Press.

Fleming, T. H., E. R. Heithaus, and W. B. Sawyer. 1977. An experimental analysis of the

food location behavior of frugivorous bats. Ecology 58: 619–627.

Fleming, T. H., E. T. Hooper, and D. E. Wilson. 1972. Three central American bat

communities: structure, reproductive cycles, and movement patterns. Ecology 53:

555–569.

Gannon, M. R., and M. R. Willig. 1997. The effect of lunar illumination on movement and

activity of the red fig-eating bat (Stenoderma rufum). Biotropica 29: 525–529.

52

Gascon, C., T. E. Lovejoy, R. O. Bierregaard, J. R. Malcolm, P. C. Stouffer, H. Vasconcelos,

W. F. Laurance, B. Zimmerman, M. Tocher, and S. Borges. 1999. Matrix habitat

and species persistence in tropical forest remnants. Biological Conservation 91 :

223–229.

Gorresen, P. M., and M. R. Willig. 2004. Landscape responses of bats to habitat

fragmentation in Atlantic forest of Paraguay. Journal of Mammalogy 85: 688–697.

Grinevitch, L., S. L. Holroyd, and R. M. R. Barclay. 1995. Sex differences in use of daily

torpor and foraging time by big brown bats (Eptesicus fuscus) during the

reproductive season. Journal of Zoology, London 235: 301–309.

Hamilton, W. J., III, and K. E. F. Watt. 1970. Refuging. Annual Review of Ecology and

Systematics 1: 263–286.

Hammond, K. A., and J. Diamond. 1994. Limits to dietary nutrient intake and intestinal

nutrient uptake in lactating mice. Physiological Zoology 67, 282–303.

Handley, C. O., Jr. 1967. Bats of the canopy of an Amazonian forest. Atlas do Simposio

sôbre a biota Amazônica 5: 211–215.

Handley, C. O., Jr., and D. W. Morrison. 1991. Foraging behavior. Pp. 137–140 in

Demography and natural history of the common fruit bat, Artibeus jamaicensis, on

Barro Colorado Island, Panamá (C. O. Handley, Jr., D. E. Wilson, and A. L.

Gardner, eds.). Smithsonian Contributions to Zoology No.511. Smithsonian

Institution Press, Washington D. C.

Hanwell, G. L., and J. M. Peaker. 1977. Physiological effects of lactation on the mother.

Symposia of the Zoological Society of London 41: 297–312.

Heithaus, E. R., and T. H. Fleming. 1978. Foraging movements of a frugivorous bat, Carollia

perspicillata (Phyllostomatidae). Ecological Monographs 48: 127–143.

Helversen, O., von, and Y. Winter. 2003. Glossophagine bats and their flowers: costs and

benefits for plants and pollination. Pp 346–397 in Bat Ecology (T. H. Kunz, and M.

B. Fenton, eds.). The University of Chicago Press.

Henry, M., D. W. Thomas, R. Vaudry, and M. Carrier. 2002. Foraging distances and home

range of pregnant and lactating little brown bats (Myotis lucifugus). Journal of

Mammalogy 83: 767–774.

Hickey, M. B. C., and M. B. Fenton. 1996 Behavioural and thermoregulatory responses of

female hoary bats, Lasiurus cinereus (Chiroptera: Vespertilionidae), to variations in

prey availability. Ecoscience 3: 414–422.

53

Hughes, P., and J. M. V. Rayner. 1993. The flight of pipistrelle bats Pipistrellus pipistrellus

during pregnancy and lactation. Journal of Zoology, London 230: 541–555.

Jones, G. 2000. The ontogeny of behavior in bats: a functional perspective. Pp 362–391 in

Ontogeny, functional ecology, and evolution of bats (R. A. Adams, and S. C.

Pedersen, eds.). Cambridge University Press.

Kalka, M., and E. K. V. Kalko. 2005. Gleaning bats as underestimated predators of

herbivorous insects: dietary composition of Micronycteris microtis

(Phyllostomidae) in Panamá. Journal of Tropical Ecology, accepted.

Kalko, E. K. V. 1998. Organisation and diversity of tropical bat communities through space

and time. Zoology 101: 281–297.

Kalko, E. K. V., D. Friemel, C. O. Handley, Jr., and H.-U. Schnitzler. 1999. Roosting and

foraging behavior of two Neotropical gleaning bats, Tonatia silvicola and Trachops

cirrhosus (Phyllostomidae). Biotropica 31: 344–353.

Kalko, E.K.V., C. O. Handley, Jr., and D. Handley. 1996. Organization, diversity, and long-

term dynamics of a neotropical bat community. Pp. 503–553 in Long-term studies

of vertebrate communities (S. M. Cody, and J. Smallwood, eds.). Academic Press,

New York.

Korine, C., E. K. V. Kalko, and E. A. Herre. 2000. Fruit characteristics and factors affecting

fruit removal in a Panamian community of strangler figs. Oecologia 123: 560–568.

Koteja, P. 1996a. Limits to the energy budget in a rodent, Peromyscus maniculatus: the

central limitation hypothesis. Physiological Zoology 69: 981–993.

Koteja, P. 1996b. Limits to the energy budget in a rodent, Peromyscus maniculatus: does gut

capacity set the limit? Physiological Zoology 69: 994–1020.

Król, E., and J. R. Speakman. 2003. Limits to sustained energy intake. VII. Milk energy

output in laboratory mice at thermoneutrality. Journal of Experimental Biology

206: 4267–4281.

Kunz, T. H., and W. R. Hood. 2000. Parental care and postnatal growth in the Chiroptera. Pp

415–468 in Reproductive biology of bats (E.G. Crichton, and P.H. Krutzsch, eds.).

Academic Press, New York.

Kurta, A. 1985. External insulation available to a non-nesting mammal, the little brown bat

(Myotis lucifugus). Comparative Biochemistry and Physiology (A) 82: 413–420.

Kurta, A., G. P. Bell, K. A. Nagy, and T. H. Kunz. 1989. Energetics of pregnancy and

lactation in free-ranging little brown bat (Myotis lucifugus). Physiological Zoology

62: 804–818.

54

Kurta, A., T. H. Kunz, and K. A. Nagy. 1990. Energetics and water flux of free-ranging big

brown bats (Eptesicus fuscus) during pregnancy and lactation. Journal of

Mammalogy 71: 59–65.

Laurance, W. F., and R. O. Bierregaard, Jr. (eds.). 1997. Tropical forest remnants: ecology,

management, and conservation of fragmented communities. The University of

Chicago Press.

Laurance, W. F., T. E. Lovejoy, H. L. Vasconcelos, E. M. Bruna, R. K. Didham, P. C.

Stouffer, C. Gascon, R. O. Bierregaard, Jr., S. G. Laurance, and E. Sampiao. 2002.

Ecosystem decay of Amazonian forest fragments: a 22-year investigation.

Conservation Biology 16:605–618.

Lewis, S. E. 1992. Behavior of tent-making bat, Uroderma bilobatum, at maternity roosts in

Costa-Rica. Journal of Mammalogy 73: 541–546.

Lewis, S. E. 1995. Roost fidelity of bats: a review. Journal of Mammalogy 76: 481–496.

Lim, B. K., and M. D. Engstrom. 2001. Bat community structure at Iwokrama forest, Guyana.

Journal of Tropical Ecology 17: 647–665.

McLean, J. A., and J. R. Speakman. 1999. Energy budgets of lactating and non-reproductive

brown long-eared bats (Plecotus auritus) suggest females use compensation in

lactation. Functional Ecology 13: 360–372.

Marimuthu, G. 1988. Mother-young relations in an insectivorous bat, Hipposideros speoris.

Current Science 57: 983–987.

Marshall, A. G. 1983. Bats, flowers and fruit: evolutionary relationships in the Old World.

Biological Journal of the Linnean Society 20: 115–135.

Medellín, R. A., and O. Gaona. 1999. Seed dispersal by bats and birds in forest and disturbed

habitats of Chiapas, México. Biotropica 31: 478–486.

Meyer, C. F. J., M. Weinbeer, and E. K. V. Kalko. 2005. Home range size and spacing

patterns of Macrophyllum macrophyllum (Phyllostomidae) foraging over water.

Journal of Mammalogy 86: 587–598.

Morrison, D. W. 1978a. Foraging ecology and energetics of the frugivorous bat Artibeus

jamaicensis. Ecology 59: 716–723.

Morrison, D. W. 1978b. Influence of habitat on the foraging distances of the fruit bat,

Artibeus jamaicensis. Journal of Mammalogy 59: 622–624.

Norberg, U. M., and J. M. V. Rayner. 1987. Ecological morphology and flight in bats

(Mammalia; Chiroptera): wing adaptations, flight performance, foraging strategy

55

and echolocation. Philosophical Transactions of the Royal Society of London (B:

Biological Sciences) 316:335–427.

Perrigo, G. 1987. Breeding and feeding strategies in deer mice and house mice when females

are challenged to work for their food. Animal Behaviour 35: 1298–1316.

Pine, R.H. 1972. The bats of the genus Carollia. Technical Monographs, Texas Agricultural

Experiment Station, Texas A and M University 8: 1–125.

Poncy, O., D. Sabatier, M.-F. Prévost, and I. Hardy. 2001. The lowland high rainforest:

structure and tree species diversity. Pp. 31–46 in Nouragues: Dynamics and plant-

animal interactions in a neotropical rainforest (F. Bongers, P. Charles-Dominique,

P.-M. Forget, and M. Théry, eds.). Kluwer Academic Publishers, Dordrecht, The

Netherlands.

Racey, P. A. 1988. Reproductive assessment in bats. Pp 31–45 in Ecological and behavioral

methods for the study of bats (T. H. Kunz, ed.). Smithsonian Institution Press,

Washington D. C.

Racey, P. A., and S. M. Swift. 1981. Variations in gestation length in a colony of pipistrelle

bats (Pipistrellus pipistrellus) from year to year. Journal of Reproduction and

Fertility 61: 123–129.

Racey, P. A., and S. M. Swift. 1985. Feeding ecology of Pipistrellus pipistrellus (Chiroptera:

Vespertilionidae) during pregnancy and lactation. I. Foraging behaviour. Journal of

Animal Ecology 54: 205–215.

Radhamani, T. R., G. Marimuthu, and N. K. Chandrashekaran. 1990. Relationship between

infant size and carrying of infants by hipposiderid mother bats. Current Science 59:

602–603.

Reis, N. R., and A. L. Peracchi. 1987. Quiropteros da regiao de Manaus, Amazonias, Brasil

(Mammalia, Chiroptera). Boletim Museu Paraense Emilio Goeldi, serie Zoologia 3:

161–182.

Reynolds, D. S., and T. H. Kunz. 2000. Changes in body composition during reproduction

and postnatal growth in the little brown bat, Myotis lucifugus (Chiroptera :

Vespertilionidae). Ecoscience 7: 10–17.

Rinehart, J. B. 2003. Offspring sex ratios provide evidence for local mate competition in

Rhinophylla pumilio in eastern Ecuador. Abstract. 30th North American Symposium

on Bat Research. Miami, Floride, USA.

Rogowitz, G. L., and P. A. McClure. 1995. Energy export and offspring growth during

lactation in cotton rats (Sigmodon hispidus). Functional Ecology 9: 143–150.

56

Rydell, J. 1993. Variation in foraging activity of an aerial insectivorous bat during

reproduction. Journal of Mammalogy 74: 503–509.

Schulze, M. D., N. E. Seavy, and F.F. Whitacre. 2000. A comparison of the Phyllostomid bat

assemblages in undisturbed neotopical forest and in forest fragments of a slash-and-

burn farming mosaic in Petén, Guatemala. Biotropica 32: 174–184.

Simmons, N. B., and R. Voss. 1998. The mammals of Paracou, French Guyana: a neotropical

lowland rainforest fauna. Part I. Bats. Bulletin of the American Museum of Natural

History 237: 0–219.

Speakman, J. R., A. Gidney, J. Bett, I. P. Mitchell, and M. S. Johnson. 2001. Limits to

sustained energy intake. IV. Effect of variation in food quality on lactating mice

Mus musculus. Journal of Experimental Biology 204: 1957–1965.

Speakman, J. R., and D. W. Thomas. 2003. Physiological ecology and energetics of bats. Pp

430–490 in Bat Ecology (T. H. Kunz, and M. B. Fenton, eds.). The University of

Chicago Press.

Studier, E. H., and D. E. Wilson. 1970. Thermoregulation in some Neotropical bats.

Comparative Biochemistry and Physiology 34: 251–262.

Swift, S. M. 1980. Activity patterns of Pipistrelle bats (Pipistrellus pipistrellus) in north–east

Scotland. Journal of Zoology, London 190: 285–295.

Thies, W. 1998. Resource and habitat use in two frugivorous bat species (Phyllostomidae:

Carollia perspicillata and C. castanea) in Panama: mechanisms of coexistence.

PhD thesis, University of Tübingen, Germany.

Thies, W., and E. K. V. Kalko. 2004. Phenology of neotropical pepper plants (Piperaceae)

and their association with their main dispersers, two short-tailed fruit bats, Carollia

perspicillata and C. castanea (Phyllostomidae). Oikos 104:362–376.

Turner, I. M. 1996. Species loss in fragments of tropical rain forest: a review of the evidence.

Journal of Applied Ecology 33: 200–219.

Tuttle, M. D. 1976. Population ecology of the gray bat (Myotis grisescens): factors

influencing growth and survival of newly volant young. Ecology 57: 587–595.

Weinbeer, M., and E.K.V. Kalko. 2004. Morphological characteristics predict alternate

foraging strategy and microhabitat selection in the orange-bellied bat,

Lampronycteris brachiotis. Journal of Mammalogy, 85: 1116–1123.

Wilde, C. J., C. H. Knight, and P A Racey. 1999 Influence of torpor on milk protein

composition and secretion in lactating bats. Journal of Experimental Zoology 284:

35–41.

57

Worthington, A. H. 1989. Adaptations for avian frugivory: assimilation efficiency and gut

transit time of Manacus vitellinus and Pipra mentalis. Oecologia 80: 381–389.

Worton, B. J. 1989. Kernel methods for estimating the utilization distribution in home-range

studies. Ecology 70:164–168.

Zortéa, M. 1995. Observations on tent-using in the carolline bat Rhinophylla pumilio in

southeastern Brazil. Chiroptera Neotropical 1: 85–88.

58

Chapitre 2

The role of habitat fragmentation and food availability in limiting populations of

understory fruit bats in French Guiana.

Rameau de Piper aduncum portant des infructescences, et détail d’une graine (dessin S. Jouard).

0,5 mm

59

THE ROLE OF HABITAT FRAGMENTATION AND FOOD AVAILABILITY IN LIMITING POPULATIONS OF UNDERSTORY

FRUIT BATS IN FRENCH GUIANA.

ABSTRACT

Bats are important components of tropical ecosystems as pollinators and seed dispersers. Yet

they have to face the challenge of performing within increasingly fragmented habitats, as

destruction of rainforests is expanding severely worldwide. We focused here on understory

fruit bats and predicted that their abundance in a fragmented forest would depend more on

habitat connectivity than on food availability, and that their decline in abundance would

continue during the decade following fragmentation. In that respect, the recently flooded area

of Saint-Eugène, French Guiana, where forest fragments are isolated by water, provides a

unique opportunity for interpreting fragmentation effects within the context of a rather neutral

matrix. In this study, we combined bat mist-net sampling and plant resource surveys

(epiphytes and Piper) conducted at 18 sites of various levels of disturbance, ranging from

undisturbed mainland to small isolated fragments. Landscape disturbance was quantified at

each site using a connectivity index and a remoteness index based on a satellite image

analysis. Mist-net sampling yielded 267 understory fruit bats, among which nearly 60% were

from the epiphyte-specialist species Rhinophylla pumilio. Due to small sample sizes, the

shrub-frugivorous bats Carollia brevicauda, C. perspicillata and Sturnira tildae were pooled

together in analyses. As predicted, the abundance of both R. pumilio and the shrub-

frugivorous bats significantly decreased with loss of habitat connectivity, while the effect of

food availability was not retained in models. Furthermore, we found that abundance of shrub-

frugivorous bats continuously declined during the 10 years following fragmentation, possibly

due to a drop in Piper availability and an apparent vulnerability to habitat remoteness. On the

contrary, abundance of R. pumilio did not significantly change after the initial population

drop following fragmentation. This might result from their small area requirement and from

the maintenance over the time of their resources (epiphytes). These results further support the

fragmentation-sensitivity of these understory fruit bats as a whole, but also underline that the

epiphyte-specialist R. pumilio appears more prone to maintain self-sustaining populations in

fragmented landscapes than the shrub-frugivorous species.

60

INTRODUCTION

Forest fragmentation is now unequivocally viewed as one of the most threatening

perturbations for tropical biodiversity (Bierregaard et al. 2001, Laurance and Bierregaard

1997, Laurance et al. 2002, Tabarelli et al. 2004). Fragmentation does not merely induce pure

habitat and species loss but also affects the integrity of ecosystem functioning through a

variety of direct and indirect mechanisms. In fragmented forests, edge length increases,

leading to forest desiccation, higher tree mortality, lower plant recruitment and invasion of

disturbance-adapted plants (Bierregaard et al. 1992, Didham and Lawton 1999, Tabarelli et

al.1999). This in turn results in sharp simplifications of animal communities, species ranking

higher in food chains being the most affected (Laurance et al. 2002, Turner 1996).

Among animals, the abundance and diversity of bats are widely altered in disturbed and

fragmented neotropical forests (Brosset et al. 1996, Cosson et al. 1999ab, Estrada and Coates-

Estrada 2001, 2002, Estrada et al. 1993, Gorresen and Willig 2004, Kalko 1998, Pons and

Cosson 2002, Schulze et al. 2000). Yet, they play a crucial role of plant pollinators

(Helversen and Winter 2003) and seed dispersers (Charles-Dominique 1986, Medellín and

Gaona 1999). Thus, identifying factors responsible for the decline of bat populations in

fragmented forests is an issue of ongoing importance in the field of bat ecology.

All bat species do not respond equally to habitat fragmentation. For instance, some small fruit

bats of the genera Carollia and Sturnira feeding on understory plants are known to occur at

high abundances in fragmented habitats compared to the large fig-eating bats of genus

Artibeus that appear more vulnerable to fragmentation. A consensus has arisen in the

literature (Brosset et al. 1996, Estrada and Coates-Estrada 2001, 2002, Estrada et al. 1993,

Schulze et al. 2000) that the maintenance of understory fruit bats in disturbed areas results

from the high densities of the shrubs they are specialized on (Piper, Solanum and Vismia)

within the second-growth vegetation. This corroborates the general finding that the most

fragmentation-tolerant animal species (including both invertebrate and vertebrate taxa) are

also those species that can tolerate or take advantage of the matrix surrounding fragments

(Gascon et al 1999, Laurance et al. 2002). According to the hypothesis that matrix exerts a

major influence on habitat use by bats, an aquatic matrix free from any regrowth may

produce a different picture of bat communities. In their flooded study area, Cosson et al.

(1999a) found that shrub-frugivores declined much more rapidly and markedly after

fragmentation than the large fig-eating Artibeus spp.

61

Several hypotheses may be invoked to explain the decline of small understory fruit bats in

fragmented habitats surrounded by a truly inhospitable matrix. This may result either directly

from the loss of habitat connectivity per se (reluctance of bats to forage in fragmented

habitats) or indirectly from a decrease in plant resources due to microclimate changes in

fragments. Beside these two non mutually-exclusive hypotheses, small understory fruit bats

may not maintain self-sustaining populations in the long term because fragmentation may

reduce the fitness of females. Lactating females of Rhinophylla pumilio, for instance, need to

reduce their foraging area and flight distances by 42% and 25% respectively, and at the same

time increase their nocturnal flight time by18-46% to cope with the physiological constraints

of producing milk and the temporal constraints of feeding their young (Chapter 1). Yet,

habitat fragmentation may exert strong demands on their energy and time budget, and

eventually may reduce breeding success. This may cause progressive population decline over

the years following fragmentation.

The objective of this study was to investigate the respective roles of habitat connectivity and

food availability in maintaining populations of understory fruit bats in the fragmented forest

of Saint-Eugène, French Guiana, whose bat community was previously surveyed by Cosson

et al. (1999a). More specifically, we predicted that (i) the abundance of understory fruit bats

depends more on habitat connectivity than on resource availability and that (ii) populations of

understory fruit bats decline over time since fragmentation. This study combines bat and food

resource surveys undertaken in forest fragments and adjacent mainland of Saint-Eugène

during the first decade following fragmentation. We particularly focused on the commonest

small understory fruit bats, namely the epiphyte-specialist Rhinophylla pumilio and the three

shrub-frugivores Carollia brevicauda, C. perspicillata and Sturnira tildae. To determine the

appropriate spatial scale of our landscape descriptors, we chose R. pumilio as a model species

and monitored its movement patterns using radio-tracking.

METHODS

Study area and time periods

Surveys were undertaken at the Saint-Eugène study area (4° 51’ N, 53° 04’ W), French

Guiana. The pristine forest surrounding Saint-Eugène was artificially fragmented by the

creation of the Petit-Saut hydroelectric dam built on the Sinnamary river, 60 km downstream

from the study area in early 1994. The subsequent flooding transformed 465 km2 of

62

continuous forest into a reservoir lake covered by 100 km2 of tiny forested fragments, mostly

<10 ha in area.

Total annual rainfall averages 3250 mm, with a main dry season from August to November

and a shorter, less marked one in early March. Descriptions of the local bat communities are

provided by Cosson et al. (1999ab) and Pons and Cosson (2002). Additional information on

the local climate, forest composition, and communities of invertebrates, birds and terrestrial

mammals can be found in Granjon et al. (1996) and Forget (2002).

All bat surveys were conducted during 1- to 1.5-month sessions during main dry seasons of

years 1995-97 and 2002-04. These two periods corresponded respectively to 2nd-4th and 9th-

11th years following fragmentation. They were termed periods of “recent” and “older”

fragmentation respectively.

Bat surveys

The bat sampling survey consisted of repeating past samplings of the 1995-97 period (recent

fragmentation) along with three additional field sessions in 2002-04 (older fragmentation)

using the same standardized methods at the same sites. We selected four mainland sites and

14 fragments (size 0.8 to 7.5 ha, Fig. 1) surveyed at least twice in 1995-97 (Cosson et al.

1999a, Cosson and Pons unpublished data). These 18 capture sites are located within a 4×4-

km area encompassing a portion of the flooded lake and the adjacent mainland. Within each

site, bats were captured simultaneously at two to five capture stations located at least 50 m

apart and at least 20 m away from the shoreline. Each capture station consisted of a group of

three mist nets (12×2.5 m, mesh 38 mm) set at ground level in T pattern when possible, or in

line otherwise. Two avoid biases due to trap-shy behavior of bats, each site was sampled a

single night per field session, but four to six times out of the six field sessions. From one field

session to the next, we tried to set stations at the same place, but this was often difficult due

to tree falls and vegetation regrowth on trails that remained unattended between field

sessions. The spatial extent occupied by capture stations at the mainland sites roughly

equaled that in fragments (except for fragment 22 which was too small to accommodate more

than two capture stations).

During capture nights, nets were opened from dusk to dawn (18:30 – 06:30) and were

continuously checked during the first and the last two hours of the night, and every 2 h

otherwise. Capture interruptions due to heavy rain were rare and short because we worked

during dry seasons. No netting was done during the 6-day periods encompassing full moon,

due to a possible slowdown in bat captures. Captured bats were kept in cloth bags before

63

Figure 1. Study area map showing the location of the 28-ha fragment were radio-tracking survey was undertaken, as well as the 18 capture sites: four mainland sites (letters) and 14 fragments (numbers).

being identified to species following a key derived from Charles-Dominique et al. (2001) and

Simmons and Voss (1998). Reproductive status of females (pregnant, lactating or non-

reproductive) was assessed by checking for the presence of a palpable fetus or of prominent

hairless nipples (Racey 1988). Juveniles were distinguished from adults according to the

degree of fusion of metacarpal epiphyses (Anthony 1988). Before being released at the

sampling site, bats were marked in the inter-scapular region using a black indelible ink to

detect possible inter-site movements and to discard recaptures from the same night. Because

the field sessions were conducted during the most marked of the two annual fruit bat

reproductive peaks (transition from dry to rainy seasons, October-November), we could

compare levels of breeding activity between fragments and mainland capture sites.

Proportions of reproductive adult females (pregnant or lactating) and of juveniles were

compared using χ² tests.

215

34

206

9 24

8 17

53

Cp12

Rg16

Oi

2219

ViN

Radio-tracking fragment

0 1km

64

Landscape descriptors

To describe the local forest structure and disturbance at each capture site, we used two

landscape descriptors, a habitat connectivity index CI and a habitat remoteness index RI,

derived from Hewison et al. (2001) and Coulon et al. (2004). CI measures the extent of

forested habitat within a certain radius around capture sites, while RI is an indicator of

isolation measuring the potential difficulty for bats to reach this site from the nearest suitable

areas. These calculations were based on a SPOT satellite image (resolution 20 m) of the study

area taken in 1996, transformed into a binary map (water vs. forest habitat) of 250×250 pixels

and exported as a binary text matrix using the software ImageJ 1.33u (National Institutes of

Health, USA; URL: http://rsb.info.nih.gov/ij/). CI and RI values were calculated in three

steps (Fig. 2). First, we assigned to each landscape unit, i.e. each map pixel corresponding to

a 20×20 m plot, an arbitrary suitability value equaling 0 or 200 (for water and forested units

respectively). Second, each suitability value was replaced by the mean suitability value of all

Figure 2. Treatment of the map for calculation of the habitat connectivity index CI and remoteness index RI. Step 1: initial binary map where each pixel (or landscape unit) can take only two values, 0 for “water” and 200 for “forest”. Step 2, we assigned to each landscape unit the mean value of all neighboring units located within a 400-m radius. Step 3 is a similar process, but values of neighboring units in step 2 are weighted by a coefficient depending on their respective distance to the considered landscape unit (see results) to transform the step-2 matrix into the CI matrix. Upper graph shows the CI profile of all landscape units transected by the segment [AB] bridging fragment 22 to the nearest landscape unit of maximum CI. Lower graph is the same, but showing the reverse values 1/(CI+1) that are summed to produce the final remoteness index RI.

0

50

100

150

200

A

B

A B

0

0.01

0.02

A B

CI

1/(C

I+1)

Step 1

Step 2

Step 3

65

neighboring landscape units within a given radius. A mean value of 200 indicates that the

considered landscape unit is completely surrounded by forested units within the chosen

radius, while a value of 100 indicated that only 1 out of 2 landscape units is forested. These

mean values denote the proportion of forested habitat within the chosen radius but do not take

into account the size of forest fragments. To overcome this limitation, we finally assigned

higher weighting coefficients to closer landscape units, and lower weighting coefficients to

farther ones, for the calculation of mean suitability values (see below). The resulting habitat

connectivity index CI ranges from 0 (no forested habitat within the chosen radius) to 200

(only forested habitat) and decreases sharply when forested habitat becomes scarce in the

immediate vicinity of the considered landscape unit.

Once the CI matrix was computed, we assigned to each capture site the CI value of their

central landscape unit. The remoteness index RI was calculated as the sum of the inverse

Σ1/(CI+1) of all landscape units a bat has to cross when reaching the capture site by flying in

a straight line from the nearest landscape unit of maximum CI (200). Thus, RI equals ~0

when the capture site has a CI=200, and otherwise increases quickly when isolated by large

numbers of landscape units with nil CI values.

Following this method, two important parameters are needed for CI and RI calculations,

namely (i) the length of the radius enclosing the so called neighboring landscape units, and

(ii) the mathematical function determining the distance-dependant weighting coefficients

assigned to these neighboring units. To define weighting coefficients, we sought a

mathematical function describing the frequency distribution of minimal distances bats cover

when they search for food. Search flights devoted to food localization are an important

component of foraging activity of understory fruit bats (e.g. Fleming et al. 1977), and their

lengths give an indication of the extent of habitat bats can investigate in a single flight to find

fruiting plants. The longest flight distances set the radius length, i.e. the distance after which

weighting coefficients would become nil.

Only a radio-tracking survey can provide all the required flight distance data. For this

purpose, we chose the epiphyte-specialist R. pumilio as a model because its particular diet

makes search flights easy to monitor. Indeed, as epiphyte infructescences are too large to be

removed and transported in feeding roosts, bats consume them right in situ (Cockle 1997).

Therefore, a part of the hanging locations reported by radio-tracking may correspond to the

location of consumed fruiting plants (Chapter 1). In turn, distances between successively

visited hanging locations may merely describe flight distances individuals cover among food

sources. In this study, we assume that the shrub-frugivorous species adopt similar movement

66

patterns within their FAs when flying from shrub to shrub in search of food. These species

shuttle back and forth between fruiting plants and feeding roosts, and then flight distances

between fruiting plants are less easily determined.

Movement pattern of R. pumilio

To assess flight distances, we radio-tracked three individuals (two males M1 and M2 and one

female F1). We chose to carry out the radio-tracking survey on a 28-ha fragment (initially

estimated as 40 ha in Cosson et al. 1999a) with a very irregular border that delineates narrow

peninsulas and zones of forest disruptions (Fig. 1). This rather flat forest patch was crossed

by numerous trails forming a 100-m spacing grid, which greatly facilitated radio-tracking.

Bats were mist-netted in this fragment in November 1999 and fitted with 0.70-g radio-

transmitters (Biotrack, UK) representing <7.5% of their body mass. Transmitters were

attached to the back of bats using surgical SkinBound® (Smith and Nephew Inc.,

Mississauga, Ontario, Canada) after a small amount of dorsal fur was trimmed. Bats were

released at the capture site within 30 min after capture. No tracking data was taken on the

capture night to avoid biases resulting from stress response to manipulation. Bats were

tracked afterward for 3-4 nights by two observers in radio-contact and each equipped with a

CE-12 receiver (Custom Electronics, Urbana, Illinois) and a four-element Yagi antenna.

Radio-tracking nights involved determining whenever possible the hanging locations of bats

by triangulation. Bats were considered to be hanging when the signal intensity was judged

constant in direction and intensity for at least 1 min. Day roosts were also located whenever

possible.

Triangulation data were computed and analyzed with the software Tracker 1.1 (Camponotus

AB, Solna, Sweden, 1994) after invalid bat locations were discarded, i.e. points >400 m away

from observer positions (maximal estimated detection range) or points situated over water

outside of the forest fragment. All direct flight distances between each pair of successive

hanging locations visited by bats were computed.

To improve the reliability of the mathematical modeling of the flight distance frequency

distribution, we supplemented our dataset with similar data collected on eleven individuals of

R. pumilio (Chapter 1). Although the latter study was undertaken in an area relatively close to

Saint-Eugène on a geographic scale (the Nouragues undisturbed forest, 110 km south-east),

possible behavioral differences between individuals from the two study areas may

compromise the adequacy of landscape descriptors. Therefore, flight distance data were

67

compared beforehand via a nested ANOVA to ensure that the study area effect was low

enough to support data pooling.

We also ensured that home ranges and foraging areas of R. pumilio were similar in size in the

two sites. To conform Chapter 1 and other studies (Meyer et al. 2005, Thies 1998, Weinbeer

and Kalko 2004), we delineated (i) home range “HR” (area used by bats for roosting during

the day and foraging during the night) by a convex polygon enclosing 100% of day roosts and

night hanging locations, and (ii) foraging areas “FAs” (areas used to forage at night) by the

95% isopleths, i.e. the curve(s) enclosing 95% of probability of presence of bats. The latter

curves were defined by the probabilistic method Adaptive Kernel (Worton 1989) with the

default smoothing coefficient of the software Tracker 1.1 (CV=0.15).

Food availability

To quantify food available to understory bats, we focused on Piper and epiphytic

Cyclanthaceae and epiphytic Philodendron spp. (Araceae) known to constitute keystone

resources for Carollia spp. and R. pumilio, respectively (Charles-Dominique and Cockle

2001, Cockle 1997, 2001, Cosson 1994, Delaval et al. 2005, Fleming 1982, 1985, Thies and

Kalko 2004). Sturnira spp. consume mostly Solanum fruits (Cosson 1994, Marinho-Filho

1991), but these plants were probably mostly restricted to tree fall gaps and were so rarely

encountered in the forest understory that we did not attempt to assess their density.

Nevertheless, Sturnira tildae also feeds regularly on Piper and epiphyte fruits (e.g. Cosson

1994, Delaval et al. 2005). Plant resources were censused within 4 to 5 200-m² plots (5×40

m) per site, uniformly distributed along the succession of capture stations. Botanical surveys

occurred in November 1999 and 2004, (5 and 10 years after completion of fragmentation)

and were associated with recent and older fragmentation periods respectively.

Only shrubby Piper individuals >50 cm tall were counted, i.e. the estimated minimum size

required for fruit production. Virtually all the species encountered in this study fitted some

usual chiropterochorous syndromes, with greenish fruiting spikes erected upward and easy to

grab. To better estimate potential fruit production, we counted on each Piper the number of

terminal branches and the number of flowering and fruiting spikes. Epiphyte infructescences

constitute the main diet of R. pumilio and are occasionally consumed by species of Carollia

and Sturnira in French Guiana. Following diet descriptions provided by Cockle (1997), we

concentrated our interest on Asplundia heteranthera, Evodianthus funifer and Thoracocarpus

bissectus (Cyclanthaceae), and several Philodendron species (Araceae), including P.

billietae, P. grandifolium, P. insigne, P. linnaei, P. pedatum, P. squamiferum, and P.

68

subgenus Pteromischum (P. duckei, P. guianense, P. placidum). Most of these are epiphytes

whose adventitious roots develop on trunks, at understory to sub-canopy level (1 to 8 m

above ground level). Therefore, we could visually census adult individuals with reasonable

accuracy, but the presence of fruits could not be documented as in Piper.

The overall spatiotemporal variability of food availability was assessed by testing the

respective effects of habitat connectivity CI and fragmentation age (recent vs. older periods)

on numbers of epiphytes or Piper spikes or branches per plot, using general linear models

(GLMs). A log-transformation successfully normalized food availability data (Kolmogorov-

Smirnov test of normality).

Determinants of bat abundances

The purpose of our study was to discriminate between the respective contribution of

landscape structure, food availability and fragmentation age in explaining the local

abundance (capture rate) of the epiphyte-specialist R. pumilio and the shrub-frugivorous bats

pooled together. Due to the ubiquity of zeroes in capture data, no data transformation could

produce normality of capture rates (mean number of bats caught per capture station and per

night). Alternatively, we did not use capture rates as an abundance index, but simply numbers

of bats caught per site and per period. In order to account for uneven sampling efforts among

the different sites and periods, we introduced in the model a variable “capture effort” (total

number of station-nights) for each site and period, sampling stations being standardized as

three mist nets monitored during an entire night. Food availability data (numbers of

epiphytes, Piper spikes and Piper branches per plot) were averaged to retain a single value by

site and per period.

We therefore used stepwise regressions (GLMs) and specified different link functions or

distributions to find which model would best approximate reality given the data we have

recorded (e.g., Poisson, Binomial, Log-ratio; see McCullagh and Nelder 1989). We started

the modeling with a Poisson distribution that is well adapted to count data such as capture

numbers (Legendre and Legendre 1998, Zar 1998). The estimated dispersion parameter

tended to be high (generally >3) whatever the model fitted, indicating overdispersion of the

data (mean>variance, whereas Poisson-distributed data have a mean=variance) and

suggesting that Poisson models are probably not adequate (McCullagh and Nelder 1989).

Alternatively, we used a Negative Binomial distribution and a Log-ratio function. Estimated

dispersion parameters were then close to 1. All two-way interactions among factors were

taken into account in a first step. We used the Akaike Information Criterion (AIC) to select

69

the best model considering fit and complexity (Johnson and Omland 2004), the best model

being the model with the lowest AIC. Statistical analyses were carried out with GenStat 6.2

(Payne et al. 2003).

RESULTS

Bat surveys.

The six field sessions totaled a capture effort of 255 station-nights (765 net-nights) and

yielded 267 small understory fruit bats (see appendix) belonging to our four target species,

namely the epiphyte-specialist R. pumilio (59.2%), and the three shrub-frugivorous C.

brevicauda (18.7%), C. perspicillata (10.9%), and S. tildae (11.2%). Only 6.5% of the shrub-

frugivores were captured in the forest fragments (n=109) against 31.6% of R. pumilio

individuals (n=158). Therefore, the abundance decrease in fragments compared to mainland

was significantly more pronounced for the former species than for the latter (χ²=24.366,

p<0.001). While shrub-frugivores were detected in 5 out of the 14 (35.7%) fragments during

the recent fragmentation period, their capture rates remained nil during the whole older

fragmentation period and for any fragment.

We did not find any significant difference in breeding activity of R. pumilio between

fragments and mainland sites, neither regarding the ratio of juveniles-to-adults (1: 4.75 and 1:

3.75 respectively, n=130, χ²=0.239, p=0.625), the ratio of reproductive-to-non-reproductive

adult females (1: 0.18 and 1: 0.40 respectively, n=48, χ²=0.879, p=0.348), nor adult sex-ratio

(females: males equals 1: 0.76 and 1: 0.85 respectively, n=106, χ²=0.064, p=0.800). Due to

restrictive capture numbers, the same statistics could not be applied to the shrub-frugivorous

bats. Nonetheless, no evidence of breeding activity was recorded for them in fragments since

only adult males were captured there.

No inter-site movement was reported in the course of our bat surveys (which were not

designed for that purpose). Conversely, two males R. pumilio were recaptured at the same

capture site they were banded at as long as 7 and 9 years before.

Movement pattern and landscape descriptors

R. pumilio individuals exhibited a foraging pattern (Table 1, Fig. 3) fairly similar to that

reported from Nouragues (Chapter 1). They used a single 4.5- to 14.1-ha FA (compared to

3.5 to 14.1 ha in Nouragues) and displayed rather short flight distances (90% of flight

70

Table 1: Descriptive results of radio-tracking sessions undertaken on three R. pumilio in Saint-Eugène, French Guyana, and comparison with the range of values obtained on 11 individuals at Nouragues (Chapter 1).

Individual identification a M1 F1 M2 Chapter 1 Survey duration (days) 11 9 8 3 – 5 Nb of tracking nights 4 3 3 3 – 5 Nb of day roost localizations 9 8 7 3 – 5 Nb of distinct day roosts 8 5 5 1 – 5 Nb of valid fixes 78 23 33 34 – 63 Home range (convex polygon) (ha) 11.1 6.2 3 2.5 – 16.9 Foraging area b (95% kernel) (ha) 7.3 14.1 4.5 3.5 – 14.1 Flight distances (m±SD) 121.4±77.6 120.3±77.0 87.1±43.9 41.1±27.7 –

117.7±77.6 a M: male, F: female. b a single foraging area for all individuals Figure 3. Spatial use by three individuals of R. pumilio radio-tracked in a 28-ha forest fragment: home range (convex polygons), foraging area (95% kernel, bold line), nocturnal fixes (small squares) and day roosts (large squares).

distances <200 m, and maximal distances=400-415 m in both cases). All tracked individuals

used day roosts located within or close to (<215 m) their respective FA, so that their HR size

(3.0 to 11.1 ha) remained in the range of values reported from Nouragues (2.5 to 16.9 ha). No

long commuting flights toward mainland or other fragments were recorded.

Mean flight distances between successive hanging locations was slightly longer in Saint-

Eugène than in Nouragues (116±72 m, n=93, and 102±75 m, n=231, respectively). However,

a nested ANOVA performed on the square-root transformed flight distances revealed that this

difference was due to a significant inter-individual variability (F=2.317, df=7, p=0.026)

rather than an effect of the study area (F=0.248, df=1, p=0.248). Therefore, we felt

comfortable in using tracking data from Nouragues together with those from Saint-Eugène to

improve our modeling of flight distances of R. pumilio.

200m 200m 200m

N N N

71

To model frequency distribution of minimal flight distances, we first pooled all of the 324

flight distance values and calculated for each 5-m distance class ranging from 15 to 415 m the

proportion of flights longer than this distance (Fig. 4). Then, we applied a logistic regression

on these values as a function of distance, which describes the probability that a given flight

will at least encompass a certain distance. Owing to the absence of long commuting flights,

flight distances displayed an homogeneous unimodal distribution, and the logistic regression

explained a high proportion of variability (R²>0.99). The logistic regression was then used to

determine the distance-dependant weighting coefficients required for calculation of the

habitat connectivity index CI. According to this function, the weighting coefficient becomes

nearly nil at a distance of 400 m. In other words, for each landscape unit, CI is calculated

over a 400 m-radius circular area (50.3 ha). The resulting CI values for the 18 capture sites

ranged from 53.2 (the smallest fragment) to 200 (one of the mainland sites) and averaged

119.7±45.8 (Appendix). The remoteness index RI was minimum (<0.01) for the latter

mainland site, peaked at 9.3 to 9.5 for two remote fragments, and averaged 2.0±3.2.

Figure 4. Graphical representation of the weighting coefficient attributed to neighboring pixels (or landscape units) as a function of their distance to the considered landscape unit when calculating habitat connectivity index CI. This curve was determined as the logistic regression of the observed frequency distribution of minimum flight distances R. pumilio covers between two successively visited hanging locations (dots). For instance, the probability that an individual covers at least 100 m from a hanging location to the next is around 0.42. The upper graph is a 3-dimentional representation of the same function, X and Y being the spatial coordinates of neighboring landscape units from the considered landscape unit.

0

0.2

0.4

0.6

0.8

1

0 100 200 300 400 500

Distance to the considered landscape unit (m)

Coe

ffic

ient

Coe

ffic

ient

Y (m)

X (m)

Coeff.=1/(1+e(0.025dist. – 2.346))R²=0.991

72

Food availability

A total of 886 Piper individuals were censused on 115 200-m² plots (14 sites per period).

Epiphyte censuses are available for only 78 of these 115 plots (6 and 14 sites in recent and

older fragmentation periods, respectively) and totaled as many as 1085 individuals. Piper

fruiting rate within plots (nb of fruiting or flowering spikes per branch) did not differ between

mainland and fragments (Mann-Whitney U=1466, df=1, p=0.211, n=109). Assuming that the

number of terminal branches indicates the potential fruiting rate (correlation: Pearson r=0.81,

p<0.001), we found that Piper resource availability was positively and significantly

influenced by the loss of habitat connectivity (decreasing CI), and also underwent a very

significant 67% decrease from recent to older fragmentation periods (Table 2, Fig. 5).

Epiphyte density did not vary with CI nor with fragmentation age (Table 2).

Determinants of bat abundances

Surprisingly, the stepwise regressions did not retain capture effort as a significant contributor

to bat capture counts. Therefore, we repeated the tests after having re-equilibrated sampling

efforts by randomly discarding from data sets several capture stations for the four most

sampled sites. However, this produced exactly the same results given that capture rates varied

little in those sites.

The regression analysis (Table 3) showed that only the forest connectivity index CI had a

significant positive effect on the abundance of R. pumilio (Table 3, Fig. 6). The best model

for R. pumilio, according to AIC values, was the model “Constant + CI” which explained

25.68% of the total deviance (GLM using a Negative binomial distribution with estimated

dispersion parameter=0.78). No other variable or two-way interactions among variables were

retained in the best model. Food availability, estimated either with the number of Piper spikes

or terminal branches or the number of epiphytes at the time of the sampling, had no

significant influence on the abundance of bats (Table 3). Likewise, we did not detect any

significant effect of the capture effort within each site and period.

The best model for the shrub-frugivorous bats, according to AIC values, was the model

“Constant + CI + RI” which explained 67.45% of the total deviance (GLM using a Negative

binomial distribution with estimated dispersion parameter=0.86). Both variables significantly

influenced the number of shrub-frugivorous bats captured at a particular site and period.

Although the Deviance ratio of CI was by far higher than that of RI, including the latter in the

model substantially improved the AIC value, i.e. AIC=44.19 when CI was alone in the

73

Table 2: Results of GLM performed to compare food availability along habitat connectivity gradients (CI) and among fragmentation ages (recent 2-4 yrs vs. older 9-11 yrs fragmentation periods). Dependant variables (number of Piper branches and number of epiphytes per 200-m² plot) were transformed following Log(value+1). Normality was verified using a Kolmogorov-Smirnov test.

df F-ratio p effect sign Piper branches (n=115) Habitat connectivity CI 1 7.27 0.008 – Fragmentation age 1 4.12 0.045 – Interaction 1 0.03 0.871 Epiphytes (n=78) Habitat connectivity CI 1 0.60 0.441 Fragmentation age 1 1.22 0.272 Interaction 1 1.09 0.299

Figure 5. Significant negative relation between habitat connectivity CI and the number of Piper branches per sampling plot, as shown by linear regressions. Piper were significantly less abundant in older fragmentation period, but the slope of the relation with CI remained unchanged.

0

1

2

50 100 150 200

○ Recent frag. ● Older frag.

Connectivity Index

Log

(nb

of P

iper

bra

nche

s + 1

)

Table 3: Sources of variation in the abundance of the epiphyte-specialist R. pumilio and of the shrub-frugivorous C. brevicauda, C. perspicillata and S. tildae within the different capture sites. Results are outputs of a GLM using a Negative Binomial distribution. Factors were sorted by order of decreasing deviation ratio; n=36 except for Epiphytes (n=20) and Piper (n=28). All two-way interactions were non significant and discarded from the table.

Factors df Dev. ratio p effect sign Rhinophylla pumilio

1. Habitat connectivity CI 1 13.80 <0.001 +

2. Fragmentation age 1 1.53 0.225

3. Habitat remoteness RI 1 1.46 0.235

4. Piper spikes per plot 1 1.31 0.263

5. Epiphytes per plot 1 0.90 0.357

6. Piper branches 1 0.83 0.370

7. Sampling effort 1 0.24 0.625

Shrub-frugivorous bats

1. Habitat connectivity CI 1 42.30 <0.001 +

2. Habitat remoteness RI 1 8.37 0.007 –

3. Epiphytes per plot 1 3.36 0.085

4. Piper branches per plot 1 2.15 0.155

5. Piper spikes per plot 1 1.69 0.205

6. Sampling effort 1 0.12 0.736

7. Fragmentation age 1 0.05 0.820

Figure 6. Effect older fragmentadetailed statistic

30

35

0

5

10

15

20

25

30

35

5

o s

Cap

ture

rate

(in

divi

dual

s / 1

0 st

atio

n-ni

ght)

□ Recent frag. × Older frag.

R. pumili

of habitat connectivity CI on capture rates of untion periods. Linear regressions are shown for bal outputs.

0

5

10

15

20

25

500 100 150 200

Connectivity I

shrub-frugivorous bat

74

derstory fruit bats during the recent and etter representation, but see Table 3 for

100 150 200

ndex

75

model, and 38.06 when RI was added. This suggests that both habitat connectivity and

remoteness are important for explaining the local abundance of these bats. Otherwise, no

other variable, including estimates of food availability or time since fragmentation and their

two-way interactions, were significant (Table 3).

Contrary to our hypothesis, the effect of fragmentation age (recent 2-4 yrs vs. older 9-11 yrs

fragmentation periods) remained low and undetected by the regression for both R. pumilio

and shrub-frugivores. For R. pumilio, however, the period effect ranked second and AIC

value for the model “Constant + CI + Period” was only slightly higher than that of the model

“Constant + CI” (39.49 and 39.04 respectively). Notwithstanding, only the abundance of the

shrub-frugivores underwent a significant decline detected by an alternative non parametric

test (sign test: p=0.039, n=18). During the older fragmentation period, they were only

reported from the three mainland capture sites with the highest habitat connectivity index and

the lowest remoteness index.

DISCUSSION

As predicted, loss of habitat connectivity was the proximal determinant of the decrease in

abundance in both R. pumilio and shrub-frugivorous bats, confirming their fragmentation-

sensitivity, while the effect of food availability was not retained in models. We found

evidence that abundance of shrub-frugivorous bats decreased during the 10 years following

fragmentation, possibly due to a concomitant drop of Piper resource availability and an

apparent reluctance to forage in remote habitats. On the contrary, R. pumilio abundance did

not change significantly, which might result from their small area requirements and from the

maintenance over the time of availability of their main food resources (epiphytes).

Rhinophylla pumilio seems less fragmentation-sensitive than their shrub-frugivorous

counterparts.

Food availability hypothesis vs. habitat connectivity hypothesis

We can reasonably conclude from our study that the understory fruit bats we focused on were

reluctant to forage in fragmented habitats mostly due to the loss of habitat connectivity per se

and not because of a confounding factor related to resource availability. While the abundance

of both the epiphyte-specialist R. pumilio and the shrub-frugivores decreased along a gradient

of connectivity loss, estimates of food availability remained stable (epiphytes) or even

76

increased (Piper) along the same gradient. This is specially true for the shrub-frugivores

Carollia spp. that virtually forgo foraging in fragments (no captures during the older

fragmentation period) where their main food resource, Piper, was still abundant (Fig. 5).

According to our personal observations, Piper density was greater on the edges along the

shoreline than farther (ca. >50 m) inside the forest. Indeed, the density of light-demanding

pioneer Piper may have increased within edges since fragmentation. Newly created edges are

characterized by greater light penetration due to (i) increased tree mortality and foliage drop

induced by the physiological stress of moisture and temperature changes (Ferreira and

Laurance 1997, Laurance et al. 1998, Lovejoy et al. 1986) and (ii) increased tree fall gaps

because of greater wind exposure (Laurance 1997). The penetration distance of these

phenomenon, ranging from 45 to 60 m from edges for reduction of canopy-foliage density

and for increase of tree fall gaps (Kapos 1989, Lovejoy et al. 1986, Malcolm 1994; see also

the review by Laurance et al. 2002) roughly fits our estimate of a 50-m large edge effect on

Piper density.

The same conclusions could not be made for S. tildae, because its main food resource,

Solanum (Cosson 1994), was too scarce at Saint-Eugène to be conveniently surveyed like

Piper. Nevertheless, similarly to Piper, many chiropterochorous Solanum are light-

demanding pioneer plants, so their density is expected to increase along edges as well.

Furthermore, S. tildae also frequently feeds on epiphytes and Piper (Charles-Dominique and

Cockle 2001, Cockle 1997, Cosson 1994, Delaval et al. 2005), making our food availability

survey appropriate for it as well.

Further decomposing fragmentation-sensitivity: the role of foraging strategies

A possible explanation for the fragmentation-sensitivity of understory fruit bats we studied

could lie in an incompatibility of their foraging strategy with the obligation to cross expanses

of matrix devoid of food sources. Basically, the foraging strategy of fruit bats comprises two

main components, namely commuting flights devoted to long, straight movements among

FAs, and search flights corresponding to phases of active food localization. The large fig-

eating Artibeus spp. have a foraging strategy based on frequent long commuting flights (up to

10 km for A. jamaicensis) and only seldom rely on search flights (Morrison 1978ab). The

huge fig crops produced by large Ficus trees (also termed “big-bang” crops) constitute a

highly patchily distributed food resource that is available for several days (Kalko 1998, Kalko

et al. 1996, Wendeln et al. 2000). Thus, for these large home-range bats, narrow habitat

disruptions do not necessarily impose substantial changes in foraging movements. They can

77

probably easily exploit several fragments within their HR. Not surprisingly, mist-net capture

rates indicated that they were less affected by fragmentation during both recent and older

fragmentation periods (Cosson et al. 1999a, Henry, Pons and Cosson unpublished data).

On the contrary, shrub-frugivorous bats like C. perspicillata feed on a spatially more

scattered resource. Piper and Solanum shrubs produce only few fruits at a time, but over

extended periods of time (“steady-state” fruit crops; e.g. Thies and Kalko 2004). This

requires more flights devoted to active food search, but also rarer and/or shorter commuting

flights between FAs (500 to1500 m only; Heithaus and Fleming 1978). By definition, search

flights would be useless within a matrix devoid of food sources, which would explain why

the abundance of shrub-frugivorous bats rapidly declined after fragmentation (Cosson et al.

1999a), though isolation distances from mainland were short (<300 m) relatively to their

movement capacity.

The foraging strategy of the epiphyte-specialist R. pumilio could even be considered as an

extreme search strategy because they use almost exclusively search flights and therefore

exploit a single small FA (Chapter 1, this Chapter). Such a strategy may reflect the well-

scattered distribution of epiphytes within the forest (although they tend to be more abundant

in humid zones bordering creeks, Cockle 1997). Radio-tracking surveys indicate that they do

not perform straightforward flights between successively visited hanging locations, but rather

follow long and sinuous flight trajectories along which they may visit many plants to search

for food (Chapter 1). A single flight lasts 8.7±5.3 min and may allow bats to investigate a

non-negligible portion of their FA at a time within continuous forests. Most of the surveyed

fragments (0.8 to 7.5 ha) are smaller than the size of their FA (3.5 to 14.1 ha, Table 1). This

might force individuals to split their FA into smaller ones distributed over two or several

contiguous fragments, resulting in regular disruptions of search flights and thus in lower

foraging efficiency.

Alternatively but not mutually exclusive, fig-eating bats are thought to separate commuting

and search flights while understory frugivores may use a mixed strategy consisting of

searching for food while commuting (Fleming et al. 1977). Nevertheless, the latter strategy

would be inefficient as well over expanses of matrix devoid of food.

Our data do not contradict the hypothesis that fragmentation-sensitivity results from an

incompatibility between foraging strategy and habitat connectivity loss. However, it is still

difficult to discriminate between this hypothesis and various alternative explanations. For

instance, it was suggested that some species may be reluctant to fly over the open matrix

because of predation risks. Accordingly, the bat falcon Falco rufigularis was often observed

78

flying or perching within the open matrix. Yet, many studies report that Carollia spp. and

Sturnira spp. cross or forage in open habitats such as man-made clearings or savannahs close

to forest edges (Bernard and Fenton 2002, 2003, Delaval et al. 2005, Simmons and Voss

1998). The lack of landmarks (e.g., trees) for sensory orientation may also be suggested as a

potential limit to the penetration of bats inside an open matrix. However, the matrix in Saint-

Eugène is covered by tall snags that would constitute valuable cues for orientation toward any

fragment.

Fragmentation age and sustainability of populations

Our data suggest that the abundance of R. pumilio was rapidly stabilized after the initial drop

in abundance following fragmentation, at least partly as a result of local reproductive

recruitment. First, in contrast to our hypothesis, the effect of fragmentation period (2-4 yrs vs.

9-11 yrs) remained non-significant. Second, we found no significant difference between

fragments and mainland capture sites in terms of population structure (adult sex-ratio) and

reproductive activity (proportions of reproductive females and juveniles). Third, the

abundance of R. pumilio was not affected by site remoteness, contrary to the other species.

Thus, a local reproductive recruitment may ensure population equilibrium jointly with

possible dispersal movements from and toward mainland. This does not support the

prediction of Chapter 1 that females of R. pumilio may not achieve reproduction equally well

in fragmented habitats. However, long term population surveys including alternative

reproduction indicators and measurements of physiological condition (e.g., hematocrit,

concentrations of stress hormones, estimates of population turnovers by capture-recapture)

may be required to confidently address conclusions in terms of fitness.

The situation is less clear for shrub-frugivores whose low capture numbers in fragments

preclude any conclusion at the species level. At first glance, they did not maintain self-

sustaining populations among remote fragments because their capture rates decreased and

eventually became nil within all surveyed fragments during the older fragmentation period.

However, this decline may also be related to a concomitant decrease in Piper resource

availability (Fig. 5). Although Piper resources remained positively influenced by the loss of

forest connectivity, it seems likely that a progressive closure of canopy foliage within edges

caused the disappearance of many light-demanding Piper plants during the past 5 years.

Many stems of dead young Piper were found during the botanical surveys of older

fragmentation period.

79

In conclusion, our results suggest that, in contrast to the shrub-frugivorous bats, R. pumilio

can be “resident” within the fragmented area at Saint-Eugène and maintain low-density

populations. The fact that R. pumilio can breed and/or forage within fragments may be partly

permitted by its small area requirements (Table 1). This is at least an ecological trait that

seems to render animal species less vulnerable to fragmentation (Laurance et al 2002).

Assuming that larger bats have larger HRs (e.g. Fenton 1997), one can expect that the 8-9 g

R. pumilio has the smallest HR of the four target species (body mass=12, 17 and 23 g for C.

brevicauda, C. perspicillata and S. tildae respectively; Charles-Dominique et al. 2001).

Although methodological differences preclude direct comparisons, it is likely that individuals

of C. perspicillata have larger HRs since females regularly commute between and among day

roosts and FAs located 1 to 2 km apart (Heithaus and Fleming 1978), i.e. 2× to 5× the

maximum flight distances of R. pumilio (Fig. 4). In a fragmented habitat in Brazil, Bernard

and Fenton (2003) estimated HR sizes as large as 155-320 ha for C. perspicillata and 160-

212 ha for C. brevicauda. These HRs covered as many as seven forest fragments of several

tens of ha each.

In addition to HR size, day roost availability might also constrain population sustainability in

fragmented habitats. Rhinophylla pumilio forms small groups (<6-7 individuals) under large

leaves whose blade is cut and modified into a shelter called a “tent”. They can roost in a

variety of large-leaved plants common in forest understories, ranging from fronds of young

palms to leaves of epiphytic Philodendron (Chapter 1; Charles-Dominique 1993, Simmons

and Voss 1998, Zortéa 1995). They often select unmodified leaves as day roosts and can

change roosts every 1-3 days (Table 1), indicating that they may be less limited by roost

availability than other species. By contrast, C. perspicillata selects roosts likely to be less

abundant (e.g., caves or hollow trees) and is considered to be a refuging species (sensu

Hamilton and Watt 1970) with large roosting colonies, high fidelity to day roosts, frequent

returns between foraging bouts, and with abundance of foraging individuals decreasing with

distance from the roost (Heithaus and Fleming 1978). This roosting behavior is compatible

with our general observation that the abundance of shrub-frugivores decreases along the

remoteness gradient RI.

Implications for conservation

As outlined by Leigh et al. (2002), forest remnants isolated in dam reservoirs such as the

fragments at Saint-Eugène allow us to study the consequences of fragmentation per se by

minimizing confounding effects due to the matrix. Water is more neutral as a matrix than

80

pastures, second growth vegetation or urbanized areas because it is truly inhospitable for

many animals and it is normally not accompanied by a variety of human-induced

disturbances such as forest burning, logging and vertebrate hunting (Tabarelli et al. 2004).

Our survey confirmed the initial finding of Cosson et al. (1999a) that the abundance of

understory fruit bats declines sharply in fragmented forests. This stands in contrast to other

studies conducted in forest fragments surrounded by abundant second-growth vegetation.

Therefore, bat surveys in Saint-Eugène indirectly underline the important role of second

growth vegetation in shaping fruit bat communities in disturbed forests. Patches of second

growth vegetation with Piper, Solanum and Vismia provide food for bats and may also

enhance overall habitat connectivity and act as corridors or stepping stones to facilitate bat

movements between forest fragments. Indeed, our data showed that both loss of habitat

connectivity and remoteness may affect bat abundance.

Without a matrix favoring bat movements across landscapes, one would expect bat

communities to be restricted to several small home-ranged species living at low population

densities, and so at greater risk of extinction. This may further exert detrimental

repercussions on the ecological processes bats are involved in, such as pollination (Quesada

et al. 2003) and seed dispersal (Chapter 3; Medellín and Gaona 1999).

REFERENCES

Anthony, E. L. P. 1988. Age determination in bats. Pp. 47–58 in Ecological and behavioral

methods for the study of bats (T. H. Kunz, ed.). Smithsonian Institution Press,

Washington, D. C.

Bernard, E., and M. B. Fenton. 2002. Species diversity of bats (Mammalia: Chiroptera) in

forest fragments, primary forests, and savannas in central Amazonia, Brazil.

Canadian Journal of Zoology 80: 1124–1140.

Bernard, E., and M. B. Fenton. 2003. Bat mobility and roosts in a fragmented landscape in

central Amazonia, Brazil. Biotropica 35: 262–277.

Bierregaard, R. O., Jr., C. Gascon, T. E. Lovejoy, and R. Mesquita (eds.). 2001. Lessons from

amazonia: the ecology and conservation of a fragmented forest. Yale University

Press, New Haven.

81

Bierregaard, R. O., Jr., T. E. Lovejoy, V. Kapos, A. A. dos Santos, and R. W. Hutchings.

1992. The biological dynamics of tropical rainforest fragments. Bioscience 42:

859–866.

Brosset, A., P. Charles-Dominique, A. Cockle, J.-F. Cosson, and D. Masson. 1996. Bat

communities and deforestation in French Guiana. Canadian Journal of Zoology 74:

1974–1982.

Charles-Dominique, P. 1986. Inter-relations between frugivorous vertebrates and pioneer

plants: Cecropia, birds and bats in French Guyana. Pp. 119–135 in Frugivores and

seed dispersal (A. Estrada, and T. H. Fleming, eds.). Dr W. Junk Publishers,

Dordrecht, The Netherlands.

Charles-Dominique, P. 1993. Tent-use by the bat Rhinophylla pumilio (Phyllostomidae:

Carolliinae) in French Guiana. Biotropica 25: 111–116.

Charles-Dominique, P., A. Brosset, and S. Jouard. 2001. Atlas des chauves-souris de Guyane,

Patrimoine naturelle No 49. Muséum National d'Histoire Naturelle, Paris.

Charles-Dominique, P., and A. Cockle. 2001. Frugivory and dispersal by bats. In F. Bongers,

P. Charles-Dominique, P.-M. Forget, and M. Théry (eds.). Nouragues: dynamics

and plant-animal interactions in a Neotropical rainforest, pp. 207–216. Kluwer

academic publishers, Dordrecht, The Netherlands.

Cockle, A. 1997. Modalités de dissémination et d'établissement de lianes de sous-bois

(Cyclanthaceae et Philodendron) en forêt guyanaise. PhD Thesis, Université Paris

VI, France.

Cockle, A. 2001. The dispersal and recruitment of Cyclanthaceae and Philodendron

(Araceae) understorey root-climbing vines. Pp 251–263 in Nouragues: Dynamics

and plant-animal interactions in a Neotropical rainforest (F. Bongers, P. Charles-

Dominique, P.-M. Forget, and M. Théry, eds.). Kluwer Academic Publishers,

Dordrecht, The Netherlands.

Cosson, J.-F. 1994. Dynamique de population et dispersion de la chauve-souris frugivore

Carollia perspicillata en Guyane Française. PhD Thesis, Université Paris XI,

France.

Cosson, J.-F., J.-M. Pons, and D. Masson. 1999a. Effects of forest fragmentation on

frugivorous and nectarivorous bats in French Guiana. Journal of Tropical Ecology

15: 515–534.

Cosson, J.-F., S. Ringuet, J. C. De Massary, O. Claessens, A. Dalecky, J. F. Villiers, L.

Granjon, and J.M. Pons. 1999b. Ecological changes in recent land-bridge islands in

82

French Guiana, with emphasis on vertebrate communities. Biological Conservation,

91: 213–222.

Coulon, A. , J.-F. Cosson, J. M. Angibault, B. Cargnelutti, M. Galan, N. Morellet, E. Petit, S.

Aulagnier, and A. J . M. Hewison. 2004. Landscape connectivity influences gene

flow in a roe deer population inhabiting a fragmented landscape: an individual–

based approach. Molecular Ecology 13: 2841–2850.

Delaval, M., M. Henry, and P. Charles-Dominique. 2005. Interspecific competition and niche

partitioning: example of a Neotropical rainforest bat community. Revue d'Ecologie

(Terre et Vie) 60 : 149–166.

Didham, R. K., and J. H. Lawton. 1999. Edge structure determines the magnitude of changes

in microclimate and vegetation structure in tropical forest fragments. Biotropica 31:

17–30.

Estrada, A., and R. Coates-Estrada. 2001. Bat species richness in live fences and in corridors

of residual rain forest vegetation at Los Tuxtlas, Mexico. Ecography 24: 94–102.

Estrada, A., and R. Coates-Estrada. 2002. Bats in continuous forest, forest fragments and in

an agricultural mosaic habitat-island at Los Tuxtlas, Mexico. Biological

Conservation 103: 237–245.

Estrada, A., R. Coates-Estrada, and D. Merrit, Jr. 1993. Bat species richness and abundance

in tropical rain forest fragments and in agricultural habitats at Los Tuxtlas, Mexico.

Ecography 16: 309–318.

Fenton, M. B. 1997. Science and the conservation of bats. Journal of Mammalogy 78: 1–14.

Ferreira, L. V., and W. F. Laurance. 1997. Effects of forest fragmentation on mortality and

damage of selected trees in central Amazonia. Conservation Biology 11: 797–801.

Fleming, T. H. 1982. Foraging strategies in plant-visiting bats. Pp. 287–386 in Ecology of

bats (T. H. Kunz, ed.). Plenum Press, New York.

Fleming, T. H. 1985. Coexistence of five sympatric Piper (Piperaceae) species in a tropical

dry forest. Ecology 66: 688–700.

Fleming, T. H., E. R. Heithaus, and W. B. Sawyer. 1977. An experimental analysis of the

food location behavior of frugivorous bats. Ecology 58: 619–627.

Forget, P.-M. (Coord.). 2002. Fragmentation de la forêt tropicale humide : le barrage de Petit-

Saut, rivière Sinnamary, Guyane Française. Revue d’Ecologie (Terre et Vie) 57 : 1–

199.

Gascon, C., T. E. Lovejoy, R. O. Bierregaard, J. R. Malcolm, P. C. Stouffer, H. Vasconcelos,

W. F. Laurance, B. Zimmerman, M. Tocher, and S. Borges. 1999. Matrix habitat

83

and species persistence in tropical forest remnants. Biological Conservation 91 :

223–229.

Gorresen, P. M., and M. R. Willig. 2004. Landscape responses of bats to habitat

fragmentation in Atlantic forest of Paraguay. Journal of Mammalogy 85: 688–697.

Granjon, L., J.-F. Cosson, J. Judas, and S. Ringuet. 1996. Influence of tropical rainforest

fragmentation on mammal communities in French Guiana: early trends. Acta

Oecologia 17: 673–684.

Hamilton, W. J., III, and K. E. F. Watt. 1970. Refuging. Annual Review of Ecology and

Systematics 1: 263–286.

Heithaus, E. R., and T. H. Fleming. 1978. Foraging movements of a frugivorous bat, Carollia

perspicillata (Phyllostomatidae). Ecological Monographs 48: 127–143.

Helversen, O., von, and Y. Winter. 2003. Glossophagine bats and their flowers: costs and

benefits for plants and pollination. Pp 346–397 in Bat Ecology (T. H. Kunz, and M.

B. Fenton, eds.). The University of Chicago Press.

Hewison, A. J. M., J. P. Vincent, J. Joachim, J. M. Angibault, B. Cargnelutti, C. Cibien.

2001. The effects of woodland fragmentation and human activity on roe deer

distribution in agricultural landscapes. Canadian Journal of Zoology 79: 679–689.

Johnson, J. B., and K. S. Omland. 2004. Model selection in ecology and evolution. Trends in

Ecology and Evolution 19: 101–108.

Kalko, E. K. V. 1998. Organisation and diversity of tropical bat communities through space

and time. Zoology 101: 281–297.

Kalko, E. K. V., E. A. Herre, and C.O. Handley, Jr. 1996. The relation of fig fruit characters

to fruit-eating bats in the New and Old World tropics. Journal of Biogeography 23:

593–607.

Kapos, V. 1989. Effects of isolation on the water status of forest patches in the Brazilian

Amazon. Journal of Tropical Ecology 5: 173–185.

Laurance, W. F., and R. O. Bierregaard, Jr. (eds.). 1997. Tropical forest remnants: ecology,

management, and conservation of fragmented communities. The University of

Chicago Press.

Laurance, W. F., L. V. Ferreira, J. M. Rankin-de Merona, and S. G. Laurance. 1998. Rain

forest fragmentation and the dynamics of Amazonian tree communities. Ecology

79: 2032–2040.

Laurance, W. F., T. E. Lovejoy, H. L. Vasconcelos, E. M. Bruna, R. K. Didham, P. C.

Stouffer, C. Gascon, R. O. Bierregaard, Jr., S. G. Laurance, and E. Sampiao. 2002.

84

Ecosystem decay of Amazonian forest fragments: a 22-year investigation.

Conservation Biology 16:605–618.

Legendre, P., and L. Legendre. 1998. Numerical Ecology, Elsevier, Amsterdam.

Leigh, E. J., J.-F. Cosson, J.-M. Pons, and P.-M. Forget. 2002. En quoi l’étude des îlots

forestiers permet-elle de mieux connaître le fonctionnement de la forêt tropicale ?

Revue d’Ecologie (Terre et Vie) 57 : 181–194.

Lovejoy, T. E., R. O. Bierregaard, Jr., A. B. Rylands, J. R. Malcolm, C. E. Quintela, L. H.

Harper, K. S. Brown, Jr., A. H. Powel, G. V. N. Powell, H. O. Schubart, and M. B.

Hays. 1986. Edge and other effects of isolation on Amazon forest fragments. Pp

257–285 in Conservation biology: the science of scarcity and diversity (M.E. Soulé,

ed.). Sinauer, Sunderland, Massachusetts.

MacArthur, R. H., and E. O. Wilson. 1967. The theory of island biogeography. Princeton

University Press, Princeton, New Jersey, USA.

McCullagh, P., and J. A. Nelder. 1989. Generalized Linear Models, 2nd edition. Chapman

and Hall, New York, USA.

Malcolm, J. R. 1994. Edge effects in central Amazonian forest fragments. Ecology 75: 2438–

2445.

Marinho-Filho, J. S. 1991. The coexistence of two frugivorous bat species and the phenology

of their food plants in Brazil. Journal of Tropical Ecology 7: 59–67

Medellín, R. A., and O. Gaona. 1999. Seed dispersal by bats and birds in forest and disturbed

habitats of Chiapas, México. Biotropica 31: 478–486.

Meyer, C. F. J., M. Weinbeer, and E. K. V. Kalko. 2005. Home range size and spacing

patterns of Macrophyllum macrophyllum (Phyllostomidae) foraging over water.

Journal of Mammalogy 86: 587–598.

Morrison, D. W. 1978a. Foraging ecology and energetics of the frugivorous bat Artibeus

jamaicensis. Ecology 59: 716–723.

Morrison, D. W. 1978b. Influence of habitat on the foraging distances of the fruit bat,

Artibeus jamaicensis. Journal of Mammalogy 59: 622–624.

Payne, R., D. Murray, S. Harding, D. Baird, D. Soutar, and P. Lane. 2003. GenStat® for

Windows(tm) (7th ed.) – Introduction. VSN International.

Pons, J.-M., and J-F. Cosson. 2002. Effect of forest fragmentation on animalivorous bats in

French Guiana. Revue d’Ecologie (Terre et Vie) 57:117–130.

Quesada, M., K. E. Stoner, V. Rosas-Guerrero, C. Palacios-Guevara, and J. A. Lobo. 2003.

Effects of habitat disruption on the activity of nectarivorous bats (Chiroptera:

85

Phyllostomidae) in a dry tropical forest: implications for the reproductive success

of the neotropical tree Ceiba grandiflora. Oecologia 135: 400–406.

Racey, P. A. 1988. Reproductive assessment in bats. Pp 31–45 in Ecological and behavioral

methods for the study of bats (T. H. Kunz, ed.). Smithsonian Institution Press,

Washington D. C.

Schulze, M. D., N. E. Seavy, and F.F. Whitacre. 2000. A comparison of the Phyllostomid bat

assemblages in undisturbed neotopical forest and in forest fragments of a slash-and-

burn farming mosaic in Petén, Guatemala. Biotropica 32: 174–184.

Simmons, N. B., and R. Voss. 1998. The mammals of Paracou, French Guyana: a neotropical

lowland rainforest fauna. Part I. Bats. Bulletin of the American Museum of Natural

History 237: 0–219.

Tabarelli, M., W. Mantovani, and C. A. Peres. 1999. Effects of habitat fragmentation on plant

guild structure in the montane Atlantic forest of southeastern Brazil. Biological

Conservation 91: 119–127.

Tabarelli, M., J. M. Cardoso da Silva, and C. Gascon. 2004. Forest fragmentation, synergisms

and the impoverishment of neotropical forests. Biodiversity and Conservation

13:1419–1425.

Thies, W. 1998. Resource and habitat use in two frugivorous bat species (Phyllostomidae:

Carollia perspicillata and C. castanea) in Panama: mechanisms of coexistence.

PhD thesis, University of Tübingen, Germany.

Thies, W., and E. K. V. Kalko. 2004. Phenology of neotropical pepper plants (Piperaceae)

and their association with their main dispersers, two short-tailed fruit bats, Carollia

perspicillata and C. castanea (Phyllostomidae). Oikos 104:362–376.

Turner, I. M. 1996. Species loss in fragments of tropical rain forest: a review of the evidence.

Journal of Applied Ecology 33: 200–219.

Weinbeer, M., and E. K. V. Kalko. 2004. Morphological characteristics predict alternate

foraging strategy and microhabitat selection in the orange-bellied bat,

Lampronycteris brachiotis. Journal of Mammalogy 85: 1116–1123.

Wendeln, M. C., J. R. Runkle, and E. K. V. Kalko. 2000. Nutritional values of 14 fig species

and bat feeding preferences in Panama. Biotropica 32: 489–501.

Worton, B. J. 1989. Kernel methods for estimating the utilization distribution in home-range

studies. Ecology 70:164–168.

Zar, J. H. 1998. Biostatistical analysis, Prentice Hall, Upper Saddle River.

86

Zortéa, M. 1995. Observations on tent-using in the carolline bat Rhinophylla pumilio in

southeastern Brazil. Chiroptera Neotropical 1: 85–88.

87

APPENDIX

Detailed bat capture data (Effort: capture effort in nb. of station-nights; Rhp: nb. of captured R. pumilio, Shf: nb. of captured shrub-frugivorous bats), landscape descriptors (CI: connectivity index, RI: remoteness index) and mean values of food availability (number of Piper branches, Piper spikes and epiphytes per survey plot) within each site and each fragmentation period.

Period Site Effort Rhp Shf P. branches P. spikes Epiphytes CI RI Recent fragmentation

5 13 2 0 146.25 14.00 - 71.46 1.97 6 12 0 0 160.75 27.50 19.21 94.67 1.17 8 6 3 1 43.00 10.00 16.56 121.43 0.40 9 4 2 0 - - - 68.65 0.43 12 3 1 0 175.25 52.75 - 92.78 0.97 16 7 7 1 81.00 17.50 12.75 129.20 0.34 17 6 8 0 73.75 20.50 15.68 122.69 0.51 19 7 3 0 98.00 7.50 - 117.13 0.47 20 8 0 1 83.50 14.00 - 118.57 0.77 21 6 1 0 - - - 87.98 9.50 22 4 2 0 57.33 15.00 - 53.24 0.63 24 6 1 3 - - - 91.63 9.28 34 4 1 0 - - - 135.20 1.78 53 6 3 1 68.67 4.67 - 81.98 7.43 Cp 11 22 38 97.57 31.57 20.97 199.17 0.03 Oi 21 16 9 48.50 8.75 - 200.00 0.00 Ro 16 28 1 87.00 19.25 - 176.19 0.20 Vi 5 1 0 41.75 3.75 10.40 193.48 0.09

Older fragmentation 5 8 0 0 54.00 6.75 17.75 71.46 1.97 6 6 0 0 14.25 1.75 19.50 94.67 1.17 8 3 3 0 14.00 0.75 10.75 121.43 0.40 9 6 1 0 - - - 68.65 0.43 12 3 0 0 46.50 2.75 11.25 92.78 0.97 16 3 2 0 25.50 6.75 12.25 129.2 0.34 17 3 0 0 31.50 3.25 11.50 122.69 0.51 19 6 4 0 25.50 4.00 14.50 117.13 0.47 20 8 3 0 23.25 1.50 20.75 118.57 0.77 21 8 0 0 - - - 87.98 9.50 22 4 0 0 32.50 7.00 12.00 53.24 0.63 24 3 0 0 - - - 91.63 9.28 34 3 2 0 - - - 135.2 1.78 53 3 1 0 30.50 1.50 9.75 81.98 7.43 Cp 17 31 50 47.13 10.50 17.13 199.17 0.03 Oi 13 3 2 7.75 4.00 6.50 200.00 0.00 Ro 9 3 0 26.00 4.50 13.25 176.19 0.20 Vi 4 4 2 34.50 2.75 21.75 193.48 0.09

Chapitre 3

Consequences of an experimental disturbance of bat activity on the seed rain pattern of some

keystone bat plants in a Neotropical rain forest.

Rameau de Solanum sp. enLycianthes pauciflora, Sol

m

2 m

88

fruit et 3 exemples de graines de Solanaceae (de g. à d. : anum coriaceum, S. rugosum ; dessin S. Jouard).

89

CONSEQUENCES OF AN EXPERIMENTAL DISTURBANCE OF BAT ACTIVITY ON THE SEED RAIN PATTERN OF SOME KEYSTONE BAT

PLANTS IN A NEOTROPICAL RAIN FOREST.

ABSTRACT

Seed dispersal by animals is a form of mutualism that plays a fundamental role in the

functioning of tropical ecosystems. There is consensus in the literature that fruit bats rank

among the most efficient seed dispersers found in Neotropical forests. However, bat

communities are known to be severely affected by the fragmentation of primary rain forest,

which is therefore likely to indirectly threaten regeneration processes of the plants they

disperse. In this study, we undertook an experimental disturbance of bat activity in a parcel of

primary rain forest (Réserve Naturelle des Nouragues, French Guiana) to determine how the

seed rain pattern would be affected in case of a reduction in bat activity. The seed rain pattern

was characterized at the community level by species diversity, and at the species level by

fundamental seed limitation, i.e. a measurement of the failure of seeds to reach all suitable

microsites for germination. A total of 50063 small endozoochorous seeds belonging to 39

species or group of species was reported from 98 1-m2 traps during 120 days. The seed rain

profile was dominated by the epiphytic Araceae and Cyclanthaceae and tree species

belonging to Cecropia and Ficus, whereas seeds from bat shrubs and treelets (Solanum,

Piper, Vismia species) were markedly seed limited. Conformingly to our hypotheses, the

experimental bat disturbance provoked a significant decrease of species diversity (-30% to -

75%). This was associated with a substantial increase of fundamental seed limitation

generalized among all of the commonest plants consumed by bats. Fundamental seed

limitation was mostly explained by restricted seed numbers (seed source limitation) rather

than lower dispersal uniformity (seed dispersal limitation) during the bat disturbance. We

conclude that bat plants with low seed productivity are more prone to dispersal failure in

fragmented areas. Conversely, bats appear as efficient dispersers, ensuring spatial uniformity

of seed dissemination in spite of disturbances affecting their abundance.

90

INTRODUCTION

Seed dispersal provided by animals is widely recognized as a crucial mechanism in tropical

ecosystems (e.g. Gautier-Hion et al. 1985, Janson 1983, Stiles 1992, Van der Pijl 1969,

Willson 1992). According to the escape hypothesis (Howe and Smallwood 1982), dispersal

allows seeds and seedlings to escape from the locally intense intraspecific competition as

well as the increased action of predators, pathogens and herbivores around parent plants.

Inasmuch as seed and seedling mortality decrease with distance from parent plants (Connell

1971, Janzen 1970), an efficient seed dispersal system is required to ensure successful

recruitment. Beyond the avoidance of distance-dependant mortality, seed dispersal may

provide plants with a variety of colonization and establishment opportunities in distant or

remote habitats. This statement, known as the colonization hypothesis (Howe and Smallwood

1982), underlines the role of random and unpredictable stochastic events in seedling survival

and establishment. Escape and colonization mediated by seed dispersal act in concert to

promote plant diversity and gene flows across landscapes.

The understanding of these plant-disperser interactions is important because their integrity is

threatened by the worldwide disturbances of tropical forests. In particular, forest disruption

and fragmentation have a negative impact on the diversity of plant and animal communities

(Bierregaard et al. 2001, Gascon et al. 2002, Laurance and Bierregaard 1997, Turner 1996).

When dispersers are affected by disturbances and experience abundance and diversity

depletion, one might expect detrimental repercussions on seed rain diversity and seed

dispersal efficiency (i.e. the probability that seeds reach a suitable place for germination). We

examined herein this hypothesis by focusing on some Neotropical plant species, from seven

families, thought to be dispersed by bats according to the current literature on bat-plant

interactions. Indeed, Neotropical fruit bats, whose important role as seed dispersers has been

clearly established, are particularly affected by forest disruption and fragmentation (Cosson et

al. 1999, Estrada and Coates-Estrada 2001, Estrada et al. 1993, Fenton et al. 1992, Schulze et

al. 2000).

The most intuitive way to test the hypothesis that reduction in bat activity affects seed rain

would be to describe spatial variations of natural seed rain patterns across fragmented and

non-fragmented areas. However, in our point of view, this method has limitations and does

not allow the isolation of the effects of fruit bat disturbance per se. A restricted seed rain in

forest fragments might also result from (i) the reduced size of parent plant populations in the

91

fragmented area due to habitat loss, or from (ii) a concomitant decrease of the activity of

other frugivores that feed on the same plants. Alternatively, we experimentally disturbed bat

activity in a primary rain forest plot to simulate a fragmentation context without modifying

either the local diversity and abundance of available seeds nor the abundance of alternative

frugivores. This approach stabilizes the potential effect of these two confounding factors and

thereby isolates the impact of bat activity per se on the seed rain pattern. Similar experiments

were also used to assess the efficiency of some seed dispersers (Bender et al. 1984).

As study models, we utilized small seeds dispersed endozoochorously by bats for two

reasons. First, these seeds are ingested and retained in the gut of dispersers and may travel

hundreds of meters before being released through defecation, thereby increasing the chance

of long range dispersal and colonization events. This is especially relevant in the context of

forest fragmentation where inter-fragment seed flows are thought to be important components

of diversity restoration. Second, small endozoochorous seeds are more abundant than larger

ones, and thus easier to survey. The seed rain can be surveyed by the use of seed traps. This

method has often been employed to study dispersal patterns of large and conspicuous seeds

(e.g. Dalling et al. 2002, De Steven and Wright 2002, Harms et al. 2000, Nathan et al. 2000,

2001, Wada and Ribbens 1997) but has also proved to be a suitable tool for assessing the seed

rain of small seeds dispersed by birds and bats. Thomas (1982) and Foresta et al. (1984) were

among the first to use seed traps to quantify bat-generated seed rain. Later studies (Gorchov

et al. 1993, Medellín and Gaona 1999, Thomas et al. 1988) found that bats disperse more

seeds than do birds in open habitats (e.g. fields). In this study, seed traps were used to

describe the seed rain pattern at the community level (seed rain diversity) and at the species

level (seed limitation). Seed limitation refers to the failure of seeds to reach all microsites

suitable for germination and establishment (Dalling et al. 2002, Eriksson and Ehrlen 1992,

Muller-Landau et al. 2002, Turnbull et al. 2000).

Our objectives were (i) to provide a brief description of the seed rain pattern of the target

plant species, (ii) to test the hypothesis that a decreased bat activity will reduce seed rain

diversity and increase seed limitation and (iii) to determine whether increased seed limitation

results from restricted seed numbers (source limitation) or restricted seed dispersal uniformity

(dispersal limitation). We will discuss the relevance of results in the context of forest

fragmentation.

METHODS

Study area

The seed rain sampling protocol was carried out at the Nouragues research station (4°50’ N,

52°42’ W) in the middle of the Réserve Naturelle des Nouragues, northern French Guiana

(Fig. 1a). This area is covered by continuous primary rain forest. The main forest type is

characterized by 30- to 35-m high trees dominated by the Caesalpinaceae, Sapotaceae and

Lecythidaceae families, and by a fairly open understory (Poncy et al. 2001). Total annual

rainfall ranges from 2500 to 3200 mm with a marked dry season from August to November.

A description of the local bat community is provided by Brosset et al. (2001) and Delaval et

al. (2005). Sampling sites were located close to a ca. 5-m wide creek in a zone of little

marked relief (Fig. 1b).

Figure 1. Location of the study area in French Guiana (A) and disposition of seed rathe study area (B). Altitude variation between consecutive isoclines is 20m.

400 m

Samplinsites

A

100 km100 km

Brazil

Surin

ame

Cayenne

Kourou

SinnamarySaint-Laurent

du MaroniAtlantic Ocean

Saint-Georges

Camopi

SaülMaripasoula

Régina

Oyapo

ck

Le M

aron

i

Les Nouragues

A B

South merica

92

in sampling sites in

N

g

93

Study species

We identified from current literature on bat-plant interactions at Nouragues and elsewhere in

French Guiana (Charles-Dominique 1986, 1995, Charles-Dominique and Cockle 2001,

Cockle 1997, 2001, Cosson 1994, Delaval 2004, Delaval et al. 2005, Geiselman et al. 2002,

Lobova et al. 2003) seven plant families that constitute keystone resources for bats and other

frugivores in our study area: Cecropia species (Cecropiaceae), epiphytic Cyclanthaceae and

Philodendron species (Araceae), Ficus species (Moraceae), Piper species (Piperaceae),

Solanum species (Solanaceae) and Vismia species (Clusiaceae). A common characteristic to

all of these groups is their year-round availability, though fruiting of some of these species

tend to peak at certain periods of the year (e.g. Piper species, Thies and Kalko 2004).

Cecropia species are Neotropical pioneer trees that play an important role in forest

regeneration in disturbed areas. Recent morphological and anatomical studies (Lobova et al.

2003) revealed that the dispersal units of Cecropia species are not seeds but fruits: the pulp

consumed by bats on Cecropia infructescences is actually derived from the enlarged fleshy

perianth. The more appropriate term “diaspore” will be used within the following text for this

species. Cecropia trees are prolific and produce large infructecences that may bear hundreds

of thousands diaspores. Cecropia are frequently dispersed by bats throughout the Neotropics

(Lobova et al. 2003). At Nouragues, C. obtusa constituted the bulk of the diet of three of the

commonest stenodermatine bats (Artibeus jamaicensis, A. lituratus, A. obscurus; Delaval et

al. 2005). The other Cecropia dominating pioneer tree communities of the study area is C.

sciadophylla, which is mostly consumed by birds (Charles-Dominique 1986) and

occasionally by the small fruit bat Rhinophylla pumilio (Lobova et al. 2003). Diaspores are

easily identifiable with their lanceolate-to-oblong shape (2.9 mm length), triangular-to-

elliptic transverse section (0.7-1.3 mm wide), brownish color and rugose surface.

Solanum, Piper and Vismia species are typical chiropterochorous understory or pioneer

shrubs. They can be abundant in secondary forests along edges (roads, open habitats). Fruits

(berries) are mostly consumed by the understory fruit bats Carollia perspicillata, C.

brevicauda (subfamilly Carolliinae), Sturnira lilium and S. tildae (subfamilly

Stenodermatinae). Solanum seeds are flat, orbicular or reniform (2-3 mm diameter) and have

a yellowish color. Piper seeds are generally smaller (<1mm wide) and are characterized by a

geometrical shape (squared or triangular aspect) and a dark color. Vismia species produce

elongated and curved, dark seeds of various size.

Araceae and Cyclanthaceae fruits are seldom consumed by the four above-mentioned fruit

bats and constitute the main diet of Rhinophylla pumilio (Charles-Dominique and Cockle

94

2001, Cockle 1997, 2001, Cosson 1994, Delaval 2004, Delaval et al. 2005). Most of the

chiropterochorous Araceae and Cyclanthaceae are epiphytes whose adventitious roots

develop on trunks between 0 and 10 m above ground level. (e.g. Evodianthus funifer,

Asplundia heteranthera and Thoracocarpus bissectus for Cyclanthaceae; Philodendron

species for Araceae). Their large infructescences produce 5 to 200×103 minute seeds which

are generally <1 mm long and <0.4 mm wide. Seeds of closely related species are often

difficult to discriminate without a constraining germination protocol. In order to classify bat-

dispersed seeds of epiphytes at Nouragues, Cockle (2001) used the following subsets: (i)

Evodianthus funifer and Asplundia heteranthera (with small, flat, yellowish seeds), (ii)

Philodendron species from the subgenus Pteromischum characterized by long seeds, 0.7-1.4

mm length, transverse section 0.2-0.3 mm diameter (Philodendron placidum, P. guianense,

P. duckei) and (iii) Philodendron species with oval seeds, 1.2-1.5 mm length, transverse

section 0.3-0.5 mm diameter (P. grandifolium, P. pedatum, P. squamiferum, P. soderstromii,

P. insigne, P. linnaei, P. billietae). Other species have distinctive seed shape or size that

allow unequivocal identification, e.g. Thoracocarpus bissectus (Cyclanthaceae) with oblong

flat seeds, 2.1-2.4 mm length, and Philodendron deflexum (Araceae) with large-sized seeds,

2.8-3.5 mm length.

The importance of figs (Ficus species, Moraceae) for bats and birds as well as for other

vertebrate taxa (marsupials, primates) has been documented in the literature (Giannini and

Kalko 2004, Kalko et al. 1996, Tello 2003, Terborgh 1986). Species of Ficus are large

canopy trees or stranglers that become free standing or epiphytic shrubs or trees, and produce

huge but short-lived crops. Owing to the asynchrony of ripening events, figs constitute an

unpredictable but year-round available resource. Like Cecropia species, Ficus dispersal units

are actually fruits and not seeds. These are mostly globose and yellowish, and varies in

diameter from 0.5 to 1.3 mm.

Design of the experiment

To study the effects of a reduction in bat activity on the seed rain pattern, we experimentally

disturbed bats by developing a massive mist-netting effort concentrated over a seed rain

sampling site. We then compared the seed rain in the experimental site with that observed on

a nearby control site. The control site allowed us to control for possible phenological

variations in the course of the study. This experiment was repeated twice under identical

conditions in March-May 2003 and 2004. We avoided dry seasons because in the absence of

95

rain, a certain proportion of defecated seeds may remain stuck in the foliage, leading to

possible underestimations of seed numbers.

We decided to work on a relatively short temporal scale with sampling sessions of 30 days

(total duration of the study: 30 days × 2 sampling sessions × 2 experiment repetitions=120

days). This appeared to us as the best compromise between several conceptual and practical

constraints. Longer sampling sessions may encompass phenological variations in seed

production likely to hide the effects of bat disturbance. Shorter sessions would yield restricted

seed numbers and preclude comparative analyses. Earlier pilot studies (M.H., unpublished

data) revealed that the coefficient of variation of mean number of collected seeds per trap

varies little beyond 15-20 days for the commonest species. Furthermore, this time scale is

relevant in the context of competitive exclusion for seedling establishment, an important

stage in the plant recruitment processes (Nathan and Muller-Landau 2000). For instance, it

reasonably fits latency periods separating seed deposition from germination (ca 20-30 days)

and from seedling stage (1.5-2 months) measured on our target epiphyte species in the same

study area (Cockle 1997).

Seed rain sampling

The seed rain was sampled by the mean of 1-m² seed traps. Traps consisted of squared plastic

sheets stretched between four trees and/or aluminum poles ca. 1 m above ground to avoid

disturbance by terrestrial animals (Appendix 1). The sheets were fitted with a central

cylinder-shaped filter (4 cm diameter, 14 cm high, stainless steel wire mesh wrapped in a thin

permeable polyester cloth) for draining water. The filter was fixed transversally to the trap

surface using a mastic resin.

Each sampling site consisted of a squared grid of 7×7=49 contiguous plots of 5×5 m (Fig. 2).

Seed traps were set as close as possible to the centre of each plot (mean error 1.6±0.7 m). All

together, the 98 traps provided an effective sampling surface of 98 m², representing 4% of the

0.245 ha area covered by the two sampling sites (35×35 m=0.1225 ha for each site).

Disturbed and control sites were separated by a narrow buffer zone (35-m wide only) so that

we assume that they share the same plant resource availability and seed sources.

Traps were visited in the middle and at the end of each sampling session (15th and 30th days

respectively). Visitation of traps consisted in collecting the entire trap contents in waterproof

bags (including leaves, branches, dust). All material removed from traps was thoroughly

rinsed above a series of three sieves so as to retain particles of 0.125 to 0.625 mm on one

hand, and 0.625 to 5 mm on the other hand. The resulting humid dust was stored in small

96

opaque plastic bags to maintain the seeds dormant, and later examined at 6 to 16×

magnification to single out tiny seeds. Intact seeds were assigned to taxonomic units (family,

genus, species or morphospecies) on the basis of their external characteristics with the help of

a seed reference collection (Muséum National d’Histoire Naturelle, Dept. Ecologie et Gestion

de la Biodiversité, Brunoy, France). The reference collection includes seeds collected on

fruiting plants or from bird and bat feces in the course of various projects undertaken at the

study site and other places in French Guiana during the past two decades.

Figure 2. Disposition of seed traps within the seed rain sampling sites.

Bat disturbance

A phenomenon known to bat biologists is the sharp decline of capture success when mist-

netting bats over several days at the same capture site. This capture depletion may result from

a combination of mist-net habituation by bats and avoidance of the area. Given the small size

of seed rain sampling sites, we assumed that setting many nets within and around a site will

locally hinder bat flights and make the area costly to cover by flying at the understory level.

Eventually, bat activity will substantially decrease during the experimental session.

Fifteen 12×2.5m nets (mesh 16mm) were used at the same time. Nets were opened during 22

and 23 whole nights (18:30 to 06:30) of the 2003 and 2004 disturbance sessions respectively,

that is to say during 73 to 77% of the total nocturnal time. Nets were checked every 30 min

during the first and the last 2 hrs of the night, and every 1.5 hrs otherwise. They were closed

during periods of abundant rainfall. Every two days, 3-4 nets were displaced to reduce a

possible effect of mist-net habituation by bats.

Captured fruit bats were kept in cloth bags and later identified, ringed with numbered plastic

wing-bands (A. C. HUGUES, England) and released at the research station, 350 m away from

Buffer zone

35 m

0

5

10

15

20

25

30

35

0 5 10 15 20 25 30 350

5

10

15

20

25

30

35

0 5 10 15 20 25 30 35

Disturbed site Control site

Sampling grid and traps

97

the sampling site. Species identification was based on information provided by Charles-

Dominique et al. (2001), Simmons and Voss (1998) and a homemade identification key.

Measures of seed species diversity

Seed species richness was considered herein as a proximal species diversity indicator. To

compare the patterns of seed species diversity across the study sessions, we first built curves

of species richness accumulation as a function of the sampling effort (number of seed traps)

for each site and each sampling session. Curves were smoothed by the mean of 100 random

reorganizations of the trap orders using the software EstimateS (version 5, R.K. Colwell,

URL: http://viceroy.eeb.uconn.edu/estimates). Our initial idea was to fit curves with

asymptotic functions (Clench model or linear dependence model; see Soberón and Llorente

1993) to predict species richness. However, these functions produced unsatisfactory fits.

Alternatively we used the accumulation rate of new species with increasing sampling effort

(i.e. the slope of accumulation curves) in lieu of diversity indicator. The species richness

accumulation curves usually exhibit a first steep section and a second flat section

corresponding to a drastic slowdown of species accumulation. In this analysis, we focused on

the latter quasi-linear section following the bending zone. A General Linear Model (GLM)

was performed to compare the observed slopes before and during the experimental

disturbance.

Measures of seed limitation

To quantify the failure of seeds to reach all suitable microsites (here materialized by traps),

we used the fundamental seed limitation index “FL” (Muller-Landau et al. 2002, Nathan and

Muller-Landau 2000) that measures the proportion of traps at which seeds do not arrive:

trapsof nb totalseedsby reached trapsof nb 1 FL −=

FL is a combination of the limitation of seed numbers (or source limitation “SL”) and the

limitation of seed dispersal (or dispersal limitation “DL”). SL refers to seed limitation due

solely to insufficient seed numbers, whereas DL refers to seed limitation due to non-uniform

distribution of seeds among traps (Muller-Landau et al. 2002).

SL was calculated as the probability of no dispersal event in a given trap, i.e. the probability

that no seeds reach a given trap during the sampling session. It is assumed that the occurrence

98

of a seed in a given trap is a rare and random event and that consequently the corresponding

probability of occurrence may follow a Poisson distribution (Muller-Landau et al. 2002).

Source limitation SL is then the Poisson probability of zero event given an expectation of [nb

of collected seeds / nb of traps] events:

= trapsof nbseeds collected of nbexp SL

Finally, DL measures the ratio between the proportion of traps effectively reached by seeds

and the proportion of traps that would be reached by seeds given a perfectly uniform seed

distribution:

SL1seeds received that trapsof proportion 1 DL −−=

The effects of experimental disturbance on seed limitation

To test the hypothesis that the experimental bat disturbance increases seed limitation, we

compared the observed variations of seed limitation values (FL, SL, and DL) in the disturbed

site to the expected variations measured in the control site. For instance, the variation of FL

in a given site is calculated as:

FL variation = FL (during bat disturbance) – FL (before bat disturbance)

The disturbance effect was considered significant when observed variation in disturbed site

significantly differed from expected variation measured in control site. In order to compare

observed and expected seed limitation variations, we produced a series of values for each site

and each species, using a bootstrap procedure. Seed limitation variations were recalculated

for 100 subsets of 20 traps randomly resampled from the initial 49 traps in each site. The

resampling procedure was actually pseudo-random because we ensured that all traps were

resampled the same number of times (±1) at each bootstrap stage. This method permit to

estimate a variance of seed limitation variations, and to perform statistical comparison tests.

As values of seed limitation variation proceed from proportions and are bounded by -1 and 1,

they were beforehand arcsine-transformed. To avoid overabundance of pseudo-replicated

values, we did not use the whole 100 bootstrap values for comparison tests, but only the

99

minimum number of values required to reach a stable coefficient of variation CV (i.e. CV

remaining around a steady plateau when adding more bootstrap values) and to obtain

normally distributed data sets (Kolmogorov-Smirnov test of normality). Therefore, we used

two-ways ANOVAs to compare observed and expected values, and to identify a possible

interaction with the study year (2003 vs. 2004). A non-significant interaction with the study

year would indicate a repeatability of the bat disturbance effects between the two experiment

repetitions. In case of significant interaction, we used Kruskal-Wallis tests in lieu of a

posteriori analysis to check for possible effect inconsistencies (i.e. a significant positive effect

on one year and a significant negative effect the other). Situations with an non-significant

effect in one of the two years were not considered as inconsistencies.

As an alternative approach to the seed limitation measurements, we assessed the effects of bat

disturbance on seed density (number of seeds collected per trap and per sampling session). To

better handle seed density values that can encompass two orders of magnitude, we applied the

transformation log10(value +1). Similarly to the previous analyses, we calculated the

variations of seed density as the difference between seed density during and before the bat

disturbance. This method assigns a value of seed density variation to each trap, and it was not

necessary to resort to bootstrapping to obtain estimates of variance. Furthermore, values of

seed density variation are characterized by a fairly symmetrical frequency distribution that do

not differ from normality for abundant species. This contrasts with the usually highly skewed

distribution of the initial seed density data (with a majority of nil or low seed numbers).

The significance level α=0.05 was used for all statistical tests. As we performed series of

similar comparison tests on several seed species, we adjusted the probability values p using

the sequential Bonferroni method for each battery of tests. All tests were performed with the

software Systat 9.0.

Seed species grouping

The statistical method described above requires to restrict analyses to the commonest species

whose seed limitation and density values are normally-distributed. In order to include rarer

plant species, we grouped some of them into groups of closely related species or groups of

species producing fruits harvested and consumed in a similar way by frugivores. Therefore,

analyses were performed on “functional units” representing either individual seed species or

groups of rarer seed species.

100

RESULTS

General seed rain description

During the study, some traps collapsed on several occasions due to branch falls, which

reduced the number of samples from 392 (49 traps × 2 sites × 2 study sessions × 2experiment

repetitions) to 381. A total of 50063 seeds <5 mm were collected and classified into 39

taxonomic units (Table 1) belonging to the seven target plant families. Overall seed rain

reached 44×103 seeds per ha and per day (s.ha-1.d-1) over the study period, but ranged from 19

to 73×103 s.ha-1.d-1, depending on the study session. The seed rain was largely dominated by

Cyclanthaceae and Araceae epiphytes (24 and 12×103 s.ha-1.d-1 respectively) followed by

Cecropia species and to a lesser extent Ficus species (6 and 1.5×103 s.ha-1.d-1 respectively).

Seeds of the typical shrub and treelet bat plants (Solanum, Piper and Vismia species)

remained rare or uncommon in samples and totaled <0.3×103 s.ha-1.d-1.

Not surprisingly, the fundamental seed limitation FL was tightly correlated to the number of

collected seeds (Fig. 3). The most abundant seed species were also the most frequently

collected ones. On one hand, Solanum, Piper and Vismia species typically displayed fairly

high FL values (>0.90) while those of some Cyclanthaceae and Araceae hardly got beyond

0.10-0.30. Seeds of Evodianthus / Asplundia even reached 100% of traps (FL=0) during the

first sampling session of 2004.

Figure 3. Mean FL per sampling session and per site plotted against the total nb of collected seeds for the 39 taxonomic units taken independently (A) and grouped by family (B). In both case, correlations are significant (pearson=-0.870; p>0.001 and pearson=-0.972; p>0.001 respectively).

Fund

amen

tal s

eed

limita

tion

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4

A 5

Piperaceae (10)

Solanaceae (11)

Vismia sp (1)

Ficus species (8)

Epiphytic Araceae (5)

Cecropia species (2)

Epiphytic Cyclanth. (2)

0

0.2

0.4

0.6

0.8

1

0 1 2 3 4 5

B

log (total nb of collected seeds) log(total nb of collected seeds)

101

Table 1 : Descriptive results of seed rain sampling sessions at Nouragues, French Guiana. For each taxonomic unit, the number of collected seeds is given with their occurrence frequency within seed traps in parentheses (proportion of traps reached by seeds). Numbers facing species indicate those reported to be dispersed by bats in our study area (1=Cockle 1997; 2=Delaval 2004). Nomenclature follows Boggan et al. (1997).

2003 2004 disturbed control disturbed control before during before during before during before during

Nb of samples 49 48 46 45 48 48 48 49 Araceae Philodendron deflexum Poepp. ex Schott 2 1 (0.02) 1 (0.02) 45 (0.02) 2 (0.02) 2 (0.04) 2 (0.02) P. fragrantissimum (Hook.) Kunth 4 (0.02) 39 (0.33) 249 (0.09) 1817 (0.67) 2 (0.02) 803 (0.23) 1033 (0.20)P. melinonii Brongn. ex Regel 1, 2 15 (0.19) 3 (0.06) 3 (0.04) Philodendron subgenus Pteromischum 1, 2 950 (0.88) 423 (0.27) 1175 (0.65) 1009 (0.38) 2449 (0.94) 1159 (0.71) 1404 (0.73) 452 (0.63)Oval-seeded Philodendron species 1, 2 161 (0.43) 141 (0.60) 19 (0.28) 134 (0.56) 30 (0.31) 54 (0.21) 4 (0.08) 12 (0.14) Total 1116 (0.94) 603 (0.75) 1444 (0.78) 3005 (0.89) 2498 (1) 1215 (0.77) 2214 (0.77) 1502 (0.76)Cecropiaceae Cecropia obtusa Trécul 1, 2 1686 (0.61) 29 (0.31) 509 (0.57) 285 (0.51) 607 (0.75) 92 (0.54) 594 (0.92) 779 (0.94)C. sciadophylla Mart. 1, 2 272 (0.51) 95 (0.25) 20 (0.26) 4 (0.07) 932 (0.52) 563 (0.35) 31 (0.31) 21 (0.24) Total 1958 (0.80) 124 (0.50) 529 (0.57) 289 (0.53) 1539 (0.85) 655 (0.69) 625 (0.94) 800 (0.94)Clusiaceae: Vismia sp1 1 (0.02) 5 (0.04) 1 (0.02) 1 (0.02) Cyclanthaceae Evodianthus funifer (Poit.) Lindm. + Asplundia heteranthera Harling 1, 2 674 (0.73) 2108 (0.71) 320 (0.50) 2624 (0.80) 5593 (1) 4393 (0.96) 7279 (1) 1881 (0.96)

Thoracocarpus bissectus (Vell.) Harling 1, 2 282 (0.45) 86 (0.46) 39 (0.26) 369 (0.49) 232 (0.71) 1837 (0.63) 53 (0.29) 322 (0.43)Total 956 (0.82) 2194 (0.81) 359 (0.63) 2993 (0.82) 5825 (1) 6230 (0.98) 7332 (1) 2203 (1) Moraceae Ficus amazonica (Miq.) Miq. 2 1 (0.02) 6 (0.07) 1 (0.02) F. guianensis Desv. ex Hamilton 99 (0.57) 186 (0.73) 128 (0.74) 278 (0.89) 69 (0.60) 16 (0.25) 300 (0.92) 49 (0.43) F. insipida Willd. 2 3 (0.04) 3 (0.04) 11 (0.04) 1 (0.02) 2 (0.02) F. nymphaeifolia Mill. 2 6 (0.08) 5 (0.06) 24 (0.09) 25 (0.23) 20 (0.21) 24 (0.27) 39 (0.41) F. trigona L. f. 36 (0.08) 12 (0.09) 53 (0.42) 11 (0.15) 34 (0.40) 17 (0.29) F. sp1 23 (0.02) 29 (0.02) 3 (0.06) F. sp2 1 (0.02) 2 (0.02) F. sp3 2 (0.04) 4 (0.02) 2 (0.04) Total 109 (0.61) 230 (0.79) 158 (0.76) 358 (0.91) 153 (0.90) 50 (0.46) 358 (0.96) 109 (0.69)Piperaceae Piper aduncum L. 2 1 (0.02) 56 (0.04) 3 (0.04) P. hostmannianum (Miq.) C. DC. 2 1 (0.02) 2 (0.04) P. trichoneuron (Miq.) C. DC. 1 (0.02) P. sp1 1 (0.02) 3 (0.02) 3 (0.02) P. sp4 1 (0.02) P. sp5 2 (0.02) P. sp6 7 (0.02) 2 (0.02) P. sp8 1 (0.02) P. sp9 1 (0.02) P. sp11 1 (0.02) Total 11 (0.08) 2 (0.02) 58 (0.09) 8 (0.04) 1 (0.02) 5 (0.04) 1 (0.02) Solanaceae Lycianthes pauciflora (Vahl) Bitter 1 (0.02) S. asperum L.C. Rich. 1, 2 1 (0.02) S. coriaceum Dunal 29 (0.18) 2 (0.04) 3 (0.04) 21 (0.16) 57 (0.23) 17 (0.15) 29 (0.19) 47 (0.22) S. leucocarpon Dunal 6 (0.02) S. rugosum Dunal 6 (0.02) 1 (0.02) 1 (0.02) S. sarmentosum L. 2 1 (0.02) 1 (0.02) S. semotum M. Nee 2 (0.04) 1 (0.02)

102

Table 1: Continued.

S. sp1 3 (0.07) S. sp6 1 (0.02) 3 (0.02) S. sp7 1 (0.02) S. sp8 2 (0.02) Total 33 (0.24) 3 (0.06) 15 (0.09) 29 (0.22) 58 (0.25) 18 (0.17) 30 (0.21) 50 (0.24)

Bat disturbance

The experimental bat disturbance led to the capture of 129 individual fruit bats belonging to

10 species (Table 2). Fruit bat assemblages differed from one year to the other. In 2003,

84.5% of them belonged to the understory fruit bat guild (i.e. mainly feeding on fruits

produced by understory plants), the other 15.5% being indexed as canopy fruit bats (i.e.

mainly feeding on fruits produced by canopy trees). The ratio was reversed in 2004, with

21.1% and 78.9% of understory and canopy fruit bats respectively, may be due to fruiting

Ficus trees attracting many Artibeus in the vicinity (Morrison 1978a). The respective

proportion of the two guilds significantly differed between the two years (Khi² test, p<0.001).

Fruit bat captures were dominated by the understory species Rhinophylla pumilio (46.5%) in

2003 and by the canopy species Artibeus jamaicensis (50,7%) in 2004. Furthermore, >50% of

the fruit bats were captured during the first disturbance week in 2003 whereas captures were

homogenously staggered along the first three weeks in 2004. Nevertheless, in both cases,

capture rate was nearly nil during the last 10 days. Only three fruit bats were recaptured once

(one R. pumilio in 2003, one A. jamaicensis in 2004 and one R. pumilio captured in both

years).

Table 2: List of captured fruit bats during the experimental mist-netting disturbance. Capture efforts are 22 and 23 whole nights with 15 nets for 2003 and 2004 respectively.

Species Guilda 2003 2004 Carolliinae subfamily

Carollia brevicauda U 5 1 Carollia perspicillata U 11 1 Rhinophylla pumilio U 27 10

Stenodermatinae subfamily Artibeus gnomus C 2 Artibeus jamaicensis C 3 36 Artibeus lituratus C 3 4 Artibeus obscurus C 3 12 Ectophylla macconnelli C 1 Sturnira tildae U 6 3 Uroderma bilobatum C 1

Total understory fruit bats U 49 15 Total canopy fruit bats C 9 56

a : U=understory fruit bats; C=canopy fruit bats.

103

The effect of bat disturbance on seed diversity and seed limitation The seed species richness (number of collected taxonomic units) per site and per sampling

session ranged from 13 to 25. Despite their relative scarcity, Piperaceae and Solanaceae

accounted together for more than half of the seed species richness (26% and 28%

respectively). Conformingly to our predictions, the richness decreased in the disturbed site

during bat disturbance (-33% and -28% for 2003 and 2004 respectively), while it remained

similar or increased in the control site (+32% and +12% respectively). Richness decrease in

the disturbed site compared to the control site is mainly explained by the disappearance of

some rare seed species (mostly from Piperaceae, Solanaceae and Araceae families).

The analysis of the species accumulation curves depicted the same trends. All species

richness accumulation curves clearly encompassed a bending zone between a first steep

section and a second flatter section (Fig. 4). However, the flatter section following the

bending zone usually did not form a plateau but followed a marked slope, precluding

satisfactory modelling with asymptotic functions (see methods). The slope of this second

curve section was highly correlated with observed species richness (Pearson correlation

coefficient=0.953; p<0.001) and was treated here as a diversity indicator. For each of the two

experiment repetitions, the slope decreased markedly during the capture of bats in the

disturbed site (-77% and -72% for 2003 and 2004 respectively), while it increased in the

control site (+24% and +73% for 2003 and 2004 respectively) (Fig. 4). For any of the four

slope comparison tests (disturbed vs. control site comparisons, before and during disturbance

in 2003 and 2004), the observed difference was significant (GLM: n=40; df=1; F=73.10 to

538.67; R²=0.658 to 0.934; p<0.001 in any case).

We also found significant seed rain modifications at the level of functional units. The

bootstrap method used for the calculation of seed limitation permitted to satisfy the normality

requirements, with 15 random resamplings, for nine functional units (Table 3). These nine

focus functional units mostly reached traps in small amounts (<10 seeds per trap and per

session) and are characterized by a mean FL ranging from 0.16 to 0.79 (Table 3). Detailed

seed limitation values per site and per session are synthesized in Appendix 2. All these

functional units (except Ficus guianensis) exhibited a significant FL increase in the disturbed

site compared to expected variations measured in the control site (Table 4, Fig. 5a). This

increase averaged 43% of the mean FL values presented in Table 3. Most of the studied

species did not respond equally to bat disturbance among the two experiment repetitions, as

revealed by the significant interactions with the study year (Table 4). However, we found no

effect inconsistencies from one year to the other (a significant positive effect on one year and

negative the other).

Figure 4. Curves of seed species richnseed traps) in disturbed and contrdisturbance and for the two experimereorganizations of the trap order. Numzone.

0

5

10

15

20

25

0 10 20

DisturbedControl

20

25

0

5

10

15

20

0 10 20

Before bat distu

Before bat distu

Nb of seed

Spec

ies r

ichn

ess a

ccum

ulta

tion

Spec

ies r

ichn

ess a

ccum

ulta

tion

0.218

ess aol sitnt re

ber

30

15

30

rban

rban

trap

0.197

ccumulation as a function of seed raes, before (left) and during (righpetitions (2003-2004). Curves were ss refer to the slope of the linear secti

40 500

5

10

0 10 20

40 500

5

10

15

20

0 10 20

5

2

ce, 2003 During bat

ce, 2004 During bat

s Nb of

0.245

in samplint) the expmoothed on followi

30

30

disturba

disturba

seed trap

0.062

40 50

5

nce, 2003

0.14

0.07

0.12

bn

n

0.033

104

g effort (nb of erimental bat y 100 random g the bending

40 50

ce, 2004

s

105

Table 3: Mean seed limitation values per session (FL, SL and DL, see Appendix 2), and frequency distribution of the size of dispersal events (nb of seeds collected per trap in one 30-days sampling session) of the nine studied functional units (sorted by order of increasing FL). Frequencies were calculated across five categories of increasing magnitude order, the modal categories being indicated in bold. Nil events were excluded from calculations but their mean occurrence frequency are indicated by the mean FL.

a Abbreviations of the studied Functional Units: EVAS=the couple Evodianthus funifer and Asplundia heteranthera; PHPT=Philodendron species of subgenus Pteromischum; CEOB=Cecropia obtusa; FIGU=Ficus guianensis; THBI=Thoracocarpus bissectus; PHOV=oval-seeded Philodendron species; CESC=Cecropia sciadophylla; SPV=typical bat shrubs and treelets (Piper, Solanum and Vismia species); FIBAT=the three Ficus species known to be consumed by bats at Nouragues (F. amazonica, F. insipida , F. nymphaeifolia; Delaval et al 2004).

Table 4: Results of the two-ways ANOVAs comparing the variations of seed limitation measurements between disturbed and control sites. Results are given for the effect of disturbance and its interaction with study year; “ns” indicates non-significant effects; signs + and – refer to the direction of the significant effects. For year interactions (indicating different effect sizes between the two years), signs are replaced by the year of the strongest disturbance effect, or by “?” in case of inter-annual inconsistencies (opposed significant effects from one year to the other). In any case, n=60 and df=1; p values were adjusted based on the sequential Bonferroni method for 9 tests. See Fig. 5 for results of a posteriori tests. Abbreviations follow Table 3.

Fondamental limitation FL Source limitation

SL Dispersal limitation DL

Functional Units F p effect F p effect F p effect EVAS disturbance 524.88 <0.001 + 0.97 ns 506.60 <0.001 + disturbance × year 391.42 <0.001 2003 2.70 ns 374.74 <0.001 2003 PHPT disturbance 636.19 <0.001 + 66.02 <0.001 + 183.35 <0.001 + disturbance × year 155.98 <0.001 2003 70.56 <0.001 2003 12.58 0.004 2003 CEOB disturbance 682.15 <0.001 + 1052.82 <0.001 + 5.31 ns disturbance × year 1.75 ns 171.26 <0.001 2003 39.07 <0.001 2004 FIGU disturbance 53.45 <0.001 −−−− 27.94 <0.001 −−−− 15.83 <0.001 −−−− disturbance × year 19.30 <0.001 2004 121.90 <0.001 2003 213.76 <0.001 ? THBI disturbance 492.05 <0.001 + 344.70 <0.001 + 0.57 ns disturbance × year 1.39 ns 14 .47 0.004 2003 3.65 ns PHOV disturbance 185.04 <0.001 + 171.62 <0.001 + 17.01 <0.001 −−−− disturbance × year 53.06 <0.001 2004 135.97 <0.001 2003 89.30 <0.001 ? CESC disturbance 103.10 <0.001 + 2.62 ns 6.53 0.044 + disturbance × year 2.51 ns 4.59 ns 14.86 <0.001 2004 SPV disturbance 877.80 <0.001 + 296.38 <0.001 + 9.52 0.010 −−−− disturbance × year 82.71 <0.001 2003 0.26 ns 0.44 ns FIBAT disturbance 380. 14 <0.001 + 203.62 <0.001 + 1.30 ns disturbance × year 122.54 <0.001 2004 13.48 <0.001 2003 5.64 ns

Seed limitation measurements Size of dispersal events FU a FL SL DL 1 2 to 10 11 to 100 101 to 1000 >1000

EVAS 0.16 0.005 0.16 11% 37% 41% 9% 2% PHPT 0.34 0.04 0.32 25% 36% 29% 10% - CEOB 0.35 0.12 0.27 28% 50% 19% 3% - FIGU 0.35 0.23 0.15 35% 56% 9% - - THBI 0.52 0.17 0.42 34% 53% 10% 2% 1% PHOV 0.66 0.46 0.30 47% 45% 8% - - CESC 0.66 0.42 0.34 47% 42% 7% 4% - SPV 0.78 0.52 0.44 44% 51% 5% - -

FIBAT 0.79 0.69 0.29 65% 32% 3% - -

106

Figure 5. Mean variations of seed limitation measurements observed during bat disturbance in both disturbed and control site. The line of slope 1 indicates expected values for no effect of bat disturbance. Letters refer to functional units, with alphabetical order respecting apparition order in Table 3 (from a=EVAS to i=FIBAT). Bold and thin characters refer to years 2003 and 2004 respectively. Dotted lines encloses non-significant results according to Two-ways ANOVAs (Table 4) and a posteriori Kruskal-Wallis tests. Most of the values of FL and SL variations lies above expected values.

Variation of seed limitation in control site

Var

iatio

n of

seed

lim

itatio

n in

dis

turb

ed si

te

-0,6 -0,4 -0,2 0 0,2 0,4 0,6

b

g

i

c h

ea

d f

g h

c b

e i f a

d

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

b g

i

c h

e

a

d

f

g

h

c

b e

i

f a

d

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

-0,6 -0,4 -0,2 0 0,2 0,4 0,6

A) Fundamental seed limitation FL

B) Seed source limitation SL b

g i

c h

e a d

f

g

h

c

b e

i

f

a d

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

-0.6 -0.4 -0.2 0 0.2 0.4 0.6

C) Seed dispersal limitation DL

107

Seed source and dispersal limitation

In this second step, we further decomposed fundamental seed limitation FL into seed source

limitation SL and seed dispersal limitation DL. The relative importance of SL and DL

depended on the focus species (Fig. 6). Functional units with FL <0.60 were mostly

dispersal-limited (high DL), while those with FL >0.60 were mostly source-limited (high

SL). Globally, these two seed limitation measurements responded differently to the bat

disturbance experiment (Table 4 and Fig. 5bc). Similarly to FL, SL significantly increased in

most of the cases (increase averaging 170% of mean SL values in Table 3) and did not

display effect inconsistencies between years. By contrast, variations of DL were

multidirectional, with either significant positive or negative effects, or non-significant effects

(increase averaging 7% only of mean DL in Table 3). Two inter-annual inconsistencies were

also encountered (Ficus guianensis and oval-seeded Philodendron with DL varying in

opposite directions from one year to the other, Fig. 5c). Only Cecropia sciadophylla and the

two commonest functional units (the couple Evodianthus / Asplundia and the Philodendron

subgenus Pteromischum) underwent a significant increase of DL during bat disturbance. Figure 6. Relation between FL and the relative importance of SL and DL, showing the functional units mostly source-limited and mostly dispersal-limited. Values proceed from the mean seed limitation measurements per sampling session and per site reported in Appendix 2. The linear regression is significant (p=0.034). Abbreviations follow Table 3.

FIBAT

SPV

PHOV

CESCFIGU

THBI

CEOB

PHPT

EVAS

y=0.6997x – 0.3652 R 2=0.4965

-0.4

-0.2

0

0.2

0.4

0 0.2 0.4 0.6 0.8 1

mos

tly d

isper

sal

limite

d m

ostly

sour

ce

limite

d

Mean FL

Mea

n SL

– M

ean

DL

108

The similar comparison tests performed on seed density values (number of seeds per trap and

per session) mostly corroborated the previous results. All species (except oval-seeded

Philodendron) that expressed a significant increase of SL also reached traps in significantly

lower amounts during the bat disturbance (Table 5, Fig. 7).

Table 5: Results of the two-ways ANOVAs comparing variations of seed density (nb of seeds per trap) between disturbed and control sites. Results are given for the effect of disturbance and its interaction with study year; “ns” indicates non-significant effects; signs + and – refer to the direction of the significant effects. Effect sizes did not differ from one year to the other (all interactions ns). In any case, n=185 and df=1; p values were adjusted based on the sequential Bonferroni method for 9 tests. See Fig. 7 for graphical representation. Abbreviations follow Table 3.

Functional Units F p effectEVAS disturbance 6.56 0.042 – disturbance × year 1.11 ns PHPT disturbance 8.69 0.036 – disturbance × year 0.32 ns CEOB disturbance 7.51 0.039 – disturbance × year 0.01 ns FIGU disturbance 5.23 0.046 + disturbance × year 5.34 ns THBI disturbance 7.85 0.039 – disturbance × year 0.42 ns PHOV disturbance 2.99 ns disturbance × year 0.61 ns CESC disturbance 5.27 ns disturbance × year 0.01 ns SPV disturbance 8.38 0.036 – disturbance × year 0.44 ns FIBAT disturbance 7.25 0.037 – disturbance × year 0.51 ns

109

Figure 7. Mean variations of nb of seeds collected per trap (±±±±1 SEM) observed during bat disturbance in both disturbed and control site. The line of slope 1 indicates expected values for no effect of bat disturbance. Data of 2003 and 2004 were pooled together because no significant interaction with study year was found. Most of functional units remained below expected values (see Table 5 for detailed statistical outputs). Abbreviations follow Table 3.

DISCUSSION

Seed rain sampling revealed a continuum of fundamental seed limitation FL among the target

keystone plants, from highly seed-limited species to poorly seed-limited species.

Conformingly to our hypotheses, we found evidences that disturbance of bat activity exerts

negative repercussions on the seed rain diversity and increases the fundamental seed

limitation (FL) of their plant resources. By focusing on the commonest functional units, we

found that this failure of seeds to reach all suitable microsites as a result of bat activity

depletion tends to proceed from a decrease of seed numbers (increased SL) rather than a less

efficient dispersal (increased DL).

PHPT

EVAS

CEOB

CESC

THBISPV

FIBAT PHOVFIGU

-0.8

-0.6

-0.4

-0.2

0

0.2

-0.4 -0.2 0 0.2 0.4

Variation of log(nb of seeds +1) in control site

Var

iatio

n of

log(

nb o

f see

ds +

1) in

dis

turb

ed si

te

110

General seed rain description

Although it was impossible to determine whether collected seeds were effectively dispersed

by bats, birds or other tree-dwelling dispersers (primates, opossums, kinkajous), we argue

that most of them proceeded from aerial dispersers. Seeds usually occurred in small amounts

(<10) and we never found evidences of abundant faecal material in traps. One can globally

distinguish three different seed rain patterns: low FL (epiphytic Cyclanthaceae and to a lesser

extent Araceae), high FL (Piper, Solanum and Vismia species), and intermediate FL

(Cecropia and Ficus trees). These discrepancies may be related to parent plant density and

productivity, as well as to dispersal modes.

The low FL of epiphytes may result from their great abundance around the sampling site and

the large number of seeds they produce per infructescence (several tens of thousands). In our

sampling sites, more than 80% of the trees support some adult epiphytes belonging to the

target species. Cockle (1997) estimated a density of up to 1000 subadult-to-adult Evodianthus

funifer and Asplundia heteranthera per ha in this part of the study area. Furthermore, the

dispersal mode provided by the common understory fruit bat Rhinophylla pumilio,

specialized on epiphytes of Araceae and Cyclanthaceae, may optimize the spatial scattering

of their seeds. Infructecences of these plants are too large to be removed by bats. Hence, R.

pumilio consumes them in situ during feeding bouts of 11±6 min according to radio-tracking

surveys (Chapter 1). Given the very rapid passage of food through their gut (5 to 30 min;

Charles-Dominique 1986, Cockle 1997), they may often defecate over their foraging area

while flying between the widely scattered epiphytes, enhancing seed dispersal uniformity.

However, foraging movements of R. pumilio are short, with small home ranges of 2.5 to 12

ha and flight distances mostly <300 m (Chapter 1). They may provide shorter dispersal

distances than those provided by other understory fruit bats like Carollia perspicillata which

exploits larger foraging areas (commuting flights regularly covering up to 1.6 km; Charles-

Dominique 1991, Heithaus and Fleming 1978).

The typical bat shrub and treelet Solanum, Piper and Vismia species showed an opposite

pattern, with fewer seeds per fruiting individual (only several tens to several hundreds at a

time) and a much lower occurrence of adult individuals in our study area, probably resulting

in the observed restricted seed rain density and high seed limitation. Only five Piper and one

Solanum individuals were censused in the 0.4 ha area enclosing the sampling sites. These

plants are usually abundant in secondary forests and along edges (e.g. Charles-Dominique

1991, Cosson 1994, Fleming 1991, Schulze 2000), but this characteristic does not apply to

the primary forest surrounding the Nouragues research station, although small patches of

111

several individuals can be found at some places. Their dispersers, mainly Carollia species at

Nouragues (Delaval et al. 2005), may also contribute to seed limitation through the repetitive

use of feeding roosts. Contrary to fruits of Araceae and Cyclanthaceae, Solanum berries and

Piper fruiting spikes are taken by bats and consumed away from the parent plants. In his

study of a Costa Rican bat-dispersed Piper, Fleming (1981) estimated that >90% of seeds

were deposited beneath feeding roosts, among which 40% remained on the unconsumed part

of the spike. A high fidelity to feeding roosts may result in a clumped or “contagious” seed

dispersal (sensu Schupp et al. 2002) and an underestimation of seed rain and dispersal

limitation as measured by seed traps. Notwithstanding, C. perspicillata regularly commutes

over hundreds of meters between its foraging areas, and searches food while commuting

(Fleming et al 1977, Heithaus and Fleming 1978), which may promote non-clumped and

long-distance dispersal events.

Finally, Cecropia and Ficus trees produce large amounts of diaspores at each fruiting event

(e.g. more than 90000 for Cecropia obtusifolia, Estrada et al. 1984) but are moderately

abundant in the pristine forest, which might explain the intermediary FL values. Botanical

surveys in this part of the study area (Poncy et al. 2001) reported only 4 adult C. obtusa and

C. sciadophylla per ha and less than 0.3 Ficus per ha.

The peculiar dispersal mode of Cecropia obtusa somehow compensates for the moderate

parent tree density. When defecated in flight by bats (mainly large canopy bats of the genus

Artibeus, Delaval et al. 2005), the feces resulting from C. obtusa consumption are split into

droplets spreading over an elliptic surface of 1×3 m (Foresta et al. 1984, M.H. pers. obs.).

Most of the diaspores are completely separated from each others (minimal inter-diaspore

distances averaging 7.1±11.2 cm for a feces impact containing 29 diaspores). Such a dispersal

peculiarity, probably resulting from the presence of external mucilage on the diaspores, may

considerably enhance their occurrence frequency in traps.

Although less abundant, Ficus diaspores were also well scattered over sampling sites. We

suppose that dispersal provided by flying frugivores is important for these tree species

because escaping from the high conspecific density around parent trees appears crucial for

seedling establishment success. At each fruiting event, Ficus trees produce massive amounts

of figs. Tello (2003) observed that diaspores of a fruiting F. pertusa were mostly deposited

under the tree crown or in the immediate neighborhood (fallen figs and figs mashed and

dropped by primates) whereas birds (Pipridae, Cotingidae, etc.) swallow a smaller proportion

of figs but defecate diaspores away from the parent tree. On Barro Colorado Island, Panama,

bats account for a great proportion (probably up to >80%) of the fig removal at some Ficus

112

trees (Korine et al. 2000). On this site, Morrison (1978a) estimated that Artibeus jamaicensis

transports and consumes figs in feeding roosts located in a 400-m array from the parent trees.

Ingested seeds may also be released much farther since these bats may commute over several

km between fruiting trees and their roosts (Morrison 1978b). The dispersal pattern of Ficus

diaspores reported in our study resembles what we would expect from flying dispersers, with

diaspores scattered in small amounts (91 to 97% of dispersal events <10 diaspores per trap

and per session, Table 3) and all along the duration of the study.

The effect of bat disturbance on seed diversity and seed limitation

Conforming to our hypothesis, the experimental bat disturbance provoked a significant

decrease of seed diversity (as indicated by both species richness and accumulation rate of

new species in samples). This suggests that the diversity of bat-generated seed rain in small

forest fragments (characterized by lower bat activity) is expected to be substantially reduced.

In the meanwhile, seed diversity increased in the control site, suggesting that bat activity

depletion in a disturbed site was compensated by greater activity in the nearby control site.

Such a scenario would not invalidate our experiment because its purpose was simply to

compare seed rain under two different levels of bat activity.

The decrease of seed species richness is clearly related to an increase of FL, i.e. a reinforced

failure of seeds to reach all traps. At least two results support this statement. First, the species

richness decrease is mainly accounted for by the disappearance from traps of the rarest

species (mostly from Araceae, Piperaceae and Solanaceae families) thereby characterized by

a high FL. Second, the bat disturbance was accompanied by a significant increase of FL

generalized among all of the commonest functional units. The only exception, Ficus

guianensis, was also to our knowledge never reported to be consumed by bats. This species

produces small reddish figs corresponding to ornithochorous syndrome while fruit bats favor

greenish ones in the Neotropics (Kalko et al. 1996, Korine et al. 2000). In the following

discussion, we discarded this species which did not exhibit any FL increase.

The general increasing trend of FL values during bat disturbance confirms the role of bats as

efficient dispersers of the plants studied. According to current theories on seed dispersal, such

a generalized FL increase may favor the maintenance of plant diversity. Indeed, the failure of

highly competitive seeds to reach suitable microsites may allow lower competitive seeds to

“win by forfeit”, which would decrease competitive exclusion (e.g. Hurtt and Pacala 1995,

Schupp et al. 2002). However, this mechanism is more likely to operate on large scale

diversity (beta diversity) to the detriment of local diversity (alpha diversity) because fewer

113

seed species reach suitable places for germination at a time (Muller-Landau et al. 2002).

Increased FL may thereby curtail the plant diversity restoration in fragmented areas.

The origin of seed limitation: restricted seed source or seed dispersal?

There are two possible explanations for the failure of seeds to reach all traps. They may be

either limited by their number (seed source limitation SL) or by the efficiency of their

dispersal (seed dispersal limitation DL). Our experiment revealed that reducing bat activity

significantly contributed to increase SL and to DL for six and three functional units

respectively. Only Philodendron subgenus Pteromischum fell into both “SL-affected” and

“DL- affected” categories.

The “DL-affected” category includes the two most abundant functional units (the couple

Evodianthus / Asplundia and the Philodendron subgenus Pteromischum) and Cecropia

sciadophylla. The latter species is mostly ornithochorous. As birds usually defecate from a

perched position whereas bats also defecate while flying (e.g. Charles-Dominique 1986,

Gorchov et al 1993), the small proportion of C. sciadophylla diaspores dispersed by bats

(Lobova et al. 2003) may contribute in greater extent to reduce DL than SL.

Among the “SL-affected” category, two functional units (oval-seeded Philodendron and the

Solanum-Piper-Vismia group) surprisingly displayed significantly lower DL measures during

bat disturbance. We believe that this unexpected amelioration of dispersal uniformity actually

reflects a drop of the proportion of “clumped” dispersal events (several seeds clumped in a

single feces) relatively to isolated events (1-2 seeds only), which results in a statistically more

uniform seed scattering.

The significant depletions of seed density globally supported the hypothesis that bat

disturbance mostly increased SL rather than DL. With one exception (oval-seeded

Philodendron), an increase of SL was associated to a seed density decrease. Although

significant, the mean loss of seed density appears low, with only minus 1-3 seeds per trap and

per session for a given functional unit (Fig. 7). However, this corresponds to a total loss of up

to 50-150 seeds among all traps during the disturbance session, that is to say >30% of the

normal seed rain density for most of the functional units.

The numerous significant inter-annual variations of bat disturbance effects on seed limitation

(Table 4) may be partly explained by the very dissimilar bat assemblages observed in 2003

and 2004 (dominated by understory and canopy fruit bats respectively). Phenological

variations (i.e. variations of seed production) may also account for a part of inter-annual

variations. The absolute effect of bat disturbance on the dispersal of a given seed species is

114

supposed to decrease with seed density, and eventually become undetectable by our statistical

tests when seeds are too rare. Note also that some statements established herein remain

limited by the difficulty of some seeds to be identified to species level. For instance, the

groups Philodendron subgenus Pteromischum and oval-seeded Philodendron combine three

and seven species respectively. We report here general trends that may not necessarily

accurately foreshadow the real response of individual plant species.

Are bats effective dispersers?

In this study, we established a link between the activity of frugivorous bats and the seed

limitation of some of their plant resources. In that respect, we can reasonably consider fruit

bats as efficient dispersers for those plants. However, there is a conceptual distinction to

make between dispersal efficiency and effectiveness (Bustamante and Canals 1995). While

dispersal efficiency refers to the propensity of dispersers to disseminate seeds in suitable

places for germination, dispersal effectiveness measures the contribution of dispersers to

seedling establishment and recruitment success. Studies linking the activity of dispersers to

plant recruitment patterns (e.g. Herrera et al. 1994) are scarce because closing the seed

dispersal loop is a tedious task for complex systems involving several to many potential

dispersers and plant competitors.

Post-dispersal seed fate was out of the scope of this experiment and further studies are

required to bridge bat activity and plant recruitment patterns. Seed limitation in itself is an

intermediary stage that does not necessarily mirror effective recruitment limitation (Nathan

and Muller-Landau 2000). Small-seeded plants, in particular, are recognize to suffer from

much lower seed-to-seedling transition probabilities than do large-seeded plants (Harms et al.

2000). Many factors are likely to severely exacerbate recruitment limitation and render

meaningless the seed limitation variations described here. For instance, seedlings may be

eliminated by superior competitors (Turnbull et al. 1999) or suffer from severe limitation of

suitable microsites for germination. The pioneer plants like Cecropia species and some Ficus

species, in particular, need light for germination (Vázquez-Yanes et al. 1996) and are then

dependent on tree-fall gaps, an infrequent habitat in primary rain forests (Dalling and Hubbell

2002, Dalling et al. 2002). Although these light conditions are not available in our understory

sampling sites, viable seeds may persist in the soil until dispersal microsites eventually

become suitable for germination. However, unlike Cecropia, Ficus diaspores appear unable

to resort to long-term dormancy (Vázquez-Yanes et al. 1996).

115

Conclusions on seed dispersal in a fragmentation context

Conforming to our hypotheses, the experimental bat disturbance led to an impoverished seed

rain diversity and an increased FL. This prefigures what we would expect to observe in forest

fragments characterized by a depletion of fruit bat activity. Furthermore, this failure of seeds

to reach all suitable microsites in case of bat activity depletion was mostly associated to an

increase of SL rather than DL. In other words, intrinsically source-limited bat plants might be

particularly prone to suffer from bat activity depletion in fragmented areas. Conversely,

dispersal limitation was not the main contributor to fundamental seed limitation. Thus, even

in case of substantial reduction of their activity (and at least up to a certain threshold that

cannot be estimated in this study) fruit bats may still offer seeds a spatially uniform

deposition pattern. This underlines the crucial utility of bats that may ensure an efficient

minimum service of seed dissemination within and between forest fragments.

REFERENCES

Bender, E. A., T. J. Case, and M. E. Gipin. 1984. Perturbation experiments in community

ecology: theory and practice. Ecology 65:1–13.

Bierregaard, R. O., Jr., C. Gascon, T. E. Lovejoy, and R. Mesquita (eds.). 2001. Lessons from

amazonia: the ecology and conservation of a fragmented forest. Yale University

Press, New Haven.

Boggan, J., V. Funk, C. Kelloff, M. Hoff, G. Cremers, and C. Feuillet. 1997. Checklist of the

plants of the Guianas (Guyana, Surinam, French Guiana), 2nd edition. Biological

Diversity of the Guianas Program. Smithsonian Institution Press, Washington D. C.

Brosset, A., P. Charles-Dominique, and A. Cockle. 2001. The bat community. Pp. 115–120 in

Nouragues: Dynamics and plant-animal interactions in a Neotropical rainforest (F.

Bongers, P. Charles-Dominique, P.-M. Forget, and M. Théry, eds.). Kluwer

Academic Publishers, Dordrecht, The Netherlands.

Bustamante, R., and M. Canals. 1995. Dispersal quality in plants: how to measure efficiency

and effectiveness of a seed disperser. Oikos 73: 133–136.

Charles-Dominique, P. 1986. Inter-relations between frugivorous vertebrates and pioneer

plants: Cecropia, birds and bats in French Guyana. Pp. 119–135 in Frugivores and

seed dispersal (A. Estrada, and T. H. Fleming eds.). Dr W. Junk Publishers,

Dordrecht, The Netherlands.

116

Charles-Dominique, P. 1991. Feeding strategy and activity budget of the frugivorous bat

Carollia perspicillata (Chiroptera : Phyllostomidae) in French Guiana. Journal of

Tropical Ecology 7: 243–256.

Charles-Dominique, P. 1995. Interactions plantes-animaux frugivores, conséquences sur la

dissémination des graines et la régénération forestière. Revue d'Ecologie (Terre et

Vie) 50: 223–235.

Charles-Dominique, P., A. Brosset, and S. Jouard. 2001. Atlas des chauves-souris de Guyane,

Patrimoine naturelle No 49. Muséum National d'Histoire Naturelle, Paris.

Charles-Dominique, P., and A. Cockle. 2001. Frugivory and seed dispersal by bats. Pp. 207–

216 in Nouragues: Dynamics and plant-animal interactions in a Neotropical

rainforest (F. Bongers, P. Charles-Dominique, P.-M. Forget, and M. Théry, eds.).

Kluwer Academic Publishers, Dordrecht, The Netherlands.

Cockle, A. 1997. Modalités de dissémination et d'établissement de lianes de sous-bois

(Cyclanthaceae et Philodendron) en forêt guyanaise. PhD Thesis, Université Paris

VI, France.

Cockle, A. 2001. The dispersal and recruitment of Cyclanthaceae and Philodendron

(Araceae) understorey root-climbing vines. Pp 251–263 in Nouragues: Dynamics

and plant-animal interactions in a Neotropical rainforest (F. Bongers, P. Charles-

Dominique, P.-M. Forget, and M. Théry, eds.). Kluwer Academic Publishers,

Dordrecht, The Netherlands.

Connell, J. H. 1971. On the role of natural enemies in preventing competitive exclusion in

some marine mammals and in rain forest trees. Pp. 298–312 in Dynamics of

populations (P. J. D. Boer, and G. Gradwell, eds.). PUDOC, Wageningen, The

Netherlands.

Cosson, J.-F. 1994. Dynamique de population et dispersion de la chauve-souris frugivore

Carollia perspicillata en Guyane Française. PhD Thesis, Université Paris XI,

France.

Cosson, J.-F., J.-M. Pons, and D. Masson. 1999. Effects of forest fragmentation on

frugivorous and nectarivorous bats in French Guiana. Journal of Tropical Ecology

15: 515–534.

Dalling, J. W., and S. P. Hubbell. 2002. Seed size, growth rate and gap microsite conditions

as determinants of recruitment success for pioneer species. Journal of Ecology 90:

557–568.

117

Dalling, J. W., H. C. Muller-Landau, S. J. Wright, and S. P. Hubbell. 2002. Role of dispersal

in the recruitment limitation of Neotropical trees. Journal of Ecology 90: 714–727.

Delaval, M. 2004. Impacts des perturbations d’origine anthropique sur les peuplements de

chauves-souris en Guyane Française. PhD Thesis, Université Paris VI, France.

Delaval, M., M. Henry, and P. Charles-Dominique. 2005. Interspecific competition and niche

partitioning: example of a Neotropical rainforest bat community. Revue d'Ecologie

(Terre et Vie) 60 : 149–166.

De Steven, D., and S. J. Wright. 2002. Consequences of variable reproduction for seedling

recruitment in three Neotropical tree species. Ecology 83: 2315–2327.

Eriksson, O., and J. Ehrlen. 1992. Seed and microsite limitation of recruitment in plant

populations. Oecologia 91: 360–364.

Estrada, A., and R. Coates-Estrada. 2001. Bat species richness in live fences and in corridors

of residual rain forest vegetation at Los Tuxtlas, Mexico. Ecography 24: 94–102.

Estrada, A., R. Coates-Estrada, and D. Meritt, Jr. 1993. Bat species richness and abundance

in tropical rain forest fragments and in agricultural habitats at Los Tuxtlas, Mexico.

Ecography 16: 309–318.

Estrada, A., R. Coates-Estrada, and C. Vásquez-Yanes. 1984. Observations on fruiting and

dispersers of Cecropia obtusifolia at Los Tuxtlas, Mexico. Biotropica 16: 315–318.

Fenton, M. B., L. Acharaya, D. Audet, M. B. C. Hickey, C. Merriman, M. K. Obrist, and D.

M. Syme. 1992. Phyllostomid bats (Chiroptera: Phyllostomidae) as indicators of

habitat disruption in the Neotropics. Biotropica 24: 440–446.

Fleming, T. H. 1981. Fecundity, fruiting pattern, and seed dispersal in Piper amalago

(Piperaceae), a bat-dispersed tropical shrub. Oecologia 51: 42–46.

Fleming, T. H. 1991. The relationhsip between body size, diet and habitat use in frugivorous

bats, genus Carollia (Phyllostomidae). Journal of Mammalogy 72: 493–501.

Fleming, T. H., E. R. Heithaus, and W. B. Sawyer. 1977. An experimental analysis of the

food location behavior of frugivorous bats. Ecology 58: 619–627.

Foresta, H., de, P. Charles-Dominique, C. Erard, and M. F. Prévost. 1984. Zoochorie et

premiers stades de la régénération naturelle après coupe en forêt guyanaise. Revue

d'Ecologie (Terre et Vie) 39: 369–400.

Gascon, C., W. F. Laurance, and T. E. Lovejoy. 2002. Pp 33–48 in How landscapes change:

human disturbance and ecosystem fragmentation in the Americas (G. A. Bradshaw,

and P.A. Marquet, eds.). Ecological studies No 162.

118

Gautier-Hion, A., J. M. Duplantier, R. Quris, F. Feer, C. Sourd, J. P. Decoux, G. Dubost, L.

Emmons, C. Erard, P. Heckestweiller, A. Moungazi, C. Roussilhon, and J. M.

Thiollay. 1985. Fruit characters as a basis of fruit choice and seed dispersal in a

tropical forest vertebrate community. Oecologia 65: 324–337.

Geiselman, C. K., S. A. Mori, and F. Blanchard. 2002 onwards. Database of neotropical

bat/plant interactions. http://www.nybg.org/ botany/ tlobova/ mori/ batsplants/

database/ dbase_frameset.htm

Giannini, N. P., and E. K. V. Kalko. 2004. Trophic structure in a large assemblage of

phyllostomid bats in Panama. Oikos 105: 209–220.

Gorchov, D. L., F. Cornejo, C. F. Ascorra, and M. Jaramillo. 1993. The role of seed dispersal

in the natural regeneration of rain forest after strip-cutting in the Peruvian Amazon.

Vegetatio 107/108: 339–349.

Harms, K. E., S. J. Wright, O. Calderón, A. Hernández, and E. A. Herre. 2000. Pervasive

density-dependent recruitment enhances seedling diversity in a tropical forest.

Nature 404: 493–495.

Heithaus, E. R., and T. H. Fleming. 1978. Foraging movements of a frugivorous bat, Carollia

perspicillata (Phyllostomatidae). Ecological monographs 48: 127–143.

Herrera, C., P. Jordano, L. López-Soria, and J. Amat. 1994. Recruitment of a mast-fruiting,

bird-dispersed tree: bridging frugivore activity and seedling establishment.

Ecological Monographs 64: 315–344.

Howe, H. F., and J. Smallwood. 1982. Ecology of seed dispersal. Annual Review of Ecology

and Systematics 13: 201–228.

Hurtt, G. C., and S. W. Pacala. 1995. The consequences of recruitment limitation: reconciling

chance, history and competitive differences between plants. Journal of Theoretical

Biology 176: 1–12.

Janson, C. H. 1983. Adaptation of fruit morphology to dispersal agents in a Neotropical

forest. Science 219: 187–189.

Janzen, D. 1970. Herbivores and the number of tree species in tropical forests. American

Naturalist 104: 501–528.

Kalko, E. K. V., E. A. Herre, and C. O. Handley, Jr. 1996. Relation of fig fruit characteristics

to fruit-eating bats in the new and old world tropics. Journal of Biogeography 23:

565–576.

119

Korine, C., E. K. V. Kalko, and E. A. Herre. 2000. Fruit characteristics and factors affecting

fruit removal in a Panamanian community of strangler figs. Oecologia 123: 560–

568.

Laurance, W. F., and R. O. Bierregaard, Jr. (eds.). 1997. Tropical forest remnants: ecology,

management, and conservation of fragmented communities. University of Chicago

Press, Chicago, Illinois.

Lobova, T. A., S. A. Mori, F. Blanchard, H. Peckham, and P. Charles-Dominique. 2003.

Cecropia as a food resource for bats in French Guiana and the significance of fruit

structure in seed dispersal and longevity. American Journal of Botany 90: 388–404.

Medellín, R. A., and O. Gaona. 1999. Seed dispersal by bats and birds in forest and disturbed

habitats of Chiapas, México. Biotropica 31: 478–486.

Morrison, D. W. 1978a. Foraging ecology and energetics of the frugivorous bat Artibeus

jamaicensis. Ecology 59: 716–723.

Morrison, D. W. 1978b. Influence of habitat on the foraging distance of the fruit bat Artibeus

jamaicensis. Journal of Mammalogy 59: 622–624.

Muller-Landau, H. C., S. J. Wright, O. Calderón, S. P. Hubbell, and R. B. Foster. 2002.

Assessing recruitment limitation: concepts, methods and case-studies from a

tropical forest. Pp. 35–53 in Seed dispersal and frugivory: ecology, evolution and

conservation (D. J. Levey, W. R. Silva, and M. Galletti, eds.). CAB International.

Nathan, R., and H. C. Muller-Landau. 2000. Spatial patterns of seed dispersal, their

determinants and consequences for recruitment. Trends in Ecology and Evolution

15: 278–285.

Nathan, R., U. N. Safriel, and I. Noy Meir. 2001. Field validation and sensitivity analysis of a

mechanistic model for tree seed dispersal by wind. Ecology 82: 374–388.

Nathan, R., U. N. Safriel, I. Noy Meir, and G. Schiller. 2000. Spatiotemporal variation in

seed dispersal and recruitment near and far from Pinus halepensis trees. Ecology

81: 2156–2169.

Poncy, O., D. Sabatier, M.-F. Prévost, and I. Hardy. 2001. The lowland high rainforest:

structure and tree species diversity. Pp. 31–46 in Nouragues: Dynamics and plant-

animal interactions in a Neotropical rainforest (F. Bongers, P. Charles-Dominique,

P.-M. Forget, and M. Théry, eds.). Kluwer Academic Publishers, Dordrecht, The

Netherlands

120

Schulze, M. D., N. E. Seavy, and D. F. Whitacre. 2000. A comparison of the Phyllostomid

bat assemblages in undisturbed Neotopical forest and in forest fragments of a slash-

and-burn farming mosaic in Petén, Guatemala. Biotropica 32: 174–184.

Schupp, E. W., T. Milleron, and S. E. Russo. 2002. Dissemination limitation and the origin

and maintenance of species-rich tropical forests. Pp. 19–33 in Seed dispersal and

frugivory: ecology, evolution and conservation (D. J. Levey, W. R. Silva, and M.

Galletti, eds.). CAB International.

Simmons, N. B., and R. Voss. 1998. The mammals of Paracou, French Guyana: a neotropical

lowland rainforest fauna. Part I. Bats. Bulletin of the American Museum of Natural

History 237: 0–219.

Soberón, J., and J. Llorente. 1993. The use of species accumulation functions for the

prediction of species richness. Conservation Biology 7: 480–488.

Stiles, E. W. 1992. Animals as seed dispersers. Pp. 87–104 in Seeds: the ecology of

regeneration in plant communities (M, Fenner, ed.). CAB International, New York.

Tello, J. G. 2003. Frugivores at a fruiting Ficus in south-eastern Peru. Journal of Tropical

Ecology 19:717–721.

Terborgh, J. 1986. Keystone plant resources in the tropical forest. Pp. 330–344 in

Conservation biology: the science of scarcity and diversity. (M. E. Soulé, ed.).

Sinauer Associates, Inc., Publishers, Sunderland, Massachusetts.

Thies, W., and E. K. V. Kalko. 2004. Phenology of Neotropical pepper plants (Piperaceae)

and their association with their main dispersers, two short-tailed fruit bats, Carollia

perspicillata and C. castanea (Phyllostomidae). Oikos 104:362–376.

Thomas, D. W. 1982. The ecology of an African savanna fruit bat community: resource

partitioning and role in seed dispersal. Ph.D. Thesis, University of Aberdeen, UK.

Thomas, D. W., D. Cloutier, M. Provencher, and C. Houle. 1988. The shape of bird- and bat-

generated seed shadows around a tropical tree. Biotropica 20: 347–348.

Turnbull, L. A., M. J. Crawley, and M. Rees. 2000. Are plant populations seed limited? A

review of seed sowing experiments. Oikos 88: 225–238.

Turnbull, L. A., M. Rees, and M. J. Crawley. 1999. Seed mass and the

competition/colonization trade-off: a sowing experiment. Journal of Ecology 87:

899–912.

Turner, I. M. 1996. Species loss in fragments of tropical rain forest: a review of the evidence.

Journal of Applied Ecology 33: 200–219.

121

Van der Pijl, L. 1969. Evolutionary action of tropical animals on the reproduction of plants.

Biological Journal of the Linnean Society 1: 85–69.

Vázquez-Yanes, C., M. Rojas-Aréchiga, M. E. Sánchez-Coronado, and A. Orozco-Segovia.

1996. Comparison of light-regulated seed germination in Ficus spp. and Cecropia

obtusifolia: ecological implications. Tree Physiology 16: 871–875

Wada, N., and E. Ribbens. 1997. Japanese maple (Acer palmatum var. Matsumurae

Aceraceae) recruitment patterns: seeds, seedlings, and saplings in relation to

conspecific adult neighbors. American Journal of Botany 84: 1294–1300.

Willson, M. F. 1992. The ecology of seed dispersal. Pp. 61–85 in Seeds: the ecology of

regeneration in plant communities (M. Fenner, ed.). CAB International.

122

APPENDIX

Appendix 1.

Representation of a 1-m² seed trap stretched between four treelets. A central filter permit water draining and a ballast attached bellow the filter gives trap a funnel-like shape. (drawing: S. Jouard).

123

Appendix 2.

Mean values (±±±±1 SD) of seed limitation measurements (FL, SL and DL) resulting from the bootstrapping procedure for each of the height sampling session××××site combinations (disturbed and control sites, before and during disturbance, and in 2003 and 2004).

2003 2004 disturbed control disturbed control before during before during before during before during

Fondamental seed limitation FL EVAS 0.26±0.05 0.28±0.07 0.45±0.08 0.21±0.07 0 0.05±0.05 0 0.02±0.03PHPT 0.11±0.04 0.72±0.06 0.33±0.06 0.57±0.12 0.07±0.05 0.27±0.06 0.28±0.08 0.35±0.10CEOB 0.40±0.05 0.70±0.08 0.44±0.09 0.49±0.08 0.27±0.06 0.45±0.10 0.07±0.05 0.05±0.04FIGU 0.44±0.09 0.26±0.06 0.26±0.09 0.13±0.09 0.36±0.07 0.72±0.06 0.09±0.05 0.56±0.07THBI 0.54±0.07 0.55±0.08 0.70±0.08 0.46±0.10 0.28±0.08 0.37±0.10 0.69±0.06 0.57±0.06

PHOV 0.60±0.09 0.40±0.07 0.69±0.10 0.43±0.06 0.65±0.08 0.76±0.09 0.91±0.05 0.84±0.07CESC 0.51±0.06 0.75±0.08 0.70±0.07 0.88±0.07 0.45±0.08 0.61±0.09 0.67±0.05 0.73±0.06SPV 0.71±0.11 0.92±0.04 0.84±0.06 0.73±0.11 0.71±0.09 0.82±0.08 0.77±0.02 0.71±0.04

FIBAT 0.87±0.06 0.89±0.03 0.91±0.06 0.88±0.10 0.70±0.06 0.79±0.07 0.71±0.05 0.55±0.06Seed source limitation SL

EVAS <0.005 <0.005 0.04±0.07 <0.005 <0.005 <0.005 <0.005 <0.005 PHPT <0.005 0.26±0.23 <0.005 0.03±0.10 <0.005 <0.005 <0.005 <0.005 CEOB 0.04±0.08 0.57±0.12 0.02±0.02 0.04±0.08 0.01±0.02 0.25±0.18 <0.005 <0.005 FIGU 0.29±0.08 0.03±0.03 0.08±0.05 0.01±0.01 0.27±0.15 0.72±0.06 0.01±0.02 0.43±0.14THBI 0.03±0.07 0.17±0.12 0.45±0.18 0.18±0.18 0.03±0.04 0.04±0.04 0.35±0.20 0.10±0.16

PHOV 0.16±0.19 0.07±0.07 0.68±0.08 0.09±0.10 0.53±0.13 0.42±0.31 0.92±0.04 0.80±0.10CESC 0.05±0.09 0.32±0.31 0.67±0.12 0.91±0.06 0.06±0.15 0.20±0.26 0.52±0.08 0.68±0.14SPV 0.49±0.24 0.90±0.06 0.41±0.38 0.45±0.15 0.35±0.24 0.72±0.13 0.47±0.09 0.35±0.10

FIBAT 0.81±0.11 0.84±0.07 0.96±0.03 0.61±0.40 0.55±0.11 0.69±0.15 0.59±0.10 0.43±0.13Seed dispersal limitation DL

EVAS 0.26±0.05 0.28±0.07 0.43±0.07 0.21±0.07 <0.005 0.05±0.05 <0.005 0.02±0.03PHBAT 0.11±0.04 0.59±0.15 0.33±0.06 0.56±0.10 0.07±0.05 0.27±0.06 0.28±0.08 0.35±0.10CEOB 0.37±0.09 0.30±0.12 0.43±0.08 0.46±0.08 0.27±0.06 0.24±0.17 0.07±0.05 0.05±0.04FIGU 0.21±0.11 0.24±0.06 0.19±0.09 0.12±0.09 0.12±0.10 0.07±0.10 0.07±0.04 0.21±0.13THBI 0.53±0.06 0.45±0.12 0.42±0.17 0.32±0.18 0.26±0.07 0.36±0.10 0.49±0.13 0.51±0.11

PHOV 0.51±0.12 0.35±0.09 0.12±0.17 0.36±0.09 0.24±0.12 0.49±0.25 0.13±0.35 0.21±0.20CESC 0.48±0.10 0.55±0.23 0.15±0.20 0.24±0.35 0.40±0.12 0.48±0.11 0.30±0.15 0.15±0.19SPV 0.37±0.17 0.21±0.27 0.57±0.33 0.51±0.12 0.48±0.27 0.31±0.18 0.55±0.08 0.56±0.07

FIBAT 0.24±0.19 0.28±0.20 0.27±0.46 0.54±0.41 0.30±0.16 0.25±0.27 0.25±0.20 0.20±0.13

124

Discussion générale

Evodianthus funifer, détail d’une infructescence, et détail d’une graine (dessin J. Jouard).

1 mm

125

DISCUSSION GENERALE

LE PATRON D’ACTIVITE DE RHINOPHYLLA PUMILIO

Dans le premier chapitre, la chauve-souris Rhinophylla pumilio, spécialiste des fruits

d’épiphytes, a été utilisée comme modèle d’étude pour caractériser le patron d’activité des

petites chauves-souris frugivores de sous-bois. Sa stratégie de quête alimentaire en forêt

intacte se limite à un enchaînement de petits vols de recherche concentrés sur une seule aire

d’alimentation de petite taille, et semble bien adaptée à la distribution très parsemée de sa

ressource alimentaire. Les femelles allaitantes transportent probablement leur progéniture

jusqu’à leur aire d’alimentation où elles les allaitent pendant la nuit. Parallèlement, elles

réduisent leurs distances de vol et leur aire d’alimentation, mais augmentent le temps passé à

voler. Ce patron d’activité semble incompatible avec la nécessité de survoler régulièrement

des zones de matrice inhospitalière qui fragmentent la forêt.

La classification des stratégies de quête alimentaire.

Rhinophylla pumilio est spécialiste des fruits d’épiphytes dont la distribution spatiale est très

diffuse. Conformément aux prédictions, R. pumilio utilise une stratégie de quête alimentaire

basée sur des vols de recherche sur une seule aire d’alimentation et la quasi-absence de longs

vols de déplacement. Cette stratégie est donc très différente de celle des grands Artibeus qui

se déplacent fréquemment sur de longues distances entre les gros Ficus en fruit correspondant

à autant d’aires d’alimentation éphémères. Mais R. pumilio se démarque également des autres

frugivores de sous-bois tels que Carollia perspicillata ou C. brevicauda qui ont également

recours, dans une moindre mesure, à des vols de déplacement entre quelques aires

d’alimentation. La stratégie de quête alimentaire de R. pumilio peut ainsi illustrer la première

extrémité d’un continuum de stratégies le long duquel la prédictibilité spatiale et temporelle

de la ressource diminue, le nombre d’aires d’alimentation et le turnover de leur utilisation

augmentent, et la fréquence des vols de déplacement entre ces aires d’alimentation

s’intensifie. La stratégie des grands Artibeus spécialistes des figues tendrait à marquer l’autre

extrémité de ce continuum, tandis que la stratégie des Carollia occuperait une place

intermédiaire. Cette classification s’affranchit de la conception binaire distinguant les

spécialistes des figues d’une part et les spécialistes des plantes de sous-bois d’autre part

(Chapitre 1), mais peut sembler encore simpliste à en juger par la diversité des chauves-souris

126

frugivores rares et encore méconnues dont les stratégies sont certainement tout aussi

multiples.

La stratégie de quête alimentaire et les capacités mnésiques.

Le premier chapitre illustre la grande capacités des chauves-souris de sous-bois à s’adapter à

une ressource alimentaire diffuse dans l’espace. A cet égard, elles doivent sans doute faire

appel à une mémoire spatiale bien développée pour mémoriser l’emplacement des plantes

susceptibles ou non d’offrir des fruits. Il peut être suggéré que la fragmentation va à

l’encontre de ces processus mnésiques.

La compréhension des processus mnésiques chez les animaux et les humains nécessite de

faire une distinction fondamentale entre la mémoire de référence (« reference memory ») et la

mémoire active (« working memory ») (Baddeley 1986). Tandis que le premier terme se

réfère au système de stockage à long terme et relativement stable de l’information, le second

terme désigne un système de stockage d’information labile et de plus faible capacité. Ce

dernier est sollicité par les opérations cognitives courantes utilisant des informations

changeantes ou temporairement utiles. La planification de déplacements entre patchs de

ressources alimentaires est un exemple typique de processus cognitif impliquant la mémoire

active. Au-delà de la simple mémorisation des coordonnées spatiales des patchs, les animaux

doivent tenir compte de leur productivité en ressources et de leur éventuel appauvrissement

lors des visites précédentes. En d’autres termes, la mémorisation à court terme des séquences

alimentaires précédentes devrait favoriser le succès de quête alimentaire. Cette « mémoire

spatiale active » (« spatial working memory ») a été mise en évidence chez plusieurs

vertébrés tels que les rats et les pigeons (Olton et Samuelson 1976, Roberts et Grant 1974),

ainsi que chez des invertébrés (abeilles, Brown et Demas 1994, Janzen 1971).

Des travaux récents de Winter et collègues (Thiele et Winter 2005, Winter et al. 2005, Winter

et Stich 2005) ont démontré l’existence d’une mémoire spatiale active performante chez des

chauves-souris nectarivores (Glossophaga soricina). Cette mémoire apparaît comme une

réponse adaptative à leur régime alimentaire spécialisé sur le nectar, une ressource souvent

diffuse dans l’espace et éphémère dans le temps (Tschapka 2004). Ces chauves-souris se

montraient efficaces à exploiter jusqu’à 40 fleurs artificielles sans que leur succès de quête

alimentaire ne soit affecté par la revisite de fleurs déjà visitées et « appauvries » dans les

instants précédents. Les espèces nectarivores présentent d’ailleurs un hippocampe

(composant du système nerveux central impliqué dans la mémoire spatiale) hypertrophié par

rapport aux chauves-souris des guildes insectivores ou carnivores (Baron et al. 1996).

127

La théorie suggère donc que les chauves-souris nectarivores, et sans doute les frugivores, sont

capables de planifier mentalement des routes de vol de façon à maximiser le nombre et la

qualité des patchs alimentaires visités tout en minimisant les distances de déplacement. Or,

ces chauves-souris basent leur mémoire spatiale sur des éléments de leur environnement

qu’elles utilisent comme repères (Winter et al. 2005). Ainsi, la fragmentation de l’habitat peu

altérer leur capacité à planifier des routes de vol en créant des ruptures dépourvues de tels

éléments de repères dans leurs aires d’alimentation.

LE DECLIN DES CHAUVES-SOURIS FRUGIVORES EN MILIEU FRAGMENTE

Conformément aux prédictions du premier chapitre, le deuxième chapitre a mis en évidence

une diminution significative de l’abondance de R. pumilio avec la perte de connectivité de

l’habitat forestier, malgré une disponibilité alimentaire inchangée. Cependant, les individus

ont maintenu une activité de reproduction dans la forêt fragmentée, et leur abondance n’a pas

diminué au long des 10 premières années de la fragmentation, sans doute grâce à la petite

taille de leur domaine vital. Les autres frugivores de sous-bois, au contraire, ont disparu des

fragments en dépit de la recrudescence des arbustes à fruits constituant leur principale

ressource alimentaire. Dans les deux cas, la disponibilité alimentaire n’a pas compensé la

perte de connectivité de la forêt.

Fragmentation et compétition interspécifique.

Les captures réalisées sur le site fragmenté de Saint-Eugène (Chapitre 2) ont révélé une

altération marquée et durable de la guilde des chauves-souris frugivores de sous-bois dans la

décennie ayant suivi le processus de fragmentation. Leur abondance ne dépendait pas de la

densité des ressources, mais plutôt du degrés de connectivité de la forêt autour des sites de

captures. Les données concernant les spécialistes des figues A. jamaicensis et A. obscurus,

non présentées dans le deuxième chapitre, corroborent les résultats de Cosson et al. (1999) et

indiquent une apparente tolérance à la fragmentation (Fig. 1).

La théorie suggère que la compétition interspécifique peut également jouer un rôle dans la

distribution différentielle des espèces entre habitat continu et fragmenté. Les diverses études

des effets de la fragmentation de l’habitat sur des communautés de chauves-souris (e.g.

Cosson et al. 1999, Estrada and Coates-Estrada 2001, 2002, Estrada et al. 1993, Gorresen et

128

Figure 1. Indice de sensibilité à la fragmentation des principales chauves-souris frugivores de sous-bois et de canopée à Saint-Eugène, Guyane Française. D’après des tests du χχχχ², les espèces constituant chacune de ces paires diffèrent significativement l’une de l’autre en terme de sensibilité à la fragmentation (* p<0,05 ; *** p<0,001 ; voir texte pour les tests statistiques).

Willig 2004, Schulze et al. 2000) visaient à identifier les traits d’histoire de vie déterminant la

capacité de certaines espèces à s’établir en milieu fragmenté. Par le fait même, elles postulent

que la fragmentation agit comme un filtre perméable aux espèces dites « tolérantes » mais

limitant l’établissement des espèces plus « sensibles ». Une autre approche consisterait à

considérer les espèces numériquement dominantes dans les habitats fragmentés comme des

compétiteurs inférieurs retranchés dans ces habitats sub-optimaux du fait d’une exclusion

compétitive de l’habitat non perturbé par de meilleurs compétiteurs. Selon ce scénario,

lorsque deux espèces sont en compétition pour une même ressource, la meilleure compétitrice

aurait une priorité sur les patchs de ressource les plus accessibles, en forêt continue, tandis

que l’autre se verrait contrainte de déporter son activité vers les habitats fragmentés, moins

propices, au prix d’une diminution de la fitness moyenne des individus.

Quelques données permettent d’étayer l’hypothèse de l’exclusion compétitive de A. obscurus

par A. jamaicensis. Ces deux espèces présentent justement un très fort taux de recouvrement

de niche alimentaire, principalement constituée de Ficus spp. et de Cecropia obtusa (Delaval

Caroll

ia sp

p

0

0,2

0,4

0,6

0,8

1

R. pu

milio

A. jamaic

encis

A. obs

curu

s

Indi

ce d

e se

nsib

ilité

à la

frag

men

tatio

n

Caroll

ia sp

p

0

0,2

0,4

0,6

0,8

1

R. pu

milio

A. jamaic

encis

A. obs

curu

s

Frugivores de sous-bois Frugivores de canopée

***

*

***

129

et al. 2005). A Saint-Eugène, A. obscurus est significativement moins affecté par la

fragmentation d’après la comparaison des proportions de ces deux espèces dans les captures

des sites de forêt continue d’une part et des sites de fragments forestiers d’autre part (Fig. 1 ;

χ²=4,268 ; n=293 ; p=0,038). Or, le sex-ratio de A. obscurus est fortement et

significativement contrebalancé en faveur des mâles sur les fragments par rapport à la forêt

continue (Fig 2b ; χ²=9,345 ; n=159 ; p=0,002) et des mesures d’hématocrite suggèrent un

moins bon succès de quête alimentaire (Fig. 2a ; Mann-Whitney U=196,5 ; n=32 ; p=0,003).

Ces paramètres ne varient pas significativement entre les fragments et la forêt continue pour

A. jamaicensis (sex-ratio : χ²=0,285 ; n=101 ; p=0,593 ; hématocrite : U=262 ; n=50 ;

p=0,515).

L’hématocrite se réfère ici à la proportion de volume sanguin occupé par les globules. Il a été

mesuré sur les individus capturés au début de la nuit (R. pumilio, C. perspicillata, A.

jamaicensis et A. obscurus) en prélevant une goutte de sang dans un capillaire et en le

centrifugeant, de façon à séparer les globules du plasma. Plus un animal est hydraté, plus la

concentration d’eau dans le sang est élevée, et plus le volume globulaire (hématocrite) est

bas. L’assimilation de l’hématocrite à une mesure du succès de quête alimentaire a été

renforcé par une expérimentation consistant à effectuer deux mesures à trois heures

d’intervalle sur des individus gardés captifs. Certains étaient régulièrement nourris à l’eau

sucrée et d’autres non nourris. L’hématocrite des premiers diminuait significativement plus

que celui des seconds (-3,18±3,73 et -0,37±3,00 respectivement ; Mann-Whitney U=151,0 ;

n=49 ; p=0,005). La baisse d’hématocrite pour les individus nourris était d’autant plus

prononcée que l’hématocrite initial était élevé (régression linéaire: n=21 ; p=0,033 ;

R²=0,216). L’hématocrite a été également assimilé à un indice de déshydratation pour le

mégachiroptère Rousettus aegyptiacus (Korine et al. 1999) car il était plus bas à l’heure du

retour au gîte le matin qu’après le départ du gîte le soir, et il tendait à augmenter pendant les

phases de repos.

Même si ces données corroborent l’hypothèse de l’exclusion compétitive, la prudence est de

mise pour interpréter de simples résultats de capture en termes de compétition

interspécifique. En pratique, mettre en évidence un tel phénomène entre espèces de chauves-

souris suppose de comparer les recouvrements de niches alimentaires intraspécifiques et

interspécifiques, à la fois en situation de sympatrie et d’allopatrie (e.g. Arlettaz et al. 1997).

A l’inverse, il peut être envisagé que R. pumilio profite d’un relâchement de la compétition

(« competition relaxation ») facilitant son établissement dans les zones très fragmentées et

130

Figure 2. Indices de santé des populations de trois espèces de chauves-souris frugivores à Saint-Eugène, Guyane Française : hématocrite (a) et sex-ratio mâles : femelles des adultes (b). L’hématocrite mesure ici la proportion de volume globulaire dans les sang. (ns=différence non significative ; * p<0,05 ; ** p<0,01 ; voir texte pour les détails et les tests statistiques).

désertées par les autres chauves-souris de sous-bois. La théorie des niches (Hutchinson 1957)

prédit que si deux espèces sont en compétition pour des ressources limitées, elle ne peuvent

coexister que s’il se met en place un système de partage des ressources par un déplacement de

niche. Ainsi, en sympatrie, la « niche fondamentale » des espèces se restreint-elle à une

« niche réalisée ». C’est peut-être le cas de R. pumilio d’une part et des Carollia d’autre part

dont le recouvrement des niches alimentaires est faible (Cosson 1994, Delaval et al 2005).

Rhinophylla pumilio est normalement spécialiste des fruits d’épiphytes, mais consomme

régulièrement des fruits des arbustes Piper, Solanum et Vismia en forêt secondaire (Cosson

0

0,5

1

1,5

2

2,5

Hém

atoc

rite

(%)

Sex-

ratio

(M :F

)

R. pumilio A. obscurus A. jamaicensis

50

54

58

62

66

70

Forêt continue Fragments

a

b

*

**

ns

ns

**

ns

131

1994). Sa niche fondamentale pourrait ressembler à celle des Carollia dont elle est

apparentée. Il serait intéressant de vérifier si R. pumilio inclus davantage de fruits d’arbustes

dans son régime alimentaire dans les fragments forestiers où les Carollia sont absentes.

Mesurer la sensibilité à la fragmentation

Les résultats précédents suggèrent que définir la sensibilité à la fragmentation sur l’unique

base de l’abondance des individus et sans tenir compte de leurs aptitudes reproductrices

pourrait s’avérer réducteur ou inexact. Ici, les données suggèrent que (i) les femelles A.

obscurus ne parviennent pas à évoluer dans le milieu fragmenté, peut-être parce qu’au

moment de l’étude (pic annuel de reproduction), leur capacité de déplacement est limité par

les contraintes de la reproduction et (ii) les individus A. obscurus exploitent moins

efficacement les fragments que la forêt continue. A. jamaicensis paraît au contraire non

affecté par la fragmentation en termes de sex-ratio et d’hématocrite. De façon similaire, il a

été prédit que les femelles reproductrices R. pumilio devaient être plus sensibles à la

fragmentation de l’habitat (Chapitre 1), alors que cette espèce à continué de se reproduire

dans les fragments forestiers et à maintenir une population apparemment en équilibre

(Chapitre 2). Les valeurs d’hématocrite, plus élevées dans les fragments (U=178,5 ; n=33 ;

p=0 ,027), corroborent la prédiction du premier chapitre et suggèrent que les femelles

auraient potentiellement une fitness moindre en forêt fragmentée qu’en forêt continue, en

dépit d’un sex-ratio équilibré entre ces deux milieux (Fig. 2b ; χ²=0,064 ; n=106 ; p=0,800).

Il paraît donc nécessaire, dans l’étude des effets de la fragmentation, de coupler les données

d’abondance des chauves-souris à d’autres paramètres, à la fois populationnels et

physiologiques.

CONSEQUENCES DU DECLIN DES CHAUVES-SOURIS FRUGIVORES SUR LA PLUIE DE

GRAINES

Les modifications de la pluie de graines subséquentes au déclin des chauves-souris frugivores

ont été étudiées dans le troisième chapitre par le biais d’une perturbation expérimentale de

l’activité des chauves-souris. Celle-ci consistait en un effort massif de captures au filet

déployé sur un site d’échantillonnage de pluie de graines (0,12 ha). La comparaison avec un

site contrôle a mis en évidence une diminution significative de 30 à 50% de la diversité des

132

graines étudiées. Cette diminution est principalement imputable à la disparition des espèces

de graines les plus rares (e.g. genres Solanum ou Piper), tandis que les graines les plus

communes ont vu leur densité diminuée substantiellement. En revanche, l’homogénéité

spatiale de la dispersion de ces graines n’a pas été modifiée. Ces résultats soulignent

l’importance des chauves-souris dans les processus de dispersion et de régénération des

espèces végétales dont elles se nourrissent. En outre, cela met en évidence les risques de

perturbation des interactions mutualistes entre plantes et chauves-souris en milieu fragmenté.

En diminuant non seulement la diversité des plantes, mais aussi des chauves-souris, la

fragmentation peut sérieusement affecter la diversité locale (diversité α) des flux de graines.

Une étude préliminaire menée sur trois sites de forêt continue et trois fragments forestiers à

Saint-Eugène confortent les conclusions du troisième chapitre (Annexe 2), malgré un effort

d’échantillonnage restreint (54 collecteurs pendant 15 jours).

Pour mener à bien des programmes de conservation de certaines plantes, il pourrait être utile

d’étudier au préalable dans quelle mesure les perturbations de l’habitat peuvent en affecter les

flux de graines. Les données rassemblées dans le troisième chapitre montrent que de telles

études ne peuvent être réalisées que pour les graines les plus abondantes ou sur de très

longues périodes d’échantillonnage avec de nombreux collecteurs. En effet, la probabilité

qu’un collecteur donné reçoive au moins une graine de l’espèce étudiée pendant la durée d’un

échantillonnage est généralement très faible, et diminue d’autant plus que la durée de

l’échantillonnage est courte. Ainsi, la densité de pluie de graines (nombre de graines

collectées par collecteur et par unité de temps) est caractérisée par une distribution de

fréquence très leptocentrique, i.e. une grande majorité de valeurs nulles ou faibles, et

quelques valeurs élevées prolongeant la « queue » de la courbe de distribution. Une telle

déviation de la normalité implique des contraintes statistiques et en particulier empêche

l’application de tests paramétriques, pourtant plus versatiles et puissants que les non

paramétriques. La puissance d’un test est la probabilité de rejeter l’hypothèse nulle quand elle

est effectivement fausse, et se réfère donc à la capacité du test à détecter une vrai différence

entre deux moyenne. Il s’agit du complément « 1-β », β étant la probabilité de faire une erreur

de type II (Sokal et Rohlf 1995). Un test moins puissant requiert de plus grandes tailles

d’échantillons et suppose donc davantage de contraintes pratiques.

A cet égard, un programme d’échantillonnage de pluie de graines doit être conçu en

considérant deux questions importantes : (i) combien de temps l’échantillonnage doit-il durer

pour obtenir des données normales (après correction logarithmique ou racine carrée) et (ii)

combien de collecteurs sont-ils nécessaires pour obtenir une puissance statistique satisfaisante

133

(entre 0,90 et 0,95) pour la comparaison de deux habitats. Dans cette étude en particulier, la

normalité des densités de pluie de graines (avec la correction log[valeur +1]) est atteinte au

bout de 45 jours d’échantillonnage (trois sessions de 15 jours) pour les principales espèces

des familles Araceae, Cecropiaceae et Cyclanthaecae. En groupant les données de nombres

de graines par collecteur pour une période de 45 jours et appliquant la méthode itérative de

Sokal et Rohlf (1995) pour le calcul du nombre d’échantillons requis selon la puissance

désirée, plusieurs courbes de puissance statistique ont été tracées (Fig. 3). En général,

détecter une vrai différence de 50% entre les moyennes des deux habitats requiert l’utilisation

de 50 à 100 collecteurs. Ce chiffre augmente de façon exponentielle lorsque l’on souhaite

davantage de précision. Par exemple, il faut environ 200 collecteurs par habitat pour détecter

une différence de 20% de la densité de graines d’Aracées. Ces chiffres peuvent être diminués

en allongeant la période d’échantillonnage.

Ces analyses montrent également que pour mesurer et comparer les densités des espèces de

graines moins communes, il convient d’effectuer des études à long terme et / ou d’utiliser

d’autres descripteurs de pluie de graines. Ces descripteurs peuvent être les indices de

limitation de la dispersion (Chapitre 3). Dans ce cas, un processus de retirage aléatoire

(« bootstrapping ») peut être nécessaire pour effectuer des tests de comparaison, car ces

indices ne produisent qu’une donnée par site d’échantillonnage et non une donnée par

collecteur.

CONCLUSIONS SUR LES IMPLICATIONS EN CONSERVATION

Au regard des résultats des trois chapitres et des perspectives examinées dans la présente

discussion, cinq recommandations sont proposées pour améliorer les études d’impacts de la

fragmentation.

1. Diversifier les études télémétriques.

La vision binaire distinguant la stratégie des chauves-souris de sous-bois d’une part et des

chauves-souris spécialistes des figues d’autre part semble simpliste. La variété des régimes

alimentaires, des tailles corporelles ou des comportements de sélection de gîtes doit être

accompagnée d’une diversité des patrons d’activité des chauves-souris frugivores. La

multiplication des études autécologiques pourrait contribuer à une meilleure anticipation des

impacts de la fragmentation sur les populations de chauves-souris.

134

Figure 3. Courbes de puissance statistique indiquant le nombre de collecteurs nécessaires en fonction de la plus faible différence à détecter entre les densités moyennes de pluie de graines de deux sites distincts. La densité de pluie de graines se réfère au nombre de graines (transformé en log) collectées par collecteur de 1 m² sur une période de 45 jours. Chaque paire de courbe délimite la zone critique sous laquelle la puissance statistique devient trop faible pour effectuer un test paramétrique fiable (puissance=0,90 et 0,95 pour les courbes inférieures et supérieures, respectivement). Exemple : détecter une différence de 50% entre la densité moyenne de pluie de graines de T. bissectus mesurée dans deux sites requiert a priori un minimum de 90-100 collecteurs par site.

0

20

40

60

80

100

120

140

160

180

200

0 20 40 60 80 100

E. funifer + A. heteranthera

T. bissectus

Toutes Cyclanthaceae

Cyclanthaceae

0

20

40

60

80

100

120

140

160

180

200

0 20 40 60 80 100

Araceae

Philodendron sous-genre Pteromischum

Toutes Araceae

0

50

100

150

200

250

300

0 20 40 60 80 100

Cecropiaceae

C. sciadophylla

C. obtusa

Toutes Cecropia spp.

Nb.

de

colle

cteu

rs

Nb.

de

colle

cteu

rs

Nb.

de

colle

cteu

rs

Plus petite différence détectable (%)

135

2. Diversifier les indicateurs de sensibilité à la fragmentation.

Il ne faut pas utiliser les simples variations d’abondance comme indicateur de sensibilité à la

fragmentation, mais trouver des indicateurs complémentaires témoignant de la stabilité des

populations et de la fitness des individus, tels que : le sex-ratio, l’activité reproductrice

(proportion de juvéniles, de femelles gestantes et allaitantes), des mesures du succès de quête

alimentaire (étude du budget temporel par télémétrie) et de la condition physiologique des

individus (e.g. dosage de la corticostérone et autres hormones de stress, etc.). Cela implique

aussi des suivis à long terme pour tenter d’évaluer les fluctuations de populations et

l’efficacité du recrutement reproductif par capture-marquage-recapture. Cette dernière

technique est normalement peu efficace compte tenu de l’étendue de l’effort de capture à

déployer, tant dans l’espace que dans le temps, pour rassembler suffisamment de données de

recaptures. Elle peut cependant s’envisager pour les espèces à refuge formant de larges

colonies pouvant faire l’objet d’un programme de suivi.

3. Privilégier les analyses de paysage aux conceptions binaires « fragments vs. forêt

continue ».

Les indices de paysage dérivés d’analyses de pixels par Systèmes d’Informations

Géographiques sont des outils pertinents d’aide à la gestion et à la prise de décision.

L’abondance des chauves-souris étudiées dans le second chapitre dépendait de l’agencement

spatial de plusieurs fragments continus. De simples modèles inspirés de la théorie de la

biogéographie insulaire (MacArthur et Wilson 1963, 1967) peuvent être inefficaces lorsque la

forme des fragments influence davantage l’abondance des animaux considérés que la taille et

le degrés isolement.

4. Tenir compte des patchs et corridors de végétation secondaire dans les analyses de

paysage.

Les chapitres 1 et 2 soulignent l’importance de la continuité des habitats pour R. pumilio, et

probablement pour les autres espèces adoptant une stratégie de quête alimentaire basée sur de

courts vols de recherche. Les corridors peuvent s’avérer des outils utiles pour le maintien des

capacités de mouvement à travers le paysage (Rosenberg et al. 1997, Wilson et Willis 1975),

notamment s’ils sont riches en espèces pionnières de Piper, Solanum et Vismia. Ils peuvent

jouer le rôle de réserves de ressources et constituer des passages à gué (« stepping stones »)

pour les chauves-souris évoluant en milieu fragmenté. Ces habitats particuliers sont à prendre

en considération dans les analyses de paysage.

136

5. Effectuer des études pilotes avant la conception de plans d’ échantillonnage de la pluie de

graines.

Des échantillonnages préliminaires impliquant au moins 50 collecteurs par habitat (e.g.

habitat intact vs. perturbé) pendant au moins 30 à 45 jours sont recommandés pour identifier

le profil général de la pluie de graines. Ces données permettraient (i) d’identifier les

principales espèces de graines, (ii) de réaliser pour ces espèces une étude de puissance

statistique afin d’identifier les meilleurs compromis entre d’une part l’effort

d’échantillonnage (nombre de collecteurs et durée d’échantillonnage) et d’autre part les

limitations logistiques, tout en conservant une puissance statistique satisfaisante (0,90 à 0,95)

pour les comparaisons de densités de graines, (iii) de déterminer, selon les espèces, la

nécessité d’utiliser des descripteurs de pluie de graines plus adaptés aux faibles effectifs

(indices de limitation de la dispersion).

RÉFÉRENCES

Arlettaz, R., N. Perrin, et J. Hausser. 1997. Trophic resource partitioning and competition

between two sibling bats species Myotis myotis and Myotis blythii. Journal of

Animal Ecology 66: 897–911.

Baddeley, A. 1986. Working memory. Oxford University Press, Oxford.

Baron, G., H. Stephan, et H. Frahm. 1996. Comparative neurobiology in Chiroptera.

Birkäuser Verlag, Berlin.

Brown, M. F., et G. E. Demas. 1994. Evidence for spatial working memory in honeybees

(Apis mellifera). Journal of Comparative Psychology 108: 344–352.

Cosson, J.-F. 1994. Dynamique de population et dispersion de la chauve-souris frugivore

Carollia perspicillata en Guyane Française. Thèse de Doctorat, Université Paris XI,

France.

Cosson, J.-F., J.-M. Pons, et D. Masson. 1999. Effects of forest fragmentation on frugivorous

and nectarivorous bats in French Guiana. Journal of Tropical Ecology 15: 515–534.

Delaval, M., M. Henry, et P. Charles-Dominique. 2005. Interspecific competition and niche

partitioning: example of a Neotropical rainforest bat community. Revue d'Ecologie

(Terre et Vie) 60 : 149–166.

137

Estrada, A., et R. Coates-Estrada. 2001. Bat species richness in live fences and in corridors of

residual rain forest vegetation at Los Tuxtlas, Mexico. Ecography 24: 94–102.

Estrada, A., et R. Coates-Estrada. 2002. Bats in continuous forest, forest fragments and in an

agricultural mosaic habitat-island at Los Tuxtlas, Mexico. Biological Conservation

103: 237–245.

Estrada, A., R. Coates-Estrada, et D. Merrit. 1993. Bat species richness and abundance in

tropical rain forest fragments and in agricultural habitats at Los Tuxtlas, Mexico.

Ecography 16: 309–318.

Gorresen, P. M., et M. R. Willig. 2004. Landscape responses of bats to habitat fragmentation

in Atlantic forest of Paraguay. Journal of Mammalogy 85: 688–697.

Hutchinson, G. E. 1957. Concluding remarks. Cold Spring Harbour Symposium on

Quantitative Biology. 22: 415–427.

Janzen, D. H. 1971. Euglossine bees as long distance pollinators of tropical plants. Science,

171: 203–205.

Korine, C., O. Zinder, et Z. Arad. 1999. Diurnal and seasonal changes in blood composition

of the free-living Egyptian fruit bat (Rousettus aegyptiacus). Journal of

Comparative Physiology – section B – Biochemical Systemic and Environmental

Physiology 169:280–286.

MacArthur, R. H., et E. O. Wilson. 1963. An equilibrium theory of insular zoogeography.

Evolution 17: 373–387.

MacArthur, R. H., et E. O. Wilson. 1967. The theory of island biogeography, Princeton

University Press.

Olton, D. S., et R. J. Samuelson. 1976. Remembrance of places past : spatial memory in rats.

Journal of Experimental Psychology: Animal Behavior Processes 2: 97–116.

Roberts, W. A., et D. S. Grant. 1974. Short term memory in the pigeon with presentation time

precisely controlled. Learning and Memory 5: 393–408.

Rosenberg, D. K., B. R. Noon, et E. C. Meslow. 1997. Ecological role of linear conservation

areas for maintaining biological diversity. BioScience 47: 677–687.

Schulze, M. D., N. E. Seavy, et F. F. Whitacre. 2000. A comparison of the Phyllostomid bat

assemblages in undisturbed neotopical forest and in forest fragments of a slash-and-

burn farming mosaic in Petén, Guatemala. Biotropica 32: 174–184.

Sokal, R. R., et L. J. Rohlf. 1995. Biometry: the principles and practice of statistics in

biological research, 3ème éd. W. H. Freeman and Company.

138

Thiele, J, et Y. Winter. 2005. Hierarchical strategy for relocating food target in flower bats:

spatial memory versus cue-directed search. Animal Behaviour 65: 315–327.

Tschapka, M. 2004. Energy density patterns of nectar resources permit coexistence within a

guild of Neotropical flower-visiting bats. Journal of Zoology, London. 263: 7–21.

Wilson, E. O., et E. O. Willis. 1975. Applied biogeography. Pp 522–534 dans Ecology and

evolution of communities (M. L. Cody, et J. M. Diamond, éds.). Belknap Press,

Cambridge, Massachusetts.

Winter, Y., et K. P. Stich. 2005. Foraging in a complex naturalistic environment: capacity of

spatial working memory in flower bats. Journal of Experimental Biology 208: 539–

548.

Winter, Y., S., von, Merten, et H.-U. Kleindienst. 2005. Visual landmark orientation by

flying bats at a large-scale touch and walk screen for bats, birds and rodents.

Journal of Neuroscience Methods 141: 283–290.

139

Annexes

Feuilles et infructescences de Cecropia sciadophylla (en haut) etC. obtusa (en bas), et détail desfruits constituant les unités dedispersion (dessin S. Jouard).

1 mm

1 mm

140

ANNEXE 1 : PUBLICATION DELAVAL ET AL. (2005).

Delaval, M., M. Henry, et P. Charles-Dominique. 2005. Interspecific competition and niche partitioning: example of a Neotropical rainforest bat community. Revue d'Ecologie (Terre et Vie) 60 : 149–166.

INTERSPECIFIC COMPETITION AND NICHE PARTITIONING:

EXAMPLE OF A NEOTROPICAL RAINFOREST BAT COMMUNITY

Marguerite DELAVAL1, Mickaël HENRY1 & Pierre CHARLES-DOMINIQUE2

1Département d’Ecologie et Gestion de la Biodiversité, UMR 5176, 4 avenue du Petit Château, F-91800 Brunoy. E-mail: marguerite.delaval@wanadoo.fr 2 UMR ECOFOG, et Station des Nouragues, CNRS, 15 Avenue André Aron, F-97300 Cayenne, Guyane Française.

SUMMARY.– To understand the organisation of a bat community and the coexistence of sympatric species, it is essential to understand how species use and share common resources. First, we describe a bat community in a primary rainforest of French Guiana. The presence of particular roosting sites, such as caves, and the absence of disturbances are important local factors in structuring communities. In the course of this study, we focused on the three most common species of three vegetarian bat guilds (understorey frugivores, canopy frugivores and nectarivores). The local coexistence of these species is possible thanks to space, food and/or time partitioning. Space partitioning is consistent with the hypothesis that smaller bats with a more manoeuvrable flight tend to occupy more cluttered space less attractive to their competitors and have smaller home range. We observed a time partitioning that is likely to reduce competition among some frugivorous bat species by reducing direct interference during foraging. Besides an interest for the field community ecology, this study of a community living in a primary forest can be used as a reference for non disturbed habitat for conservation purposes.

RESUME.– Compétition interspécifique et division des niches : l’exemple du peuplement de chauves-souris d’une forêt tropicale humide. Afin de comprendre l’organisation des communautés de chauves-souris ainsi que la coexistence de nombreuses espèces sympatriques, il est essentiel de déterminer comment les espèces utilisent et se partagent les ressources. Dans un premier temps, nous avons décrit un peuplement de chauves-souris, en forêt primaire de Guyane Française. Les facteurs locaux tels la présence de gîtes particuliers comme les grottes, ou l’absence de perturbation anthropique ont une influence sur la structure et la composition des communautés. Au cours de cette étude, nous nous sommes concentrés sur les trois espèces les plus communes de chaque guilde végétarienne (frugivores de sous-bois, frugivores de canopée et nectarivores). Leur coexistence est possible grâce au partage des ressources alimentaires, de l’espace et du temps. Au sein d’une même guilde, les plus petites espèces ont une bonne manoeuvrabilité et semblent occuper d’avantage les espaces fermés moins attractifs pour leurs compétiteurs. Elles ont aussi un domaine vital plus petit que les grandes espèces. De même, le décalage des rythmes d’activités que nous avons observé, peut réduire la compétition entre les chauves-souris frugivores en diminuant les interférences directes lors de l’alimentation. En plus d’un intérêt pour l’écologie des communautés, cette étude peut servir de référence en biologie de conservation car elle permet d’avoir un bon état des lieux des peuplements de chiroptères présents en forêt primaire.

________________________________________________

Bats are of paramount importance in Neotropical rainforest ecosystems because of their abundance, diversity and ecological roles. Bats account for 50% of the Neotropical mammals species and, constitute the most important order of mammals in Neotropical rainforest (Emmons & Feer, 1990). About 70 bat species may coexist in a given forest site in French Guiana, corresponding to as many species as all other mammals species (Simmons & Voss 1998; Brosset et al., 2001). Bats pollinate many plants and contribute to forest regeneration by dispersing seeds (Gardner, 1977; Heithaus, 1982; De Foresta et al., 1984; Charles-Dominique, 1995). Zoochory is particularly widespread within pioneer plants and nearly half of the most abundant species are bat dispersed (Charles-Dominique, 1986). The development of flight, as well as numerous morphological and sensory adaptations such as the sophisticated echolocation system, allow bats to access a wide range of habitats and exploit a great variety of food resources. For instance, diet of tropical bats is unequalled in variety compared to other mammals since it ranges from fruits, pollen, nectar, and leaves to small vertebrates, blood, and insects (Kalko, 1997).

Factors determining species composition of a bat community in a given region are poorly known. Spatial heterogeneity has long been recognized as an important factor promoting diversity of animals and

141

plants, especially in species rich tropical lowland forest (Kalko & Handley, 2001). Local factors, such as food and shelter availability, and the degree of disturbance, may also be strong pressures for determining species composition (Bonaccorso, 1979; Fenton et al., 1992). However, some studies (Willig & Moulton, 1989; Arita, 1997) indicate that historical factor of colonisation are also important in determining the structure of bat faunas.

It is usually admitted that in order to coexist, two sympatric species should differ to a certain degree in their ecological niches. Niche differentiation has long been interpreted in terms of partitioning of food resource, e.g. foraging strategies and diet composition (McNab, 1971). However, it has been shown in several recent studies that differentiation may also include spatial segregation in habitat use and roost site selection (Marshall, 1983; Marinho-Filho & Sazima, 1989; Marinho-Filho, 1991; Saunders & Barclay, 1992; Kalko, 1995; Kalko et al., 1996b).

Studies of bat community structure are often limited to disturbed habitats (Reis & Peracchi, 1987; Simmons & Voss, 1998; Bernard, 2001; Estrada & Coates-Estrada, 2001; Bernard & Fenton, 2002). Our study concerns a bat community in a primary rainforest in French Guiana and thus can be used as a reference for non disturbed habitat. A better understanding of mechanisms, such as space, time and food partitioning, which determine the coexistence of many sympatric species, has profound implications, not only for the field of community ecology but also for conservation biology. This is especially important in tropical forest ecosystem increasingly threatened by habitat alteration, fragmentation and deforestation.

The objectives of this study are (i) to describe the composition and organisation of a Neotropical primary rainforest bat community and illustrate how its guilds share habitats, (ii) to investigate the components of partitioning at intra-guild level in terms of space, time and food resources, by focusing on frugivorous and nectarivorous bats.

MATERIAL AND METHODS

STUDY SITE The study was carried out at the Nouragues biological station in the centre north of French Guiana (4°50’ N,

52°42’ W). This study area, in the middle of the Nouragues natural reserve, is included in a large block of dense continuous tropical primary rainforest where human interference has been absent or negligible over the past two centuries (Brosset et al., 2001). The first roads and human settlements are at about 25 km from the station which is only accessible by helicopter. The average annual temperature is 26.3°C and the average annual rainfall ranges from 2500 to 3200 mm, with a marked dry season from August to November.

Caesalpinaceae, Sapotaceae and Lecythidaceae families dominate the local canopy tree community, with a great abundance of Eperua falcata for Caesalpinaceae (Poncy et al., 2001). The primary forest is more precisely a mosaic of plant community units varying in structure and composition (Aubreville, 1938; Oldeman, 1990; Riéra, 1998). Basically, the mature forest can be divided into two subsets according to canopy height (Poncy et al., 2001). The dominant forest type is characterised by 30-35-m high trees with emergent trees reaching more than 50 m and a fairly open understorey. This forest matrix encloses patches of liana forest (10-25m high) characterised by a very dense understorey and mainly composed of trees overloaded by liana stems. Liana forest forms a ca. 30 ha block surrounded by several coalescent patches which overlap the southern part of the study area.

BAT SAMPLING PROTOCOL AND DETERMINATION OF DIET Bat assemblages were sampled in four different habitats: high forest (HF) corresponding at the more stratified

habitat; liana forest (LF) an habitat with a lot of tree fall; creek corridors (CC) (above a 5-10 m wide creek going through the study area) and edges of forest clearing (EC) (along the border of the 1.5 ha clearing were the actual camp was built) which are both open habitat. The borders of the camp are colonized by pioneer vegetation. All bat surveys were carried out in the 1.3×1.1 km quadrat delimited by a network of small trails surrounding the camp. Captures sessions were conducted during three dry seasons (October-December 2000, July-October 2001 and July-September 2002) and one wet season (February-May 2001). Bats were captured at understorey level using mist nets (10×2 m or 12×2 m, mesh size 16 mm) in the four above-mentioned habitats. Each station was sampled during two consecutive nights before nets were removed. Four nets were also set at sub-canopy, between 8 and 10 m, and four others in canopy, between 15 and 20 m, in the edge of the camp (EC habitat) by the mean of vertical ropes.

For each captured individual, we recorded the netting station and time of capture, the species and sex and we measured body weight (± 0.25 g) and forearm length (± 0.05 mm). Females were indexed as non reproductive, pregnant, or lactating, according to palpation of abdomen and aspect of nipples, (Anthony, 1988; Racey, 1988). Taxonomic nomenclature of bats followed Charles-Dominique et al. (2001). After identification, frugivorous and nectarivorous bats were ringed with numbered plastic wing-bands (rings A. C. HUGUES England). Each individual was

142

kept in a cloth bag at least one hour before being released. The faecal material remaining in bags was preserved in alcohol in order to later (i) identify the seeds and pollen it might contain and (ii) check for the presence of arthropod fragments. Furthermore, additional pollen grains were collected on the snout of nectarivorous bats by dabbing the fur with the sticky surface of a transparent tape fixed beforehand on a microscope slide. Some faeces contained only vegetal fragment (with no seed, pollen nor insects fragments) which was likely to correspond to the pulp of big fruits (with big seeds), or to nectar, for frugivores or nectarivorous bat faeces respectively.

Seed identification was based on external criteria such as size, shape, colour, surface texture, etc. Identification was facilitated by comparing seeds with a seed reference collection (Muséum National d’Histoire Naturelle, laboratoire d’Ecologie Générale, Brunoy, France). Seeds were identified either to species or to genus/family level when morphological variations between species or genera were too subtle. Unknown seeds were classified in different morphospecies. Pollen grains collected on snouts and faecal samples were cleaned by acetolysis and identified by M.P. Ledru by comparison with a collection reference (Université de Montpellier 2, France).

DATA ANALYSES Structure and composition of community Capture effort was expressed in meters of net-hour (mnh). Capture rate (number of captures/mnh) was used

hereafter as an index of abundance. One of the most informative graphical representations of community structures is the species accumulation

curve (curve representing the cumulative number of reported species as a function of the total number of captured individuals). Species richness can be estimated through mathematical extrapolation of such cumulative curves. In addition, these curves give some useful parameters such as the community diversity (slope of the curve) or completeness of sampling protocol (Moreno & Halffter, 2000, 2001; Willott, 2001). Cumulative curves were smoothed by the mean of 100 random reorganisations of capture orders (EstimateS Software, Colwell, 1997), and fitted with the Clench model (Moreno & Halffter, 2000). This model assumes that the probability of adding new species to the list decreases with the number of species already recorded, but increases over time: S(t) = at / (1+bt) where t is a measure of capture effort (in our case the number of individuals), S(t) is the predicted number of species at t, a represents the rate of increase at the beginning of the sampling, and b is species accumulation. Maximal species richness is given by the predicted asymptotic value a/b.

To describe the composition of the bat community and its spatial structuration across the landscape, we performed a correspondence analysis (Statbox Pro 5.0). This analysis relates species to their habitats. For that purpose, data were transformed using refocusing (mean fixed to 20) and reweighting (variance fixed to 1), and variables were doubled in higher and lower values according to Ponge & Delhaye (1995). Bat species defined by their capture rate were taken as main or active variables. Habitat, vertical level, bat species richness and abundance were taken as additional, i.e. passive variables.

Space, time and food partitioning among vegetarian bats In the following analyses, we focused on vegetarian bats. Each captured species was assigned a guild

according to our observations and to information available literature (Bonaccorso, 1979; Brosset & Charles-Dominique, 1990; Cosson, 1994; Kalko, 1995; Simmons & Voss, 1998; Brosset et al., 2001). We laid emphasis on the three guilds characterised by partially or totally vegetarian food habits: (1) canopy frugivores, which forage mostly on fruits that grow in the trees of the canopy and sub-canopy level of forests, (2) understorey frugivores which forage mostly on fruits of shrubby understorey plants, (3) nectarivores which consume nectar and/or pollen.

Capture rates were computed for each species/habitat combination, in order to determine habitat preferences of the most common species. As a complementary approach, re-capture rates were compared between species. High recapture rates indicated sedentary habits over the study area. Lower recapture rates described species exploiting a larger home range and/or changing foraging site more frequently (Fleming et al., 1972; Heithaus et al., 1975; Laval & Fitch, 1977; Pedro & Taddei, 1997).

After analysing spatial variations, we checked for possible interspecific differences of activity time budget. Captures were grouped per species and per 1-hour period after dusk in order to construct actograms describing the mean activity pattern of the most common species for the first half of the night.

Finally, we tackled the food partitioning question by computing diet breadth (sensu Colwell & Futuyma, 1971) and diet overlap (sensu Pianka, 1973) among vegetarian bat species. Calculation of these parameters was based on the occurrence of seeds, pollen or both items in faecal samples. When bats consume a big fruit (with big seed) or nectar, only vegetal matter is found in faeces. As a consequence, overlap and breath were only calculated on consumption of little fruit (with little seeds) or pollen. Pollen grains removed from snouts were indexed as independent items only if different from faecal pollen grains. For clarity reasons, we will not distinguish snout pollen grains from faecal grains in the following text. Calculations of niche breadth/overlap based on less than 10 items per species were disregarded.

143

RESULTS

SPECIES RICHNESS For a total capture effort of 322.3 × 103 mnh, we captured 2063 individuals that were distributed over

63 species, 32 genera, and 9 families (Table I). Phyllostomidae was the most abundant family in species number (46 species) and in individuals (88 % of captures).

According to species accumulation curves (Figure 1), species richness was estimated to a total of 67 species, including 55 species foraging in the understorey among which 26 were vegetarian. The estimated completeness of our bat survey was always higher than 94 %.

Figure 1.— Smoothed average curves produced by 100 random reorderings and estimation of species richness by extrapolation using clench model (dotted lines).

At ground level, the 10 most frequently captured species (more than 30 captures for each species)

were nine Phyllostomidae (3 Glossophaginae, 2 Carollinae, 3 Stenodermatinae, 1 Phyllostominae) and one Mormoopidae. They accounted for 81.7 % of all captures. In term of guild, nectarivores were largely dominant, followed respectively by canopy frugivores, understorey frugivores and animalivorous bats (Table I).

HABITAT PARTITIONING The first axis constituted by the correspondence analysis represented the vertical foraging gradient

while the second axis described forest structure complexity, increasing from creek corridors (open habitat) to high forest (stratified forest) (Figure 2). Bat species richness gradient fitted axis 1 while the capture rate gradient followed axis 2. Species richness increased from canopy to ground level and capture rate was higher in open habitat (creek corridors and edge of forest clearing) than in stratified forests (high forest and liana forest).

Aerial insectivores (Molossidae and Emballonuridae) were associated with sub-canopy and canopy levels whereas gleaning insectivores (Phyllostominae) tended to be associated with forest understorey (Figure 2). Considering ground level capture, most of the guilds were obviously associated with one or two particular habitats (Figure 3). Dependence between guilds and habitats was significant (Chi-square test of independence, χ² = 346; df = 18; P < 0.001).

In addition, canopy frugivores and understorey frugivores preferred creek corridors and forest edge respectively. Nectarivores, on the contrary, were ubiquitous in forest interior and avoided open habitats. However, a notable interspecific variation of foraging habitat exists within guilds (Figures 4a to 4c). Large canopy frugivores were found to more frequently forage either in open habitat, such as Artibeus jamaicensis (ARJA) or Artibeus lituratus (ARLI), or in the stratified forest, such as Artibeus obscurus (AROB) (Figure 2). In fact, it exist a strong spatial partitioning between A. obscurus on the one hand, that preferred edges and

All guilds in canopy and understorey

All guilds in understorey

Vegetarian guild in understorey

0

10

20

30

40

50

60

70

1 501 1001 1501 2001

Total number of individuals

Cum

ulat

ive

num

bero

fspe

cies All guilds in canopy and understorey

All guilds in understorey

Vegetarian guild in understorey

0

10

20

30

40

50

60

70

1 501 1001 1501 2001

Total number of individuals

Cum

ulat

ive

num

bero

fspe

cies

144

TABLE I

Bat species captured in primary forest at the Nouragues station (French Guiana). The total number of captures is indicated for each understorey habitat (HF: high forest, LF: Liana forest, EC: edge of forest clearing, CC: creek

corridor, and C: canopy). The mean forearm length and mean weight are also indicated, along with the most abundant item in the diet of vegetarian species (P: pollen, F: fruit)

Code of

species Guilda HF LF EC CC C Total Mean forearm

length (mm)±SE

Body mass (g) ± SE

Most abundant vegetal item

Glossophaginae Lionycteris spurrelli

LISP N 284 28 2 7 8 329 35.2 ± 1.9 8.6± 1.2 Eperua falcata (P)

Lonchophylla thomasi

LOTH N 118 70 2 1 191 32.1± 2.7 6.6± 0.7 Eperua falcata (P)

Anoura geoffroyi ANGE N 141 4 2 10 157 42.5± 1.5 14.7± 1.7 Eperua falcata (P)

Choeroniscus minor

CHMI N 2 2 33.3± 0.3 7.8± 0.8 Eperua falcata (P)

Carolliinae Rhinophylla pumilio

RHPU UF 95 13 38 4 4 154 33.4± 0.9 8.5± 1.6 Philodendron spp. (F)

Carollia perspicillata

CAPE UF 48 19 27 1 1 96 41.3± 1.4 15.5± 1.9 Piper spp., solanum spp. (F)

Carollia brevicauda

CABR UF 8 9 4 21 36.2± 0.8 11.4± 1.5 Piper spp., solanum spp. (F)

Stenodermatinae Artibeus jamaicensis

ARJA CF 87 6 1 42 32 168 67.8± 2.4 57.0± 7.8 Cecropia obtusa (F)

Artibeus obscurus AROB CF 87 9 23 1 28 148 60.8± 3.1 36.3± 4.7 Cecropia obtusa (F)

Artibeus lituratus ARLI CF 18 6 1 7 15 47 71.6± 2.4 67.4± 5.5 Cecropia obtusa (F)

Chiroderma villosum

CHVI CF 2 1 44 47 46.1± 1.4 22.5± 3.4 Ficus spp. (F)

Chiroderma trinitatum

CHTR CF 2 1 33 36 39.1± 1.1 14.2± 3.8 Ficus spp. (F)

Platyrrhinus helleri PLHE CF 6 7 1 1 21 36 38.3± 1.3 13.4± 1.8 Cecropia obtusa (F)

Artibeus gnomus ARGN CF 10 5 3 1 10 29 37.4± 1.2 9.9± 2.1 Ficus spp. (F) Ectophylla macconnelli

ECMA CF 17 11 28 31.4± 1.0 7.3± 1.0 Ficus spp. (F)

Ametrida centurio AMCE CF 20 20 30.6± 2.8 9.9± 1.5 Eperua falcata (P)

Artibeus concolor ARCO CF 3 2 10 15 47.9± 1.8 18.7± 2.1 Eperua falcata (P)

Uroderma bilobatum

URBI CF 7 3 10 43.3± 1.4 17.4± 1.6

Vampyressa brocki VABR CF 1 8 9 33.1± 0.7 10.0± 1.3

Sturnira tildae STTI UF 2 4 9 3 1 19 46.6± 1.4 22.2± 2.8 Philodendron spp.,C. obtusa (F)

Phylloderma stenops

PHST UF 6 2 8 69.0± 2.2 40.1± 1.5

Vampyrodes caraccioli

VACA UF 1 1 58.4 46.0

Phyllostominae Tonatia silvicola TOSI GI 28 8 2 38 59.2± 2.1 35.0± 5.0 Mimon crenulatum MICR GI 26 1 1 1 29 47.2± 1.5 11.2± 1.6 Tonatia saurophila TOSA GI 25 1 26 56.3± 1.3 24.8± 2.3

Micronycteris schmidtorum

MISC GI 13 3 1 17 34.7± 0.8 5.5± 0.3

Lonchorhina inusitata

LOIN GI 7 3 1 2 13 53.5± 1.9 16.8± 2.6

Micronycteris microtis

MIMC GI 5 2 7 33.3± 0.8 5.3± 0.3

Micronycteris hirsuta

MIHI GI 5 1 1 7 44.0± 1.0 12.8± 3.1

145

Code of species

Guilda HF LF EC CC C Total Mean forearm length (mm)±SE

Body mass (g) ± SE

Most abundant vegetal item

Mimon bennetti MIBE GI 5 1 6 52.1± 1.6 20.1± 2.6

Micronycteris megalotis

MIME GI 2 1 3 33.1± 1.3 5.3± 0.3

Tonatia schulzi TOSC GI 3 3 43.3± 0.8 17.2± 0.2

Micronycteris minuta

MIMI GI 2 2 35.3± 1.4 5.2± 0.1

Glyphonycteris daviesi

GLDA GI 1 1 2 53.3± 1.4 17.8± 1.7

Micronycteris brosseti

MIBR GI 1 1 34.7 6.3

Tonatia brasiliense TOBR GI 1 1 34.6 8.7

Phyllostomus elongatus

PHEL IN 11 2 3 16 66.4± 1.7 36.9± 4.3 Eperua falcata (P)

Phyllostomus latifolius

PHLA IN 13 13 58.6± 1.1 27.7± 2.8 Eperua falcata (P)

Phyllostomus discolor

PHDI IN 11 11 60.7± 1.9 34.1± 2.6 Eperua falcata (P)

Trinycteris nicefori TRNI IN 10 10 37.8± 0.9 8.2± 0.7 Eperua falcata (P)

Phyllostomus hastatus

PHHA IN 7 1 8 83.5± 2.2 79.8± 5.1 Eperua falcata (P)

Glyphonycteris sylvestris

GLSY IN 3 1 4 38.6± 0.7 7.5± 0.5 Eperua falcata (P)

Trachops cirrhosus TRCI C 8 1 2 11 63.2± 1.4 36.3± 4.3

Chrotopterus auritus

CHAU C 1 1 2 80.0± 4.3 61.8

Vampyrum spectrum

VASP C 1 1 102.7

Desmodontinae

Desmodus rotundus DERO H 8 1 1 10 57.0± 2.7 25.9± 1.9

Emballonuridae

Cormura brevirostris

COBR AI 2 1 7 10 46.6± 1.0 8.4± 1.3

Saccopteryx bilineata

SABI AI 2 5 7 47.0± 1.7 7.6± 1.0

Saccopteryx leptura SALE AI 2 5 7 38.3± 1.4 5.1± 0.6

Peropteryx macrotis

PEMA AI 2 2 43.6± 3.4 5.5± 2.1

Peropteryx kappleri PEKA AI 1 1 51.2 9.4

Mormoopidae

Pteronotus parnellii

PTPA AI 153 27 3 2 185 61.6± 3.0 22.2± 4.1

Thyropteridae

Thyroptera tricolor THTR AI 2 2 35.7± 1.7 3.9± 0.7

Furipteridae

Furipterus horrens FUHO AI 1 1 34.5 3.9

Molossidae

Molossus rufus MORU AI 12 12 50.8± 1.4 33.8± 8.5

Molossops abrasus MOAB AI 3 3 43.9± 2.3 26.4± 2.1

Eumops auripendulus

EUAU AI 2 2 59.2± 1.7 29.7± 0.6

Eumops hansae EUHA AI 2 2 38.8± 0.8 14.1± 0.8

Nyctinomops laticaudatus

NYLA AI 2 2 43.7± 2.5 11.7± 0.4

Molossops paranus MOPA AI 1 1 32.5

Vespertillionidae

Myotis riparius MYRI IA 3 1 2 6 35.9± 1.3 5.5± 1.0

Eptesicus chiriquinus

EPCH IA 1 2 3 47.4± 0.7 11.1± 0.9

Myotis nigricans MYNI IA 1 1 33.2± 0.3 3.7

146

Code of species

Guilda HF LF EC CC C Total Mean forearm length (mm)±SE

Body mass (g) ± SE

Most abundant vegetal item

Nb. of captures 1294 234 133 79 323 2063

Nb. of species 48 26 25 17 37 63

Effort (net-hour 10-

3) 116.2 33.8 15.7 2.8 153.8 322.3

a Guilds are as following: CF is canopy frugivore, UF is understorey frugivore, N is nectarivore, IN is insectivore-nectarivore, GI is gleaning insectivore, AI is aerial insectivore, H is hematophage and C is carnivore. high forest understorey, and on the other hand A. jamaicensis and A. lituratus that were abundant over creek corridors (Figure 4a).

Understorey frugivores Carollia brevicauda (CABR), Carollia perspicillata (CAPE) and Rhinophylla pumilio (RHPU) appeared to forage in different habitat types, respectively liana forest, and open habitat (Figure 2). Among the understorey frugivores, both C. perspicillata and R. pumilio favoured forest edge, but R. pumilio also exploited creek corridors (Figure 4b).

From the correspondence analysis, we could infer that the nectarivorous species Anoura geoffroyi (ANGE), Lionycteris spurrelli (LISP) and Lonchophylla thomasi (LOTH) foraged mainly in the understorey of stratified forest (Figure 2). However, L. thomasi appeared to prefer liana forest (Figure 4c).

Recapture probabilities also varied among species of the same guild (Table II). Lonchophylla thomasi and Artibeus obscurus individuals were significantly (or nearly) more often recaptured than the other common species of their respective guilds. Three understorey frugivores (Carollia perspicillata, Carollia brevicauda and Rhinophylla pumilio) had a high capture rate.

TABLE II Results of χ2 tests (χ2 value and probability P) comparing the proportion of recaptures to the total number of captures for each vegetarian species. Recapture rates are indicated in parentheses for each species. Abbreviations correspond to

the first two letters of genus and species names (see Table I)

Canopy frugivores Understorey frugivores Nectarivores

ARLI (6.8)

ARJA (1.9)

AROB (21.6)

CAPE (22.1)

CABR (35.7)

RHPU (30.3) LISP (4.9) LOTH

(23.8) ANGE (7.7)

ARJA 2.7 P > 0.05

AROB 3.6 P ≥ 0.05

22.9 P < 0.001

CAPE 3.4 P > 0.05

21.5 P < 0.001

0.003 P > 0.05

CABR 5.1 P < 0.05

24.4 P < 0.001

0.8 P > 0.05

0.7 P > 0.05

RHPU 6.5 P < 0.05

33.6 P < 0.001

1.3 P > 0.05

0.8 P > 0.05

0.01 P > 0.05

LISP 0.3 P > 0.05

2.4 P > 0.05

21 P <0.001

17.3 P < 0.001

15.2 P <0.001

36.7 P < 0.001

LOTH 4.4 P < 0.05

26.3 P < 0.001

0.1 P > 0.05

0.05 P > 0.05

0.5 P > 0.05

0.7 P > 0.05

26.5 P < 0.001

ANGE 0.03 P > 0.05

5.2 P < 0.05

7.2 P < 0.05

6.2 P < 0.05

7.2 P < 0.05

14.1 P < 0.001

1.2 P > 0.05

9.3 P <0.005

147

Figure 2.— Correspondence analysis separating species according to their habitat preferences. Diagrams represent the relative disposition of open space (dashed zones) and understorey, sub-canopy and canopy strata (superposed, sensu Oldeman, 1990) within each of the four sampled habitats (HF: high forest, LF: liana forest, CC: creek corridors, EC: edge of forest clearing). Species abbreviations correspond to the two first letters of genus and species names (see Table I), and the most common species of vegetarian bat guilds are shown in bold. Symbols left to species abbreviations indicate the guild to which the species belongs: • canopy frugivores; * understorey frugivores; ▲ nectarivores; - aerial insectivores and none symbol for gleaner animalivores.

Figure 3.— Repartition among guilds in different habitats of a tropical rainforest.

VABR••••

URBI••••TRNI

TRCI

TOSI

TOSATHTR-

STTI*

SALESABI-

RHPU*

PTPA-

PLHE

PHST* PHLAPHHA

PHELPHDI

PEMA

NYLA-

MYRI-MORU-

MOAB-

MISC

MIMI

MIME

MIHI

MICR

MIBE

LOTH▲

LOIN

LISP▲

GLSY

GLDA

EUHA-

EUAU-

EPCH-

ECMA••••

DERO

COBR-

CHVI••••

CHTR••••

CHMI▲ CHAU

CAPE*

CABR*

AROB••••

ARLI••••

ARJA••••

ARGN••••

ARCO••••

ANGE▲

AMCE••••

Canopy

Sub canopy

Axis 1(% variance = 4.38)

Axi

s 2

(% v

aria

nce

= 3.

74)

Edge of foret Clearing

High Forest

Richness

Capture rate

Liana Forest

Creek Corridors

Ground level

MIMC

HFHFECEC

CCCCHFHF

LFLFHFHF HFHF

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Indi

vidu

als

perc

enta

ge

Understory frugivores

Canopy frugivores

Nectarivores

Animalivores

Understory frugivores

Canopy frugivores

Nectarivores

Animalivores

Highforest

Lianaforest

Edge offorestclearing

Creekcorridors

148

Figure 4.— Mist-net capture rates (number of captures per meters of net-hour 10-4) in the four surveyed habitats. Results are presented by species (most common species only) for three guilds (a: canopy frugivores; b: understorey frugivores and c: nectarivores).

30

0

5

10

15

20

25

Carolliaperspicillata

Carolliabrevicauda

Rhinophyllapumilio

3030

0

5

10

15

20

25

Carolliaperspicillata

Carolliabrevicauda

Rhinophyllapumilio

B

30

0

5

10

15

20

25

Anourageoffroyi

Lionycterisspurrelli

Lonchophyllathomasi

3030

0

5

10

15

20

25

Anourageoffroyi

Lionycterisspurrelli

Lonchophyllathomasi

C

High forestLiana forestEdge of forest clearingCreek corridors

High forestLiana forestEdge of forest clearingCreek corridors

0

5

10

15

20

25

30

150

155

Artibeusjamaicensis

Artibeuslituratus

Artibeusobscurus

0

5

10

15

20

25

30

150

155

Artibeusjamaicensis

Artibeuslituratus

Artibeusobscurus

A

149

TIME PARTITIONING Most species exhibited a similar activity pattern during the first half of the night: a first peak during

the first two hours after sunset, between 18:30 and 20:30 and a second although less-pronounced peak between 22:00 and 23:00.

The three largest canopy frugivorous species showed activity peaks at different times of the night (Figure 5a). The activity pattern of Carollia brevicauda strikingly contrasted with that of the other two Carollinae understorey frugivores, with peaks coinciding with the slowdown periods observed for Carollia perspicillata and Rhinophylla pumilio (Figure 5b). By contrast, nectarivores had an activity mainly concentrated around the first peak (Figure 5c).

FOOD PARTITIONING A total of 787 faecal samples were collected. Seeds, pollen and insect fragments were found in

respectively 33, 31 and 21 % of the faeces. A total of 31 % faeces contained only vegetal fragments, probably corresponding to the pulp of large fruits or nectar. Seeds and pollens were classified into 60 and 37 taxa or morphospecies respectively.

Faeces contained mostly seeds of Cecropia obtusa, Philodendron spp. and Ficus spp. seeds (in 39, 13, and 13% of seeds samples respectively). Pollen species were dominated by Eperua falcata (41 %). The other two most frequently encountered species were Caryocar spp. and Inga Cayenensis (in 5 and 4% of pollens samples respectively). More detailed results on diet of vegetarian bats are given in Table I.

Estimations of niche breadth and overlap values were obtained for 13 species (Table III). The three most frequent canopy frugivores (Artibeus jamaicensis, Artibeus lituratus, and Artibeus obscurus) exploited poorly diversified plant resources (mainly Cecropia obtusa and Ficus spp.), resulting in relatively low niche breadth values (0.24 to 0.60) and fairly high niche overlap values (0.88 to 0.98). Similarly, the 3 commonest nectarivores showed both high niche breadth values (0.82 to 1.09) and high overlap values (0.82 to 0.88). Conversely, the three most abundant understorey frugivores (Carollia brevicauda, Carollia perspicillata, and Rhinophylla pumilio), fed on a variety of shrub fruits (Solanum spp., Piper spp.) and epiphytes (15 species of Cyclanthaceae and Araceae) and expressed high niche breadth values (0.93 to 1.29) and low niche overlap values (lower than 0.26). Despite the fact that bats were specialised on one type of food, understorey frugivores appeared more generalist than canopy frugivores. Mean niche overlap was lower for fruit resources than flower resources (0.31 and 0.83 respectively) (Table III).

We found many cases of “deviant” or “atypical” feeding behaviour such as small quantities of insect remains present in 4 % of faeces of frugivores and in 35 % of faeces of nectarivores. We also found some seeds ingested by nectarivores (2 %) and pollen ingested by insectivores-nectarivores (54 %, see table I). Consumption of insects by frugivores almost exclusively concerned understorey species (Carollia brevicauda, Carollia perspicillata, Rhinophylla pumilio and Sturnira tildae).

DISCUSSION

COMMUNITY STRUCTURE Bat species show a strong vertical stratification. This stratification is evident when we consider that

11 species captured in canopy have not been encountered in understorey. Conversely, 26 understorey species have never been captured in canopy. Others studies have shown even stronger discrepancies between strata (Handley, 1967; Bernard, 2001; Kalko & Handley, 2001).

This study establishes a species richness of 67 bat species in the Nouragues primary rainforest, Nevertheless, captures and identification of echolocation signals that have been conducted at the Nouragues station since 1987 have revealed the existence of 76 species in this area. Indeed, 13 others species have been contacted scarcely: Pteronotus gymnonotus, P. personatus, Sturnira lilium, Anoura caudifera, Macrophyllum macrophyllum, Molossus molossus, Molossops greenhalli, Molossops planirostris captured by nets, Diclidurus scutatus captured in its roost (Brosset & Charles-Dominique, 1990; Cockle, 1997; Charles-Dominique comm. pers.); Eptesicus furinalis, Peropteryx trinitatis, Rynchonycteris naso identified by their echolocation signal (Leblanc, 2002). Although the estimator we used seems to have under estimated species richness, nevertheless it predicted a higher asymptote than other models (Moreno & Halffter, 2000).

150

Figure 5.— Capture rates (capture per 10-5 mnh) of three most common species of canopy frugivores (a), understorey frugivores (b) and nectarivores (c). Captures rates are indicated every hour after sunset.

0

20

40

60

80

100

120

140

18h

19h

20h

21h

22h

23h

00h

01h

L.spurrelli

A. geoffroyiL. thomasi

02h

C

0

2

4

6

8

10

12

14

16

18

18h

19h

20h

21h

22h

23h

00h

01h

A. jamaicensisA. lituratusA. obscurus

02h

A

C. brevicauda (y2)

R. pumilio (y1)

C.perspicillata (y1)B

0

5

10

15

20

25

0

0.5

1

1.5

2

2.5

3

3.5

18h

19h

20h

21h

22h

23h

00h

01h

02h

y2y1

151

TABLE III

Niche breadth and niche overlap values based on the occurrence of seeds and pollen in the faeces of bats from Nouragues, French Guiana. Niche overlap values were calculated either with seeds only (in italic), pollen only (in

bold) or both seeds and pollen when the number of items was greater than 10. Niche breadth values are indicated for each species. Abbreviations correspond to the first two letters of genus and species names (see Table I)

The bat community under study shows interesting species abundance profiles which reveal the

importance of local factors in structuring communities. In understorey, the most captured species are generally frugivores (Artibeus jamaicensis, Artibeus lituratus, Artibeus obscurus, Carollia brevicauda, Carollia perspicillata, Rhinophylla pumilio), as shown by many documented studies in Northern Amazonia: French Guiana: (Simmons & Voss, 1998; Cosson et al., 1999); Brazil: (Reis & Peracchi, 1987; Bernard, 2001; Bernard et al., 2001; Bernard & Fenton, 2002); Guyana: (Lim & Engstrom, 2001); Mexico (Estrada & Coates-Estrada, 2001); Panama (Kalko et al., 1996a) or Central America (Handley et al., 1991). Nectarivores never occupy any of the first four abundance ranks. Conversely, in our study, the three most common nectarivorous bats (Anoura geoffroyi, Lionycteris spurrelli, Lonchophylla. thomasi) occupied the four primary ranks together with the insectivore Pteronotus parnellii. Moreover, the species generally found as dominant in abundance are in our case relegated to the 5th-14th ranks. The presence of caves in the northern part of our study area is undoubtedly responsible for the bat community structure depicted by mist-net captures. Brosset and Charles-Dominique (1990) estimated that several thousands Anoura geoffroyi, Pteronotus parnellii and Lionycteris spurrelli may roost in these caves. This merely illustrates how the presence of particular roosting sites may influence the local bat community profile. Similarly, particular human structures potentially accommodating large bat colonies and abundance of pioneer plant species available in secondary forest, may explain the abundance of some Artibeus and Carollia species in fragmented or altered habitat observed in most others studies (Reis & Peracchi, 1987; Brosset et al., 1996; Kalko et al., 1996a; Simmons & Voss, 1998; Cosson et al., 1999; Estrada & Coates-Estrada, 2001; Bernard & Fenton, 2002).

SPACE, TIME AND FOOD PARTITIONING Our results show high niche overlap values for the exploitation of flower resource. Heithaus et al.

(1975) found similar results in Costa Rica and interpreted them as indicators of superabundance of flower resource. A greater partitioning of fruit by bats suggests that competition for these resources may be stronger. Abundance and diversity of fruit resources may in turn play a greater role in determining vegetarian bat species diversity than diversity of pollen resources (Heithaus et al., 1975).

We found evidences of space, time or food partitioning within each of the most common guilds.

Canopy frugivores Understorey frugivore Nectarivore

AMCE ARCO ARJA ARLI AROB PLHE CABR CAPE RHPU STTI LISP LOTH ANGE 0.132 0.373 0.644 0.389 0.392 0 1.021 1.324 1.258 0.641 1.089 0.09 0.956

AMCE 0.907 0.015 0.078 0.092 0 0.266 0.654 0.514 0.132 0.841 0.745 0.924 ARCO 0.351 0.444 0.47 0.39 0.348 0.639 0.513 0.431 0.841 0.729 0.861 ARJA 0.888 0.882 0.88 0.236 0.059 0.016 0.701 0.051 0.018 0.006 ARLI 0.893 0.987 0.99 0.281 0.102 0.034 0.791 0.093 0.051 0.061 AROB 0.886 0.981 0.99 0.289 0.119 0.055 0.801 0.11 0.068 0.084 PLHE 0.883 0.989 0.997 0.267 0.06 0 0.795 0.034 0 0 CABR 0.255 0.286 0.288 0.29 0.351 0.311 0.531 0.244 0.258 0.326 CAPE 0.947 0.073 0.082 0.082 0.08 0.263 0.478 0.198 0.637 0.535 0.635 RHPU 0.839 0 0 0.011 0 0.236 0.117 0.282 0.581 0.506 0.536 STTI 0.139 0.099 0.154 LISP 0.847 0.904 0.944 0.837 0.871 LOTH 0.748 0.771 0.822 0.838 0.815 ANGE 0.94 0.91 0.849 0.883 0.825

152

CANOPY FRUGIVORES The main canopy frugivores (Artibeus jamaicensis, Artibeus lituratus, Artibeus obscurus) had very

high diet overlap values resulting from a poorly diversified diet (Cecropia obtusa and Ficus spp. mainly). These species may tolerate high diet overlap since they feed on abundant resources. Ficus species show asynchronous and massive fruiting events that provide a high quantity of resource throughout the year, so, canopy frugivores are considered as not resource limited (Morrison 1978; Bonaccorso & Gush 1987; Kalko et al. 1996b). Similarly, Cecropia obtusa fruits are produced during a long period of the year and bats have been recorded to collect every night the entire ripe parts of infrutescences of fruiting trees (Charles-Dominique 1986). Diet overlap is likely to represent species diet even if faeces analysis did not include large-seeded fruits since bats usually drop the seeds after eating the pulp, leaving no item for identification in their faeces. However, several large-seeded fruits are known to be consumed by the three large Artibeus species (e. g. Parinari spp, Couepia spp., Licania spp., Andira spp., Manilkara spp, Cariocar glabrum, Dipteryx odorata) (Charles-Dominique, pers. obs.).

In addition, species show numerous differences in foraging strategy that are likely to reduce the potential high interspecific competition generated by an absence of food partitioning. i) Artibeus obscurus appears to be more sedentary (as shown by its high recapture rate) than Artibeus jamaicensis and Artibeus lituratus. ii) It exploits different habitats than A. jamaicensis and A. lituratus. The latter species are basically restricted to creek corridors, an uncluttered habitat ideal for penetrating and prospecting forest interior for such large bats. On the contrary, A. obscurus, has been found to exploit a wider range of habitats. Compared to the other two Artibeus species, A. obscurus is smaller and is dotted with a higher manoeuvrability (sensu Norberg & Rayner, 1987) as indicated by its higher forearm length / body weight ratio. As a consequence, it can express greater aptitudes to fly and it can better exploit the cluttered understorey than its competitors. iii) A. obscurus presents a first peak activity at a time when its competitors are not foraging in the understorey but are more likely foraging in group on fruiting trees (Fleming, 1982; Handley et al., 1991). iv) A. obscurus roosts in small groups whereas A. jamaicensis individuals constitute large colonies compatible with synchronised massive displacements toward fruiting trees at dusk (Fleming, 1982; Handley et al., 1991).

UNDERSTOREY FRUGIVORES Contrary to canopy frugivores, the main understorey frugivores show low diet overlap values (<

0.236) that indicate a high food partitioning. Rhinophylla pumilio is quasi-specialised on epiphyte fruits (Araceae, Cyclanthaceae) whereas Carollia perspicillata and Carollia brevicauda primarily feed on fruits from Solanum and Piper species and only occasionally on epiphytes (see also Cosson, 1994 and Cockle, 1997). In our study, the estimated diet overlap between C. perspicillata and C. brevicauda is lower (0.351) than the values found in others studies (Gorchov et al., 1995; Cockle, 1997). R. pumilio favours creek corridors compared to forest interior, following the gradient of epiphyte abundance generally encountered in tropical forests (Cockle, 1997). However, its presence in forest edges may reflect the research of pioneer shrub fruits, the main resource consumed by C. perspicillata.

In addition to a food partitioning, we found that Carolliinae species showed different nocturnal activity.

Frugivorous bats exhibit different flight, and thus feeding, activities. These differences are likely to reduce interspecific interferences when food resources are abundant but patchily distributed (Bonaccorso, 1979). Reducing crowding at resource trees is likely an efficient feeding strategy and it may also reduce the probability of detection by arboreal or aerial predators (Humphrey & Bonaccorso, 1978).

NECTARIVORES The three most abundant nectarivores express a very high diet overlap that could result in a high

interspecific competition. However, several factors are likely to reduce this potentially high competition. Lonchophylla thomasi differs from Anoura geoffroyi and Lionycteris spurrelli by i) its habitat preferences (it forages in liana forest while the others forage in high forest), ii) its roosting habits (it roosts in hollow trees while the others roost in cave (Simmons & Voss, 1998; Brosset et al., 2001)), iii) its higher fidelity to foraging habitat (as indicated by the high recapture rate) and iv) its diet, which includes more insects than that of its competitors. The relative importance of insects in its diet (61.3 % of faeces) may result from a combination of energetic constraints and competitive exclusion. Bats complement their diet with insects, which are rich in protein (Gardner, 1977; Herrera et al., 2001). L. thomasi also differs from the other two species by v) its exploitation of forest habitats. Creek corridors may particularly attract A.geoffroyi and L. spurrelli because Eperua falcata trees, which produce their main food resource, are more abundant along

153

creeks (Riera & Joly, 1996). Although Eperua falcata also constitutes the core diet of L. thomasi, this bat is more restricted to liana forest. It has a higher manoeuvrability than its competitors and is associated with the most closed vegetation. It can therefore exploit a habitat which is less attractive to its competitors.

We have not found any evidence of a time partitioning between nectarivorous species. However, our results show that these species tend to concentrate their activity in the early period of the night. A similar pattern has been found for frugivorous bats (Marinho-Filho & Sazima, 1989). This may be due to the fact that the quantity of favoured fruits or flowers tends to decrease over the night. This pressure may have selected for a concentration of activity in the first hours of the night (Marinho-Filho & Sazima, 1989).

Our study confirms that rainforests house an important community of bat species. By combining different methods (captures, habitat characterisation, and faeces analysis), we analysed in details the different factors that could explain species coexistence. Within a guild, species exhibit numerous and complex differences in their diet composition, feeding activities, foraging habits, type of habitats explored.

Given the crucial role of bats in the tropical rainforest ecosystem and the increasing habitat degradation observed in tropical forests (Wilson, 1988; FAO, 1997), it is important to provide a ‘reference’ picture of a non-disturbed bat community, in order to understand the impact of disturbances on bat communities and to elaborate efficient conservation programs.

ACKNOWLEDGMENTS

We gratefully thank Gilles Peroz for his technical assistance, Sylvie Jouard for seed identification and Marie-Pierre Ledru for pollen identification. Many thanks to Doris Gomez, Konstantinos Theodorou, Martine Perret, Sandra Ratiarison, and Jean-François Ponge for their helpful suggestions. This study was conducted at the Nouragues Station, UPS 656 CNRS.

REFERENCES ANTHONY, E.L.P. (1988).— Age determination in bats. Pp. 47-58, in: Kunz TH (ed.), Ecological and behavioral

methods for the study of bats. Smithsonian Institution Press, Washington D. C. ARITA, H.T. (1997). Species composition and morphological structure of the bat fauna of Yucatan, Mexico. J. Anim.

Ecol., 66: 83-97. AUBREVILLE, A. (1938). La forêt coloniale: les forêts de l'Afrique Occidentale Française. Annales de l'Académie des

Sciences Coloniales, 9: 1-245. BERNARD, E. (2001). Vertical stratification of bat communities in primary forest of Central Amazon, Brazil. J. Trop.

Ecol., 17: 115-126. BERNARD, E., ALBERNAZ, A.L.K.M. & MAGNUSSON, E. (2001). Bat species composition in three localities in the

Amazon Basin. Studies on Neotropical Fauna and Environment, 36: 177-184. BERNARD, E. & FENTON, M.B. (2002). Species diversity of bats (Mammalia: Chiroptera) in forest fragments, primary

forests, and savannas in central Amazonia, Brasil. Can. J. Zool., 80: 1124-1140. BONACCORSO, F.J. (1979). Foraging and reproductive ecology in a Panamanian bat community. Bull. Florida State

Mus., Biol. Sci., 24: 359-408. BONACCORSO, F.J. & GUSH, T.J. (1987). Feeding behaviour and foraging strategies of captive Phyllostomid fruit bats

an experimental study. J. Anim. Ecol., 56: 907-920. BROSSET, A. & CHARLES-DOMINIQUE, P. (1990). The bats from French Guiana: a taxonomic, faunistic and

ecological approach. Mammalia, 54: 509-560. BROSSET, A., CHARLES-DOMINIQUE, P. & Cockle, A. (2001). The bat community. PP. 115-120, In: Bonger F,

Charles-Dominique P, Forget P-M & Théry M (eds.), Nouragues. Dynamics and plant-animal interactions in neotropical rainforest. Kluwer Academic Publishers.

BROSSET, A., CHARLES-DOMINIQUE, P., COCKLE, A., COSSON, J.F. & MASSON, D. (1996). Bat communities and deforestation in French Guiana. Can. J. Zool., 74: 1974-1982.

CHARLES-DOMINIQUE, P. (1986). Inter-relations between frugivorous vertebrates and pioneer plants: Cecropia, birds and bats in French Guyana. Pp. 119-135, in: Estrada A & Fleming TH (eds.), Frugivores and seed dispersal. Dr W. Junk Publishers.

CHARLES-DOMINIQUE, P. (1995). Interactions plantes-animaux frugivores, conséquences sur la dissémination des graines et la régénération forestière. Rev. Ecol. (Terre Vie), 50: 223-235.

CHARLES-DOMINIQUE, P., BROSSEt, A. & JOUARD, S. (2001). Atlas des chauves-souris de Guyane. Museum National d'Histoire Naturelle, Paris.

COCKLE, A. (1997). Modalités de dissemination et d'établissement de lianes de sous-bois (Cyclanthaceae et Philodendron) en foret Guyanaise.Thèse de doctorat, Univesité de Paris VI.

154

COLWELL, R.K. (1997). EstimateS: Statistical estimation of species richness and shared from samples. http://viceroy.eeb.uconn.edu/estimates.

COLWELL, R.K. & FUTUYMA, D.J. (1971). On the measurement of niche breadth and overlap. Ecology, 53: 567-576. COSSON, J.F. (1994). Dynamique de population et dispersion de la chauve-souris frugivore Carollia perspicillata en

Guyane Française. Thèse de doctorat, Université Paris Sud XI, Paris. COSSON, J.F., RINGUET, S., CLAESSENS, O., MASSARY, J.C., DALECKY, A., VILLIERS, J.F., GRANJON, L. & PONS, J.M.

(1999). Ecological changes in recent land-bridge islands in French Guiana with emphasis on vertebrate communities. Biol. Cons., 91: 213-222.

DE FORESTA, H., CHARLES-DOMINIQUE, P., ERARD, C. & PREVOST, M.F. (1984). Zoochorie et premier stades de la régénération naturelle après coupe en forêt Guyanaise. Rev. Ecol. (Terre Vie), 39: 369-400.

EMMONS, L. & FEER, F. (1990). Neotropical rainforest mammals. A field guide. The University of Chicago Press, Chicago.

ESTRADA, A. & COATES-ESTRADA, R. (2001). Species composition and reproductive phenology of bats in a tropical landscape at Los Tuxtlas, Mexico. J. Trop. Ecol., 17: 627-646.

FAO. (1997). State of the world's forests. Forestry Department, Food and Agricultural Organization, Rome. FENTON, M.B., ACHARYA, L., AUDET, D., HICKEY, M.B.C., MERRIMAN, C., OBRIST, M.K., SYME, D.M. & ADKINS, B.

(1992). Phyllostomid bat (Chiroptera: Phyllostomidae) as indicators of habitat disruption in the Neotropics. Biotropica, 24: 440-446.

FLEMING, T.H. (1982). Foraging strategies of plant-visiting bats. In: Kunz TH (ed.), Ecology of bats. Plenum Press, New York and London.

FLEMING, T.H., HOOPER, E.T. & WILSON, D.E. (1972). Three central American bat communities: structure, reproductive cycles, and movement patterns. Ecology, 53: 555-569.

GARDNER, A.L. (1977). Feeding habits. In: Baker RJ, Jones JK & Carter DC (eds.), Biology of bats of the New World family Phyllostomatidae. Part II. Special publications. The Museum Texas Tech University.

GORCHOV, D.L., CORNEJO, F., ASCORRA, C.F. & JARAMILLO, M. (1995). Dietary overlap between frugivorous birds and bats in the Peruvian Amazon. Oikos, 74: 235-250.

HANDLEY, C.O. (1967). Bats of the canopy of an Amazonian forest. Atlas do Simposio sôbre a biota Amazônica, 5: 211-215.

HANDLEY, C.O., WILSON, D.E. & GARDNER, A.L. (1991). Demography and natural history of the common fruit bat, Artibeus jamaicensis, on Barro Colorado Island, Panamà. Smithsonian Contribution to Zoology. Number 511 edn, Washington, D. C.

HEITHAUS, E.R. (1982). Coevolution between bats and plants. Pp. 327-367 in: Kunz TH (ed.), Ecology of bats. Plenum Publishing Corp.

HEITHAUS, E.R., FLEMING, T.H. & OPLER, P.A. (1975). Foraging patterns and resource utilization in seven species of bats in a seasonal tropical forest. Ecology, 56: 841-854.

HERRERA, L.G., HOBSON, K.A., MIRON, L., RAMIREZ, N., MENDEZ, G. & SANCHEZ-CORDEREO, V. (2001). Sources of protein in two species of phytophagous bats in a seasonal dry forest: evidence from stable-isotope analysis. J. Mammal., 82: 352-361.

HUMPHREY, S.R. & BONACCORSO, F.J. (1978). Population and community ecology. Pp. 406-441 in: Baker RJ, Jones JK & Carter DC (eds.), Biology of the New world family Phyllostomatidae. Part III. Texa Tech Press, Lubbock.

KALKO, E.K.V. (1995). Echolocation signal design, foraging habitats and guild structure in six Neotropical sheath-tailed bats (Emballonuridae). Symp. zool. Soc. London, 67: 259-273.

KALKO, E.K.V. (1997). Diversity in tropical bats. Pp. 13-43, in: Ulrich H (ed.), Tropical biodiversity and systematics. Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn.

KALKO, E.K.V. & HANDLEY, C.O. (2001). Neotropical bats in the canopy: diversity, community structure, and implications for conservation. Plant Ecol., 153: 319-333.

KALKO, E.K.V., HANDLEY, C.O. & HANDLEY, D. (1996a). Organization, diversity, and long-term dynamics of a neotropical bat community. Pp. 503-553, in: Long-term studies of vertebrate communities. Academic Press.

KALKO, E.K.V., HERRE, E.A. & HANDLEY, C.O. (1996b). Relation of fig fruit characteristics to fruit-eating bats in the New and Old World tropics. J. Biogeography, 23: 565-576.

LAVAL, R.K. & FITCH, H.S. (1977). Structure, movements and reproduction in three Costa Rica bat communities. Museum of Natural History of the University of Kansas. Occasional papers., 69: 1-28.

LEBLANC, F. (2002). Caractérisation des cris ultrasonores des chiroptères de Guyane. D.E.R., Faculté des Sciences et techniques de Limoges.

LIM, B.K. & ENGSTROM, M.D. (2001). Bat community structure at Iwokrama forest, Guyana. J. Trop. Ecol., 17: 647-665.

MARINHO-FILHO, J. (1991). The coexistence of two frugivorous bat species and the phenology of their food plants in Brazil. J. Trop. Ecol., 7: 59-67.

155

MARINHO-FILHO, J. & SAZIMA, I. (1989). Activity patterns of six Phyllostomid bat species in southeastern Brazil. Rev. Brasil. Biol., 49: 777-782.

MARSHALL, A.G. (1983). Bats, flowers and fruit: evolutionary relationships in the Old World. Biol. J. Linn. Soc., 20: 115-135.

MCNAB, B.K. (1971). The structure of tropical bat faunas. J. Mammal., 52: 352-358. MORENO, C.E. & HALFFTER, G. (2000). Assessing the completeness of bat biodiversity inventories using species

accumulation. J. Appl. Ecol., 37: 149-158. MORENO, C.E. & HALFFTER, G. (2001). On the measure of sampling effort used in species accumulation curves. J.

Appl. Ecol., 38: 487-490. MORRISON, D.W. (1978). Foraging ecology and energetic of the frugivorous bat Artibeus jamaicensis. Ecology, 59:

716-723. NORBERG, U.M. & RAYNER, J.M.V. (1987). Ecological morphology and flight in bats (Mammalia; Chiroptera): wing

adaptation, flight performance, foraging strategy and echolocation. B. Biological sciences, 316: 335-427. OLDEMAN, R.A.A. (1990). Forest: elements of silvology. Springer-Verlag. PEDRO, W.A. & TADDEI, V.A. (1997). Taxonomic assemblage of bats from Panga Reserve, southern Brazil:

abundance patterns and trophic relations in the phyllostomidae (Chiroptera). Bol. Mus. Mello Leitao, 6: 3-21. PIANKA, E.R. (1973). The structure of lizard communities. Ann Rev. Ecol. Syst., 4: 53-74. PONCY, O., SABATIER, D., PREVOST, M.F. & HARDY, I. (2001). The lowland high rainforest: structure and tree

species diversity. Pp. 31-46, in: Bonger F, Charles-Dominique P, Forget P-M & Théry M (eds.), Nouragues. Dynamic and plant-animal interactions in a neotropical rainforest. Kluwer Academic Publishers.

PONGE, J.F. & DELHAYE, L. (1995). The heterogeneity of humus profiles and earthworm communities in a virgin beech forest. Biol. Fertil. Soils, 20: 24-32.

RACEY, P.A. (1988). Reproductive assessment in bats. Pp. 31-45, in: Kunz TH (ed.), Ecological and behavioral methods for the study of bats. Smithsonian Institution Press, Washington D. C.

REIS, N.R. & PERACCHI, A.L. (1987). Quiropteros da regiao de Manaus, Amazonias, Brasil (Mammalia, Chiroptera). Bol. Mus. Par. Emilio Goeldi, 3: 161-182.

RIÉRA, B. (1998). Caractérisation d'une mosaïque forestière et de sa dynamique en forêt tropicale humide sempervirente. Biotropica, 30: 251-260.

RIERA, B. & JOLY, A. (1996). Eperua falcata: un cas de distribution spatiale en relation avec les changements climatiques récents. Pp. 247-248, in: Dynamique à long terme des écosystèmes forestiers intertropicaux. Actes Symp., ORSTOM/CNRS, Bondy.

SAUNDERS, M.B. & BARCLAY, R.M. (1992). Ecomorphology of insectivorous bats: a test of predictions using two morphologically similar species. Ecology, 73: 1335-1345.

SIMMONS, N.B. & VOSS, R. (1998). The mammals of Paracou, French Guyana: a neotropical lowland rainforest fauna. Part I. Bats. Bull. Amer. Mus. Nat. Hist., 237: 0-219.

WILLIG, M. & MOULTON, M.P. (1989). The role of stochastic and deterministic processes in structuring neotropical bat communities. J. Mammal., 70: 323-329.

WILLOTT, S.J. (2001). Species accumulation curves and the measure of sampling effort. J. Appl. Ecol., 38: 484-486. WILSON, E.O. (1988). The current state of biological diversity. In: E. O. Wilson and F. M. Peter (eds), Biodiversity.

National Academic Press, Washington, D. C.

156

ANNEXE 2 :COMMUNICATION AFFICHÉE PRÉSENTÉE À L’ATBC 2003.

(Annual Symposium of the Association for Tropical Biology and Conservation, Aberdeen, Ecosse, UK). ARE ANIMAL-DISPERSED SEEDS LESS EFFICIENTLY SCATTERED IN FOREST FRAGMENTS? A TEST USING BAT FRUIT-PLANTS IN FRENCH GUIANA. Mickaël HENRY, Jean-Marc PONS and Pierre CHARLES-DOMINIQUE Muséum National d’Histoire Naturelle 4, Av. du petit château ; 91800 Brunoy ; FRANCE Abstract We tested the prediction that fragmentation has a negative effect on fluxes of animal-dispersed seeds by focusing on fruit-plants consumed by bats. Density and diversity of seed rain were lower in the small forest fragments than in the control mainland sites, raising questions about plant recruitment efficiency in fragmented landscapes. Introduction Interactions between fruit plants and seed dispersers constitute a form of mutualism. Fruits provide dispersers with energy while dispersers carry the seeds away from parent trees and deposit them in places potentially more suitable for their germination and growth. Disturbances like fragmentation were shown to affect faunas, resulting in impoverished animal communities. Thus, the mutualistic plant-animal interactions such as seed dispersal are likely to be altered in fragmented landscapes. We tested the prediction that fragmentation has a negative effect on fluxes of animal-dispersed seeds by using the chiropterochorous seeds (i.e. bat-dispersed seeds) as a study model. Indeed, fruit bats generally swallow the tiny seeds and defecate them not only under feeding roosts but also while flying. This dispersal mode generates a “seed rain” that reaches a variety of sites. This is one of the reasons why bats are thought to hold a crucial role in seed dispersal and hence in the regeneration process of many tropical plants. The bat-generated seed rain can be sampled by the mean of seed collectors at ground level. In this study, “Chiropterochorous seeds” will refer to the seeds likely to be ingested by bats according to the seed collection of the French Museum of Natural History, built from a large amount of bat fecal samples from French Guiana. Material and methods The study was undertaken at St-Eugène (French Guiana, South America), a pristine forest area recently fragmented by the completion of a hydroelectric dam (1994/95). The subsequent flooding created many forest patches of various size. The seed rain was sampled using 54 seed collectors (Fig. 1) distributed over 6 survey sites: 3 small forest fragments (2 – 7.5 ha) and 3 control sites on the adjacent continuous forest (Fig. 2). Collectors were checked every 1 – 3 days during a total of 15 days in november 2002 (dry season). All feces droplets containing chiropterochorous seeds were collected. In case of rain wash out, the whole collector contents were removed and later examined in the lab. In order to relate values of seed rain density to bat abundance, bats were mist-netted in the same 6 sites (36 mist-net-hours per site). Results and Discussion A total of 1104 tiny chiropterochorous seeds belonging to 7 species were collected. Two of these species were unknown but included in the

157

chiropterochorous species because they were found in bat feces during the study. These are indexed as “morphospecies 1 and 2”. The cumulative species richness curve (Fig. 3) obtained from mainland samples is clearly steeper and reaches a higher asymptotic value than does the curve obtained from forest fragments. This underlines a substantial depletion of seed richness and diversity in fragments, and is to be related to a sharp decrease of seed rain density generalized among taxa (Fig. 4). These contrasting seed rain patterns are consistent with the prediction that fragmentation alters the dispersal of zoochorous seeds. Furthermore, a significant correlation between the abundance of fruit-bats and the seed rain density (Fig. 5) suggests that the decrease and impoverishment of seed fluxes in forest fragments may partly result from the alteration of bat communities (that was clearly demonstrated on the same study area by Cosson et al., 1999 1). This study should be replicated and/or extended to other taxa (e.g. seeds dispersed by birds and rodents) so as to better estimate the extent to which dispersal limitation could turn to dispersal failure in small forest fragments. 1 COSSON J.-F., J.-M. PONS and D. MASSON, 1999. Effects of fragmentation on frugivorous and nectarivorous bats in French Guiana. Journal of Tropical Ecology, 15:515-534.

Fig. 1. One of the 54 collectors used to sample the seed rain. Collectors consisted in a 80×80 cm plastic sheet (0.64 m2) tightened between 4 trees, with a central metallic filter for water draining off.

158

Fig. 2. Study area: the fragmented forest of St-Eugène. Open and closed symbols indicate mainland and fragment survey sites respectively. Fig. 3. Cumulative seed species richness curves as a function of the number of seed trap replicates. Curves were smoothed by the mean of 100 randomizations and fitted with a Clench model curve (continuous lines). Numbers refer to the asymptotic values.

1 km

N

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30

p

� Mainland � Fragments 7.7

5.4

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30

p

� Mainland � Fragments 7.7

5.4

Cum

ulat

ive

spec

ies r

ichn

ess

Nb of seed traps

159

Fig. 4. Comparison of the seed rain densities (nb of seeds per 15 days – collector) estimated in the mainland and on small forest fragments. Asterisks refer to the results of non parametric Mann-Whitney comparison tests. Fig. 5. Correlation between seed rain density (nb of seeds per 15 days – survey site) and fruit-bat abundance (nb of captures per 100 m of net – night) over the study area. Each dot represents one of the 6 surveyed sites. Note that bats were much more abundant in mainland sites than in forest fragments.

0

1

2

3

4 � Mainland � Fragments*P ≤ 0.05 **P≤0.01 ***P ≤0.001

All see

ds **

All Cec

ropia

seeds

***

Morpho

specie

s 1 *

Cecrop

ia ob

tusa *

Cecrop

ia sci

adop

hylla

**Othe

r see

ds *

Thor

acoc

arpus

bisse

ctus,

Solan

um su

rinam

ense,

S.arg

entum

, Morp

hosp

. 2

0

1

2

3

4 � Mainland � Fragments*P ≤ 0.05 **P≤0.01 ***P ≤0.001

All see

ds **

All Cec

ropia

seeds

***

Morpho

specie

s 1 *

Cecrop

ia ob

tusa *

Cecrop

ia sci

adop

hylla

**Othe

r see

ds *

Thor

acoc

arpus

bisse

ctus,

Solan

um su

rinam

ense,

S.arg

entum

, Morp

hosp

. 2

01234567

0 1 2 3 4 5 6 7

� Mainland � Fragments Pearson = 0.858 ; P = 0.029

Log (Bat abundance +1)

Log

(nb

of se

eds +

1)

160

ANNEXE 3 : LISTE DES COMMUNICATIONS SCIENTIFIQUES EN CONGRES.

1. HENRY, M., 2005. Seed dispersal patterns of keystone bat and bird plants in a

pristine forest of French Guiana. Communication orale. Annual Symposium of the

Association for Tropical Biology and Conservation, Uberlândia, Brésil.

2. HENRY, M. et E.K.V. KALKO, 2005. Effect of lactation on the nocturnal activity

pattern of a small neotropical fruit bat, Rhinophylla pumilio (Carolliinae), in

French Guiana. Communication orale. 18th Annual Conference of the Society for

Tropical Ecology (GTÖ), Berlin, Allemagne.

3. HENRY, M., J.-M. PONS et J.-F. COSSON, 2004. Modifications d’une communauté

de chauves-souris frugivores suite à la fragmentation d’une forêt tropicale

primaire:dix ans d’inventaires au barrage de Petit-Saut (Guyane Française).

Communication orale. 27ème réunion du Groupe de Biologie et Génétique des

Populations, Paris, France.

4. HENRY, M., J.-M. PONS et P. CHARLES-DOMINIQUES, 2003. Are animal-

dispersed seeds less efficiently scattered in forest fragments? A test using bat

fruit-plants in French Guiana. Communication affichée. Annual Symposium of the

Association for Tropical Biology and Conservation, Aberdeen, Ecosse, UK.

5. COSSON, J.-F., J.-M. PONS, M. HENRY et R. KIRSCH, 2002. Effects of forest

fragmentation on understorey frugivorous bats in St Eugène, French Guiana.

Communication orale. Annual Symposium of the Association for Tropical Biology

and Conservation, Panama City, Panama.

THE DECLINE OF FRUIT BAT POPULATIONS IN FRAGMENTED NEOTROPICAL FORESTS – CONSEQUENCES ON SEED DISPERSAL.

ABSTRACT Bats are involved in the pollination and seed dispersal of many tropical plants. Yet, they are worldwide threatened by the increasing forest fragmentation processes resulting from urbanisation and changes in agricultural practices. This study aims at investigating the causes of the fragmentation-sensitivity of some understory fruit bats and the possible consequences on seed rain patterns in fragmented rainforests of French Guiana. These topics were examined in three separate chapters whose objectives were (i) to characterize the range size and foraging strategy of a model species from the understory fruit bat guild, with a particular emphasis on the physiological constraints of lactation as a critical aspect of breeding success; (ii) to compare the respective contributions of habitat connectivity and food availability in maintaining populations of understory fruit bats in a fragmented forest; and (iii) to determine how the seed rain pattern would be in turn affected in case of reduction in fruit bat activity. In the first chapter, the epiphyte-specialist Rhinophylla pumilio was chosen as a study model. Its foraging strategy in an undisturbed forest was mostly restricted to short (40-120 m) search flights in a single small foraging area (3.5-14.1 ha), which seems well suited to the scattered distribution of their main food resource. Lactating females most probably transported their young and nursed it in their foraging area at night. This was associated with a decrease in flight distances and size of foraging area, and an increase in total flight time throughout the night. Because it is based on search flights, the overall foraging strategy of R. pumilio, and specially in lactation, appears incompatible with the need to regularly cross expanses of inhospitable matrix in fragmented forests. Conformingly, the second chapter has shown that in a fragmented forest, R. pumilio individuals were significantly less abundant in areas of lower forest connectivity, in spite of unmodified availability of the epiphyte resource. Yet, they maintained a reproductive activity within the fragmented forest, leading to unchanged abundance over the 10 years since fragmentation, probably thanks to small area requirements. On the contrary, the other understory fruit bats eventually disappeared from fragments despite the significant recrudescence of their food resource, mostly including pioneer shrubs. In both cases, food availability did not compensate for the loss of forest connectivity. The seed rain modifications following such a reduction in bat abundance were investigated in the third chapter through an experimental disturbance of bat activity consisting in massive mist-net captures over a seed rain sampling site (0.12 ha). A comparison with a control site led to the conclusion that a reduction in bat activity results in significantly lower seed rain diversity and density, but that seed rain uniformity remained unchanged. Bats appeared as efficient dispersers, ensuring spatial uniformity of seed dissemination in spite of disturbances affecting their abundance This study designated the loss of forest connectivity per se as the proximal extrinsic factor responsible for the fragmentation-sensitivity of understory fruit bats, and stressed the possible contribution of their foraging strategy, unsuited to tolerate habitat disruptions within their foraging area. Enhancing forest connectivity in fragmented forests should greatly promote bat abundance and seed rain density and diversity. Key words: phyllostomid fruit bats – French Guiana – tropical rainforest – Rhinophylla pumilio – foraging strategy – lactation – habitat fragmentation – landscape analysis – seed rain – seed dispersal limitation.

RÉSUMÉ Les chauves-souris participent activement à la pollinisation et la dispersion des graines de nombreuses plantes tropicales. Cependant, elles sont menacées par la fragmentation grandissante des forêts, résultant du développement de l’urbanisation et des nouvelles pratiques agricoles. La présente étude a pour but de documenter les causes de la vulnérabilité des chauves-souris frugivores de sous-bois face à la fragmentation, et les conséquences que cela implique sur le patron de pluie de graines en forêt fragmentée de Guyane Française. Cette problématique est abordée en trois chapitres distincts, dont les objectifs sont : (i) de caractériser les capacités de mouvement et la stratégie de quête alimentaire d’une espèce modèle appartenant à la guilde des frugivores de sous-bois, en mettant l’accent sur les contraintes physiologiques de l’allaitement puisque celles-ci sont directement reliées au succès reproducteur, (ii) de confronter les contributions respectives de la connectivité de la forêt et de la disponibilité alimentaire au maintien des populations des frugivores de sous-bois en milieu fragmenté, (iii) de déterminer comment le patron de pluie de graines est à son tour affecté par la réduction de l’abondance des chauves-souris frugivores. Dans le premier chapitre, la chauve-souris Rhinophylla pumilio spécialisée sur les fruits d’épiphytes a été utilisée comme modèle d’étude. Sa stratégie de quête alimentaire en forêt intacte se limite à un enchaînement de petits vols de recherche (40-120 m) concentrés sur une seule aire d’alimentation de petite taille (3,5-14,1 ha), ce qui semble bien adapté à sa ressource très parsemée dans l’espace. Les femelles allaitantes transportent probablement leur progéniture jusqu’à leur aire d’alimentation où elles les allaitent pendant la nuit. Parallèlement, elles réduisent leurs distances de vol et leur aire d’alimentation, mais augmentent le temps passé à voler. Ce patron d’activité semble incompatible avec la nécessité de traverser régulièrement des zones de matrice inhospitalière qui fragmentent la forêt. Conformément à ces observations, le second chapitre a mis en évidence une diminution significative de l’abondance de R. pumilio avec la perte de connectivité de la forêt, malgré une disponibilité alimentaire inchangée. Cependant, les individus ont maintenu une activité de reproduction dans la forêt fragmentée, et leur abondance n’a pas diminué au long des 10 premières années de la fragmentation, sans doute grâce à la petite taille de leur domaine vital. Les autres frugivores de sous-bois, au contraire, ont disparu des fragments en dépit de la recrudescence des arbustes à fruits constituant leur principale ressource alimentaire. Dans les deux cas, la disponibilité alimentaire n’a pas compensé la perte de connectivité de la forêt. Les modifications de la pluie de graines subséquentes à une telle réduction de l’abondance des chauves-souris frugivores ont été étudiées dans le troisième chapitre par le biais d’une perturbation expérimentale de l’activité des chauves-souris. Celle-ci consistait en un effort massif de captures au filet déployé sur un site d’échantillonnage de pluie de graines (0,12 ha). La comparaison avec un site contrôle a mis en évidence une diminution significative de la diversité et de la densité de la pluie de graines, mais pas de son uniformité, soulignant l’efficacité de la dispersion par les chauves-souris. Globalement, cette étude désigne la perte de connectivité de la forêt comme principal facteur extrinsèque de la vulnérabilité des frugivores de sous-bois à la fragmentation, et suggère que leur stratégie de quête alimentaire n’est pas adaptée aux ruptures d’habitat dans leurs aires d’alimentation. Rétablir une meilleure connectivité dans les forêts fragmentées devrait favoriser l’abondance des chauves-souris et la diversité et la densité de la pluie graines. Mots clés: chiroptères frugivores phyllostomidés – Guyane française – forêt tropicale humide – Rhinophylla pumilio – stratégie de quête alimentaire – allaitement – fragmentation de l’habitat – analyse de paysage – pluie de graines – limitation de la dispersion des graines.