19
Comparative morphology of populations of  Monstera  Adans. (  Araceae) from natural forest fragments in Northeast Brazil using elliptic Fourier Analysis of leaf outlines I. M. Andrade 1 , S. J. Mayo 2 , D. Kirkup 2 & C. Van Den Berg 1 Summary.  A comparative study of the leaf outline morphometrics of  Monstera adansonii  var.  klotzschiana ,  M. adan- sonii  var. laniata  and  M. praetermissa  was carried out. The study focused on populations in isolated montane humid (brejo) forests of Ceará state in Northeast Brazil and compared them with populations from Amazonia and the Brazilian Atlantic Forest. Digitised outlines were prepared from a total of 1,695 eld-collected leaf images from 20 populations, and elliptic Fourier analysis was used to generate matrices of coef cients, from which six shape  variables (principal components) were extracted using Principal Components Analysi s. Intra-population variability and inter-population differences were analysed with multivariate distance methods. Separate analyses were carried out for each of three leaf size classes (juvenile, submature, mature) because of the strong heteroblasty typical of this genus. Juvenile leaves were the least variable size class within populations of  M. adansonii  var.  klotzschiana . The shape variables expressed very similar types of variation in all three size classes. The Ceará brejo populations of  M. adansonii  var.  klotzschiana  showed signicant differences between mature leaf outlines in all pairwise comparisons; the Pacatuba population was the most distinct. The Ceará populations did not cluster together exclusively. In all three size classes, populations clustered together into their taxonomic groups, most clearly so in mature leaves. No correlation between morphological and geographic distance matrices was found, nor between morphological and molecular distance. The study showed that leaf outline shape is a practicable and useful quantitative trait for studying morpho logica l variability at species, varieta l and population levels. Key Words.  Araceae , Brazil, brejo forest, Ceará, elliptic Fourier Analysis, Monstera , morphometrics. Introduction The context of this study is the  brejo forest  a mosaic of isolated humid for est area s in Nor theast Brazil scattered within the regions predominant semi- ari d caatinga biome wherever rainfall broug ht by humid winds from the Atlantic Ocean provides suitable microclimates in mountainous terrain (Sales et al.  1998; Rodal  et al.  1998; Sampaio  et al .  2002; Borges-Nojosa & Caramaschi  2003; Pôrto  et al.  2004;  Andrade  et al .  2007). The brej o forests are today classied as part of the Brazilian Atlantic forest in the category of seasonal semi-deciduous montane forest (IBGE  1985,  2005). In Ceará State there are seven such areas (Figuei redo 1997) situated roughly halfway bet ween Amaz oni a and the Atlanti c fore st, Brazil s major humid forest regions. The brejo for ests are of part icular bio geo grap hic int ere st bec aus e the y are tho ugh t to be nat ura l frag men ts rep res ent ing rel ict s of previo us expansions of the Atl ant ic and Ama zon for est s (Ab Sáber  1982; Andr ade- Lima 1982; Borg es-No josa & Caramaschi  2003). The results of Olive ira  et al .  (1999), in a st udy of a Quater na ry polle n prole from cen tra l Bah ia (re lati vel y nea r the focus of our study in Ceará), suggest that humid and cooler climates existed in Northeast Brazil until around 4,500 yrs BP, and they discussed two hypotheses which postulate corridors of humid forest linking the Amazon and Atlantic forests; one along the northeastern coast of Brazil from Maranhão to the region around Salvador in Bah ia and the oth er acr oss centr al Bra zil thr oug h a network of gallery fores ts. The exist ence of such connections and associated patterns of gene  ow over Accepted for publication June 2008. 1 Programa de Pós-Graduação em Botânica, Departamento de Ciências Biológicas, Universidade Estadual de Feira de Santana (UEFS), Avenida Universitária s/n, 44031-460, Feira de Santana, Bahia, Brazil. 2 Herbarium, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3AB, UK. e-mail: [email protected] Electronic supplementary material The online version of this article (doi:10.1007/s12225-008-9032-z) contains supplementary material, which is available to authorized users. KEW BULLETIN VOL.  63: 193211 (2008) © The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

Embed Size (px)

Citation preview

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 1/19

Comparative morphology of populations of Monstera  Adans.

( Araceae) from natural forest fragments in Northeast Brazilusing elliptic Fourier Analysis of leaf outlines

I. M. Andrade1, S. J. Mayo2, D. Kirkup2 & C. Van Den Berg1

Summary. A comparative study of the leaf outline morphometrics of  Monstera adansonii  var.  klotzschiana ,  M. adan- 

sonii  var. laniata  and  M. praetermissa  was carried out. The study focused on populations in isolated montane humid(brejo) forests of Ceará state in Northeast Brazil and compared them with populations from Amazonia and theBrazilian Atlantic Forest. Digitised outlines were prepared from a total of 1,695  field-collected leaf images from 20populations, and elliptic Fourier analysis was used to generate matrices of coef ficients, from which six shape

 variables (principal components) were extracted using Principal Components Analysis. Intra-population variability and inter-population differences were analysed with multivariate distance methods. Separate analyses were carriedout for each of three leaf size classes (juvenile, submature, mature) because of the strong heteroblasty typical of this genus. Juvenile leaves were the least variable size class within populations of  M. adansonii  var.  klotzschiana . Theshape variables expressed very similar types of variation in all three size classes. The Ceará brejo populations of  M.

adansonii  var.  klotzschiana  showed significant differences between mature leaf outlines in all pairwise comparisons;the Pacatuba population was the most distinct. The Ceará populations did not cluster together exclusively. In allthree size classes, populations clustered together into their taxonomic groups, most clearly so in mature leaves. Nocorrelation between morphological and geographic distance matrices was found, nor between morphological andmolecular distance. The study showed that leaf outline shape is a practicable and useful quantitative trait forstudying morphological variability at species, varietal and population levels.

Key Words.  Araceae , Brazil, brejo forest, Ceará, elliptic Fourier Analysis,  Monstera , morphometrics.

IntroductionThe context of this study is the   “brejo forest ” —   amosaic of isolated humid forest areas in Northeast Brazil scattered within the region’s predominant semi-arid caatinga biome wherever rainfall brought by humid winds from the Atlantic Ocean providessuitable microclimates in mountainous terrain (Saleset al.   1998; Rodal   et al.   1998; Sampaio   et al.   2002;Borges-Nojosa & Caramaschi   2003; Pôrto   et al.   2004; Andrade   et al.   2007). The brejo forests are today classified as part of the Brazilian Atlantic forest in the

category of seasonal semi-deciduous montane forest (IBGE   1985,   2005). In Ceará State there are sevensuch areas (Figueiredo 1997) situated roughly halfway between Amazonia and the Atlantic forest, Brazil’smajor humid forest regions.

The brejo forests are of particular biogeographicinterest because they are thought to be natural fragmentsrepresenting relicts of previous expansions of the Atlanticand Amazon forests (Ab’Sáber   1982; Andrade-Lima1982; Borges-Nojosa & Caramaschi  2003). The resultsof Oliveira   et al.   (1999), in a study of a Quaternary pollen profile from central Bahia (relatively near thefocus of our study in Ceará), suggest that humid andcooler climates existed in Northeast Brazil until around4,500 yrs BP, and they discussed two hypotheses whichpostulate corridors of humid forest linking the Amazon

and Atlantic forests; one along the northeastern coast of Brazil from Maranhão to the region around Salvador inBahia and the other across central Brazil through anetwork of gallery forests. The existence of suchconnections and associated patterns of gene   flow over

Accepted for publication June 2008.1 Programa de Pós-Graduação em Botânica, Departamento de Ciências Biológicas, Universidade Estadual de Feira de Santana (UEFS), Avenida Universitária s/n,

44031-460, Feira de Santana, Bahia, Brazil.2 Herbarium, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3AB, UK. e-mail: [email protected]

Electronic supplementary material The online version of this article (doi:10.1007/s12225-008-9032-z ) contains supplementary material, which is available toauthorized users.

KEW BULLETIN VOL.  63: 193–211 (2008)

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 2/19

 wide areas of eastern South America is consistent withrecent results from lizard and amphisbaenid biogeogra-phy (Borges-Nojosa & Caramaschi  2003), and our ownresults using AFLP molecular markers in   Monstera 

adansonii   Schott var.   klotzschiana   (Schott) Madison(Andrade  et al.  2007).

This paper is part of a population level study aimed at comparing morphological and molecular patterns in indi- vidual species of the family   Araceae   occurring in brejoforests in the Brazilian state of Ceará and in the Atlanticand Amazon forests; a molecular study of   Monstera 

adansonii    var.   klotzschiana   has already been published(Andrade et al. 2007).

Monstera adansonii   var.  klotzschiana  was chosen as atypical representative of the herbaceous brejo  flora forthree main reasons. Firstly as a root-climber of humidtropical forests it is intolerant of low humidity and isconfined ecologically to this habitat   — it has not beenobserved in caatinga or seasonally dry forests. On the

other hand, it is rarely found in undisturbed primary forest, occurring typically in secondary moist tropicalforest with a well-developed canopy. Flowering is most conspicuous where light levels are higher, at the edgeof clearings and along forest edges. We observed that in Ceará this taxon only occurs near waterfalls andstreams that are permanent throughout the year.Second, this taxon is widespread in the eastern half of tropical South America (Madison   1977), allowing comparison of the Ceará populations with those in Amazonia and the Atlantic forest. It is one of severalspecies of   Araceae   which show a disjunct biogeo-graphical range between these two major humid forest 

biomes (Mayo 1983). Finally, previous work (Andrade& Mayo 1998) provided a good understanding of themorphology of the plant in its natural habitat.

Our aim was to estimate quantitatively morpho-logical divergence between the populations with a view to comparison with genetic patterns estimated withmolecular AFLP markers. Using leaf outline morpho-metrics as data for estimating morphological variation(see Methods for further explanation) we focused on thefollowing specific objectives: 1. Comparison of the Ceará brejo populations of  Monstera adansonii  var.  klotzschiana ;2. Estimation of the morphological differences betweenthese populations; 3. Comparison of the Ceará popula-

tions with those in the Atlantic and Amazon forests; 4.Comparison of leaf outline morphology in var.  klotzschi- 

ana   with that of the related taxa in Northeast Brazil,M. adansonii   var.   laniata   (Schott) Madison and   M.

 praetermissa  E. G. Gonç. & Temponi.

Materials and Methods

Leaf outline morphometrics.   Two constraints led tothe choice of leaf outline data for the following 

reasons. Firstly, to obtain results comparable with ourmolecular study (Andrade   et al.   2007), adequately sampled morphological data from the same popula-tions were needed in a form which could be analysedusing similar quantitative techniques. Second, limitedresources for   fieldwork and the large geographical

distances between study populations meant that many localities could be sampled only once. These factorsfavoured leaves since, unlike inflorescences, they werenearly always available in suf ficient numbers and easily sampled and preserved. Using a digital camera, leaf outline shapes and dimensions were easily captured inthe  field. Digitisation of the leaf outlines from imagesand obtaining mathematical descriptions (using ellip-tic Fourier Analysis) from the resulting coordinatedata was made easy by freely available software(TpsDig: Rohlf (2004); Morpheus   et al.: Slice (2008);SHAPE: Iwata & Ukai (2002); download from Rohlf 2006).

Taxonomic determination of  Monstera  species nor-mally demands observation of characters from inflor-escences as well as vegetative organs (Madison  1977).Fertile voucher specimens were gathered from sam-pled populations whenever found, and identifiedusing the standard revision (Madison   1977); these were deposited permanently as herbarium specimens(see Appendix and Electronic Supplementary Material).Combinations of vegetative characters were used toidentify   field-collected specimens of the three taxa of Monstera   studied.  M. praetermissa   is easily distinguishedfrom   M. adansonii   in non-flowering plants as follows:the mature plant is altogether smaller in stature; the

leaves become dark brown to blackish after drying; theinternodes are much more slender, darker green andlack the abundant whitish speckling of  M. adansonii   —

instead the epidermis is shallowly verrucate, the verrucae being the same colour as the rest of theinternode. The petiole sheath margins are very quickly marcescent, decomposing into fragile black membra-naceous fragments which are eventually deciduous; inM. adansonii  the sheath margins are persistent, greenand herbaceous. The leaf perforations in  M. praetermissa 

are relatively few and large, extending almost to the very margin of the blade, which often becomes partly pinnatifid by the rupture of the remaining marginal

strip. In  M. adansonii   leaf perforation patterns are very  variable, ranging from absent to extremely abundant and with the individual perforations varying widely insize and shape. However, they never extend right to the very margin and the blade thus does not becomepinnatifid.

 Within Monstera adansonii , the differentiation of var.klotzschiana  and var.  laniata  is problematic in fragmen-tary herbarium material, but easier in living   fieldpopulations. The most reliable diagnostic charactersare in the   flowers: in var.   klotzschiana   the stigma is

194   KEW BULLETIN VOL. 63(2)

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 3/19

elliptic and raised up on a shallow but distinct stylarprominence at the gynoecium apex, whereas in var.laniata   the stigma is linear and sessile on a truncate,flattened gynoecial apex. Living non-flowering plantscan be distinguished in the   field as follows: var.klotzschiana   prefers more shaded and protected hab-

itats under forest canopies and has leathery matureleaf blades which are more than twice as long as wideand ovate-elliptic to oblong-elliptic in shape; the leaf blade base is less asymmetric, usually acute on oneside and obtuse to rounded on the other. Var.   laniata 

typically grows in more open exposed habitats such ascoastal scrub on rocks or even on fence posts in humidtropical pasture; the mature leaf blades are rather thinin texture, less than twice as long as wide and broadly ovate to ovate-elliptic, with strongly asymmetric leaf blade bases acute on one side and usually truncate tosubcordate on the other. Var.   laniata   is quite rare innon-Amazonian Brazil, its main range being Central

 America, Pacific tropical South America, western Amazonia, Venezuela, the Guianas and northern Amazonian Brazil. Madison (1977) cited no recordsfor eastern or northeastern Brazil, but subsequent fieldwork uncovered a form of var.   laniata   in the Atlantic forest of southern Bahia and Pernambuco;the Bahian plants are notable for their very numerousand irregular leaf perforations.

The degree to which leaf outline shape is geneti-cally determined is an important issue from a taxo-nomic standpoint. The experimental studies neededto investigate this, involving comparative growth trialsin an experimental garden, were beyond the scope of 

this project. However, there is evidence which suggeststhat leaf shape has a genetic basis both in general andin the particular case of   Monstera . Kessler & Sinha(2004) reviewed studies of genetic control of leaf shape in model organisms and discussed a range of genetically mediated processes involved in establishing the mature shape of the leaf blade. Studies by Furuta  et 

al.   (1995), Yamanaka   et al.   (2001) and Iwata   et al.

(2002) used principal component shape variablesderived from matrices of Elliptic Fourier coef ficientsto describe leaf shape quantitatively in selectedcultivars and varieties of soybean and citrus. All thesestudies found significant genetic components in the

shape variables, and were furthermore able to show that the variables which describe symmetric changes inleaf shape have a much higher genetic component that the asymmetrical ones. Yoshioka   et al.   (2004)obtained similar results using petal shape in primulacultivars. The heteroblasty shown by all   Monstera 

species also indicates a major role for genetic deter-mination in leaf shape in these plants. Juvenile leavesalways differ considerably in shape, size and presenceor pattern of perforations and undergo a predictableseries of shape changes as the plant develops from

seedling to adult; this was used by Madison (1977) asthe basis of his sectional classification of the genus.

Leaf outline capture.  1,695 leaves were sampled fromnatural populations at 32 sites in Brazil and FrenchGuiana, grouped into 20 populations (Fig.  1, Table 1,

Electronic Supplementary Material). At each site,leaves at various stages of the growth cycle werecollected according to availability. To avoid possiblebias caused by clonal growth, care was taken to samplefrom plants that were widely distant and to avoidsampling from plants on the same host tree; thisreduced the scope for making larger samples in somepopulations. Digital images were made using a FujiFinepix A303 digital camera while the leaves were stillfresh. After trimming off most of the petiole, theleaves were attached with map pins to a backboardmade of rubberised carpet tiles with the abaxial sideexposed and photographed using a 15 cm measuring 

rule as a scale. One source of variation in shapecapture arose from the fact that the leaves were not strictly two-dimensional, but undulated to variousdegrees especially along the margin. To reduce thiserror, the pinning was carried out so that the leaf undulation was evenly distributed around the marginand the leaves were photographed directly overhead.

The images were prepared for outline digitisationusing the Gimp software package (GIMP   2006).Monstera   leaves are distichous and normally oblique,i.e. one half of the leaf blade is wider than the other, with successive leaves showing mirror-symmetry. Eachleaf was rotated so that the apex faced to the left with

the midrib more-or-less horizontal and   flipped whennecessary to bring the broadest half of the blade ontothe lower side. The paint tool was used to adjust theimage to provide suf ficient uniformity of tone along the margin. Minor damage to the margin caused by herbivores was also edited out.

The prepared images were then opened in tpsDig (Rohlf  2004, ver. 1.40). The petiole-leaf blade junctionand leaf blade apex were digitised as landmarks 1 and2 respectively. To remove   fine scale   “noise”  along themargin, the image was smoothed 8 to 10 times using the   “Smooth image”  tool. The thresholded image was viewed and checked using tpsDig ’s Image tools to

capture the outline as precisely as possible by optimis-ing the threshold value. Editing-out of shadows wassometimes necessary at this stage. After these steps,the outline was digitised using the   “Outline object ”tool. All the outlines of a population were thus writteninto a single   “.tps”  file for further processing.

Size classification of leaves.  A methodological compli-cation was the conspicuous change in leaf shape and size(heteroblasty) observed in   Monstera adansonii   during development from juvenile to adult plants (Andrade &

195COMPARATIVE MORPHOLOGY OF POPULATIONS OF   MONSTERA ADANS.

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 4/19

Mayo  1998), which is even more pronounced in someother species of the genus (Madison   1977; Ray   1983).The total sample of leaves was therefore divided into

three size classes for analysis so that leaves at approxi-mately the same heteroblastic stage would be compared.

In  Monstera adansonii , this shape change is gradualduring the progression from the smallest to largest leaves, but in  M. praetermissa  a sharp change in shape was observed at about 15 cm length. Using this as aguide, we selected 15 cm leaf blade length as theboundary between juvenile and submature leaves. Theboundary between submature and mature was set at 30 cm leaf blade length after examination of 167 fertileherbarium specimens at the Kew herbarium (K) of  M.

adansonii  var. klotzschiana  and var. laniata , to establish an

approximate lower limit to the leaf blade size in matureflowering shoots.

Elliptic Fourier Analysis.  Elliptic Fourier Analysis (EFA)is a method for quantitatively characterising closedoutline shapes and has been used widely in many biological   fields, including taxonomy (e.g. Rohlf & Archie 1984; Ferson et al. 1985; McLellan 1993; Cannon& Manos 2001; Rumpunen & Bartish 2002; Jensen et al.

2002; Tatsuta et al. 2004; Yoshioka et al. 2004; Ravelosonet al.   2005). It is particularly useful for shapes whichpossess few clearly homologous landmarks, as in thepresent case, where only the petiole-leaf blade junctionand the leaf blade apex can be regarded as consistent,biologically meaningful homologous sites. We used EFA 

KBAS, LBAIlh,

LBAUb

KCEma, KCEpa

KFG, LFG

KCEub, PCEib, PCEub

KPA

KPER, LPET

KBAR, PBAR

KESli, KESst

KRJ

KSP

N

1000 km

PCEbat

Fig. 1.  Geographical locations of Monstera  populations studied in South America. Monstera adansonii  var.   klotzschiana. KBAR:Bahia, Recôncavo; KBAS: Bahia, Ubaitaba; KCEma: Ceará, Serra de Maranguape; KCEpa: Ceará, Serra de Aratanha, Pacatuba;KCEub: Ceará, Serra de Ibiapaba; KESli: Espírito Santo, Linhares; KESst: Espírito Santo, Santa Tereza; KFG: French Guiana; KPA:Pará, near Belém; KPER: Pernambuco, Recife; KRJ: Rio de Janeiro, near Rio de Janeiro city; KSP: São Paulo, Bertioga.  Monsteraadansonii  var.  laniata. LPET: Pernambuco, Tapacurá; LBAIlh: Bahia, Ilhéus; LBAUb: Bahia, Ubaitaba; LFG: French Guiana. Monstera

 praetermissa. PBAR: Bahia, Serra da Copioba; PCEbat: Ceará, Serra de Baturité; PCEib: Ceará, Serra de Ibiapaba (Ibiapina); PCEub:Ceará, Serra de Ibiapaba (Ubajara).

196   KEW BULLETIN VOL. 63(2)

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 5/19

Table 1. Geographical details of sampled populations (20 populations, 1695 leaves) of  Monstera adansonii  var. klotzschiana, M. adanso

Code and taxon Latitude longitude Population detailsCountry orBrazilian state Total

KBAR   Monstera adansonii  var.   klotzschiana    12°50  ′20  ″S 39°16  ′19  ″ W Recôncavo (mean value for 2 localities) Bahia 43 KBAS " 14°18  ′02  ″S 39°19  ′01  ″ W Ubaitaba Bahia 13 KCEma " 3°48  ′18  ″S 38°42  ′46  ″ W Serra de Maranguape (mean value for

2 localities)Ceará 62

KCEpa " 3°57  ′21  ″S 38°37  ′44  ″ W Serra de Aratanha, Pacatuba Ceará 75 KCEub " 4°01  ′19  ″S 40°51  ′43  ″ W Serra de Ibiapaba (Ibiapina, São Benedito,

Ubajara) (mean value for 3 localities)Ceará 179

KESli " 19°9  ′3  ″S 40°4  ′15  ″ W Linhares Espírito Santo 14 KESst " 19°57  ′57  ″S 40°31  ′58  ″ W Santa Tereza Espírito Santo 40 KFG " 4°37  ′26  ″N 52°32  ′05  ″ W Petit Saut, Kaw, Matoury, Arataí (mean

 value for 4 localities)French Guiana 189

KPA " 1°30  ′32  ″S 48°17  ′59  ″ W Acará, Marituba Pará 110 KPER " 8°2  ′45  ″S 34°47  ′00  ″ W Recife (mean value for 2 localities) Pernambuco 61 KRJ " 22°58  ′08  ″S 43°14  ′38  ″ W Rio de Janeiro city area (mean value for

2 localities)Rio de Janeiro 114

KSP " 23°51  ′14  ″S 46°8  ′20  ″ W Bertioga São Paulo 71

LBAIlh  Monstera adansonii  var.   laniata    14°46  ′19  ″S 39°13  ′19  ″ W Ilhéus Bahia 21 LBAUb " 14°18  ′2  ″S 39°19  ′1  ″ W Ubaitaba Bahia 44 LFG " 4°54  ′32  ″N 52°16  ′26  ″ W Cayenne, Fort Diamant, Colline de

Montravel (mean value for 3 localities)French Guiana 316

LPET " 8°0  ′0  ″S 35°10  ′58  ″ W Tapacurá Pernambuco 37

PBAR   Monstera praetermissa    12°46  ′41  ″S 39°03  ′38  ″ W Serra da Copioba Bahia 8 PCEbat " 4°16  ′33  ″S 38°57  ′26  ″ W Serra de Baturité Ceará 39 PCEib " 3°55  ′51  ″S 40°52  ′33  ″ W Serra de Ibiapaba (Ibiapina) Ceará 171 PCEub " 3°49  ′58  ″S 40°56  ′23  ″ W Serra de Ibiapaba (Ubajara) Ceará 88

 ©T  h   e B  o a r   d   of   T  r   u s  t   e  e  s  of    t  h   e R 

 o  y  a l   B  o t   a ni    c  G a r   d   e n s  ,K  e w ,2  0  0  8 

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 6/19

to produce a matrix of 160 coef ficients from which sixshape descriptor variables, the   first six principal com-ponents, were derived.

The Fourier analysis was carried out using theMorpheus   et al.  package (Slice   2008). Before running the Elliptic Fourier Analysis module, the outlines were

standardised using the option   “Bookstein superimposi-tion”, which requires that the orientation, scale, star-tingPt and location settings be switched off. The outlinesare then superimposed using two landmarks placed at the base and apex of the leaf blade which equalises thelength and orientation of the interval between theselandmarks. In contrast, the default standardisation inMorpheus involves equalising the length and orienta-tion of the   first ellipse of all outlines. The justificationfor using Bookstein superimposition is that the twolandmarks at either end of the midrib represent themost consistent homologies recognisable in the leaves.

The Fourier analysis was carried out to a limit of 40

harmonics, resulting in a matrix of 164 elliptic Fouriercoef ficients. The Fourier procedure reconstructs theoutlines to a level of approximation limited only by the number of pixels used in the original digitisation,so the decision on how many harmonics to set is tosome extent arbitrary (Rohlf & Archie   1984). Aftercomparing the reconstructed outlines by eye with theoriginals we considered that with 40 harmonics, noobvious artefactual distortions would be introducedinto the analyses. The  first 4 coef ficients (harmonic 0)are discarded because they correspond to an arbitrary starting point of the outline tracing sequence ratherthan any feature of the leaf shape, so the  final matrix

consisted of 160 variables (1st  to 40th harmonic, eachharmonic consisting of 4 values).

Multivariate data analyses.  Multivariate analyses of theresulting coef ficient matrices were carried out using NTSYSpc ver. 2.20d (Rohlf  2005). Principal component analysis of the coef ficient matrices was used to reducethe dimensionality of the data to a much smallernumber of uncorrelated shape descriptor variables. Eachof these variables was reified using the proceduredescribed by Yoshioka   et al.   (2004). First, the mean vector of the Fourier coef ficients was computed. Eacheigenvector was then multiplied by integer factors ( f  )from 1 to 5. The factored eigenvectors were thenseparately added or subtracted from the mean vector.The mean vector corresponds to the mean shape of thedata set. The effect of adding or subtracting a particularfactored eigenvector is to exaggerate or diminish itseffect on the mean shape while holding all the othereigenvectors constant. The change in principal compo-nent values corresponds to a change of   f    standarddeviations on the principal component concerned,provided the principal component values have beensuitably scaled. The resulting vectors were then plotted

using the   “Fourplot ” module in NTSYSpc, which recon-structs the outline from the set of Fourier coef ficientsprovided. These outlines visualise the shape changesexpressed independently by each of the shape variables(principal components) we used. Figures 2, 3 and 4 show this transformation from  −2 to + 2 standard deviations.

 We selected the  first six principal components (PCs)to be treated as variables in a Canonical Variates Analysis(CVA); together these accounted for 96.6% of the variance in the original covariance matrix. The reduc-tion to six PCs was necessary because once the leaf samples had been divided into their size classes, somepopulation samples were reduced, in one case to only 8 individuals. The aim of the CVA, carried out using NTSYSpc, was to show the maximum separation of thecentroids (multivariate means) of each population. Theresulting eigenvectors were used to project the matrix of principal component scores for all the leaves analysed,onto canonical variate axes. The matrix of pairwise

generalised distance between the population centroids was used to generate Neighbor-Joining (NJ) trees, alsousing NTSYSpc. As a comparison with the pattern basedon generalised distance, the mean vectors of the sixshape variables were computed for each population andtheir pairwise Euclidean distances computed withNTSYSpc. The resulting distance matrix was then usedto carry out a Principal Coordinates Analysis (PCoA),incorporating a minimum spanning tree (MST).

To estimate intra-population variability, the Euclid-ean distance between all pairs of individuals for eachpopulation was computed, using the matrix of 6principal component scores from the combined PCA 

of all populations. Analyses of each size class werecarried out separately. The pairwise inter-individualdistances of each population were then summarised by calculating the mean, standard deviation and coef fi-cient of variation (Raveloson  et al. 2005). The pairwisedifferences in median Euclidean distance in the threeCeará populations were tested using the non-paramet-ric Kruskal-Wallis test as implemented in PAST (Ham-mer et al.  2001).

Similarity relationships between pairs of popula-tions (based on the matrix of principal component scores) were tested using multivariate proceduresimplemented in PAST: NPMANOVA is a non-paramet-

ric analogue of MANOVA which calculates an  F  value with significance values computed by 5000 replicatepermutations of group membership (Table 2).

Results

The shape variables.   In all three size classes, the   first three principal components accounted for over 90%of the total variance (Mature: 92.29%, Submature:

198   KEW BULLETIN VOL. 63(2)

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 7/19

      1   X

  -      0 .      4

      9

  -      0 .      2

      4

      0 .      0

      1

      0 .      2

      5

      0 .      5

      0

  -      0 .      1

      9

  -      0 .      0

      9

      0 .      0

      1

      0 .      1

      1

      0 .      2

      1

      1   X

  -      0 .      4

      8

  -      0 .      2

      3

      0 .      0

      1

      0 .      2

      6

      0 .      5

      0

      0 .      2

      2

      0 .      1

      0

      0 .      0

      1

      0 .      1

      2

      0 .      2

      4

      1   X

  -      0 .      4

      7

  -      0 .      2

      3

      0 .      0

      2

      0 .      2

      6

      0 .      5

      1

      0 .      2

      5

      0 .      1

      2

      0 .      0

      1

      0 .      1

      4

      0 .      2

      7

      1   X

  -      0 .      4

      7

  -      0 .      2

      2

      0 .      0

      3

      0 .      2

      7

      0 .      5

      2

  -      0 .      2

      9

  -      0 .      1

      4

      0 .      0

      1

      0 .      1

      5

      0 .      3

      0

      1   X

  -      0 .      4

      9

  -      0 .      2

      3

      0 .      0

      3

      0 .      2

      9

      0 .      5

      4

  -      0 .      3

      2

  -      0 .      1

      6

      0 .      0

      0

      0 .      1

      7

      0 .      3

      3

      1   X

  -      0 .      5

      0

  -      0 .      2

      5

  -      0 .      0

      1

      0 .      2

      4

      0

 .      4      8

      0 .      1

      0

      0 .      0

      4

      0 .      0

      2

      0 .      0

      8

      0 .      1

      4

      1   X

  -      0 .      4

      9

  -      0 .      2

      4

      0 .      0

      0

      0 .      2

      4

      0

 .      4      9

      0 .      1

      5

      0 .      0

      7

      0 .      0

      2

      0 .      1

      1

      0 .      1

      9

      1   X

  -      0 .      4

      6

  -      0 .      2

      2

      0 .      0

      3

      0 .      2

      7

      0

 .      5      2

      0 .      2

      8

      0 .      1

      4

      0 .      0

      1

      0 .      1

      5

      0 .      2

      9  -      0

 .      4      8

  -      0 .      2

      3

      0 .      0

      1

      0 .      2

      6

      0

 .      5      0

      0 .      2

      2

      0 .      1

      0

      0 .      0

      1

      0 .      1

      3

      0 .      2

      4  -      0 .      4

      7

  -      0 .      2

      2

      0 .      0

      3

      0 .      2

      8

      0

 .      5      3

      0 .      3

      4

      0 .      1

      7

      0 .      0

      0

      0 .      1

      7

      0 .      3

      5

      1

  -      0 .      5

      0

  -      0 .      2

      5

  -      0 .      0

      1

      0 .      2

      4

      0 .      4

      9

  -      0 .      1

      0

  -      0 .      0

      4

      0 .      0

      1

      0 .      0

      7

      0 .      1

      2

      1   X

  -      0 .      4

      9

  -      0 .      2

      4

      0 .      0

      0

      0 .      2

      5

      0 .      4

      9

  -      0 .      1

      4

  -      0 .      0

      6

      0 .      0

      2

      0 .      1

      0

      0 .      1

      8

      1   X

  -      0 .      4

      6

  -      0 .      2

      2

      0 .      0

      2

      0 .      2

      7

      0 .      5

      1

  -      0 .      2

      6

  -      0 .      1

      2

      0 .      0

      1

      0 .      1

      5

      0 .      2

      8  -      0

 .      4      8

  -      0 .      2

      3

      0 .      0

      1

      0 .      2

      6

      0 .      5

      0

  -      0 .      2

      0

  -      0 .      0

      9

      0 .      0

      1

      0 .      1

      2

      0 .      2

      3

      1   X

  -      0 .      4

      6

  -      0 .      2

      1

      0 .      0

      3

      0 .      2

      7

      0 .      5

      2

  -      0 .      3

      2

  -      0 .      1

      5

      0 .      0

      1

      0 .      1

      7

      0 .      3

      3

- 1 s.d.- 2 s.d. Mean shape + 2 s.d.+ 1 s.d.

Mature (54.67% of total variance)

Submature (71.28% of total variance)

Juvenile (62.85% of total variance)

Fig. 2.   Reification of principal component 1 as shape variable  Anisotropy . Derived from principal component analysis of ellipticFourier coefficient matrices of leaf outlines of Monstera adansonii  var. klotzschiana, M. adansonii  var. laniata and  M. praetermissa.Each outline is a 40-harmonic reconstruction from elliptic Fourier analysis data. Each row shows the trend of shape change alongprincipal component 1, in separate PCAs of mature (>=30 cm long), submature (15 to 29.9 cm long) and juvenile (< 15 cm long)leaves. From left to right the outlines represent the principal component scores corresponding to: (mean − 2 standard deviations),(mean  −  1 s.d.), mean, (mean + 1 s.d.), (mean + 2 s.d.). Computed with NTSYSpc ver. 2.2.

199COMPARATIVE MORPHOLOGY OF POPULATIONS OF   MONSTERA ADANS.

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 8/19

 S  u b m a t   ur  e (  2 1 .4 

 0  %  of   t   o t   al  v  ar i   an c  e )  

 J  uv  eni  l   e (  2  9 .2  3 

 %  of   t   o t   al  v  ar i   an c  e )  

-0.47 -0.22

-0.47 -0.23

-0.47 -0.23

-0.47 -0.22

0.47 -0.22

1

-0.47 -0.22 0.03 0.29 0.54

.19

.09

.02

.13

.24

1

X

-0.47 -0.23 0.02 0.27 0.51

.20

.09

.02

.13

.24

-0.48 -0.23 0.01 0.26 0.50

.22

.10

.01

.13

.24

1

-0.48 -0.23 0.01 0.26 0.51

.23

.11

.01

.13

.25

-0.48 -0.23 0.02 0.27 0.52

.25

.12

.00

.12

.25

1

X

-0.48 -0.22 0.03 0.28 0.53

-0.19

-0.08

0.02

0.13

0.23

1

-0.47 -0.23 0.02 0.26 0.51

-0.19

-0.09

0.02

0.12

0.23

1

X

-0.48 -0.23 0.01 0.26 0.50

-0.20

-0.09

0.01

0.12

0.23

1

X

-0.48 -0.23 0.02 0.26 0.51

-0.21

-0.10

0.01

0.12

0.23

1

X

-0.49 -0.23 0.02 0.27 0.53

-0.23

-0.11

0.00

0.11

0.23

F  i        g.

 3  .

R  e i   fi    c  a  t  i    on of     p

r  i   n c i     p a l    c  om  p on e n t  2  a  s  s h   a   p e v  a r  i    a  b  l    e 

D i   mi    d  i    a  t   e .D  e r  i   v  e  d  f   r   om

  pr  i   n c i     p a l    c  om  p o

n e n t   a n a l     y  s i    s  of    e l   l   i     p t  i    c 

F   o ur  i    e r   c  o e f   fi    c i    e n t  m a  t  r  i    c  e  s  of   l    e  a f    o u t  l   i   n e  s  of   M on s  t   e r   a  a  d   a n s  oni   i   v  a r  .k  l    o t  z  s  c h  i    a n a  ,M. a  d   a n s  oni   i   v  a r  .l    a ni    a 

 t   a  a n d  M.  pr   a  e  t   e r  mi    s  s  a .

F   or  f    ur   t  h   e r   e x   pl    a n a  t  i    on

 s  e  e F  i     g.2 .

 ©T  h   e B  o a r   d   of   T  r   u s  t   e  e  s  of    t  h   e R  o  y  a l   B  o t   a ni    c  G a r   d   e n s  ,K  e w ,2  0  0  8 

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 9/19

 S  u b m a t   ur  e (  2 

.4  5  %  of   t   o t   al  v  ar i   an c  e )  

 J  uv  eni  l   e (   3 . 3 

 3  %  of   t   o t   al  v  ar i   an c  e )  

1

X

-0.49 -0.25 -0.00 0.24 0.49

-0.19

-0.08

0.02

0.13

0.23

1

X

-0.48 -0.24 0.00 0.25 0.49

-0.19

-0.09

0.02

0.12

0.23

1

X

-0.48 -0.23 0.01 0.26 0.50

-0.20

-0.09

0.01

0.12

0.23

1

X

-0.47 -0.22 0.02 0.26 0.51

-0.21

-0.10

0.01

0.12

0.23

1

X

-0.46 -0.21 0.03 0.27 0.52

-0.21

-0.10

0.01

0.12

0.24

1

X

-0.49 -0.25 -0.00 0.24 0.48

.21

.09

.02

.14

.25

1

X

-0.48 -0.24 0.00 0.25 0.49

.21

.10

.02

.13

.25

1

X

-0.48 -0.23 0.01 0.26 0.50

.22

.10

.01

.13

.24

1

X

-0.47 -0.22 0.02 0.27 0.51

.22

.10

.01

.13

.24

1

X

-0.46 -0.22 0.03 0.28 0.52

.22

.11

.01

.13

.24

-0.46 -0.21

-0.46 -0.22

-0.47 -0.23

-0.48 -0.23

-0.49 -0.24

.24

.12

.01

.14

.26

F  i        g.4  .R  e i   fi    c  a  t  i    on o

f     pr  i   n c i     p a l    c  om  p on e n t   3  a  s  s h   a   p e v  a r  i    a  b  l    e  Ov  a  t   e .D  e r  i   v  e  d  f   r   om  pr  i   n c i     p a l    c  om  p on e 

n t   a n a l     y  s i    s  of    e l   l   i     p t  i    c F   o ur  i    e r  

 c  o e f   fi    c i    e n t  m a  t  r  i    c  e  s 

 of   l    e  a f    o u t  l   i   n e  s  of   M on s  t   e r   a  a  d   a n s  oni   i   v  a r  .k  l    o t  z  s  c h  i    a n a  ,M. a  d   a n s  oni   i   v  a r  .l    a ni    a  t   a  a n d  M.  pr   a  e  t   e r  mi    s  s  a .F   or  

f    ur   t  h   e r   e x   pl    a n a  t  i    on

 s  e  e F  i     g.2 .

 ©T  h   e B  o a r   d   of   T  r   u s  t   e  e  s  of    t  h   e R  o  y  a l   B  o t   a ni    c  G a r   d   e n s  ,K  e w ,2  0  0  8 

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 10/19

95.13%, Juvenile: 95.41%) and described distinct andeasily recognised trends of leaf outline shape change which are illustrated in Figs.   2,   3   and  4. The fourth,fi

fth and sixth principal components expressed ob-scure shape variations that were not easily interpreted visually.

The   first principal component is named here theAnisotropy   shape variable (Fig.   2), because its most conspicuous characteristic is the increase in length/ width ratio. This shape variable also expresses anasymmetry at the leaf base which becomes morepronounced in the transition from juvenile to mature

leaves. The second principal component, here termed Dimidiate    shape variable (Fig.   3) is characterisedmainly by the increasing difference in the width and

area of the two sides of the leaf blade, one sideeventually becoming very narrow. A secondary featureis a slight basiscopic migration of the axis of widest  width. The third principal component, the   Ovate 

shape variable (Fig.  4), is characterised chiefl y by themigration of the widest width axis giving a trend fromelliptic to ovate or vice versa (the orientation of theprincipal component axes between analyses is arbi-trary). A slight inverse dimidiate trend is also evident,

Table 2.  Inter-population differences between nine populations of Monstera adansonii   var.   klotzschiana  based on matrix of sixshape variables derived from elliptic Fourier analysis of mature leaf outlines.

KBAR KBAS KCEma KCEpa KCEub KFG KPA KPER KSP

KBAR    –

KBAS 0.009   –

KCEma 0.016 n.s.   –

KCEpa 0.000 n.s. 0.000   –

KCEub 0.048 0.027 0.003 0.000   –

KFG n.s. 0.011 0.011 0.000 0.009   –

KPA n.s. 0.027 0.011 0.000 n.s. n.s.   –

KPER n.s. 0.006 n.s. 0.000 0.002 0.020 0.042   –

KSP n.s. 0.002 0.000 0.000 n.s. 0.004 n.s. 0.001   –

 Values shown are p -values (probability that two groups are the same) derived from Non-Parametric MANOVA implemented in PAST ver.1.34 (Hammer et al.  2001); 5,000 replicates computed. See Table  1  for key to population codes.

0

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

PCEub PCEib LFG KCEpa KCEub KFG KPER LPET KCEma KPA LBAUb KBAR KSP

Populations

   M  e  a  n   i  n   t  e  r  -   i  n   d   i  v   i   d  u  a   l   E  u  c   l   i   d  e  a  n   d   i  s   t  a  n  c  e  p  e  r  p  o  p  u   l  a   t   i  o  n

mature submature juvenile

Fig. 5.   Intra-population variability of leaf shape in  Monstera adansonii   var.   klotzschiana,   M. adansonii   var.   laniata   and   M. praetermissa. For each population, the mean Euclidean distance between pairs of individuals is shown for three size classes (left bar :mature leaves; middle bar : submature leaves; right bar  [when data available]: juvenile leaves). Data is matrix of individual scores onsix shape variables (principal components) derived from EFA coefficients of leaf outlines. 95% confidence limits shown. See Fig. 1for key to population codes.

202   KEW BULLETIN VOL. 63(2)

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 11/19

i .e. the more ovate the leaf becomes, the less

dimidiate: this secondary trend is more distinct inthe submature and juvenile size classes.

Comparison of intrapopulational variability in Ceará

populations of   Monstera adansonii   var.   klotzschiana . Juvenile leaves were only available for the Ubajarapopulation (KCEub), so comparisons of the three Ceará populations were made with submature and matureleaves. Using mean Euclidean distance between pairs of individuals as a measure of population variability,submature leaves were more variable than mature leaves(Fig.   5), with the largest discrepancy in size class variability shown in the Pacatuba population (KCEpa).

Comparison of the mature leaves showed significantly greater variability in the Maranguape (KCEma) popu-lation compared to the other two (Pacatuba: KCEpa;Ubajara: KCEub), and non-significant difference in the variability of the Pacatuba and Ubajara populations.

Differences Between the Ceará Populations of Monstera 

adansonii  var. klotzschiana . The three Ceará populationsdid not cluster together to form an exclusive group inany analysis and the Pacatuba population (KCEpa) was more distinct in a CVA of the three populations(Fig.   6); the Maranguape and Ubajara populations

clustered closer together in both this and the NJ

analysis (Fig.   10). Pairwise tests between the threepopulations (Table   2) showed significant differencesbetween all populations for mature leaves. Submatureleaves were less distinct, with significant differences inall tests only between the Pacatuba and Ubajarapopulations. Multivariate analyses (CVA and NJ)showed somewhat different patterns of similarity between the mature (Fig.   7) and submature (Fig.   8)leaf classes; submature leaves of the Pacatuba popula-tion were closer to var.   laniata   than to other var.klotzschiana  populations.

Comparison of the Ceará Populations with those

of Monstera adansonii   var.   klotzschiana ,  var.   laniata   andM. praetermissa   in the Atlantic and Amazon Forests.

Comparison of intra-population variability, as estimatedby pairwise Euclidean distance between individuals,showed that in most populations for which we had datafrom all size classes, mature leaves are less variable thansubmature ones (Fig.   5). In most cases, juvenilepopulations of var.   klotzschiana  were less variable thanin the other two size classes.

Multivariate analysis (CVA, NJ) of the three sizeclasses produced quite similar results. Most populationsof var.  klotzschiana  grouped together, and populations

CV1

-0.0045 -0.0040 -0.0036 -0.0031 -0.0026

CV2

0.0027

0.0034

0.0041

0.0047

0.0054

Fig. 6.  Canonical variates analysis of leaf outlines from populations of  Monstera adansonii  var.  klotzschiana from Ceará. Derivedfrom six shape variables extracted from elliptic Fourier coefficient data from mature leaves (≥  30cm). Grey circles: Maranguape(KCEma); white squares: Pacatuba (KCEpa); black diamonds: Ubajara (KCEub). Computed in NTSYSpc ver. 2.2.

203COMPARATIVE MORPHOLOGY OF POPULATIONS OF   MONSTERA ADANS.

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 12/19

CV1

-3.95 -2.56 -1.17 0.21 1.60

CV2

-2.50

-1.52

-0.55

0.42

1.40

KBAR

KBAS

KCEma

KCEpa

KCEub

KFGKPA

KPER

LPET

KSP

LBAUb

LFG

PBAR

PCEib

PCEub

      1

  -      0 .      4

      7

  -      0 .      2      2

      0 .      0      2

      0 .      2      7

      0 .      5

      2

      5      1      4      7

      1   X

  -      0 .      4

      7

  -      0 .      2

      2

      0 .      0

      3

      0 .      2      7

      0 .      5

      2

 .      2      3

 .      1      1

 .      0      2

 .      1      4

 .      2      6

  -      0 .      4      6

  -      0 .      2

      2

      0 .      0      3

      0 .      2      8

      0 .

 .      3      0

 .      1      5

 .      0      0

 .      1      5

 .      3

      1

  -      0 .      4      6

  -      0 .      2      1

      0 .      0      4

      0 .      2      8

      0 .      5

      3

  -      0 .      2

      8

  -      0 .      1

      4

      0 .      0

      0

      0 .      1

      5

      0 .      2

      9

      1   X

  -      0 .      4      7

  -      0 .      2      2

      0 .      0      2

      0 .      2      7

      0 .      5      1

 .      2      8 .      1      4

 .      0      0 .      1      5

 .      2      9

      1

  -      0 .      4

      6

  -      0 .      2      2

      0 .      0

      3

      0 .      2      8

      0 .      5

      3

      9      5      5

      1   X

  -      0 .      4      7

  -      0 .      2

      3

      0 .      0

      2

      0 .      2

      7

      0 .      5      1

      2      5      1      2      0      1      1      3      2      6

      1

  -      0 .      4

      8

  -      0 .      2

      3

      0 .      0

      2

      0 .      2

      7

      0 .      5

      2

 .      2      6 .      1      3

 .      0      0 .      1      4

 .      2      7

      1   X

  -      0 .      4

      8

  -      0 .      2      3

      0 .      0

      2

      0 .      2

      7

      0 .      5

      2

  -      0 .      2

      5

  -      0 .      1

      2

      0 .      0

      1

      0 .      1

      3

      0 .      2

      6

      1   X

  -      0 .      4

      8

  -      0 .      2

      3

      0 .      0

      1

      0 .      2      6

      0 .      5

      1

      0 .      2

      2

      0 .      1

      1

      0 .      0

      1

      0 .      1

      2

      0 .      2

      4

      1

  -      0 .      4

      7

  -      0 .      2

      2

      0 .      0      2

      0 .      2

      7

      0 .      5      1

      2      2

      1      0      2

      1      4

      2      6

  -      0 .      4      8

  -      0 .      2

      4

      0 .      0      1

      0 .      2      6

      0 .      5      0

      6      1      3      0

      1      4      7

      1   X

  -      0 .      4

      8

  -      0 .      2

      4

      0 .      0

      1

      0 .      2      6

      0 .      5      0

      3      1      1      3      5

      1   X

  -      0 .      4

      7

  -      0 .      2

      2

      0 .      0

      2

      0 .      2

      7

      0 .      5

      1

 .      2      5

 .      1      3

      0 .      0

      0 .      1

      3 .      2

      6

      1   X

  -      0 .      4      8

  -      0 .      2      3

      0 .      0      2

      0 .      2      7

      0 .      5      1

      Y

  -      0 .      2      4

  -      0 .      1      2

      0 .      0      1

      0 .      1      3

      0 .      2      6

CV1

-3.87 -1.94 -0.01 1.91 3.84

CV2

-2.88

-1.85

-0.82

0.21

1.25

PC1

PC2

PC3

PC4

PC5

PC6

Fig. 7. Canonical variates analysis of mature leaf outlines in nine populations of Monstera adansonii  var. klotzschiana (KBAR, KBAS,KCEma, KCEpa, KCEub, KFG, KPA, KPER, KSP), three populations of  M. adansonii   var.   laniata   (LBAUb, LFG, LPET) and threepopulations of M. praetermissa (PBAR, PCEib, PCEub). Plot shows mean population shapes. CVA computed from six shape variables(principal components) extracted by principal components analysis from a matrix of elliptic Fourier coefficients. Lower figure showsvector diagram of eigenvectors in relation to the first two canonical variates axes. Computed with NTSYSpc ver. 2.2.

204   KEW BULLETIN VOL. 63(2)

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 13/19

CV1

-1.95 1.03 2.51 4.00

CV2

-1.90

0.14

1.16

2.1

8

KBAR

KCEma

KCEpa

KCEub

KESli

KESst

KFG

KPA

KPER

LPET

KRJ

KSP

LBAIlh

LBAUb

LFG

PCEbat

PCEib

PCEub

      X

  -      6 .      8

      2

  -      3 .      3

      5

      0 .      1

      1

      3 .      5      8

      7 .      0

      4

      0      7      6      8      1

      1   X

  -      5 .      2

      4

  -      2 .      5

      3

      0 .      1

      8

      2 .      8

      9

      5 .      6

      0

      6      0      3      4      1      8

      1   X

  -      6 .      4

      4

  -      3 .      0

      1

      0 .      4

      2

      3 .      8

      5

      7 .      2

      8

      7      2      3      9      4

  -      4      3

 .      1      0

  -      2      1

 .      0      4

      1 .      0

      2

      2      3

 .      0      8

      4      5

 .      1      3

      6

  -      0 .      4      8

  -      0 .      2      3

      0 .      0      1

      0 .      2

      6

      0 .      5

      1

      3      1      1      3      5

      X

  -      0 .      4

      8

  -      0 .      2

      4

      0 .      0

      1

      0 .      2      6

      0 .      5

      0

      0      0      1      1      2

      X

  -      0 .      4

      7

  -      0 .      2

      2

      0 .      0

      2

      0 .      2

      7

      0 .      5

      2

      4      1      0      1      3      6

      1   X

  -      0 .      4

      8

  -      0 .      2

      4

      0 .      0

      1

      0 .      2

      6

      0 .      5

      1

      4      1      1      3      5

      1

  -      0 .      4

      7

  -      0 .      2

      2

      0 .      0

      3

      0 .      2

      7

      0 .      5

      2

      3      1      1      4      6

      1 X

  -      0 .      4

      9

  -      0 .      2

      5

      0 .      0

      0

      0 .      2

      5

      0 .      5

      0

 .      1      1

 .      0      4

 .      0      2

 .      0      9

 .      1      6

      1   X

  -      0 .      4

      8

  -      0 .      2      4

      0 .      0      1

      0 .      2

      6

      0 .      5      0

 .      1      6 .      0      6

 .      0      3 .      1      2

 .      2      1

      1   X

  -      0 .      4      8

  -      0 .      2

      4

      0 .      0      1

      0 .      2      6

      0 .      5

      0

      6      7      2      2      1

  -      0 .      4

      9

  -      0 .      2      4

      0 .      0      0

      0 .      2      5

      0 .      5

      0

      1

  -      0 .      4

      9

  -      0 .      2

      4

      0 .      0

      1

      0 .      2

      5

      0 .      5

      0

      9      9      1      1      1

      1

  -      0 .      4

      9

  -      0 .      2

      4

      0 .      0

      0

      0 .      2

      5

      0 .      5

      0

      7      0      8

      0      1      0      9

      1   X

  -      0 .      4

      9

  -      0 .      2

      4

      0 .      0

      1

      0 .      2

      5

      0 .      5

      0

      0      9      1      2      2

  -      0 .      4

      6

  -      0 .      2

      2

      0 .      0

      3

      0 .      2

      7

      0 .      5

      2

      9      4      1      6      1

      1   X

  -      0 .      4

      6

  -      0 .      2

      1

      0 .      0

      3

      0 .      2

      8

      0 .      5

      3

      7      3      1      4      8

CV1

-1.95 -0.46 1.03 2.51 4.00

CV2

-1.90

-0.85

0.20

1.25

2.30

PC1

PC2

PC3

PC4

PC5

PC6

Fig. 8. Canonical variates analysis of submature leaf outlines in eleven populations of Monstera adansonii  var. klotzschiana (KBAR,KCEma, KCEpa, KCEub, KESli, KESst, KFG, KPA, KPER, KSP), four populations of M. adansonii   var.   laniata  (LBAIlh, LBAUb, LFG,LPET) and three populations of M. praetermissa  (PCEbat, PCEib, PCEub). For further explanation see Fig. 7.

205COMPARATIVE MORPHOLOGY OF POPULATIONS OF   MONSTERA ADANS.

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 14/19

CV1

-3.80 -2.50 -1.19 0.11 1.41

CV2

-2.95

-1.54

-0.13

1.29

2.70

KBAR

KCEub

KESst

KFG

KPA

KRJ

KSP

LBAUb

LFG

PCEbat

PCEib

      1   X

  -      0 .      4

      9

  -      0 .      2

      4

      0 .      0

      1

      0 .      2      6

      0 .      5

      0

      0 .      1      7

      0 .      0      8

      0 .      0      2

      0 .      1      1

      0 .      2      0

      X

  -      0 .      4      7

  -      0 .      2

      2

      0 .      0

      2

      0 .      2      7

      0 .      5      1

      2      3      1      1      0      1      1      2      2      4

      1   X

  -      0 .      4

      9

  -      0 .      2

      4

      0 .      0

      1

      0 .      2

      6

      0 .      5

      0

      1      8

      0      8

      0      1

      1      0

      1      9

      1   X

  -      0 .      4

      8

  -      0

 .      2      3

      0 .      0

      1

      0 .      2

      6

      0 .      5

      0

      2      1      1      1

      1      3      5

  -      0 .      4

      9

  -      0 .      2      4

      0 .      0

      1

      0 .      2

      5

      0 .      5      0

      1      3

      0      6

      0      1

      0      8

      1      6

      X

  -      0 .      4

      8

  -      0 .      2

      4

      0 .      0

      1

      0 .      2      6

      0 .      5      0

      0  1   X

  -      0 .      4

      8

  -      0 .      2

      4

      0 .      0

      1

      0 .      2      6

      0 .      5

      0

      1   X

  -      0 .      4

      7

  -      0 .      2      2

      0 .      0

      2

      0 .      2      7

      0 .      5      1

      2      1      1      0      0      1      1      3      2      4

      X

  -      0 .      4      6

  -      0 .      2

      1

      0 .      0

      4

      0 .      2

      9

      0 .      5      3

      2      7

      1      3

      0      1

      1      5

      2      9

      1 X

  -      0 .      4      9

  -

      0 .      2      4

      0 .      0

      0

      0 .      2      5

      0 .      5      0

      1      2      0      5      0      2      0      9      1      6

      1   X

  -      0 .      4

      7

  -      0 .      2

      3

      0 .      0

      2

      0 .      2

      6

      0 .      5

      1

      1      8

      0      8

      0      2

      1      2

      2      2

CV1

-3.65 -2.12 -0.59 0.93 2.46

CV2

-2.87

-1.52

-0.17

1.18

2.53

PC1

PC2

PC3

PC4

PC5

PC6

Fig. 9.  Canonical variates analysis of juvenile leaf outlines in seven populations of  Monstera adansonii   var.   klotzschiana  (KBAR,KCEub, KESst, KFG, KPA, KRJ, KSP), two populations of   M. adansonii   var.   laniata   (LBAUb, LFG) and two populations of  M.

 praetermissa (PCEbat, PCEib). For further explanation see Fig. 7.

206   KEW BULLETIN VOL. 63(2)

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 15/19

of var.   laniata   and   Monstera praetermissa   formed sepa-rate groups (Figs. 7, 8, 9 and 10). These taxonomically correlated clusters were most distinct in mature leaves(Fig.   7), with all populations of var.   klotzschiana 

forming a single cluster.

Pairwise NPMANOVA tests of mature-leaf popula-tions of var.   klotzschiana   (Table   2) showed the most significant differences in pairs involving the Pacatubapopulation (KCEpa). The results of these tests werebroadly consistent with a PCoA ordination of themean vectors of the six shape variable scores for each

of the nine var.  klotzschiana  populations (Fig. 11). Thelatter plot emphasised the disparity between the threebrejo populations in Ceará.

Mantel tests computed in NTSYSpc ver. 2.2 foundno significant correlation between leaf outline shape(measured as Euclidean distance between populationmean vectors of the six shape variables) and geo-graphical distance between populations [10,000 ran-dom permutations;   p   (random   Z   <= observed   Z ) =0.6305,   p   (random   Z   >= observed   Z ) = 0.3695; matrixcorrelation (normalised Mantel statistic Z ): r  = 0.06712]or between leaf outline and genetic distance [data from

 Andrade   et al.   2007, genetic distance estimated aspairwise F ST  ; 10,000 random permutations;  p   (randomZ  <= observed   Z ) = 0.3308,   p   (random   Z   >= observedZ ) = 0.6692; matrix correlation (normalised Mantelstatistic Z ):  r   =  −0.1158].

DiscussionNo strong geographical patterns emerged from theanalyses of var. klotzschiana  populations. This contrasted with the results obtained in our AFLP analysis (Andrade

et al. 2007), which showed a clear indication of greatergenetic diversity in Northeast Brazilian and Amazonianpopulations and a significant correlation of genetic andgeographical distance. The Ceará brejo populations of  var. klotzschiana  emerge from this study as distinct fromone another, at least using mature leaves. Althoughthere is considerable overlap (Fig.  5), their shapes aresignificantly different (Table   2). The Maranguapepopulation is more variable but the Pacatuba popula-tion is the most distinct within Ceará (Figs.   5,   8,   10).The study shows that in var.   klotzschiana , leaf outlinemorphology expresses a degree of inter-population

Coefficient

0.00 0.65 1.29 1.94 2.58

 KBAR

KPA

KSP

 KBAS

 KCEub

 KCEma

 KPER

KFG

 KCEpa

 LPET

 LBAUb

LFG

 PBAR

 PCEib

 PCEub

Fig. 10.   Inter-population morphological distance of mature leaf outlines in nine populations of  Monstera adansonii   var.klotzschiana (KBAR, KBAS, KCEma, KCEpa, KCEub, KFG, KPER, KSP), three populations of M. adansonii  var.   laniata (LBAUb, LFG,LPET) and three populations of   M. praetermissa   (PBAR, PCEib, PCEub). Neighbor-Joining tree computed from a matrix ofgeneralised distance between population centroids derived from canonical variates analysis of six shape variables. Shape variables(principal components) computed by principal components analysis from a matrix of elliptic Fourier coefficients. Computed withNTSYSpc ver. 2.2.

207COMPARATIVE MORPHOLOGY OF POPULATIONS OF   MONSTERA ADANS.

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 16/19

divergence which is often significant (Table 2). All themost significant inter-population differences involvedCeará populations, and the degree of morphologicaldivergence between the Ceará brejo populations isgreater than between most population pairs from themajor forest blocks in Amazonia and the AtlanticForest). The Ceará populations do not cluster togeth-er as an exclusive group, but rather are mixed with

populations from Amazonia and the Atlantic forest  without an obvious pattern (Fig.   10). Broadly speak-ing, the leaf outline data was successful as a markerfor the taxa (species and varieties) but showed noclear geographical patterns at population level within the best sampled taxon, var.  klotzschiana . Thiscontrasts with our AFLP results (Andrade et al. 2007),in which the molecular data revealed clear geograph-ical structure.

This study demonstrates that leaf outline shapealone is suf ficiently effective as a taxonomic marker tocluster the populations of these three taxa into theircorrect hierarchical taxonomic order successfully,

using mature leaves. Although population differences were weaker, significant divergence was observed. Leaf outline shape, treated quantitatively can be a usefultool for taxonomic comparisons down to populationlevel. This is somewhat surprising to anyone familiar with the great variability of the leaves of   Monstera 

adansonii . Even more surprising is that the smaller leaf size classes are almost as successful in discriminating the taxa.

In var.  klotzschiana  the generally lower variability of  juvenile leaf shape, combined with its power todiscriminate the three taxa, is potentially of practical

 value. In the damaged brejo forests, we observed that  where the forest had lost its continuous canopy through human disturbance (e.g. São Benedito),root-climbers like   Monstera   were represented almost entirely by terrestrial and rupicolous juvenile forms,because of the absence of suitable host trees. In thissituation, discrimination of juvenile forms of different species is of practical importance in making invento-

ries of endangered biodiversity.Clearly, a wider range of quantitative characters isneeded to  “capture” the morphology more completely.Leaf shape plays a significant (but by no meansexclusive) role in the taxonomic definition of the taxa we studied and therefore our  finding that leaf outlineshape discriminates the taxa could be seen as a circularargument. In our defence however, we would point out that our   field identification of vegetative individualsrelied on a range of other characters including habitat,stem morphology, leaf texture, leaf perforation pat-terns and colour of the dried leaves. Furthermore thisstudy used quantitative estimates of variability and

difference in leaf shape and thus at the very least it provides verification of grouping behaviour of amorphological feature that is usually deployed intui-tively in alpha taxonomy. Working from  field-collecteddigital images proved to be a practical and effective way to gather large samples from which quantitative datacould be captured semi-automatically. If extended toother quantitative characters recoverable from images,this approach could bring detailed and sophisticatedmorphological studies at population level within thereach of many workers, making it possible to work fromlarge samples without permanently damaging popula-

PCo1

-0.0026 -0.0014 -0.0003 0.0009 0.0021

PCo2

-0.0021

-0.0011

-0.0002

0.0008

0.0018

KBAR

KBAS

KCEma

KCEpa

KCEub

KFG

KPA KPER

KSP

Fig. 11.   Inter-population morphological distance of mature leaf outlines in nine populations of  Monstera adansonii   var.klotzschiana (KBAR, KBAS, KCEma, KCEpa, KCEub, KFG, KPER, KSP). Principal coordinates analysis (PCoA) with minimum spanningtree superimposed computed from a matrix of Euclidean distance between population mean vectors, derived from a matrix of sixshape variables. Shape variables (principal components) computed by principal components analysis from a matrix of elliptic Fourier

coefficients. Computed with NTSYSpc ver. 2.2.

208   KEW BULLETIN VOL. 63(2)

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 17/19

tions by removing significant numbers of plants forherbarium vouchering. This is increasingly a concernfor biologists who study surviving populations in smalland threatened habitats like the brejo forests of Northeast Brazil.

Acknowledgements We are most grateful to the following organisations andpersons for their essential help: FUNCAP (FundaçaoCearense de Amparo à Pesquisa), FAPESB (Fundaçãode Amparo a Pesquisa do Estado da Bahia, WWF Brasil,Kew Latin America Research Fellowships Programme(Royal Botanic Gardens, Kew), Margaret Mee Fellow-ships Programme (Fundação Botânica Margaret Mee,Rio de Janeiro, Royal Botanic Gardens, Kew), Minis-tério do Meio Ambiente (IBAMA, permit number 001/2004, 042/04), Conselho Nacional de Desenvolvi-mento Científica e Tecnológica (CNPq, MCT permit 

number, 168/01), Dr Ana Maria Giulietti, Dr RobynCowan, Anna Haigh, Dr Jean-Jacques de Granville andthe staff of the CAY herbarium in French Guiana,Bertrand Goguillon and the staff of IRD (FrenchGuiana) for permission to collect, Dr Denis Barabé,Thales Alves Ribeiro, Sr. José Teodoro Soares, Rectorof the Universidade Estadual do Vale do Acaraú (UVA),Sobral-CE.

Appendix

List of voucher specimens.

 Voucher specimens are deposited in the following herbaria: EAC = Universidade Federal do Ceará,Fortaleza, Ceará, Brazil; HUEFS = Universidade Estadualde Feira de Santana, Bahia, Brazil; K = Royal BotanicGardens, Kew, United Kingdom; UVA = Herbarium of the Universidade Estadual Vale do Acaraú, Sobral, Ceará.Population codes are given in bold.

Monstera adansonii   Schott var.   klotzschiana   (Schott)MadisonKBAR : Brazil, Bahia, Serra da Copioba:  I. M. Andrade 

2771 (HUEFS),   I. M. Andrade   2772 (HUEFS),   I. M.

Andrade   2783 (HUEFS),   I. M. Andrade   2784 (EAC),I. M. Andrade  1996 (HUEFS).KBAS: Brazil, Bahia, Ubaitaba:   I. M. Andrade   2075(HUEFS),   I. M. Andrade   2077 (EAC),   I. M. Andrade 

2078 (HUEFS),  I. M. Andrade  2079 (HUEFS).KCEma: Brazil, Ceará, Maranguape:  I. M. Andrade  867(UVA).KCEpa : Brazil, Ceará, Pacatuba:   I. M. Andrade   685(HUEFS),  I. M. Andrade  706 (EAC),  I. M. Andrade  707(HUEFS).KCEub: Brazil, Ceará, São Benedito:  I. M. Andrade  602(EAC),  I. M. Andrade  605 (EAC),   I. M. Andrade  2613

(HUEFS). Ubajara: I. M. Andrade  2677 (HUEFS), I. M.

Andrade  2680 (HUEFS),  I. M. Andrade  2709 (HUEFS),I. M. Andrade  2722 (HUEFS).KESli:  Brazil, Espírito Santo, Linhares:   I. M. Andrade 

1618 (EAC),  I. M. Andrade  1622 (EAC),  I. M. Andrade 

1628 (EAC).

KESst: Brazil, Espírito Santo, Santa Tereza:  I. M. Andrade 764 (HUEFS),  I. M. Andrade  767 (HUEFS).KFG: French Guiana, Petit Saut:   I. M. Andrade   2800(EAC, HUEFS, K), I. M. Andrade  2802 (EAC, HUEFS, K).KPA : Pará, Acará:   I. M. Andrade   2385 (HUEFS),   I. M.

Andrade  2388 (HUEFS).KPER : Brazil, Pernambuco, Recife:  I. M. Andrade  1881(HUEFS).KRJ: Brazil, Rio de Janeiro, Vista Chinesa:  I. M. Andrade 

2222 (HUEFS),   I. M. Andrade   2228 (HUEFS),   I. M.

Andrade  2230 (HUEFS).KSP:   Brazil, São Paulo, Bertioga:   I. M. Andrade  2121(HUEFS),  I. M. Andrade  2137 (HUEFS).

Monstera adansonii  Schott var.  laniata  (Schott) MadisonLBAIlh:  Brazil, Bahia, Ilhéus, Centro de Pesquisas doCacau (CEPEC):  I. M. Andrade  1983 (UVA).LBAUb:   Brazil, Bahia, Ubaitaba:   I. M. Andrade   1962(UVA).LFG: French Guiana, Cayenne: I. M. Andrade  2815 (K).LPET:  Brazil, Pernambuco, Tapacurá:   I. M. Andrade 

1933 (UVA).

Monstera praetermissa  E. G. Gonç. & TemponiPBAR:  Brazil, Bahia, Serra da Copioba:   I. M. Andrade 

2740 (UVA).PCEbat: Brazil, Ceará, Serra de Baturité,  I. M. Andrade 

s.n. (UVA).PCEib: Brazil, Ceará, Ibiapina, I. M. Andrade  2529 (UVA).PCEub: Brazil, Ceará, Ubajara, I. M. Andrade  2630 (UVA).

References Ab’Sáber, A. N. (1982). The paleoclimate and paleo-

ecology of Brazilian Amazônia. In: G. T. Prance(ed.), Biological diversification in the Tropics.Columbia University Press, New York. Pp. 41   – 59.

 Andrade, I. M. & Mayo, S. J. (1998). Dynamic shoot morphology in   Monstera adansonii   Schott var.klotzschiana   (Schott) Madison (Araceae ). Kew Bull.532: 399   –  417.

 ____, ____, Van den Berg, C., Fay, M., Chester, M.,Lexer, C. & Kirkup, D. (2007). Genetic variation inpopulations of   Monstera adansonii   Schott (Araceae )from natural forest fragments in Northeast Brazilestimated with AFLP molecular markers. Ann. Bot.100: 1143   – 1154.

 Andrade-Lima, D. de (1982). Present-Day Forest Refuges in Northeastern Brasil. In: G. T. Prance

209COMPARATIVE MORPHOLOGY OF POPULATIONS OF   MONSTERA ADANS.

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 18/19

(ed.), Biological Diversification in the Tropics.Columbia University Press, New York. Pp. 245   –

251.Borges-Nojosa, D. M. & Caramaschi, U. (2003). Com-

posição e Análise Comparativa da Diversidade e das A finidades Biogeográ ficas dos Lagartos e Anfisbení-

deos (Squamata) dos Brejos Nordestinos. In: I. Leal, J. M. C. Silva & M. Tabarelli (eds.), Ecologia eConservação da Caatinga. ed. 1, 1: 489  –  540. Recife.

Cannon, C. H. & Manos, P. S. (2001). Combining andcomparing morphometric shape descriptors with amolecular phylogeny: the case of fruit type evolu-tion in Bornean  Lithocarpus  ( Fagaceae ). Syst. Biol. 50(6): 860   – 880.

Ferson, S., Rohlf, F. J. & Koehn, R. K. (1985).Measuring shape variation of two-dimensional out-lines. Syst. Zool. 34(1): 59   – 68.

Figueiredo, M. A. (1997). Vegetação. In: FundaçãoInstituto de Planejamento do Ceará (IPLANCE).

 Atlas do Ceará. Fortaleza. Pp. 28   – 29.Furuta, N., Ninomiya, S., Takahashi, S., Ohmori, H. &

Ukai, Y. (1995). Quantitative evaluation of soybean(Glycine max  (L.) Merr.) leaflet shape by principalcomponent scores based on elliptic Fourier descrip-tors. Breeding Sci. 45: 315   – 320.

GIMP (2006).  http://www.gimp.org . Version 2.2.4 by S. Kimball & P. Mattis.

Hammer, Ø., Harper, D. A. T. & Ryan, P. D. (2001).PAST: Palaeontological Statistics software package foreducation and data analysis. Palaeontologia Electron-ica 41: 9;  http://folk.uio.no/ohammer/past 

IBGE (1985). Atlas nacional do Brasil. IBGE. Rio de

 Janeiro. ____ (2005)  http://www.ibge.gov.br/home/presidencia/

noticias/noticia_impressão.php?id_noticia 06/07/2005

Iwata, H. & Ukai, Y. (2002). SHAPE: A computerprogram package for quantitative evaluation of biological shapes based on elliptic Fourier descrip-tors. J. Hered. 93: 384   – 385.

 ____, Nesumi, H., Ninomiya, S., Takano, Y. & Ukai, Y.(2002). The evaluation of genotype x environment interactions of citrus leaf morphology using imageanalysis and elliptic Fourier descriptors. Breeding Sci. 52: 243   – 251.

 Jensen, R. J., Ciofani, K. M. & Miramontes, L. C. (2002).Lines, outlines, and landmarks: morphometric analy-ses of leaves of   Acer rubrum ,   Acer saccharinum 

(Aceraceae ) and their hybrid. Taxon 51: 475   – 492.Kessler, S. & Sinha, N. (2004). Shaping up: the genetic

control of leaf shape. Curr. Opin. Pl. Biol. 7: 65  – 72.Madison, M. (1977). A revision of  Monstera  (Araceae ).

Contr. Gray Herb. 207: 1   – 101.Mayo, S. J. (1983). Aspectos da fitogeografia das Araceae 

bahianas. In: Anais XXXIV Congresso Nacional de

Botânica do Brasil, Porto Alegre, Brasil 2: 215   –

227. Diretoria da Sociedade, Porte Alegre.McLellan, T. (1993). The roles of heterochrony and

heteroblasty in the diversification of leaf shapes inBegonia dregei  (Begoniaceae ). Amer. J. Bot. 807: 796   –

804.

Oliveira, P. E., Barreto, A. M. F. & Suguio, K. (1999).Late Pleistocene/Holocene climatic and vegeta-tional history of the Brazilian caatinga: the fossildunes of the middle São Francisco River. Palaeo-geogr., Palaeoclimatol., Palaeoecol. 152: 319   –

337.Pôrto, K. C., Cabral, J. J. P. & Tabarelli, M. (2004).

Brejos de altitude em Pernambuco e Paraíba:história natural, ecologia e conservação. Ministériodo Meio Ambiente, Brasília.

Raveloson, H., Le Minor, J.-M., Rumpler, Y. &Schmittbuhl, M. (2005). Shape of the lateral mandib-ular outline in   Lemuridae : a quantitative analysis of 

 variability using elliptical Fourier analysis. FoliaPrimatol. 76: 245   – 261.

Ray, T. (1983).  Monstera tenuis   (Chirravaca, Mano deTigre, Monstera). Pp. 278   –  280 In: D. H. Janzen,Costa Rican Natural History. University of ChicagoPress, Chicago.

Rodal, M. J. N., Sales, M. F. & Mayo, S. J. (1998).Florestas Serranas de Pernambuco: Localização dosremanescentes dos brejos de altitude. ImprensaUniversitária, Universidade Federal Rural de Per-nambuco, Recife.

Rohlf, F. J. (2004). TpsDig Version 1.40. Digitizing software.  http://life.bio.sunysb.edu/morph/

 ____ (2005). NTSYSpc: Numerical Taxonomy System ver. 2.20d, Exeter Publishing, Ltd. Setauket, NY.http://www.exetersoftware.com/cat/ntsyspc/ntsyspc.html

 ____ (2006). Morphometrics at SUNY Stony Brook: website http://life.bio.sunysb.edu/morph/

 ____ & Archie, J. W. (1984). A comparison of Fouriermethods for the description of wing shape inmosquitoes ( Diptera: Culicidae ). Syst. Zool. 33(3):302   –  317.

Rumpunen, K. & Bartish, I. V. (2002). Comparison of differentiation estimates based on morphometric andmolecular data, exemplified by various leaf shape

descriptors and RAPDs in the genus   Chaenomeles (Rosaceae ). Taxon 51: 69   – 82.

Sales, M. F., Mayo, S. J. & Rodal, M. J. N. (1998).Plantas vasculares das  florestas serranas de Pernam-buco: Um checklist da   flora ameaçada dos brejosde altitude, Pernambuco Brasil. Imprensa Universi-tária, Universidade Federal Rural de Pernambuco,Recife.

Sampaio, V. S. B., Giulietti, A. M., Virginio, J. & Gamarra-Rojas, C. F. L. (2002). Vegetação e  flora da caatinga.

210   KEW BULLETIN VOL. 63(2)

© The Board of Trustees of the Royal Botanic Gardens, Kew, 2008

8/12/2019 Andrade Et Al. - 2008 - Comparative Morphology of Populations of Monstera Adans-libre

http://slidepdf.com/reader/full/andrade-et-al-2008-comparative-morphology-of-populations-of-monstera-adans-libre 19/19

 Associação Plantas do Nordeste (APNE), CentroNordestino de Informação Sobre Plantas (CNIP),Recife.

Slice, D. (2008). Morpheus et al. Multiplatform softwarefor morphometric research.   http://www.morphometrics.org/id6.html

Tatsuta, H., Mizota, K. & Akimoto, S.-I. (2004).Relationship between size and shape in the sexually dimorphic beetle   Prosopocoilus inclinatus   (Coleoptera: 

Lucanidae ). Biol. J. Linn. Soc. 81: 219   – 233.

 Yamanaka, N., Ninomiya, S., Hoshi, M., Tsubokura, Y., Yano, M., Nagamura, Y., Sasaki, T. & Harada, K.(2001). An informative linkage map of soybeanreveals QTLs for  flowering time, leaflet morphology and regions of segregation distortion. DNA Res. 8:61   – 72.

 Yoshioka, Y., Iwata, H., Ohsawa, R. & Ninomiya, S.(2004). Analysis of petal shape variation of  Primula 

sieboldii  by elliptic Fourier descriptors and principalcomponent analysis. Ann. Bot. 94: 657   – 664.

211COMPARATIVE MORPHOLOGY OF POPULATIONS OF   MONSTERA ADANS.