Diversity and Distribution
of Amphibians in Luxembourg
Laura R. Wood
Thesis submitted for the degree of
Doctor of Philosophy
February 2011
The Durrell Institute of Conservation and Ecology
University of Kent, Canterbury, CT2 7NR, England
i
Research and printing funded by:
Project collaborators:
ii
Abstract
Amphibians are highly threatened on a global scale, with more species considered
to be in decline than in any other vertebrate class. There are still gaps in our
knowledge of amphibian ecology, distribution and status, even in developed
countries where investment in research and scientific output are high. In
Luxembourg there is an extensive protected area (PA) network, which has been
created to meet obligations under European and national law. Luxembourg has 12
amphibian species, five of which are present in more than a quarter of surveyed 1
km grid squares and seven that are only present in < 15%, including two in < 1%.
The national PA system was found to cater well for the common species – records
of their presence both inside and outside these areas indicated that they were
accommodated by, but not restricted to, PAs. Record centre data from pre-1985
and 1996-2005 indicated that common species‟ range sizes had increased or
remained stable, while four of the less common and rare species had decreased,
although Alytes obstetricans had more than doubled its range size. Since 2005 the
three very rare species (Hyla arborea, Bufo calamita and Bombina variegata)
have occupied only four sites between them, all of which are protected. The most
important component of local landscape variables in predicting species records
was „forest and elevation‟. Amphibian species richness in breeding ponds was
found to show potential as an indicator of aquatic plant species richness, but no
relationship with invertebrate species richness was found. Occupancy modelling
was used to compare the detectability of different species and to evaluate survey
methods. With sexes and life stages combined, netting was not found to be
reliable for any species; aural survey was good for anurans with extended
breeding periods; trapping was the most successful method for newts and visual
searches achieved moderate to high detection rates for all species. Occupancy
modelling has the potential to transform the design of survey protocols for all taxa
with imperfect detection rates. The greater investment required to collect higher
quality data for modelling was considered worthwhile, due to the time it could
save through improved survey protocols. Amphibians‟ cryptic and condition-
dependent behaviour often makes them difficult to survey for, thus necessitating
careful survey design. Models developed here are an important step towards
advancing ecological survey protocol design and maximising returns from survey
efforts.
Keywords: amphibian; landscape covariates; Luxembourg; macroinvertebrates;
occupancy modelling; plants; protected areas; range size.
iii
Acknowledgements
I have had so much help and support from so many colleagues, friends and family
in the last few years, which I can only begin to address here. Firstly, my
collaborators and funders in Luxembourg, most especially Edmée Engel (Musée
national d‟histoire naturelle) and Dr Laurent Schley (Administration de la Nature
et des Forêts), who conceived the project, acted as in-country guides and
fieldwork „fixers‟ and provided generous financial support, and of course the FNR
(Fonds National de la Recherche Luxembourg) for funding my studentship and
attendance at conferences. I am also very grateful to the Amphibian Conservation
Research Trust (ACRT) for a further grant to fund fieldwork equipment.
At the Durrell Institute of Conservation and Ecology (DICE) I‟ve had extensive
and very patient support from my supervisor Prof. Richard Griffiths and also Dr
David Sewell and Dr Bob Smith with the applied modelling and mapping. I am
also very grateful to Guru Guillera-Arroita, José Lahoz-Monfort and Rachel
McCrea at the National Centre for Statistical Ecology for their specialist help with
occupancy modelling, and for not making me feel too stupid! Other good friends
and colleagues at DICE, most notably Nina Cornish, Leida dos Santos, Steve
Green, Brett Lewis, Susy Paisley and Claire Raisin, have been a tremendous
source of encouragement, entertainment, blanket dances, pizza and gin. Mrs
Roche and the menagerie at Thanington Court (including some of the afore-
mentioned) provided the most wonderful home / retreat in Canterbury and a
plentiful supply of funny animal and river antics.
During fieldwork I was very lucky to have some fantastic assistants, namely Liza
Glesener, Jan Herr, Véronique Ludwig, Masse Sylla, Heléne Smuk-Matringe and
Lucia Cabete Rodrigues, who made the fieldwork not only possible, but
enjoyable, and made me feel at home in Luxembourg.
Last but not least, I would like to say a special thank you to Grannie Banks who
instilled in me a love of natural history and my whole family who‟ve shown an
un-wavering belief in me and my PhD quest. Especially my parents Phil and
Jennie, sisters Amy and Sally, and „little‟ cousin Alicia whose bespoke newt traps
and company and help with fieldwork in Luxembourg were invaluable.
iv
Contents
Abstract ............................................................................................................... ii
Acknowledgements ............................................................................................. iii
Contents.............................................................................................................. iv
Abbreviations and Acronyms ............................................................................ viii
List of Tables ....................................................................................................... x
List of Figures .................................................................................................... xii
1. General introduction ........................................................................................ 1
1.1 Amphibian global status ............................................................................. 1
1.2 Luxembourg ............................................................................................... 3
1.2.1 Location and geography....................................................................... 3
1.3 Conservation legislation ............................................................................. 4
1.3.1 Influence of political boundaries .......................................................... 6
1.3.2 IUCN .................................................................................................. 7
1.4 Pond ecology.............................................................................................. 8
1.4.1 Amphibian surveying .......................................................................... 8
1.5 Project objectives ....................................................................................... 9
2. General methodology ..................................................................................... 11
2.1 Historical data - MNHN ........................................................................... 11
2.1.1 Map data ........................................................................................... 12
2.2 Study pond selection ................................................................................ 13
2.3 Amphibian surveying ............................................................................... 14
2.3.1 Visual encounter survey (VES) .......................................................... 15
2.3.2 Funnel trapping ................................................................................. 15
2.3.3 Dip netting ........................................................................................ 15
2.3.4 Aural ................................................................................................. 16
2.4 Data analysis ............................................................................................ 16
3. Landscape features as predictors of amphibian presence................................. 17
3.1 Introduction .............................................................................................. 17
3.1.1 Recording centres .............................................................................. 17
v
3.1.2 Landscape and amphibians ................................................................ 18
3.2 Methods ................................................................................................... 19
3.3 Results ..................................................................................................... 20
3.3.1 Historical data descriptives ................................................................ 20
3.3.2 Landscape predictors of amphibian species‟ occurrence ..................... 23
3.3.3 Landscape predictors of amphibian diversity ..................................... 27
3.4 Discussion ................................................................................................ 28
4. Amphibian range size change and the effectiveness of Luxembourg‟s protected
areas .................................................................................................................. 32
4.1 Introduction .............................................................................................. 32
4.1.1 Protected areas................................................................................... 32
4.1.2 Range size ......................................................................................... 34
4.2 Methods ................................................................................................... 35
4.2.1 Historical data ................................................................................... 35
4.2.2 Protected areas................................................................................... 35
4.2.3 Protected area species richness analyses ............................................ 37
4.2.4 Range size ......................................................................................... 37
4.3 Results ..................................................................................................... 38
4.3.1 Amphibians and protected areas ........................................................ 38
4.3.2 Range size change ............................................................................. 40
4.4 Discussion ................................................................................................ 43
4.4.1 Protected areas and amphibian species richness ................................. 43
4.4.2 Range size changes ............................................................................ 46
5. Amphibians as indicators of plant species richness ......................................... 47
5.1 Introduction .............................................................................................. 47
5.2 Methods ................................................................................................... 50
5.2.1 Study area.......................................................................................... 50
5.2.2 Plant survey method .......................................................................... 50
5.2.3 Amphibian data ................................................................................. 51
5.2.4 Analysis ............................................................................................ 51
5.3 Results ..................................................................................................... 52
5.3.1 Plant community descriptives and pond size effect ............................ 52
5.3.2 Amphibians as indicators of plant richness and vegetation cover ....... 54
5.3.3 Wetland plant richness and amphibian occupancy.............................. 55
vi
5.4 Discussion ................................................................................................ 57
6. Amphibian species richness as an indicator of macroinvertebrate family
richness .............................................................................................................. 61
6.1 Introduction .............................................................................................. 61
6.2 Methodology ............................................................................................ 63
6.2.1 Study area.......................................................................................... 63
6.2.2 Macroinvertebrate survey method ...................................................... 64
6.2.3 Amphibian data ................................................................................. 65
6.2.4 Analysis ............................................................................................ 65
6.3 Results ..................................................................................................... 67
6.3.1 Macroinvertebrate descriptives .......................................................... 67
6.3.2 Pond size and macroinvertebrate numbers and family richness .......... 69
6.3.3 Macroinvertebrates and wetland plant richness and extent ................. 69
6.3.4 Amphibians as indicators of macroinvertebrate diversity ................... 70
6.3.5 Macroinvertebrate richness and amphibian species occupancy ........... 71
6.3.6 Specific relationships between macroinvertebrate and amphibian
species ....................................................................................................... 73
6.4 Discussion ................................................................................................ 74
7. Absent or not detected? Development of an occupancy modelling approach to
amphibian survey design .................................................................................... 77
7.1 Introduction .............................................................................................. 77
7.1.1 Comparing survey methodologies ...................................................... 77
7.1.2 Detectability ...................................................................................... 78
7.1.3 Occupancy modelling ........................................................................ 79
7.1.4 Covariates ......................................................................................... 80
7.2 Methods ................................................................................................... 81
7.2.1 Study area and amphibian data .......................................................... 81
7.2.2 Covariate measurements .................................................................... 81
7.2.3 Program PRESENCE analyses .......................................................... 84
7.2.4 Comparing survey methods ............................................................... 85
7.2.5 Predictors of occupancy and detection ............................................... 86
7.2.6 Model selection ................................................................................. 86
7.2.7 Number of survey visits required ....................................................... 88
7.3 Results ..................................................................................................... 89
vii
7.3.1 Detectability by different survey methods .......................................... 89
7.3.2 Predictors of occupancy and detection ............................................... 91
R. temporaria ......................................................................................... 92
Pelophylax spp. ...................................................................................... 96
A. obstetricans ........................................................................................ 98
I. alpestris .............................................................................................. 99
T. cristatus ........................................................................................... 101
L. helveticus ......................................................................................... 105
L. vulgaris ............................................................................................ 109
7.4 Discussion .............................................................................................. 113
7.4.1 Covariates of detection .................................................................... 114
7.4.2 Covariates of occupancy .................................................................. 114
7.4.3 Occupancy modelling ...................................................................... 115
8. General discussion ....................................................................................... 117
8.1.1 Future data collection ...................................................................... 120
9. References ................................................................................................... 122
10. Appendices ................................................................................................ 147
Appendix 1 – Amphibian nomenclature ....................................................... 148
Appendix 2 – Contributors to the MNHN database....................................... 150
Appendix 3 – Study ponds ........................................................................... 151
Appendix 4 – Species dates .......................................................................... 152
Appendix 5 – Species distribution maps ....................................................... 153
Appendix 6 – Plants species surveyed .......................................................... 155
Appendix 7 – Macroinvertebrate families surveyed ...................................... 159
Appendix 8 – Occupancy modelling covariates ............................................ 161
Appendix 9 – Publications ........................................................................... 163
Wood, L. R., Griffiths, R. A. Groh, K., Engel, E. & Schley, L. (2008)
Interactions between freshwater mussels and newts: a novel form of
parasitism? Amphibia-Reptilia, 29, 457-462. ............................................ 163
Wood, L. R., Griffiths, R. A. & Schley, L. (2009) Amphibian
chytridiomycosis in Luxembourg. Bulletin de la Société des Naturalistes
Luxembourgois, 110, 109-114. ................................................................. 170
viii
Abbreviations and Acronyms
AIC Delta AIC
ACT Administration du Cadastre et de la Topographie
AIC Akaike‟s Information Criterion
AICc Akaike‟s Information Criterion – adjusted for small sample size
Ao Alytes obstetricans (European midwife toad)
BAP Biodiversity Action Plan
Bb Bufo bufo (common toad)
Bc Bufo calamita (natterjack toad)
BMWP Biological Monitoring Working Party
BS Bray-Curtis Dissimilarity Index
Bv Bombina variegata (yellow-bellied toad)
ĉ C-hat, variance inflation factor (in occupancy models)
CBD Convention on Biological Diversity
EC European Commission
EU European Union
GAA Global Amphibian Assessment
Ha Hyla arborea (European tree frog)
Ia Ichthyosaura alpestris (Alpine newt)
IBEM Indice de Biodiversité des Etangs et Mares
IUCN International Union for Conservation of Nature
KARCH Koordinationsstelle für Amphibien- und Reptilienschutz in der
Schweiz (Coordination centre for the protection of amphibians and
reptiles in Switzerland)
km kilometres
km2 kilometres squared
Lh Lissotriton helveticus (palmate newt)
Lv Lissotriton vulgaris (smooth newt)
m metres
m.a.s.l. metres above sea level
m2 metres squared
MNHN(L) Musée national d‟histoire naturelle (Luxembourg)
NARRS National Amphibian and Reptile Recording Scheme
ix
NPS National Pond Survey
p Detectability (in occupancy models)
PA Protected Area
PC Principal Component
PCA Principal Component Analysis
PLOCH Plans d‟eau suisses
PNPN Plan national concernant la protection de la nature
P.spp Pelophylax species (green frogs: P. lessonae and P. kl. esculentus)
PSYM Predictive System for Multimetrics
QAIC Quasi Akaike‟s Information Criterion
QAICc Quasi Akaike‟s Information Criterion (adjusted for small sample
size)
RAVON Reptilien Amfibieën Vissen Onderzoek Nederland (Reptile,
Amphibian and Fish Conservation Netherlands)
RIVPACS River Invertebrate Prediction and Classification System
Rt Rana temporaria (common frog)
SAC Special Area of Conservation
s.e. Standard error
SPSS Statistical Package for the Social Sciences
Ss Salamandra salamandra (fire salamander)
Tc Triturus cristatus (great crested newt)
UNEP United Nations Environment Programme
VES Visual Encounter Survey
WDPA World Database of Protected Areas
ψ Occupancy
x
List of Tables
Table 1.1 Luxembourgish amphibians on the Bern Convention and Habitats
Directive appendices and their IUCN Red List status and global population trend.
............................................................................................................................ 5
Table 2.1 Landscape features drawn in ArcMap and extracted to quantify habitat
in each 1 km grid square. ................................................................................... 12
Table 2.2 Survey methods appropriate to each amphibian species by life stage and
sex. .................................................................................................................... 16
Table 3.1 Multivariate factor loadings of principal components with eigenvalues
> 1. .................................................................................................................... 23
Table 3.2 Final models from stepwise logistic regression for anurans. ............... 24
Table 3.3 Final models from stepwise logistic regression for newts. .................. 26
Table 3.4 Multiple logistic regression chi-square values for landscape features
and amphibian species richness. ......................................................................... 27
Table 4.1 Mann-Whitney U tests comparing amphibian species richness in PAs
and non-PAs. ..................................................................................................... 38
Table 4.2 Index of relative range size change: proportion of grid squares occupied
in the early and later survey periods. .................................................................. 43
Table 5.1 The five most commonly occurring wetland plants across all study
ponds (n = 43). ................................................................................................... 53
Table 5.2 Relationships between individual amphibian and plant species, before
and after sequential Bonferroni correction. ......................................................... 57
Table 6.1 The amphibian species present in ponds surveyed for
macroinvertebrates. ............................................................................................ 63
Table 6.2 The five most numerous macroinvertebrate families across all 11 ponds
surveyed. ........................................................................................................... 67
Table 6.3 The most commonly occurring macroinvertebrate families across all 11
ponds surveyed. ................................................................................................. 68
Table 6.4 Pearson's product-moment correlation results for wetland vegetation
and macroinvertebrates, including all ponds surveyed for invertebrates (n = 11). 70
Table 6.5 One-way ANOVA values comparing the invertebrate predator total
count and the number of families and species in ponds occupied / not occupied by
moderately common amphibian species. ............................................................ 72
Table 6.6 Moderately common macroinvertebrates exhibiting a significant
relationship with amphibian species‟ occupancy (Nponds = 11) before and after
sequential Bonferroni correction. ....................................................................... 74
Table 7.1 Site-specific covariates concerning the occupancy of other species and
the method of normalisation. .............................................................................. 82
Table 7.2 Site-specific habitat covariates and the method of normalisation. ....... 83
Table 7.3 Survey-specific covariates and the method of normalisation............... 84
xi
Table 7.4 Most parsimonious models for R. temporaria surveys in 2007 and 2008
by three survey methods..................................................................................... 93
Table 7.5 Most parsimonious models for Pelophylax spp. surveys in 2007 and
2008 by four survey methods. ............................................................................ 96
Table 7.6 Most parsimonious models for A. obstetricans 'year 1' data using aural
surveys............................................................................................................... 99
Table 7.7 Most parsimonious models for I. alpestris surveys in 2007 and 2008 by
three survey methods. ...................................................................................... 100
Table 7.8 Most parsimonious models for T. cristatus surveys in 2007 and 2008 by
three survey methods. ...................................................................................... 102
Table 7.9 Most parsimonious models for L. helveticus surveys in 2007 and 2008
by three survey methods................................................................................... 106
Table 7.10 Most parsimonious models for L. vulgaris surveys in 2007 and 2008
by three survey methods................................................................................... 110
Table 10.1 Nomenclature of study species (following Speybroeck et al., 2010).
........................................................................................................................ 149
Table 10.2 Study ponds in 25 m and 200 m groupings. .................................... 151
Table 10.3 Survey dates in 2007. ..................................................................... 152
Table 10.4 Survey dates in 2008. ..................................................................... 152
Table 10.5 Survey dates in 2009 – additional ponds used in „year 1‟ data for A.
obstetricans. ..................................................................................................... 152
Table 10.6 All wetland plants surveyed and the percentage of study ponds in
which each plant occurred (names following Stace, 1997). ............................... 155
Table 10.7 Non-wetland plants surveyed and the percentage of study ponds in
which each plant occurred (names following Stace, 1997). ............................... 157
Table 10.8 The macroinvertebrates found and their frequency of occurrence
(names following ZipcodeZoo.com, 2009). ...................................................... 159
Table 10.9 Summary of survey-specific covariate values for each study year... 161
Table 10.10 Summary of site-specific covariates for each study year. .............. 162
Table 10.11 Site-specific variables concerning the presence / non-detection of
other species. ................................................................................................... 162
xii
List of Figures
Figure 1.1 Luxembourg. ...................................................................................... 3
Figure 2.1 Study pond locations in Luxembourg. .............................................. 14
Figure 3.1 The number of amphibian species recorded in each 1 km grid square,
(a) 1955-2005, (b) 1996-2005. ........................................................................... 21
Figure 3.2 Comparison of the mean area or length of landscape features in
squares surveyed for amphibians between 1996 and 2005 and those not surveyed.
.......................................................................................................................... 22
Figure 3.3 The percentage of 1 km grid squares surveyed (n = 654) between 1996
and 2005 occupied by each amphibian species. .................................................. 22
Figure 3.4 Scree plot of the eigenvalues of all components. ............................... 24
Figure 3.5 Mean values of the significant principal components predicting anuran
species‟ occupancy in 1 km grid squares. ........................................................... 25
Figure 3.6 Mean values of the significant principal components predicting newt
species‟ occupancy in 1 km grid squares. ........................................................... 27
Figure 4.1 Map of Luxembourg: grid squares (1 km2) randomly selected and
retained for analysis. .......................................................................................... 36
Figure 4.2 Amphibian species richness in 1 km grid squares (all records 1996 –
2005) and all statutory protected areas in Luxembourg. ...................................... 39
Figure 4.3 The frequency distribution of amphibian species richness in non-PA
grid squares and (a) ≥ 50% PA, (b) ≥ 80% PA, (c) ≥ 90% PA and (d) ≥ 98% PA
grid squares. ....................................................................................................... 40
Figure 4.4 The percentage of 1 km grid squares at each PA coverage level
occupied by each amphibian species (all records 1996 – 2005). ......................... 41
Figure 4.5 Range sizes of pond-breeding amphibians recorded in the early and
later survey periods from 104 grid squares (a) untransformed 1 km squares and (b)
as logit-transformed proportions of the total. ...................................................... 42
Figure 4.6 Species percentage occupancy in grid squares surveyed in both
periods. .............................................................................................................. 42
Figure 5.3Figure 5.1 There was no correlation between the number of amphibian
species detected and the size of the study pond (r = 0.28, n = 34, p > 0.05). ....... 53
Figure 5.2 Diversity of wetland plant species in study ponds and the frequency of
occurrence (including plants not identified to species and plants at ≤ 5% and > 5%
cover). ............................................................................................................... 53
Figure 5.3 The number of wetland plant species (≤ 5% and > 5% cover) plotted
against pond size (logarithmic scales; y = 0.2675x + 0.8013). ............................ 54
Figure 5.4 The mean number (±s.e.) of wetland plant species and all plant species
(wetland and non-wetland) found in ponds according to amphibian species
diversity (plants of ≤ 5% and > 5% cover).......................................................... 54
Figure 5.5 The extent (%) of wetland plant cover (±s.e.) and the number of
amphibian species present. ................................................................................. 55
xiii
Figure 5.6 Mean number of wetland plant species in ponds occupied / not
occupied by moderately common amphibian species, plants of ≤ 5% and > 5%
cover included. A. obstetricans data include the additional ponds surveyed in
2009. .................................................................................................................. 55
Figure 5.7 Occupancy of Potamogetan natans in ponds by A. obstetricans
occupancy. ......................................................................................................... 56
Figure 6.1 The number of ponds that macroinvertebrate families occurred in. ... 68
Figure 6.2 The relationship between pond size and macroinvertebrate family
richness (y = 0.0033x + 3.1533) and the total count of macroinvertebrates (y =
0.1261x + 6.2569) (logarithmic scales; including pond 17). ............................... 69
Figure 6.3 The number of macroinvertebrate families in surveyed ponds by (a)
the number of wetland plant species present and (b) the extent of wetland plant
cover. ................................................................................................................. 70
Figure 6.4 (a) The mean count of macroinvertebrates and (b) the mean number of
macroinvertebrate families found in ponds according to the number of amphibian
species present. .................................................................................................. 71
Figure 6.5 Mean number of macroinvertebrate families in ponds occupied / not
occupied by moderately common amphibian species. ......................................... 72
Figure 6.6 The mean number of predatory invertebrate families with (a) the
number of amphibian species present and (b) amphibian species‟ occupancy. .... 73
Figure 7.1 „Detectability‟ (±s.e.) of anurans by survey method, ψ(.)
p(survey_method, max_temp). ............................................................................ 90
Figure 7.2 'Detectability' (±s.e.) of newts by survey method, ψ(.)
p(survey_method, max_temp). ............................................................................ 91
Figure 7.3 The mean number of adult and larval B. bufo seen or captured during
surveys, by day of year. ..................................................................................... 92
Figure 7.4 The mean number of adult and larval R. temporaria seen or captured
during surveys, by day of year. .......................................................................... 93
Figure 7.5 Detection rates (±s.e.) of R. temporaria by the most parsimonious
models, where water temperature was a covariate of detection. .......................... 94
Figure 7.6 Detection rates (±s.e.) of R. temporaria by the most parsimonious
models, where „clear shore‟ was a predictor of detection rate. ............................ 94
Figure 7.7 Detection rates (±s.e.) of R. temporaria by top models, where „day of
year‟ was a covariate of detection. ..................................................................... 95
Figure 7.8 Detection rates (±s.e.) of R. temporaria by trapping in 2008, where
time taken was a covariate of detection. ............................................................. 95
Figure 7.9 The mean number of adult and larval Pelophylax spp. seen or captured
during surveys by day of year. ........................................................................... 96
Figure 7.10 Detection rates (±s.e.) of Pelophylax spp. by the most parsimonious
models, where „day of year‟ was a covariate of detection rate. ........................... 97
Figure 7.11 Detection rates (±s.e.) of Pelophylax spp. by the most parsimonious
models, where water temperature was a covariate of detection. .......................... 98
xiv
Figure 7.12 The mean number of adult and larval I. alpestris seen or captured
during surveys by day of year. ......................................................................... 100
Figure 7.13 The detection rate (±s.e.) of VES for I. alpestris in 2008, where
„pm_timetaken‟ was a covariate of detection. ................................................... 101
Figure 7.14 Detection rates (±s.e.) of I. alpestris by top models where the day of
year was a covariate of the detection rate. ........................................................ 101
Figure 7.15 The detection rate (±s.e.) of VES for T. cristatus in 2007, where the
percentage of clear shore was a covariate of detection. ..................................... 103
Figure 7.16 The detection rate (±s.e.) of T. cristatus in the top 2007 trapping
model, where „max_temp‟ was a covariate of detection. ................................... 103
Figure 7.17 Naïve and model-estimated occupancy (ψ) rates (±s.e.) of T. cristatus
by survey method and year. .............................................................................. 105
Figure 7.18 Detection rates (±s.e.) of L. helveticus by top models where water
temperature was a covariate of detection rate. .................................................. 107
Figure 7.19 Detection rates (±s.e.) of L. helveticus by top models, where „time
taken‟ was a covariate of detection. .................................................................. 108
Figure 7.20 Naïve and model-estimated occupancy (ψ) rates (±s.e.) of L.
helveticus by survey method and year. ............................................................. 109
Figure 7.21 Detection rate (±s.e.) of L. vulgaris by the top trapping model in
2008, where „pm_timetaken‟ was a covariate of detectability. .......................... 110
Figure 7.22 Detection rates (±s.e.) of L. vulgaris by top models, where „day of
year‟ was a covariate of detection. ................................................................... 111
Figure 7.23 Occupancy of 2007 study ponds by T. cristatus and L. vulgaris. ... 112
Figure 7.24 Naïve and model-estimated occupancy (ψ) rates (±s.e.) of L. vulgaris
by survey method and year. .............................................................................. 112
Figure 10.1 Distributions of amphibian species in Luxembourg in 1 km squares,
1996-2005, data from the MNHN database. ..................................................... 153
1. General introduction
1
1. General introduction
General introduction
1.1 Amphibian global status
Amphibian populations globally have experienced alarming declines (Alford and
Richards, 1999). It has been estimated that up to a third of all known amphibian
species are facing decline or even extinction. They are more threatened than either
birds or mammals (Stuart et al., 2004). Since declines were first detected in the early
1980s and their extent and severity was collectively realised at the First World
Congress of Herpetology in 1989 (Barinaga, 1990; Skelly et al., 2003) the possible
causes have been a hot topic of research (Collins and Halliday, 2005). However the
threats facing amphibians are often complex (Daszak et al., 1999; Corser, 2001;
Kiesecker et al., 2001) and they are difficult to differentiate because they interact
with each other (Blaustein et al., 2003; Collins and Storfer, 2003).
Many of the challenges faced by amphibians are directly or indirectly caused by
humans and have occurred too rapidly for species to adapt. The extinction rate across
all living organisms is 100 to 1000 times greater now than before global colonisation
by humans (Pimm et al., 1995), but it soared in the 20th century largely due to habitat
destruction (Pimm et al., 2006). Amphibians often have specific habitat requirements
and, compared to other vertebrates, cannot – or do not – disperse over great distances
(Duellman and Trueb, 1986), either due to physical limitation or inflexible
behavioural patterns (Stebbins and Cohen, 1995). The key threats directly inflicted
by human activities include habitat loss and degradation, for example by clear-felling
of mature forests, draining of wetlands and urban expansion (e.g. Semlitsch, 2000;
Wood et al., 2003; Hamer and McDonnell, 2008), environmental pollution (e.g.
Beebee et al., 1990; Rouse et al., 1999) and over-harvesting for food (e.g. Warkentin
et al., 2009). Equally, indirect anthropogenic impacts include predation by
introduced species (e.g. Kats and Ferrer, 2003), disease (e.g. Carey et al., 1999; Lips
et al., 2006), increased ultra-violet radiation (e.g. Blaustein et al., 2003) and climate
change (e.g. Pounds et al., 1999). The threats interact in a variety of ways (Pounds,
2001; Green, 2003): for example, low pH might make amphibians more susceptible
to UV-B radiation (Long et al., 1995), while UV-B radiation increases mortality
from disease (Kiesecker and Blaustein, 1995; Blaustein et al., 2003). Equally,
1. General introduction
2
amphibian harvesting and trade not only deplete the source population, but
potentially lead to non-native introductions and spread of disease (Picco and Collins,
2008).
Heightened awareness of amphibian declines has stimulated greater research activity
in all areas of amphibian biology, improving understanding of population and
behavioural ecology, examining direct and indirect threats and developing
management strategies to tackle the known underlying causes. Species richness and
threatened species are not evenly distributed around the world – areas such as
Central and South America have the highest species richness and the highest number
of endangered species (Myers et al., 2000). However, building a more complete
knowledge of amphibian ecology in all types of habitat and developing efficient
survey and analysis methodologies are important to advance understanding. Detailed
field surveys and long-term datasets are central to achieving this (Alford and
Richards, 1999; Houlahan et al., 2000).
While there is greater species diversity in the tropics, there are more ecologists in
temperate regions (Collen et al., 2008), especially North America and Europe. This
expertise bias has lead to a huge skew towards developed countries in conservation
research output (Memmott et al., 2010; Milner-Gulland et al., 2010) and in practical
conservation efforts such as captive breeding programmes (Griffiths and Pavajeau,
2008). Despite these biases, occasionally there are still surprising findings in well-
studied regions, such as the discovery in 1980 of live Mallorcan midwife toads
(Alytes muletensis), which were previously known only from the fossil record (Buley
and Garcia, 1997). Even in developed countries, there are still knowledge gaps in
ecosystem ecology and a lack of data on some taxa, which impede conservation.
Moreover, the relatively simple species assemblages in Europe represent an
opportunity to develop and improve field methods and analytical approaches to the
study and monitoring of herpetofauna.
1. General introduction
3
1.2 Luxembourg
1.2.1 Location and geography
The present study was conducted in Luxembourg, a small landlocked country
bordered by Belgium, France and Germany (Figure 1.1) in the middle of the
temperate region of Europe (Martínez Rica, 1997). Despite its small size (2586 km2)
Luxembourg has very diverse geology and landscapes (Wolff, 2009), ranging from
the foothills of the Ardennes in the north to lowland agriculture and former wetlands
in the south. One of the major costs of the country‟s economic success has been the
loss of biodiversity during the progression from a largely rural economy to a major
European financial centre, via steel production (Chamber of Commerce of the Grand
Duchy of Luxembourg, 2009). The booming economy precipitated a rapid expansion
of housing and business developments and also a vastly expanded transport network
(Wolff, 2009), particularly in the centre and south. Although Luxembourg is in the
top ten highest human population densities in Europe (Gaston et al., 2008), it
maintains a high percentage of forest cover (37.0%, mostly in the north) and a good
protected area coverage (20.2%). The forested areas generally have a better
conservation status than farmed and aquatic ecosystems (Wolff, 2009).
Figure 1.1 Luxembourg.
According to local assessments approximately 55% of Luxembourgish mammals,
40% of birds, 30% of reptiles and 70% of amphibians are threatened (Wolff, 2009),
although biodiversity and nature conservation have recently attracted greater
France
Germany
Belgium
Luxembourg
N 300 km
1. General introduction
4
political attention and consequently better funding and better informed actions. The
key threats to amphibians and other wildlife in Luxembourg are considered to be
agricultural chemicals, wetland drainage and hedgerow removal (Wolff, 2009).
There are 12 native species of amphibian extant in Luxembourg, for which the
nomenclature is detailed in Appendix 1, and possibly one additional species (Rana
dalmatina) that has not yet been confirmed as present. Eleven of the known species
usually breed in ponds (Rana temporaria, Pelophylax spp., Hyla arborea, Bufo bufo,
Alytes obstetricans, Bufo calamita, Bombina variegata, Ichthyosaura alpestris,
Triturus cristatus, Lissotriton helveticus and L. vulgaris), while Salamandra
salamandra favours streams (Kwet, 2009).
1.3 Conservation legislation
The legal protection and conservation of animals, plants and habitats in Luxembourg
derive from both European and national law. The key international legislation that
dictates the protection afforded to species and habitats in Europe are the Convention
on the Conservation of European Wildlife and Natural Habitats (the 'Bern
Convention'; Council of Europe, 1979) and the Council Directive 92/43/EEC on the
Conservation of natural habitats and of wild fauna and flora (the 'Habitats
Directive'; Council of Europe, 1992). Adopted in 1992, the Habitats Directive is the
European Union‟s (EU) practical interpretation of its conservation obligations under
the Bern Convention (Joint Nature Conservation Committee, 2010), both include
appendices listing the species and habitats in need of protection, which also dictate
the level of protection that must be provided. EU member states are required to
implement a number of practices, including to:
Maintain or restore „favourable conservation status‟ for protected habitats and
species.
Help to build a protected area (PA) network across Europe using statutory site
designations.
Implement suitable management and planning for existing and proposed PAs.
Monitor species and habitats.
Enforce „strict protection‟ of the most highly threatened species, which are
listed in annex IV of the Habitats Directive (Table 1.1).
1. General introduction
5
Under European legislation all native amphibian species are protected from
exploitation that could endanger whole populations (Council of Europe, 1979). Six
species occurring in Luxembourg benefit from additional „strict protection‟ under
Annex IV of the Habitats Directive, and T. cristatus and Bombina variegata are both
further protected under Annex II (Table 1.1), which stipulates that the sites where
they occur be designated as Special Areas of Conservation (SACs). National
legislation affords a high level of protection to all amphibian species in Luxembourg
under a 2004 law „Loi du 19 janvier 2004 concernant la protection de la nature et
des ressources naturelles‟ (Grand-Duché de Luxembourg, 2004) and a more recent
amendment where species not previously covered were added „Règlement grand-
ducal du 9 janvier 2009 concernant la protection intégrale et partielle de certaines
espèces animales de la faune sauvage‟ (Conseil d‟Etat [Luxembourg], 2009).
Table 1.1 Luxembourgish amphibians on the Bern Convention and Habitats
Directive appendices and their IUCN Red List status and global population trend.
Species Bern
Convention appendices
Habitats Directive annexes
Red List
status
Global population
trend
R. temporaria III V LC stable
Pelophylax spp.: P. kl. esculentus III V LC decreasing
P. lessonae III IV LC decreasing
H. arborea II, III IV LC decreasing
Bufo bufo III LC stable
A. obstetricans II, III IV LC decreasing
Bombina variegata II, III II, IV LC decreasing
Bufo calamita II, III IV LC decreasing
T. cristatus II, III II, IV LC decreasing
I. alpestris III LC decreasing
L. vulgaris III LC stable
L. helveticus III LC stable
S. salamandra III LC decreasing
Bern Convention (Council of Europe, 1979): Appendix II: no deliberate capture, keeping,
killing, damage to / destruction of habitat, disturbance, destruction or taking of eggs, or
possession or trade; Appendix III: no exploitation endangering the population and trade must be regulated; Species and Habitats Directive (Council of Europe, 1992): Annex II: species
whose conservation requires SAC designations; Annex IV: species of community interest,
requiring strict protection; Annex V: species of community interest, may need management
measures; LC = „least concern‟ (IUCN et al., 2008, criteria version 3.1).
1. General introduction
6
Luxembourg‟s own conservation targets, set out under the „Plan national concernant
la protection de la nature‟ (PNPN; Ministère de l'Environnement, 2007) echo those
of the Habitats Directive and the EU Convention on Biological Diversity (CBD;
which is discussed further in section 4.1.1) – firstly to stop biodiversity loss by 2010
and secondly to preserve and re-establish ecosystem services and processes. Progress
has been made on a number of targets, such as designation of more PAs and habitat
improvement for threatened species (Wolff, 2009). Conservation action plans have
recently been drawn up for many of the priority plant and animal species, including
three amphibians, Bufo calamita, H. arborea and T. cristatus, and work is already
underway to improve management of the sites they occupy, create new potential
breeding sites and a reintroduction programme has started for H. arborea (MNHNL -
Groupe herpétologique, 2009a, b; Proess, 2009). Monitoring and management of
acutely threatened species or habitats in Luxembourg are implemented via national
and regional government agencies or consultants directly contracted by the
government. This ensures that adequate data are available to inform management
decisions at specific locations, for example a population size estimate for the donor
site of H. arborea for reintroduction is essential to avoid depletion. It would be
useful to have such detailed information for all threatened species, including those
remaining at multiple sites but thought to be in decline (e.g. A. obstetricans and T.
cristatus).
1.3.1 Influence of political boundaries
EU member states each apply the European Directives according to their own
constitution (Jongman et al., 2008), which leads to slight variation between
countries. The system of PAs has promoted the conservation of patches of land that
are frequently isolated from other similar habitat (Harrop, 1999) or that become less
relevant as conservation needs shift (Bennett, 2003; Hannah et al., 2007). The next
phase for EU conservation is to increase connectivity between existing reserves
using habitat corridors, which add value to PAs by making it possible for species to
travel across landscapes (Haslett et al., 2008; Haslett et al., 2010; Samways et al.,
2010).
1. General introduction
7
The drive for coherence and connectivity of PAs within and between EU countries
was pre-empted by several decades by an agreement between Belgium, the
Netherlands and Luxembourg: The Benelux Agreement concerning Nature
Conservation and Landscape Protection (Benelux Economic Union, 1982). Formal
coordination of conservation efforts between these small countries improves
efficiency by working in the context of a larger area.
1.3.2 IUCN
The IUCN Red List (International Union for Conservation of Nature) is an
assessment of species‟ population trends and risk of extinction, its aim is to
document every known species‟ conservation status. It classifies species into „threat
categories‟ by application of quantitative criteria that can be fitted across a broad
spectrum of taxa, informed by extensive consultation with experts. The system is
designed to produce an impartial evaluation of extinction risk, that is widely
understood and comparable across taxa and geographic areas (IUCN, 2001). The
degree of legal protection afforded to a species is often determined by its Red List
status, which can also prompt research and conservation action, and, as in Europe,
oblige countries to take measures to safeguard species and habitats.
For the Global Amphibian Assessment (GAA) in 2004 amphibians became one of
the first taxonomic groups to be fully assessed under the IUCN Red List criteria,
which was updated in 2006 and 2008, and covers all of the 6260 known species
worldwide (IUCN, 2009). Like 42% of amphibian species in the world (IUCN,
2009), most of the species occurring in Luxembourg currently have a downward
global population trend: 8 out of 12 species were assessed to be decreasing across
their global range, while the other four species are stable (Table 1.1). Despite the
negative population trends, in the global Red List all Luxembourgish amphibian
species are categorised as „Least Concern‟ in the 2008 assessment, meaning that they
are considered to be widespread and/or abundant, although „a taxon may require
conservation action even if it is not listed as threatened‟ (IUCN, 2001). However,
according to national assessments by IUCN criteria, the three amphibian species with
action plans are considered at high risk within Luxembourg: H. arborea is highly
1. General introduction
8
threatened, B. calamita is at risk of extinction and T. cristatus is categorised as
threatened (MNHNL - Groupe herpétologique, 2009a, b; Proess, 2009).
1.4 Pond ecology
Assessments and monitoring of European amphibian species tend to be undertaken
during their aquatic phase, when they converge on breeding ponds and behave more
conspicuously, and therefore become relatively easy to survey. At a regional level
ponds are very important for biodiversity – on average ponds hold a greater number
of species than other waterbodies (Davies et al., 2008) and due to their relatively
small catchment areas ponds within the same area can host quite different species
assemblages (Nicolet et al., 2004). Amphibians need both terrestrial and aquatic
habitats to complete their life cycles, making them vulnerable to environmental
changes in two places (Pope et al., 2000; Marsh and Trenham, 2001), but this
sensitivity to change also makes them candidates for use as surrogate measures of
biodiversity (Oertli et al., 2005). When general information on the „value‟ of a site is
required, proven surrogates, or indicators, may be used instead of undertaking
surveys for every taxon, although even very well-fitting surrogates give only a broad
generalisation of what is actually present (Bilton et al., 2006). Amphibians are
regularly touted as potential surrogates for species richness of other taxa or for
general ecosystem health (e.g. Lips, 1998; Loyola et al., 2007). However studies
examining European pond-breeding species and other aquatic groups have generally
found that while there is potential for such relationships, further study is required
(Oertli et al., 2005; Sewell, 2006).
1.4.1 Amphibian surveying
Amphibians are often cryptic both in behaviour and colouration (Vitt and Caldwell,
2009), with activity dependent on environmental variables such as temperature and
rainfall (e.g. Heyer et al., 1994; Brooke et al., 2000; Grafe and Meuche, 2005). This
makes them difficult to survey because of the confounding effect of imperfect
detection (Altwegg et al., 2008). In long-term projects or databases it is particularly
important that the effectiveness of survey protocols is thoroughly examined, because
they may be used without further appraisal for many years (Crouch and Paton, 2002;
Kaiser, 2008). Selecting an inefficient monitoring strategy can be costly in terms of
1. General introduction
9
both finances and data quality (Paszkowski et al., 2002; Field et al., 2005), so the
survey method costs and benefits should be carefully considered alongside project
aims (Quinn et al., 2007).
How a method is implemented is as important as the choice of method itself
(Paszkowski et al., 2002; de Solla et al., 2005; Gooch et al., 2006). For example, call
surveys have been used routinely for anurans in North America and Canada, but the
recommendations for survey length derived from them tend to vary widely, from
three minutes (Shirose et al., 1997) up to fifteen minutes of survey effort (Pierce and
Gutzwiller, 2004). Dip-netting is commonly used to survey for newts and all
amphibian larvae, but opinion differs on its usefulness. Sometimes netting is done
very rigorously covering 100% of the pond (e.g. Arntzen, 2002), while other surveys
are more restricted, limiting netting to a set number of sweeps to minimise damage
and disturbance to the vegetation (Griffiths and Raper, 1994). Although most
European amphibian species are well-studied, there is still room for improvement at
all stages of survey design and analysis. Use of appropriate methods is important to
ensure that the data collected are appropriate to study aims and that inaccuracies
such as false absences are minimised. Development of survey protocols and analysis
techniques is essential in all areas of biodiversity monitoring, even where methods
are already established and species are already well described.
1.5 Project objectives
The present study addresses some core questions in amphibian ecology and survey
methodology, and tests a new approach to statistical analysis of survey data:
Does the distribution of amphibians in Luxembourg – as assessed from
historical records - exhibit any relationships with local landscape variables?
(Chapter 3.)
Have the range sizes of amphibians changed in recent decades? (Chapter 4.)
How much of Luxembourg‟s amphibian diversity is captured within its
protected areas? (Chapter 4.)
Are amphibians useful indicators of plant or invertebrate species richness?
(Chapters 5 and 6.)
1. General introduction
10
What combination of survey methods is needed to reliably detect the various
species as being present in a pond? (Chapter 7.)
The pros and cons of different data types, survey methods and analyses are discussed
in each chapter. Further to the studies included within the thesis chapters, two papers
published in the course of the work are included in Appendix 9. The first is not
directly relevant to the thesis subject, however was an interesting observation of the
phenomenon of freshwater mussels attached to newts‟ feet. The second paper reports
on the results of screening for the amphibian fungal disease chytridiomycosis in
Luxembourg – background levels of the disease-causing fungus Batrachochytrium
dendrobatidis were found, which has implications for amphibian conservation and
survey protocols, discussed in chapters 7 and 8.
2. General methodology
11
2. General methodology
General methodology
The data in Chapters 3 and 4 were taken from a national database collated by the
Musée national d‟histoire naturelle (Luxembourg) (MNHN[L]), and data for
Chapters 5 – 7 were collected specifically for the present study on pond-breeding
amphibians in Luxembourg. The species are listed and their nomenclature is detailed
in Appendix 1.
2.1 Historical data - MNHN
Historical data on amphibian presence throughout Luxembourg were retrieved from
the database held by the MNHN. Sixty-five ambiguous, incomplete and duplicated
records were deleted, and the dataset was refined to a maximum of one record per
species / year / 1 km grid square; leaving a total of 1317 records between 1955 and
2005. Records from 2006 onwards had not all been submitted to the MNHN
database when these data were extracted, so are not used here. One kilometre grid
squares were the finest resolution possible with this dataset whilst maintaining its
explanatory power (Luoto et al., 2007). Six figure grid references (describing 1 km
grid squares) were found and checked for all records, most importantly for those
referred to by place names and less accurate grid references in the original database.
Green frogs, Pelophylax lessonae and P. kl. esculentus, were originally recorded
separately, however the names are likely to have been used interchangeably by many
recorders due to the difficulty of distinguishing them in the field (Appendix 1).
Records have therefore been merged for present use and they are collectively
referred to as Pelophylax spp. (or P.spp) here and throughout following chapters.
The MNHN data (Chapters 3 and 4) are presence-only records – i.e. there are also
unsurveyed, unrecorded grid squares where the study species might also be present.
False absences are a flaw in this dataset and they cannot be addressed due to the
nature of the data collection, however this issue is taken up in Chapter 7.
2. General methodology
12
2.1.1 Map data
Base maps were drawn in ArcMap (v9.3.1.1850) with data from the Luxembourg
government‟s cartography department, the „Administration du Cadastre et de la
Topographie‟ (ACT). The landscape features and protected areas projected onto the
map are listed in Table 2.1. Most map files had been updated by ACT in 2006.
Similar layers were merged together using ArcMap‟s „geoprocessing wizard‟, for
example streams and rivers were merged to create „running water‟.
A grid of 1 x 1 km squares was applied to the map and each square partly or wholly
in Luxembourg was assigned a unique identification code and its within-country area
was measured. Landscape layers were converted to „shapefiles‟ and the area, or the
length of line features, within each grid square was extracted. The identification
codes were used to import extracted data into a spreadsheet detailing the contents of
each grid square. Measurements of features in squares not wholly within
Luxembourg were adjusted according to their within-country area (e.g. 20,000 m2
coniferous forest cover in a square only half in Luxembourg was reduced by a factor
of 0.5 to 10,000 m2).
Table 2.1 Landscape features drawn in ArcMap and extracted to quantify habitat in
each 1 km grid square.
Type Feature Measurement unit
Geographical & habitat
National & regional boundaries n/a
Waterbodies - running: streams & rivers - still: ponds, lakes & uncoded ‘basins’ - water treatment basins
Length (m) Area (m
2)
Forest - coniferous - broadleaved - mixed
Area (m2)
Area (m2)
Area (m2)
Roads – all widths Length (m)
Buildings – all types (including domestic, agricultural, industrial & undefined)
Area (m2)
Elevation Mean (m.a.s.l.)
Protected Areas (PAs)
Nature reserves Area (m2)
Natura 2000 sites - Habitats Directive - Birds Directive
Area (m2)
Area (m2)
2. General methodology
13
Preliminary models in Chapter 3 also included a soil type covariate, however it
produced uninformative models which are not reported. This may have been because
it was a categorical variable, and using only the commonest soil type in each square
produced a very simplified, over-generalised soil map. Data on agricultural uses of
land surrounding ponds were unavailable.
2.2 Study pond selection
The present study was conducted in Luxembourg (see section 1.2). Study ponds
were selected from historical data collated by the Luxembourg Musée national
d‟histoire naturelle (MNHN; see Appendix 2) between 1997 and 2006, initially to
ensure inclusion of ponds where three species thought to be in steep local decline
had been recorded: Hyla arborea, Alytes obstetricans and Triturus cristatus. Other
ponds were found on topographical maps (1:20,000) to give a good spread across
central and southern Luxembourg (Figure 2.1) and to facilitate surveys at several
ponds in one evening.
The definition of a „pond‟ was taken from the National Pond Survey (NPS) (Biggs et
al., 1998, pp.6):
„a body of water, of man-made or natural origin, between 1 m2 and 2 ha,
which usually holds water for at least four months of the year‟.
Study ponds are listed in Appendix 3, pond area (m2) was estimated in the field and
confirmed by measurement on maps in ArcGIS. Thirty-one ponds were studied in
2007, and in 2008 26 of the same ponds were surveyed: access permission was
withdrawn at one pond (number 8) and four others were dropped to allow inclusion
of three extra ponds to boost data on ponds occupied by A. obstetricans. In 2009 a
short fieldwork season was undertaken to include nine ponds from the north of the
country, and only A. obstetricans data were collected in that final season.
In Chapters 5 and 6 ponds separated by less than 25 m were classified as a single
pond because this distance is readily traversed by all of the amphibian species found
in Luxembourg. For occupancy modelling, landscape features within 100 m of ponds
2. General methodology
14
were examined, therefore ponds separated by less than 200 m were grouped for
Chapter 7.
Figure 2.1 Study pond locations in Luxembourg.
2.3 Amphibian surveying
Four standard methods were used to survey for amphibians: visual encounter surveys
(VES; for adults, larvae and eggs), funnel traps, dip-netting and aural surveys – these
are described in detail below. Where possible in 2007 and 2008 all survey methods
were employed at each pond on at least five repeat visits between late February and
early August (Appendix 4), but occasionally a survey method had to be dropped
when conditions were unsuitable (e.g. water too shallow or warm for trapping). In
2009 only aural surveys were conducted and five survey rounds were completed
between 12th May and 2
nd June.
The appropriate survey methods for each species are detailed in Table 2.2, where
possible, the species, sex and life stage of animals found were recorded. Only
definite identifications were retained in the final data. Larvae of the Lissotriton newt
species could not be distinguished, and females of the same species were only
10 km
N
2. General methodology
15
included where they could be identified in the hand. A review of the advantages and
disadvantages of the following survey methods is available from Griffiths and Raper
(1994).
2.3.1 Visual encounter survey (VES)
After dark a powerful torch (Clulite© Clubman Standard CB1, 0.5 million
candlepower) was used to survey for amphibians in the water. All VESs were
conducted by one person (LW) walking slowly around accessible parts of the pond
perimeter, sweeping the torch beam from the bank to approximately 2 m into the
pond and counting amphibians seen in the light. Where identifiable, adults, larvae
and eggs were recorded (Table 2.2). Torching took up to 20 minutes for the biggest
ponds. Night-time torch survey data and visual sightings made during other daytime
surveys were included as VES.
2.3.2 Funnel trapping
Funnel traps were set after VES. Traps were made from 2 litre plastic drinks bottles
with the spout cut off and inverted (following Griffiths, 1985). Up to 25 traps were
set in each pond, or 30 traps between paired ponds, at 2 m intervals along the
accessible margin, in groups of 5 for larger ponds. As ponds dried up and decreased
in size the number of traps was reduced to maintain the 2 m spacing.
Traps were checked and emptied after dawn the following morning, an average of 10
hours 20 minutes after setting (maximum 14.5 hours, minimum 6.5 hours). The
length of time traps were in situ was generally shortened throughout the season as
hours of darkness decreased and water temperature increased. Trapping ceased
during June of both 2007 and 2008 when the water temperature became too high to
safely use traps in any pond.
2.3.3 Dip netting
Dip-netting was undertaken in the morning, after traps had been removed to the
bank. A full net sweep was made from the bank at each trap location (i.e. up to 25
sweeps per pond, or 30 sweeps for paired ponds). The net was swept in a continuous
movement across the pond substrate in an arc (radius approximately 1 m). Netting
2. General methodology
16
was limited to 25 or 30 sweeps to minimise damage to the ponds. The net used had a
0.3 cm gauge mesh, 30 cm long net bag and a square frame 25 cm across.
2.3.4 Aural
Adult anurans were monitored by call surveys. This was conducted simultaneously
with other night time surveys (VES and trap setting). The seven anuran species
present in Luxembourg have distinctive calls (Rana temporaria, Pelophylax spp.,
Hyla arborea, Bufo bufo, Alytes obstetricans, Bufo calamita, and Bombina
variegata). Where A. obstetricans were not already calling, their distinctive call was
mimicked over short periods during the survey visit to elicit response calls from
females and competitive calls from males.
Table 2.2 Survey methods appropriate to each amphibian species by life stage and
sex.
Species Eggs Larvae
Adults Males Females
VES Trap Net VES Trap Net VES Trap Net Aural VES Trap Net Aural
Rt
P.spp
Ha
Bb
Bc
Ao
Bv
Ia
Tc
Lh
Lv
See Appendix 1 for amphibian species name abbreviations; „Adults‟ includes recently
metamorphosed and juvenile individuals; VES = visual encounter survey; newt eggs were not differentiated between and therefore were excluded from analysis; L. helveticus and L.
vulgaris larvae are indistinguishable in the field, and females can only be distinguished in
the hand and are therefore excluded from VES.
2.4 Data analysis
Data analysis is explained in each chapter. Unless otherwise stated, analyses were
performed in SPSS v.16.
3. Amphibians and landscape features
17
3. Landscape features as predictors of amphibian presence
Amphibians and landscape features
Summary: The data collated by biological record centres are an
important resource for examining broad scale relationships and trends.
An understanding of species‟ relationships with their wider habitat helps
to inform management and conservation decisions. Between 1996 and
2005 amphibians were surveyed in 23.5% of 1 km grid squares in
Luxembourg: surveyed squares had a greater average forest cover than
unsurveyed squares, but there was no difference in markers of human
development. Principal component analysis grouped habitat variables
into four components (PCs): 1. forest and elevation, 2. water, 3. human
developments, 4. protected areas. Overall PC1 was the most useful in
predicting amphibian presence records. Bufo bufo and Alytes obstetricans
were positively associated with PC1, while Pelophylax spp., Triturus
cristatus and Lissotriton vulgaris were negatively associated with the
same variable. B. bufo and A. obstetricans records were also positively
related to human development, whereas Ichthyosaura alpestris and L.
helveticus were negatively associated with markers of development.
Suitable terrestrial habitat surrounding amphibian breeding ponds is, of
course, crucial to their persistence in an area, and knowing how habitat
features influence amphibians is key to their conservation. Here broad
categories of habitat are examined.
3.1 Introduction
3.1.1 Recording centres
Recent technological advances have vastly increased the availability of fine scale
data from ecological surveys, along with the facility to store, share and map them
(Phillips et al., 2006; Gaston et al., 2008). Large databases are essential in the
production of atlases showing species‟ ranges, for example the „Atlas of Amphibians
and Reptiles in Europe‟ by Gasc et al. (1997) gives broad distribution details of all
herpetofauna species occurring in Europe. Compiling such an extensive atlas is a big
task, requiring collation of existing datasets and commissioning of additional
research to fill the gaps.
Some interest groups have exploited modern computing power more than others. For
example, a concerted effort by the ornithological community in many parts of the
world has led to the development of „atlassing‟, whereby volunteers systematically
survey map squares and submit their data to a central database. Atlassing has
3. Amphibians and landscape features
18
immensely expanded knowledge of bird distributions (Donald and Fuller, 1998), but
unfortunately no other taxa have been subjected to the same thorough and sustained
recording effort. Consequently atlas data are usually presence-only (i.e. empty grid
squares could indicate a species‟ absence or a lack of surveys, and abundance is not
routinely recorded) and therefore cannot be used for population monitoring (Donald
and Fuller, 1998).
Many countries have regional and national recording centres that attempt to collate
biological records voluntarily submitted and to make the data available for research
and conservation decision-making. Sometimes these records are collected following
a standardised scheme, but more often the databases contain incidental records. Data
collected unsystematically are inherently biased and should therefore be used
carefully and critically (Elith et al., 2006). Presence-only data collected by ad-hoc
and variable methods, such as those in record centre databases, tend to be rather
patchy, both in geographical location and across time. The data may be subject to
multiple biases, with recorders perhaps more likely to report rare species than
common ones and sampling effort being concentrated in species-rich areas close to
recorders‟ homes (Dennis et al., 1999; Dennis and Thomas, 2000).
The Musée national d‟histoire naturelle (MNHN) is the national biological recording
centre in Luxembourg and holds a large database of animal and plant records
submitted by professional and amateur naturalists (all contributors are listed in
Appendix 2). Amphibian records date back to 1955, but increase greatly in number
after 1970 and again during the 1990s. The data have previously been summarised in
an atlas of Luxembourgish amphibians (Proess, 2003b), which is a useful overview
of species‟ distributions, but they have not been analysed in any detail. The atlas by
Proess (2003b) maps species records before and after 1996 at a fairly broad scale (5
km grid squares) and includes some general information on amphibian natural
history and the geography of Luxembourg.
3.1.2 Landscape and amphibians
Examination of the relationships between organisms and their habitats allows
generalisation about the places they are likely to occur – record centre data can be
3. Amphibians and landscape features
19
very useful for looking at general trends. Surrounding landscape features may be just
as important in determining amphibian presence and abundance as the more
immediate variations in habitat conditions (Van Buskirk, 2005). The influence of
landscape features on amphibian species richness and abundance is strongest near
breeding habitat: features such as forest cover and wetland area have been found to
be positively correlated with both measures, while markers of human development
and activity, such as road density and soil nitrogen levels, are negatively correlated
(Houlahan and Findlay, 2003). All of the amphibian species occurring in
Luxembourg spend much of their time away from water; many forage and over-
winter on land and immature animals often do not return to water until they are ready
to breed.
In the current chapter the influence of landscape features on amphibian occurrence is
considered, with data collated by the MNHN. Habitat features are also examined in
Chapter 7 as covariates in occupancy models using the more detailed data collected
specifically for this purpose. The present chapter aims to address the following
questions:
Are all areas of Luxembourg represented in the MNHN amphibian record
database?
Which broad landscape variables are useful in predicting each species‟
occurrence?
Which landscape variables are indicative of high amphibian species richness?
3.2 Methods
The maps and methods of data extraction are given in the General Methods chapter
(Chapter 2). The habitat features extracted from maps (Table 2.1) were organised
into eleven continuous scale variables (coniferous forest, broadleaved forest, mixed
forest, nature reserves, Natura 2000 sites, buildings, roads, water treatment basins,
still water, running water and mean elevation). Prior to any analyses the variables
were transformed by log10(x+1) to reduce the disproportionate influence of very high
or zero values.
3. Amphibians and landscape features
20
Only amphibian data collected between 1996 and 2005 were used here, to ensure the
relevance of the investigation. Firstly the landscape features in surveyed 1 km grid
squares were compared to unsurveyed grid squares using MANOVA, followed by
one-way ANOVA to explore where the significant differences lay.
A number of the habitat variables exhibited some degree of correlation with each
other, therefore to avoid multicollinearity confounding the results, principal
component analysis (PCA) was applied. Components with eigenvalues > 1 (Field,
2005) were retained for logistic regression.
Forward stepwise (likelihood ratio) logistic regression was used to explore the
derived principle components as predictors of the presence / absence (non-records)
of each amphibian species and the recorded amphibian species richness in surveyed
grid squares only. A record for one species was assumed to indicate that general
amphibian surveys had been conducted; Salamandra salamandra was excluded
because it requires an entirely different set of survey methods. The Wald test was
chosen to test the significance of regression coefficients and the Hosmer and
Lemeshow test to check goodness of fit. To maintain sufficient statistical power in
the logistic regression analyses, only species recorded in ≥ 5% of surveyed squares
were used.
3.3 Results
3.3.1 Historical data descriptives
Surprisingly, 76.6% of the 1 km grid squares wholly or partly in Luxembourg (n =
2789) have no amphibian records in the database between 1955 and 2005. This could
indicate either that amphibians have never been found there, or simply that no
surveys have been conducted. A further 19.4% of grid squares had only been
surveyed in one year (out of 51), while very few (0.14%) had records from ten or
more years. Over the 51 years, between one and nine amphibian species were
recorded in each of the surveyed grid squares (Figure 3.1a). Coverage in the most
recent complete decade from the database (1996-2005) is similar to that in the entire
database, with 76.5% of grid squares having no amphibian records, while the
3. Amphibians and landscape features
21
surveyed squares held between one and eight amphibian species in this period
(Figure 3.1b).
(a) 1955-2005
(b) 1996-2005
Figure 3.1 The number of amphibian species recorded in each 1 km grid square, (a)
1955-2005, (b) 1996-2005.
Multivariate analysis (MANOVA) revealed that there is a significant difference in
the landscape features between the 1 km grid squares that were surveyed between
1996 and 2005 and those that were not (F12,2774 = 3.75, p < 0.001). Squares where
amphibian surveys were carried out have, on average, more broadleaved forest cover
(F1,2786 = 19.98, p < 0.001) and more „total‟ forest cover (F1,2786 = 9.49, p < 0.005),
greater Natura2000 coverage (F1,2786 = 4.52, p < 0.05) and more water treatment
works (F1,2786 = 7.06, p < 0.01; Figure 3.2). However, there was no difference in the
measures of development between surveyed and unsurveyed squares (buildings:
F1,2786 = 3.28, p > 0.05; roads: F1,2786 = 0.19, p > 0.05).
Unsurveyed squares had a significantly smaller area within the boundaries of
Luxembourg (F1,2785 = 25.41, p < 0.001), indicating that more were on the border:
14.8% of unsurveyed squares were on the border, compared to 9.8% of the surveyed
squares.
N
Number of amphibian species recorded:
10 km
3. Amphibians and landscape features
22
Figure 3.2 Comparison of the mean area or length of landscape features in squares
surveyed for amphibians between 1996 and 2005 and those not surveyed.
= not surveyed, = surveyed; * p < 0.05, ** p < 0.01, *** p < 0.005; measurements adjusted for grid square area within-country; dotted bars are linear features measured in
metres (rather than m2).
Figure 3.3 The percentage of 1 km grid squares surveyed (n = 654) between 1996
and 2005 occupied by each amphibian species.
= not occupied, = occupied; see Appendix 1 for species abbreviations.
All of the amphibian species present in Luxembourg were recorded in at least two 1
km grid squares between 1996 and 2005. However the three rarest species (Hyla
arborea, Bufo calamita and Bombina variegata) were only recorded in less than 3%
of squares (Figure 3.3) and were therefore excluded from further analysis.
***
***
*
**
0
500
1000
1500
2000
2500
3000
3500M
ean m
2(±
s.e
.)
0
10
20
30
40
50
60
70
80
90
100
Rt P.spp Ha Bb Ao Bc Bv Ia Tc Lh Lv Ss
% o
f surv
eyed s
quare
s
3. Amphibians and landscape features
23
3.3.2 Landscape predictors of amphibian species’ occurrence
PCA reduced the habitat variables to four principal components (PCs) with
eigenvalues > 1 (Table 3.1). This cut-off point is supported by the scree plot which
has an „elbow‟ after four components (Figure 3.4; Everitt and Dunn, 2001). Overall
the four extracted PCs explain 61.41% of the total variance. Grouped by highest
loading the variables fit tidily into the four components, which can be interpreted as
(1) Forest and elevation, (2) Water, (3) Human development and (4) Protected areas
(PAs).
Water treatment basins were included in the principal component analysis based on
personal observation of amphibians reproducing in them. Many of the water
treatment works are „natural‟, a series of ponds planted with reeds (Phragmites), and
they account for 40% of all still water basins in Luxembourg. Although they
associate slightly more with human developments (PC3) than other waterbodies,
water treatment works did not load heavily onto any component.
Table 3.1 Multivariate factor loadings of principal components with eigenvalues > 1.
Variable Component number
1 2 3 4
Coniferous forest 0.829
Mixed forest 0.761
Broadleaved forest 0.695 0.169 -0.173
Mean elevation 0.515 -0.465 0.399
Still water 0.820 0.140 0.140
Running water 0.114 0.754
Roads 0.823
Buildings -0.319 0.711
Water treatment 0.259 0.331
Nature reserves 0.864
Natura 2000 sites 0.818
Eigenvalue 2.439 1.745 1.332 1.240
% of variance 22.170 15.859 12.111 11.269
Cumulative % 22.170 38.029 50.140 61.409
Very low factor loadings (< 0.1) are suppressed; loadings > 5 are shown in bold typeface; 4 of 11 components had an eigenvalue > 1; rotation method: Oblimin with Kaiser
Normalization.
3. Amphibians and landscape features
24
Figure 3.4 Scree plot of the eigenvalues of all components.
Table 3.2 Final models from stepwise logistic regression for anurans.
Species Predictor Step
Regression coeff.
Wald statistic (squared)
Odds ratio
e s.e. 2 df p
R. temporaria Constant - 0.19 0.08 5.87 1 - -
PC2: water 1 -0.25 0.09 7.39 1 0.007 0.78
Notes: H&L 2
8 = 8.42, p = 0.394; model 2
1 = 7.54, p < 0.01
Pelophylax spp. Constant - -0.66 0.09 57.43 1 - -
PC1: forest & elev. 1 -0.79 0.10 56.61 1 < 0.001 0.46
Notes: H&L 2
8 = 15.63, p = 0.048; model 2
1 = 62.29, p < 0.005
Bufo bufo Constant - -1.02 0.09 119.46 1 - -
PC1: forest & elev. 1 0.36 0.11 9.77 1 0.002 1.43
PC3: human dev. 2 0.32 0.11 8.98 1 0.003 1.38
Notes: H&L 2
8 = 7.29, p = 0.506; model 2
2 = 16.12, p < 0.001
A. obstetricans Constant - -2.15 0.14 238.53 1 - -
PC1: forest & elev. 1 0.55 0.17 10.03 1 0.002 1.72
PC3: human dev. 2 0.32 0.14 5.02 1 0.025 1.38
Notes: H&L 2
8 = 6.75, p = 0.563; model 2
2 = 13.85, p < 0.005
Step = logistic regression step that added covariate; Goodness of fit statistic: H&L = Hosmer and Lemeshow; Odds ratio: values < 1 indicate that the PC and the species are less likely to
occur together, a value of 1 would indicate no effect and a value > 1 indicates a positive
association.
One landscape PC was added by stepwise logistic regression to explain the
distribution of Rana temporaria: the water principal component (PC2) was
significantly negatively associated with R. temporaria presence, although the final
model explains only 56.7% of the variation (Table 3.2 and Figure 3.5a).
Pelophylax spp. records were best explained by the inclusion of the forest and
elevation PC (Table 3.2), but the model has a high Hosmer and Lemeshow score and
is only a borderline good fit even though the final model explains 70.3% of the
0.0
0.5
1.0
1.5
2.0
2.5
1 2 3 4 5 6 7 8 9 10 11
Eig
envalu
e
Component number
3. Amphibians and landscape features
25
variation. The species‟ negative association with forest and elevation (PC1) is very
clear (Figure 3.5b).
Two principal components made a well-fitted model for Bufo bufo (Table 3.2),
which accounted for 71.7% of the variation: PC1 representing forest and elevation
and PC3 representing human development were both positively associated with Bufo
bufo (Figure 3.5c).
Occupying just 79 of the 654 grid squares surveyed, Alytes obstetricans can be
described as rare in the study area (along with Triturus cristatus, Lissotriton vulgaris
and Salamandra salamandra; Figure 3.3). Like Bufo bufo, the presence of A.
obstetricans was best predicted by PCs 1 and 3 (Figure 3.5d), with the final model
explaining 87.9% of the variation.
PC
valu
e (
±s.e
.)
(a) R. temporaria
(b) Pelophylax spp.
PC
valu
e (
±s.e
.)
(c) Bufo bufo
(d) A. obstetricans
Figure 3.5 Mean values of the significant principal components predicting anuran
species‟ occupancy in 1 km grid squares.
* p < 0.05, *** p < 0.005, indicating the step significance (Wald statistic); = species
absent / unrecorded, = species present.
*
-0.2
0.0
0.2
PC2: water
***
-0.4
-0.2
0.0
0.2
0.4
PC1: forest & elevation
***
***
-0.2
0.0
0.2
0.4
PC1: forest & elevation
PC3: human development
***
*
-0.2
0.0
0.2
0.4
0.6
PC1: forest & elevation
PC3: human development
3. Amphibians and landscape features
26
Table 3.3 Final models from stepwise logistic regression for newts.
Species Predictor Step
Regression coeff.
Wald statistic (squared)
Odds ratio
e s.e. 2 df p
I. alpestris Constant - -0.74 0.08 77.76 1 - -
PC3: human dev. 1 -0.22 0.10 5.43 1 0.020 0.80
Notes: H&L 2
8 = 4.92, p = 0.766; model 2
1 = 5.50, p < 0.019
T. cristatus Constant - -3.08 0.21 220.73 1 - -
PC1: forest & elev. 1 -0.77 0.18 18.16 1 < 0.001 0.46
PC4: PAs 2 0.37 0.16 5.19 1 0.023 1.45
Notes: H&L 2
8 = 10.02, p = 0.264; model 2
2 = 22.91, p < 0.001
L. helveticus Constant - -0.46 0.08 32.16 1 - -
PC3: human dev. 1 -0.22 0.09 5.84 1 0.016 0.80
Notes: H&L 2
8 = 8.00, p = 0.434; model 2
1 = 5.96, p < 0.05
L. vulgaris Constant - -2.31 0.14 260.02 1 - -
PC1: forest & elev. 1 -0.72 0.14 26.17 1 < 0.001 0.49
Notes: H&L 2
8 = 7.05, p = 0.531; model 2
1 = 25.62, p < 0.001
Step = logistic regression step that added covariate; Goodness of fit statistic: H&L = Hosmer
and Lemeshow.
PC3 was negatively related to Ichthyosaura alpestris presence, resulting in a model
accounting for 67.9% of the variance. Every unit increase in human development
markers decreased the odds of this species‟ presence by a factor of 0.80 (Table 3.3
and Figure 3.6a).
The presence of T. cristatus in grid squares was best modelled by PCs 1 and 4
(Figure 3.6b), explaining 94.6% of the variance: it has a negative relationship with
PC1 representing forest and elevation, while the likelihood of T. cristatus presence
increases by a factor of 1.45 in association with PC4, protected areas.
Like I. alpestris, presence of Lissotriton helveticus was also best predicted by a
negative relationship with PC3 (Table 3.3 and Figure 3.6c), the final model
explained 61.2% of the variance.
PC1 showed a negative association with L. vulgaris: there is a highly significant
difference in forest cover and elevation between grid squares where L. vulgaris was
recorded and where it was not recorded (Table 3.3 and Figure 3.6d). The final model
accounted for 90.4% of the variance.
3. Amphibians and landscape features
27
PC
valu
e (
±s.e
.)
(a) I. alpestris
(b) T. cristatus
PC
valu
e (
±s.e
.)
(c) L. helveticus
(d) L. vulgaris
Figure 3.6 Mean values of the significant principal components predicting newt
species‟ occupancy in 1 km grid squares.
* p < 0.05, *** p < 0.005, indicating the step significance (Wald statistic); = species
absent / unrecorded, = species present.
3.3.3 Landscape predictors of amphibian diversity
None of the four principal components contributed significantly to a logistic
regression model of the influence of landscape features on amphibian species
richness (Table 3.4).
Table 3.4 Multiple logistic regression chi-square values for landscape features and
amphibian species richness.
Predictor -2 Log
Likelihood
Likelihood Ratio Tests
2 df p
Constant 2665.00 - - -
PC1: Forest & elev. 2026.00 6.19 7 0.518
PC2: Water 2025.00 5.48 7 0.602
PC3: Human dev. 2024.00 4.85 7 0.678
PC4: PAs 2026.00 6.16 7 0.521
*
-0.2
0.0
0.2
PC3: human development
***
*
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
PC1: forest & elevation
PC4: PAs
*
-0.2
0.0
0.2
PC3: human development
***
-0.6
-0.4
-0.2
0.0
0.2
0.4
PC1: forest & elevation
3. Amphibians and landscape features
28
3.4 Discussion
Many previous studies examining the effect of landscape variables on amphibian
occupancy have looked at breeding ponds and the surrounding terrestrial habitat (e.g.
Beebee, 1977, 1979, 1980, 1981, 1985; Pavignano et al., 1990; Marnell, 1998; Van
Buskirk, 2005). The current chapter uses a large set of presence-only data and is
lacking the detail included by others with small sets of presence-absence (or
detection–non-detection) data, but while both data types are susceptible to errors
caused by non-detection (Gu and Swihart, 2004), here detail is forfeited for sample
size.
The records held by the MNHN indicate comprehensive coverage of the whole
country at the 5 km grid square level, as used in Luxembourg‟s 2003 atlas of
amphibian distributions (Proess, 2003b). However it was surprising that at a higher
magnification (1x1 km grid) many squares (76.6%) have never been surveyed or had
amphibians reported. In a small wealthy country with a strong network of
professional and amateur naturalists it could perhaps have been expected that a
greater part of the country would be surveyed, or that the unsurveyed squares would
mostly be urban, but this was not the case. A scarcity of still waterbodies may
explain the high number of unsurveyed squares: the maps show still water in
relatively few grid squares (14.9%), although 84.6% of squares have running and/or
still water. However, the maps were not ground-truthed and might be missing some
of the very small or temporary pools and ditches used by amphibians.
The 5 km scale used in the Luxembourg atlas (Proess, 2003b) is similar to that
recently used by other European countries with active herpetological research and
monitoring groups. For example, on their herpetology group websites the
Netherlands (RAVON, 2008) and Switzerland (KARCH, 2002) report 5 km grid
square data for all of their amphibian species, while larger countries Spain
(Ministerio de Medio Ambiente y Medio Rural y Marino, 2008) and Italy (Societas
Herpetologica Italica, 2009) report 10 km square data. In the UK the National
Amphibian and Reptile Recording Scheme aims to collect data at a 1 km scale by
allocating map squares to surveyors (NARRS, 2011). However, so far
comprehensive data at a 5 km scale are publically available only from a few counties
3. Amphibians and landscape features
29
(e.g. Clemons, 1998; Guest and Harmer, 2006). Each country‟s most common
species show records from almost every grid square countrywide; the scales selected
probably reflect practical constraints such as display space and clarity more than
limitations of the data. Distribution maps for each amphibian species found in
Luxembourg are given in Appendix 5.
The principal component representing mostly the three forest types and elevation
(PC1) was selected by logistic regression as predicting records of five of the eight
amphibian species modelled. There is extensive forest coverage (38.6%) in
Luxembourg, predominantly in the hills and valleys of the north and in patches
between the towns and farms that dominate the central and southern areas.
Pelophylax spp. were found to be negatively predicted by PC1, which is consistent
for a lowland species (IUCN, 2006) with a preference for open landscapes (Van
Buskirk, 2005) and open waterbodies where they can bask in the sun (Sinsch, 1984).
Pelophylax spp. are very widespread in Luxembourg (Figure 10.1b, Appendix 5) and
their conspicuous behaviour makes them unlikely to be missed by surveyors
(Beebee, 1980; but see 'detectability rates' given in Chapter 7). The broad habitat
categories created by principal component analysis mean that the selected variables
are not always what would be expected by experienced field surveyors. For example
Pelophylax spp. are generally considered to be strongly linked to waterbodies, but
the PCA did not identify this with the current dataset.
T. cristatus and L. vulgaris were also negatively related to PC1 (forest and elevation)
– the similarity of their habitat preferences have often been noted (e.g. Dolmen,
1980; Denton, 1991), although usually researchers conclude that some scrub or
woodland in the terrestrial habitat near to breeding ponds is essential for both species
(Beebee, 1977, 1981; Denton, 1991; Van Buskirk, 2005). The negative relationship
here may be due to the scale of the comparison between occupied and unoccupied
squares – many grid squares had very high (≈ 100%) forest cover and were on hills
(hence forests and elevation loading onto the same component), which are unsuitable
for T. cristatus (Schoorl and Zuiderwijk, 1980) and make the relationship appear
biased towards a preference for less forest.
3. Amphibians and landscape features
30
Bufo bufo and A. obstetricans were significantly positively associated with
forest/elevation (PC1) and human development (PC3). Forests provide ideal
terrestrial habitat for both of these species, which are aquatic only for relatively short
periods (Bousbouras and Ioannidis, 1997; AmphibiaWeb, 2010) and as adults spend
most or all of the year foraging or sheltering terrestrially. A. obstetricans in
particular require many small holes, cracks and fissures (Marquez, 1992) which are
abundant under tree roots and leaf litter, although they may also inhabit grassland.
Previous work has also suggested a strong association between Bufo bufo and urban-
suburban ponds in England, Switzerland and Jersey (Beebee, 1979, 1981; Van
Buskirk, 2005; Wilkinson, 2007). However, human developments generally have a
negative impact on amphibian dispersal (Holenweg Peter, 2001; Schmidt and
Zumbach, 2008) and Bufo bufo suffer particularly high mortality on roads as they
migrate to and from breeding ponds (Hels and Buchwald, 2001; Orłowski, 2007),
indicating that the positive relationship exists despite hazards encountered during
migration and dispersal. In the course of amphibian surveys during 2007 and 2008,
Bufo bufo were regularly seen on the cycleways beside major roads, where there are
also deep pools to collect water runoff. These artificial habitats might contribute to
the association between Bufo bufo and human development in Luxembourg. A.
obstetricans is tolerant of a wide range of habitats, including urban areas (IUCN,
2006), but a significant association with human-influenced landscapes was not
expected. This may be due to bias in the data, whereby animals are more likely to be
seen or heard near to human habitations and records of rare species are more likely
to be reported (Dennis and Thomas, 2000).
I. alpestris and L. helveticus showed a negative association with human
developments. The preference of newts for open habitat, without buildings and roads
has been noted before (e.g. Beebee, 1979; Van Buskirk, 2005). Although the forest
PC was not selected by these species‟ regression models, both I. alpestris and L.
helveticus are common in open deciduous forest (Kwet, 2009), being more tolerant
of the low pH caused by leaf litter than T. cristatus and L. vulgaris (Marnell, 1998).
Two PCs only occurred once in the final models: PC2 (water) was negatively
associated with R. temporaria, although this model only predicts square occupancy
3. Amphibians and landscape features
31
at slightly above chance levels and therefore is not very informative. The poor model
performance may simply indicate that none of the variables measured or principal
components were good predictors of R. temporaria, but water was a marginally
better predictor than chance alone. It is counter-intuitive that a pond-breeding
amphibian should occur less often in grid squares with more water, however both the
original running water and still water variables that loaded heavily on to the water
PC had lower average values in grid squares where R. temporaria was recorded. The
protected area PC was added at the second regression step for T. cristatus, showing a
positive relationship; the relationship between amphibian species‟ distributions and
PAs is examined in the next chapter (Chapter 4).
Suitable terrestrial and aquatic habitats are necessary for amphibian populations to
thrive. Most species, including those in Luxembourg, are on land for the majority of
the year. Terrestrial habitat also influences conditions in breeding ponds, for
example rain falling through pine trees is more acidic than rain falling through birch
woodlands (Beebee et al., 1990) and road run-off contains different pollutants to
agricultural run-off. In this chapter detail was compromised to use a large dataset:
the exact locations of records in the MNHN database are not known and therefore
the landscape features measured here share the same map square as the record rather
than being within an evenly measured distance from it. Also, here it is unknown
whether the MNHN records are breeding sites or just incidental records, because the
life-stage columns in the original database were often incomplete. Despite the coarse
nature of the MNHN dataset, the findings presented here are broadly in agreement
with previous studies, where forest cover and markers of human development were
found to be important predictors of amphibian diversity.
Characteristics of amphibian breeding ponds are explored in a smaller, more detailed
dataset in later chapters (Chapters 5 and 6) and local terrestrial habitat covariates are
examined as predictors of species‟ occupancy in ponds during the breeding season
(Chapter 7).
4. Range sizes and protected areas
32
4. Amphibian range size change and the effectiveness of
Luxembourg’s protected areas
Range sizes and protected areas
Summary: Based on the principle of providing protection from the
threats imposed by human activities, protected areas (PAs) have been
promoted and enforced by national and international legislation as a key
strategy in biodiversity conservation. A total of 20.2% of Luxembourg‟s
area has a statutory PA designation. The five common amphibian species
in Luxembourg (Rana temporaria, Pelophylax spp., Bufo bufo,
Ichthyosaura alpestris and Lissotriton helveticus) were found to be
equally distributed inside and outside PAs. Controlling for differences in
survey effort, their range sizes either increased or remained the same
between pre-1985 and 1996-2005. Over the same period the less
common Alytes obstetricans records more than doubled, while Triturus
cristatus and L. vulgaris were underrepresented in PAs and their range
sizes shrank. The rarest species to be assessed, Hyla arborea and
Bombina variegata, showed alarming range size reductions, but all
remaining sites are within PAs. PAs form a significant part of most
amphibian species‟ ranges in Luxembourg and therefore serve an
important function in their conservation.
4.1 Introduction
4.1.1 Protected areas
Protected areas (PAs) ideally perform two key functions: to hold a sample of
biodiversity and to protect this sample from external pressures (Margules and
Pressey, 2000). However they tend to occur on land that is of little commercial value
(e.g. at high elevation or too water-logged for agriculture), which can create a bias in
the types of habitat, and therefore species, that are conserved (Jackson and Gaston,
2008). Another bias in the creation of PAs arises from the focus on rare or threatened
species or remarkable habitats, paying little attention to the common species that
may also be in sharp decline and not well accommodated within them (Devictor et
al., 2007), although Devictor et al. (2007) also noted that PAs could act as important
„spatial refuges‟ for common species in decline. As the spaces between PAs continue
to undergo development and to support less biodiversity, so PAs are increasingly
isolated from similar habitat patches (Gaston et al., 2008).
4. Range sizes and protected areas
33
There is surprisingly little information available on the degree to which species are
restricted to PAs (Gaston et al., 2006) – while PAs may be valuable as population
strongholds, isolation threatens long-term survival. In a meta analysis of „BAP‟
(Biodiversity Action Plan) species in Britain, Jackson and Gaston (2008) found that
more than half were mostly or completely restricted to PAs.
Signatories – including Luxembourg – to the Convention on Biological Diversity
(CBD), which followed the United Nations 1992 Conference on Environment and
Development, pledged to aim to significantly reduce or halt the rate of biodiversity
loss by 2010, at national, regional and global scales. The CBD aims to “set the
global biodiversity agenda” through several key principles: communication,
education, raising public awareness and engaging key players and stakeholders
(Convention on Biological Diversity, 2002).
In the same year as the CBD the European Union (EU) launched the Natura 2000
network, a combination of the existing „Birds Directive‟ (Council Directive
2009/147/EC) and a new complementary „Habitats Directive‟ (Council Directive
92/43/EEC). Under the Natura 2000 directives, EU member states are “obliged to
maintain or re-establish in a favourable conservation status” (Grand-Duché de
Luxembourg, 2004) a long list of species and habitats at sites approved by the
European Commission (EC). Countries are required to protect 18% of their land area
under Natura 2000, an arbitrary figure representing the political agenda rather than
useful or realistic goals (Gaston et al., 2008), with sanctions imposed by the EC for
failure to comply (Joint Nature Conservation Committee, 2010).
National law in Luxembourg defines all PAs under three titles: Natura 2000 sites,
National Protected Areas and Local Protected Areas (Grand-Duché de Luxembourg,
2004), although currently none are designated under the latter. The Natura 2000
coverage comes close to EC targets, at 17.5% of the country‟s area, and statutory
PAs of all designations cover 20.2% of the country. Luxembourg has the third
highest percentage of PA cover out of the 48 countries in geographical Europe,
which is remarkable for the country with the seventh highest human population
density (Gaston et al., 2008). Thirty-eight sites are designated under the Habitats
4. Range sizes and protected areas
34
Directive and thirteen under the Birds Directive, covering 352.0 km2 and 160.2 km
2
respectively, with the bird zones almost entirely inside the Habitats areas (Ministère
de l'Environnement, 2003).
4.1.2 Range size
Contracting range sizes of species under threat have regularly been documented and
occur in various patterns according to the types and scale of the threat. A species‟
„niche breadth‟, or its degree of habitat specialisation, determines its sensitivity to
habitat alteration. The way in which range size is measured affects if or how changes
are detected (Gaston, 1996): for example measuring the „extent‟ of a species‟ range
(i.e. the area within a polygon drawn around all populations) will not detect losses
from the centre of its range.
Range sizes are often thought to reduce from the edges, where populations might be
smaller, with less immigration and less access to other suitable habitat. At a very
broad scale a threat occurring across a gradient can shift a species‟ range; for
example agricultural intensification spreading across a continent (Thomas et al.,
2008) or climate change pushing species‟ entire ranges towards the poles (e.g.
Parmesan and Yohe, 2003; Root et al., 2003; Thomas et al., 2004a; Anderson et al.,
2009). However, local habitat loss and fragmentation are better represented by the
more fine scale approach of counting occupied grid squares, which is more sensitive
to local changes (Swihart et al., 2003).
Previous studies have used atlas data to roughly quantify changes in species‟
distributions. Record centre (or museum) data are generally more available than
detailed long-term datasets (Skelly et al., 2003) and are a useful source for
identifying trends across taxa and for individual species. Amphibian range sizes
cannot be examined over a short period because of inter-year variations in their site
occupancy (Alford and Richards, 1999; Skelly et al., 2003). Telfer et al. (2002)
described a method to calculate a relative change index for groups of species
surveyed simultaneously – for example all birds or all carabid beetles. Using this
method on British census data, it has been demonstrated that the median range
change of butterflies was a 13% loss of 10 km squares, including some species
4. Range sizes and protected areas
35
increasing by 164%, while others declined and two species became extinct (Thomas
et al., 2004b).
Luxembourg has a large historical dataset available, as used in the previous chapter.
Here I explore the MNHN data in relation to PAs and species range sizes to address
the following questions:
Is there greater amphibian species richness inside PAs than outside?
Are amphibians in Luxembourg restricted to PAs?
Have amphibian species‟ range sizes changed between pre-1985 and 1996-
2005?
4.2 Methods
4.2.1 Historical data
See General Methods (section 2.1.1) for description of how the large dataset collated
by the MNHN was prepared for use here.
The ten most contemporary complete survey years from the MNHN database were
included in the present chapter (1996-2005 inclusive), with data up to 1985 included
for comparison of range sizes. Amphibian species were given a single occupancy
score for every 1 km grid square: „1‟ for recorded and „0‟ for not-recorded, and
species richness was given as the total number of amphibian species recorded in that
period.
4.2.2 Protected areas
Amphibian data from protected and non-protected areas (PAs) in Luxembourg were
compared. Grid squares on the country borders contain less than 1 km2 of land
within country, but are retained as long as they fit the PA percentage criteria. The
percentage of Luxembourgish land within each grid square covered by a PA was
extracted from maps using ArcMap (v9.3.1.1850; see section 2.1.1).
Grid squares containing no PA, ≥ 50%, ≥ 80%, ≥ 90% and ≥ 98% PA were grouped.
Sections of road or river are sometimes not included in a PA even if they run through
it, therefore ≥ 98% PA was taken to mean that a square was completely protected.
4. Range sizes and protected areas
36
(a) Non-PA and ≥ 50% PA
(b) Non-PA and ≥ 80% PA
(c) Non-PA and ≥ 90% PA
(d) Non-PA and ≥ 98% PA
Figure 4.1 Map of Luxembourg: grid squares (1 km2) randomly selected and
retained for analysis.
= protected areas; = non-PA squares, = PA squares.
Spatial autocorrelation of mapped data violates the independence and equal variance
assumptions of standard statistical methods (Dormann et al., 2007). In this case
neighbouring grid squares cannot be considered independent sampling units because
of the local dispersal of amphibians (Proess, 2003a). Therefore random selection was
applied to each percentage group to determine which squares would be retained for
analysis: a list of random numbers was generated (Microsoft Excel „randbetween‟
10 km
N
4. Range sizes and protected areas
37
function) and the corresponding grid squares were drawn in that order; squares
whose neighbour had already been drawn were excluded. The non-PA set varies
slightly between comparisons because the PA squares were drawn first each time.
The grid squares retained for analysis are shown in Figure 4.1.
The exact positions of amphibian records within grid squares are not known, due to
the low resolution of the MNHN dataset, therefore here it is assumed that amphibian
records originate either from within the PA or are at least close enough to benefit
from the protection it offers (e.g. terrestrial habitat or a source population).
4.2.3 Protected area species richness analyses
The amphibian species richness in non-PA grid squares was compared to that in ≥
50%, ≥ 80%, ≥ 90% and ≥ 98% PA squares with Mann-Whitney U tests. Individual
species‟ occupancy of the sets of non-PA and PA grid squares were compared with
Pearson‟s chi-square test in 2x2 contingency tables, and acceptable alpha levels were
adjusted with a sequential Bonferroni correction.
4.2.4 Range size
Following Telfer et al. (2002), the dataset was examined for changes in species‟
range size between early and later survey periods, up to 1985 and 1996-2005
respectively. Telfer et al.‟s method was developed specifically for use with
biological atlas data where similar (but not identical) surveys were conducted across
both time periods. It accounts for variable geographic coverage and recording effort,
so is ideal to use with data held by recording centres. This method examines only
grid squares that were surveyed in both periods and compares their species
assemblages between early and late.
A key assumption is that recorders tried to record all species in the taxonomic group
on any given survey, therefore species-specific surveys in the original dataset were
excluded from range size analyses. Telfer et al. (2002) acknowledge that
detectability differs between species within a taxonomic group, but these differences
remain the same over time. Most of the amphibian species occurring in Luxembourg
have overlapping habitat and survey requirements and therefore it is reasonable to
4. Range sizes and protected areas
38
assume that they would have been recorded simultaneously. However Salamandra
salamandra use different habitats and require different, separate, surveys and
therefore cannot be assumed to have been surveyed alongside the other species and
were excluded from range size analyses.
There are 2789 1 km grid squares wholly or partially in Luxembourg; only 104 of
these (3.73%) had amphibian surveys recorded in both the early and later time
periods compared here – these squares were retained for analyses. All amphibian
species occurring locally are represented in both periods. The proportion of grid
squares occupied in each survey period was used to calculate an index of relative
change, by squaring the difference between periods. Range size change is also
presented as a percentage increase or decrease.
4.3 Results
4.3.1 Amphibians and protected areas
The mean size of the 186 PAs in Luxembourg is just 5.4 km2 (±3.1 km
2 s.e.), ranging
from less than 0.01 km2 to 360 km
2 (calculated from data on the World Database of
Protected Areas (WDPA) website, IUCN and UNEP, 2009).
Table 4.1 Mann-Whitney U tests comparing amphibian species richness in PAs and
non-PAs.
% PA PA squares non-PA squares
U p n median n median
≥ 50 81 2.0 144 2.0 5545.00 0.520
≥ 80 46 1.5 151 2.0 3297.00 0.583
≥ 90 36 1.5 151 2.0 2571.50 0.597
≥ 98 20 2.0 152 2.0 1431.50 0.661
There is no difference in amphibian species richness between non-PA squares and
squares with ≥ 50%, ≥ 80%, ≥ 90% and ≥ 98% PA (Table 4.1, Figure 4.2 and Figure
4.3). Some amphibian species were recorded in too few grid squares to statistically
compare the non-PA and PA squares. For example, between 1996 and 2005
Bombina variegata was recorded in only two squares, both of which contained a
high percentage of PA (Figure 4.4e) but give insufficient data for comparison. Some
species were never recorded in high percentage PA squares and therefore were not
4. Range sizes and protected areas
39
compared with non-PA squares (e.g. there are no Triturus cristatus or Lissotriton
vulgaris in the ≥ 98% PA grid square sample).
Figure 4.2 Amphibian species richness in 1 km grid squares (all records 1996 –
2005) and all statutory protected areas in Luxembourg.
= protected areas.
Chi-square was applied to test the difference between each species‟ occupancy in
non-PA squares and the four PA categories: most species were unaffected by PAs,
either negatively or positively. Figure 4.4 appears to show small differences between
non-PA and PA squares for several species, however most were non-significant – a
larger sample size in each of the PA categories may have improved the statistical
power. Hyla arborea, Bufo calamita and Bombina variegata were too rare to show
any differences.
Pre-Bonferroni correction, significantly more S. salamandra records were from grid
squares with a high percentage (≥ 80%) cover of PA than those with no PA at all, but
there was no difference between grid squares of medium PA (≥ 50%) cover and no
PA (Figure 4.4l; ≥ 98% PA: 2 = 3.94, p < 0.05, n = 172; ≥ 90% PA:
2 = 5.90, p <
N
1 to 8 amphibian species
10 km
4. Range sizes and protected areas
40
0.05, n =187; ≥ 80% PA: 2 = 4.44, p < 0.05, n =197; ≥ 50% PA:
2 = 2.18, p > 0.05,
n = 225).
G
rid s
quare
s (
%)
Grid s
quare
s (
%)
Number of amphibian species
Figure 4.3 The frequency distribution of amphibian species richness in non-PA grid
squares and (a) ≥ 50% PA, (b) ≥ 80% PA, (c) ≥ 90% PA and (d) ≥ 98% PA grid
squares.
= non-PA squares, = PA squares.
Lissotriton helveticus was recorded in a greater proportion of grid squares with some
PA coverage than those with no PA, but this trend was only significant in the
comparison of non-PA squares with ≥ 50% PA squares (2 = 3.92, p < 0.05, n = 225;
Figure 4.4j), and none of the differences reported here stand up to sequential
Bonferroni correction.
4.3.2 Range size change
104 squares surveyed in both the early and later survey periods (≤ 1985 and 1996-
2005) were retained for range change analysis. In Figure 4.5 data points are quite
evenly spread about the intercept line, indicating that about half were recorded in a
higher number of squares in the later survey period, while others‟ range size had
shrunk or remained the same, and this is confirmed in the index of range size change
(Table 4.2). Bufo calamita was excluded from further analysis with Telfer‟s method
because it was present in too few grid squares (< 5) in the earlier period. A linear
relationship in range size between the survey periods indicates stability – i.e. species
0
10
20
30
40
50
1 2 3 4 5 6 7
(a) ≥ 50% PA
0
10
20
30
40
50
1 2 3 4 5 6 7
(b) ≥ 80% PA
0
10
20
30
40
50
1 2 3 4 5 6 7
(c) ≥ 90% PA
0
10
20
30
40
50
1 2 3 4 5 6 7
(d) ≥ 98% PA
4. Range sizes and protected areas
41
that were rare initially stayed rare and common species stayed common (Figure 4.5b;
y = 0.47 + 1.28x, R2 = 0.712).
The greatest range size changes between the early and later survey periods were in
the relatively common species, Pelophylax spp. and L. helveticus, which both
increased their range sizes. Increases were also seen in Rana temporaria, A.
obstetricans and Ichthyosaura alpestris, while Bufo bufo did not change and H.
arborea, Bombina variegata, T. cristatus and L. vulgaris all decreased in range size
(Table 4.2 and Figure 4.6).
Square
s o
ccupie
d (
%)
Protected Area (%)
Figure 4.4 The percentage of 1 km grid squares at each PA coverage level occupied
by each amphibian species (all records 1996 – 2005).
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
R. temporaria
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
Pelophylax spp.
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
H. arborea
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
Bufo bufo
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
A. obstetricans
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
Bufo calamita
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
Bombina variegata
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
I. alpestris
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
T. cristatus
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
L. helveticus
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
L. vulgaris
0
20
40
60
0 ≥50 ≥80 ≥90 ≥98
S. salamandra
4. Range sizes and protected areas
42
(a)
(b)
Figure 4.5 Range sizes of pond-breeding amphibians recorded in the early and later
survey periods from 104 grid squares (a) untransformed 1 km squares and (b) as
logit-transformed proportions of the total.
Dashed lines = intercept, for comparison of both periods on the same scale; solid line =
linear regression; see Appendix 1 for species name abbreviations.
Figure 4.6 Species percentage occupancy in grid squares surveyed in both periods.
≤ 1985, = 1996-2005; see Appendix 1 for species name abbreviations.
Ao
Bb
Bc Bv
Ha
P.spp
Rt
Ia
Tc
Lh
Lv
0
10
20
30
40
50
60
0 20 40 60
Num
ber
of r
ecord
ed 1
km
square
s i
n
late
r peri
od (
1996-2
005)
Number of recorded 1km squares in early period
(≤1985)
Ao
Bb
Bv
Ha
P.sppRt
Ia
Tc
Lh
Lv
-5
-4
-3
-2
-1
0
-4 -3 -2 -1 0
Logit-t
ransfo
rmed range s
ize
pro
port
ions in late
r peri
od (
1995-2
005)
Logit-transformed range size proportions in early period
(≤1985)
0
5
10
15
20
25
30
35
40
45
Rt P.spp Ha Bb Ao Bv Ia Tc Lh Lv
Surv
eyed g
rid s
quare
s o
ccupie
d (
%)
Amphibian species
4. Range sizes and protected areas
43
Table 4.2 Index of relative range size change: proportion of grid squares occupied in
the early and later survey periods.
Species
Percentage of surveyed squares occupied Direction and
% change
Index of relative change
(squared residual) ≤ 1985 1996-2005
R. temporaria 35.5 41.3 + 16.33 0.0034
Pelophylax spp. 18.1 38.4 + 112.00 0.0412
H. arborea 6.5 2.9 - 55.56 0.0013
Bufo bufo 23.2 23.2
0.00 0.0000
A. obstetricans 6.5 15.9 + 144.44 0.0089
Bombina variegata 4.3 1.4 - 66.67 0.0008
I. alpestris 27.5 37.0 + 34.21 0.0089
T. cristatus 23.9 13.8 - 42.42 0.0103
L. helveticus 29.0 42.8 + 47.50 0.0190
L. vulgaris 26.8 19.6 - 27.03 0.0053
4.4 Discussion
4.4.1 Protected areas and amphibian species richness
The small average size of protected areas in Luxembourg may limit their potential to
provide refuge to viable populations of threatened species due to the greater „edge
effects‟ they are subject to (Janzen, 1983, in Jackson and Gaston, 2008) and the
fragmentation caused by development of intervening land. On the other hand,
multiple small reserves can perform better than a single large one, as long as the
nearby unprotected land still provides some habitat (Hansen and DeFries, 2007).
There is greater variation in environmental pressures between multiple small sites
than within one large PA, which reduces the likelihood of a single catastrophic event
simultaneously affecting all populations (Goodman, 1987).
In Luxembourg there is no difference in species richness in squares with and without
statutory PA designation. All amphibian species in Luxembourg occur in both PA
and non-PA grid squares, except the rarest, which cannot be assessed with the
current coarse dataset, but are known to only occur in PAs. It is very encouraging
that while some populations of amphibians are safeguarded by PAs, they are
generally not restricted to them – although PAs are an essential conservation tool,
most of Europe is not protected (Papageorgiou and Vogiatzakis, 2006) and therefore
it is ideal that species thrive both inside and outside. Recent EU conservation policy
4. Range sizes and protected areas
44
to increase connectivity between PAs (Samways et al., 2010) will benefit both the
common and rare species. As shown on the maps in Figure 4.1 and Figure 4.2 the
PAs in Luxembourg form some large areas where many are adjacent to one another,
so the average size could be much larger than figures from the WDPA suggest
(IUCN and UNEP, 2009). Their listing as separate PAs reflects local circumstances,
such as land ownership and natural boundaries more than whether or not it borders
another PA.
The two least common newt species, T. cristatus and L. vulgaris, are not well
represented in the high PA cover squares, although in the previous chapter (Chapter
3) the principal component representing PAs was significantly positively related to
T. cristatus records in 1 km grid squares. This under-representation may be due to
the small sample size, but could also indicate that few PAs cater for the habitat
requirements of these species. The rarest anurans, H. arborea, Bombina variegata
and Bufo calamita, occupied very few grid squares in the first place and therefore are
not well represented by the dataset excluding neighbouring squares; their recent
history in Luxembourg is described below.
In the late 19th and early 20
th centuries, Bufo calamita is thought to have been present
throughout Luxembourg, even abundant in some places (Junck et al., 2003). Since
MNHN records began in 1955, it has only been recorded in ten 1 km grid squares at
five different locations, but between 2001 and 2005 it was known from only two
sites (a total of five 1 km grid squares) in central Luxembourg, both are former
quarries within PAs and are separated by approximately 25 km in a straight line. The
remaining populations are both small and are geographically and genetically isolated
from each other (Frantz et al., 2009).
In the five years to 2000 the MNHN database shows diminishing numbers of H.
arborea in fourteen 1 km grid squares at several sites in central Luxembourg and a
more stable population at one south-east site. However, since 2000 there are only
two records entered and the species is thought to have gone extinct at all of the
central sites, remaining only at a single pond in the south east. This last breeding
pond supports a large population and is surrounded by statutory protected areas, but
4. Range sizes and protected areas
45
it is isolated with the next nearest standing waterbodies both more than 1.5 km away,
separated by agricultural land and roads. H. arborea are reported to be declining
throughout their range, especially in western Europe, and the main cause is thought
to be intensive land use (Stumpel, 1997).
The only population of Bombina variegata remaining in Luxembourg today has been
regularly recorded since the early 1970s in a large nature reserve in the very south of
the country. Historically Luxembourg was on the north-western edge of the species‟
European distribution (AmphibiaWeb, 2010), but it is threatened at its range margins
by low precipitation and rising temperatures (Gollmann et al., 1997). The MNHN
database includes records from up to ten 1 km grid squares in the centre and south of
the country, however these are mostly from the early 1980s. Luxembourg‟s only
extant population is rumoured to have been unofficially re-introduced by a member
of the public. There are thriving populations several kilometres away in France (R.
Proess, pers. comm., 2007), but major roads and a large town in the intervening
space make natural recolonisation unlikely.
The approach taken here to compare grid squares with some level of statutory
protection to un-protected squares is useful for looking at large datasets.
Unfortunately, more specific grid references were not available for most records, and
therefore it was often not possible to identify exactly whether they were from within
PAs or nearby. Jackson and Gaston (2008) applied similar comparisons to look at the
level of restriction of „biodiversity features‟ to PAs, but with datasets from a larger
geographic area and taxonomic groups containing many more species than the
Amphibia of Luxembourg. This method is a more realistic analysis of the efficiency
of PAs than the computerised random re-distributions of PAs used in other work
(e.g. Araújo, 1999), which do not account for the fact that PAs are based on real
landscape features and many other practical and logistical constraints. However a
major weakness of all these approaches is the use of presence-only data (Altwegg et
al., 2008), a problem that cannot be addressed in the historical data, but that could be
built into recording schemes to improve the data quality in the future. A further
parameter of interest would be the age of each PA, however these data were not
available for the current study.
4. Range sizes and protected areas
46
4.4.2 Range size changes
Comparing the range sizes of Luxembourg‟s amphibians between the early and later
survey periods demonstrates, on average, no change in range size between the two
periods. However, some species exhibit a sharp decline: H. arborea and Bombina
variegata both declined between the early and later periods and as described above
are thought to have declined even further since. On the other hand, A. obstetricans,
which is also one of the rarer species, shows more than a doubling in range size
between the two periods.
A weakness of using atlas data for calculating changes in range size is the unknown
degree of bias within the data – for example rare species may be more likely to be
reported as „one off‟ sightings than common species, and records tend to be clustered
near the homes of the most active recorders (Dennis and Thomas, 2000). Telfer et
al.‟s (2002) method employed here addresses these potential biases by only
comparing grid squares that were surveyed in both periods, however this also
drastically reduces the sample size. The comparison could be extended to more than
two survey periods, but in the current study that would leave too few grid squares for
meaningful analysis and therefore also exclude more species. A larger dataset from a
wider area (e.g. Western Europe) could have increased the statistical power and
reduced the possible bias from relatively few recorders in a small area.
In this chapter and the previous one a sweeping generalisation is made about the
survey effort: that recorders tried to record all species on each survey. This does not
assume equal detectability between species (Boulinier et al., 1998), but it does
assume that suitable methods were used and that species‟ detectability rates remain
similar across time (see Chapter 7).
5. Wetland plants and amphibians
47
5. Amphibians as indicators of plant species richness
Wetland plants and amphibians
Summary: Plants are an important feature of amphibian breeding ponds,
providing shelter, food and egg-laying substrates. The strong reliance of
amphibians on aquatic plants means that it is often suggested that they
may be used as a surrogate measure of plant diversity in ponds. The
species richness of wetland plants and amphibians were assessed in 43
ponds. The plant species richness was found to increase with pond size
(following the „species – area‟ relationship), and there was also a positive
relationship between the number of plant species and the number of
amphibian species recorded. However, relationships between plant
species and individual amphibian species were weaker. Although
amphibian assemblages showed potential as indicators of plant species
richness, their role as surrogate measures of wider biodiversity is limited.
5.1 Introduction
Waterbodies are an important feature in any landscape, harbouring different fauna
and flora to the surrounding terrestrial areas and acting as corridors for dispersal of
organisms. At a regional level ponds harbour a higher proportion of the total aquatic
biodiversity than other aquatic habitats (Williams et al., 2003; Davies et al., 2008)
and should therefore receive a greater amount of attention in monitoring schemes.
A greater variety of resources, both trophic (i.e. food) and non-trophic (i.e. habitat),
is available to consumers when there is more variety among primary producers
(Tilman, 1986; Auderset Joye et al., 2002; Zhao et al., 2006). Amphibians are
dependent on plants in many ways, such that the plant assemblage at a site could be
expected to influence amphibian presence. Plants provide an egg-laying substrate,
particularly important for newts that wrap eggs individually in leaves (e.g. Duellman
and Trueb, 1986; Norris and Hosie, 2005), but also for amphibian species that attach
egg masses or strings to underwater stems or whose spawn floats in large clumps
protected from drifting apart by plants (Stebbins and Cohen, 1995; Egan and Paton,
2004). Newt eggs folded in leaves also benefit from protection from ultraviolet
radiation, mechanical damage and predation (Miaud, 1993, 1994; Marco et al., 2001;
Orizaola and Brana, 2003). Greater plant variety increases the availability of suitable
oviposition sites (Dvorak and Gvozdik, 2009) and newts show selectivity for certain
plants and leaves for egg-wrapping (Hosie and Potter unpublished data cited by
5. Wetland plants and amphibians
48
Norris and Hosie, 2005). In addition to their role in protecting amphibian eggs,
plants provide shelter to larvae and adults from predation (Babbitt and Jordan, 1996;
Babbitt and Tanner, 1997; Tarr and Babbitt, 2002) and also habitat for their
invertebrate prey.
The use of easy-to-survey taxa as a metric for those that are more difficult or costly
to survey is a popular and useful idea in ecology (Landres et al., 1988), but one that
must be applied only to well tested relationships between species and / or habitats
(Caro and O'Doherty, 1999). Caro and O‟Doherty (1999) classified so-called
surrogate species into three categories: indicators, umbrellas and flagships. Where
indicators are a species chosen to represent trends among biota sharing a habitat,
umbrellas are species whose distribution determines the extent of protected areas and
flagships are species promoted to garner public interest and support for conservation.
Indicators can be a very useful tool in improving the efficiency with which surveys
are done and maximising the scope of conclusions drawn from data collected in the
field (Balmford et al., 1996). In temperate zones amphibians have rarely been
considered as indicators because there are too few species (Sewell, 2006) but they
have shown potential for this role (Lebboroni et al., 2006). In the current study
amphibians could be considered as an indicator for wetland plants, or vice versa.
Amphibian surveys require greater time investment than plant surveys. However, in
Europe plant surveys require greater identification expertise due to the relatively
high number of species. Here amphibians are examined as indicators of plant
diversity, since the literature favours the use of higher taxonomic groups to indicate
biodiversity in lower groups (e.g. Balmford et al., 1996; Balmford and Gaston, 1999;
Gaston, 2000; Grelle, 2002). Selection of an indicator species or group should be
objectively determined by what it is intended for (Dale and Beyeler, 2001), but in the
current project it is clearly determined by the focus of the overall study.
An indicator species or group must be carefully selected and robust across the whole
geographic area to which it will be applied (Qian, 2007) – since different metrics
may apply at, for example, different altitudes (Menetrey et al., 2004) – and also
reflect changes in condition in the same way as the group or species it is
5. Wetland plants and amphibians
49
representing. The intended purpose of an indicator should be considered carefully
when it is selected and tested to avoid „erroneous assumptions‟ (Sahlén and
Ekestubbe, 2001) about its relationship to the organisms or conditions it is
indicating. A well-selected indicator would meet these criteria set out by Pearson
(1994): taxonomy fully known and described, easy to survey for and present in a
broad range of habitats. The criteria are elaborated by Sewell and Griffiths (2009) in
their review of the indicator literature, they found eight criteria for good indicators
that were repeatedly identified by other authors: ease and cost effectiveness,
correlation with other groups, sensitivity, susceptibility to a range of stressors,
distinguishable responses to manmade and natural stress, wide distribution, a group
of species rather than a single species and ecological relevance. Hereafter the term
„indicator‟ is used to describe a taxon or taxonomic group used as a proxy measure
of others.
Some studies have found a positive correlation between plant and amphibian species
richness in waterbodies. In very large scale studies amphibians have been found to
be strongly positively correlated with plant species richness, for example in reserves
across China (Zhao et al., 2006; Qian, 2007). These findings have been replicated at
both regional and local scales: amphibian species richness increases with vegetation
cover within study sites confined to a single valley or estate (Pavignano et al., 1990;
Hartel et al., 2007). Triturus cristatus presence has been identified as a potential
indicator of good aquatic plant richness in small waterbodies in both Sweden
(Gustafson et al., 2006) and England (Sewell, 2006).
Luxembourg covers a fairly small area (2585 km2) (see General methodology, Figure
2.1), therefore this study can be considered to be on a local scale. The objective of
surveying plants at ponds where the amphibian assemblage is known was to explore
the following relationships:
Is there a correlation between pond size and wetland plant species richness?
Are amphibians a good indicator of plant species richness?
Does any amphibian species live only in high or low plant diversity ponds?
Are there any strong associations between particular plant taxa and
amphibians?
5. Wetland plants and amphibians
50
5.2 Methods
5.2.1 Study area
The study area is described in the General Methodology (section 2.2). Most
vegetation surveys were conducted in 2008 (ponds 1-34, between 3-20 June 2008),
except pond 8 which was surveyed on 17 June 2007 and not resurveyed in 2008 due
to access permission being withdrawn, and the additional nine ponds (35-43) that
were surveyed 23-25 May 2009 and used only in analyses specifically about Alytes
obstetricans (Appendix 3).
5.2.2 Plant survey method
The vegetation in each study pond was surveyed from the bank and by wading to
produce an exhaustive list of the plant species present and an estimate of each
species‟ percentage cover within the high water level of the pond.
Vascular plants were mostly identified to species level using standard field guides
(Haslam et al., 1982; Hubbard and Hubbard, 1984; Rose, 1989, 2006), however
some were only identified to family or genus due to difficulty distinguishing similar
species (e.g. some grasses). Non-vascular plants were simply categorised into broad
groups (e.g. algae, mosses).
Plant species abundance was estimated as a percentage of the total pond surface area.
Where a plant covered less than 5% of the pond it was recorded as „< 5‟: these
species were only included in species richness analyses. The total „percentage‟ cover
of plants at any one pond can exceed 100% due to the overlapping of submerged,
floating and emergent plants. As all areas of each pond were surveyed, data were not
adjusted for sampling effort.
The plants surveyed were divided into wetland and non-wetland species according to
whether Stace (1997) described their normal habitat as wetland of any type (e.g.
“wetland”, “marsh”, “lake”). Plants normally only associated with dry ground were
assumed to be remnants from recent changes at the site or opportunist colonisers left
over from recent years when many ponds shrunk during hot, dry periods. Non-
5. Wetland plants and amphibians
51
wetland plant species were excluded from analyses. Common and scientific names
follow Stace (1997). Plant diversity data were also incorporated as a covariate in
occupancy modelling (Chapter 7).
5.2.3 Amphibian data
Amphibian survey methods are described in the General Methodology (Chapter 2).
For the present analyses amphibian species were recorded as a binary variable for
present / absent (1 / 0 respectively). Multiple visits to each site minimise the
possibility of non-detection of species (Zanini et al., 2009); nevertheless, as
illustrated by discrepancies in the naïve and model-estimated occupancy rates and
low detectability modelled in Chapter 7, there could be errors in ponds identified as
not occupied by any species. However, according to the final occupancy models in
Chapter 7 an adequate number of visits were made for 95% confidence that absences
were true, and for all survey methods cumulative detection probabilities were high
across multiple surveys and (in some cases) more than one season (Pellet and
Schmidt, 2005).
5.2.4 Analysis
Ponds within 25 m of each other were grouped for the current analyses. Plant data
were described by the number of species present in study ponds and their
commonness. Due to many wetland plant species being present only at low
percentage covers (< 5%) the diversity data would be significantly altered by
excluding low prevalence plants and those not identified to species level (t42 = 5.27,
p < 0.001), therefore they were retained for richness analyses but removed for cover
analyses.
Partial correlations were used to test the relationship between wetland plant species
richness and pond size, and between amphibians and plant richness and vegetation
cover. Correlation analyses between amphibians and plants were controlled for pond
size, with the log pond size as a covariate.
For the seven commonest amphibian species an ANCOVA was computed to
compare the wetland plant species richness between occupied and unoccupied ponds,
5. Wetland plants and amphibians
52
with pond size as a covariate. An a priori power analysis was calculated in
„G*POWER‟ (Erdfelder et al., 1996) to check that there was sufficient power ( >
0.8, with a medium effect size specified [w = 0.3]) for the test.
Pearson‟s chi-square in 2x2 contingency tables was applied to test specific
associations of moderately common amphibian species (Bufo bufo, T. cristatus, L.
vulgaris and A. obstetricans, present at 38%, 38%, 65% and 33% of ponds
respectively) with individual wetland plant species. The other seven amphibian
species (of 11 encountered) occurred at too many (> 91%) or too few (< 3%) ponds
to be tested for associations. A sequential Bonferroni correction was applied to
reduce the likelihood of Type I errors resulting from repeated tests (Rice, 1989).
5.3 Results
5.3.1 Plant community descriptives and pond size effect
A total of 113 species of wetland plant and 45 non-wetland plant species were
recorded in the ponds surveyed (Appendix 6). Fifty-two of the wetland plant species
recorded were only observed at very low coverage (i.e. less than 5% of a pond; n =
47) and / or could not be identified to species level (n = 12); most of the non-wetland
plants recorded did not exceed 5% cover at any pond (n = 42, of 45 species). Beyond
site descriptions, all non-wetland plant species were excluded from analyses.
The total number of wetland plant species (≤ 5% and > 5% cover) ranged from 3 to
33 (mean = 14.1, s.d. ± 7.5, mode = 9; Figure 5.2). The number of wetland plant
species in the study ponds, identified to species level and covering > 5% of a pond‟s
area, ranged between 0 and 18 (mean = 5.1, s.d. ± 3.1, mode = 5).
Thirty one point nine percent of the wetland plant species surveyed were seen in only
one pond. However, some wetland plant species were very common and found in
more than 50% of the study ponds (Table 5.1).
5. Wetland plants and amphibians
53
There was a significant positive correlation between pond size and wetland plant
species richness (r = 0.47, n = 43, p < 0.001;
Figure 5.3Figure 5.1 There was no correlation between the number of amphibian
species detected and the size of the study pond (r = 0.28, n = 34, p > 0.05).
Figure 5.2 Diversity of wetland plant species in study ponds and the frequency of
occurrence (including plants not identified to species and plants at ≤ 5% and > 5%
cover).
Table 5.1 The five most commonly occurring wetland plants across all study ponds
(n = 43).
Species Study ponds occurring in
n %
Soft-rush (Juncus effusus) 27 62.8
Common duckweed (Lemna minor) 26 60.5
Floating sweet grass (Glyceria fluitans) 22 51.2
Creeping buttercup (Ranunculus repens) 22 51.2
Bottle sedge (Carex rostrata) 19 44.2
1.0
10.0
1.0 10.0
Num
ber
of w
etlant
pla
nt
specie
s (
log)
Pond size (log)
0
2
4
6
8
10
12
14
1-5 6-10 11-15 16-20 21-25 26-30 31-35
Num
ber
of ponds
Total number of wetland plant species
5. Wetland plants and amphibians
54
Figure 5.3 The number of wetland plant species (≤ 5% and > 5% cover) plotted
against pond size (logarithmic scales; y = 0.2675x + 0.8013).
5.3.2 Amphibians as indicators of plant richness and vegetation cover
When controlling for pond size effects, the total number of plant species in ponds
increased significantly with the number of amphibian species present (r = 0.57, n =
34, p < 0.005; Figure 5.4). This relationship held when non-wetland plant species
were excluded (r = 0.56, n = 34, p < 0.005).
Although plant species richness was positively related to the number of amphibian
species present, ponds with a higher coverage of vegetation generally supported
fewer amphibian species, although this trend was not significant (r = -0.32, n = 34, p
> 0.05; Figure 5.5).
Figure 5.4 The mean number (±s.e.) of wetland plant species and all plant species
(wetland and non-wetland) found in ponds according to amphibian species diversity
(plants of ≤ 5% and > 5% cover).
= wetland plant species, = all plant species.
1.0
10.0
1.0 10.0
Num
ber
of w
etlant
pla
nt
specie
s (
log)
Pond size (log)
0
10
20
30
40
50
60
3 (n=6) 4 (n=3) 5 (n=9) 6 (n=12) 7 (n=2) 8 (n=2)
Mean n
um
ber
of pla
nt
specie
s (
±s.e
.)
Number of amphibian species present
5. Wetland plants and amphibians
55
Figure 5.5 The extent (%) of wetland plant cover (±s.e.) and the number of
amphibian species present.
5.3.3 Wetland plant richness and amphibian occupancy
A priori power analyses run in „G*POWER‟ (Erdfelder et al., 1996) returned > 0.8
for the amphibian species tested, indicating sufficient power to detect an effect if one
existed. Four species were not analysed here due to occupancy at too few (Hyla
arborea, Bufo calamita, Bombina variegata) or too many (Ichthyosaura alpestris)
ponds for comparison. Controlling for pond size effect, on average ponds occupied
by moderately common amphibian species contained more wetland plant species
than unoccupied ponds (Figure 5.6). This trend is statistically significant for T.
cristatus (F1,31 = 4.87, p < 0.05) and Lissotriton vulgaris (F1,31 = 5.65, p < 0.05).
Figure 5.6 Mean number of wetland plant species in ponds occupied / not occupied
by moderately common amphibian species, plants of ≤ 5% and > 5% cover included.
A. obstetricans data include the additional ponds surveyed in 2009.
= occupied, = not occupied; * = significant at p < 0.05; see Appendix 1 for amphibian
species name abbreviations.
0
20
40
60
80
100
3 (n=5) 4 (n=3) 5 (n=7) 6 (n=13) 7 (n=4) 8 (n=2)
Exte
nt of w
etland p
lant
cover (%
)
Number of amphibian species present
**
0
5
10
15
20
25
Rt P.spp Bb Ao Tc Lh Lv
Mean n
um
ber
of w
etland
pla
nt
specie
s (
±s.e
.)
Amphibian species
5. Wetland plants and amphibians
56
Specific relationships between plant species and amphibian species
Nineteen significant relationships were observed between wetland plant species and
amphibian species (Table 5.2), but a sequential Bonferroni correction applied to
adjust the alpha level for simultaneous tests leaves just one significant relationship.
The Bonferroni correction is very conservative, however it is necessary to avoid
incorrect conclusions.
Potamogetan natans was significantly positively associated with A. obstetricans (2
= 13.66, p = 0.0002, significant at the Bonferroni–adjusted alpha level of 0.0004):
both were absent in 60.5% of ponds, they occurred together in 20.9% of ponds and
singly in just 7.0% (P. natans) and 11.6% (A. obstetricans) of ponds (Figure 5.7).
Figure 5.7 Occupancy of Potamogetan natans in ponds by A. obstetricans
occupancy.
= P. natans absent, = P. natans present.
0
20
40
60
80
present absent
All p
onds (
%)
A. obstetricans
5. Wetland plants and amphibians
57
Table 5.2 Relationships between individual amphibian and plant species, before and
after sequential Bonferroni correction.
Pearson’s 2
Uncorrected p value
(alpha 0.05)
Relationship Direction
Sequential Bonferroni corrected
alpha
A. obstetricans
Potamogetan natans 13.66 < 0.0005 + 0.000446 *
Lysimachia nummularia 7.14 < 0.01 - 0.000455
Salix cinerea 6.29 < 0.05 - 0.000472
Ranunculus repens 6.24 < 0.05 - 0.000476
Iris pseudacorus 6.11 < 0.05 - 0.000481
Dactylis glomerata 6.03 < 0.05 - 0.000485
Bufo bufo
Carex rostrata 7.54 < 0.01 - 0.000450
Lemna minor 5.38 < 0.05 - 0.000495
Glyceria fluitans 5.38 < 0.05 - 0.000500
Mentha aquatica 4.11 < 0.05 + 0.000532
T. cristatus
Juncus inflexus 6.84 < 0.01 + 0.000459
Iris pseudacorus 6.84 < 0.01 + 0.000463
Eleocharis palustris 6.35 < 0.05 + 0.000467
Salix alba 4.44 < 0.05 + 0.000510
Rumex conglomeratus 4.11 < 0.05 + 0.000515
Mentha aquatica 4.11 < 0.05 + 0.000521
Lemna trisulca 4.11 < 0.05 + 0.000526
L. vulgaris
Myostis scorpioides 5.71 < 0.05 + 0.000490
Salix alba 4.89 < 0.05 + 0.000505
* = significant after correction; relationship direction: + positive, - negative.
5.4 Discussion
More than 70% of the plant species recorded during vegetation surveys were wetland
plants (according to Stace, 1997). Pond 34 held the highest number of wetland plants
with 33 species, while ponds 14 and 26 held just three wetland plant species each –
at pond 14 no plant covered 5% of the total area (study ponds listed in Appendix
3). The most speciose pond, 34, was in fact a cluster of small ephemeral pools
around a larger pond, at a site that is managed for its herpetological interest. The
least speciose were both unmanaged woodland ponds with complete canopy cover
and thick leaf litter in the water. The gamma (γ), or regional, diversity of wetland
plants recorded in these 43 Luxembourgish ponds (γ diversity = 113 species) is much
5. Wetland plants and amphibians
58
greater than was found in the 20 ponds surveyed by Williams et al. (2003) in
southern England (γ diversity = 67).
The number of wetland plant species increased with pond size. This has been
observed previously in small-scale studies of pond plants (e.g. Sewell, 2006) and the
species-area relationship is well documented, originally in Arrhenius‟ species-area
equation (1921) and more recently in the species-area-habitat model by Triantis et al.
(2003). Oertli et al. (2002) found pond size to be unimportant for aquatic plant
diversity, noting that several small ponds could hold more diverse flora than a
solitary big pond, but they conclude that in general the species-area relationship
appears to hold true for aquatic plants in ponds. Gee et al. (1997) suggest that it is in
fact the size of the vegetated area in the pond, rather than the pond size per se, that
determines its biodiversity.
Wetland plant species richness was observed to increase with the number of
amphibian species present in a pond. The additional habitat complexity afforded by a
variety of plants increases the ecological niches available, allowing a greater number
of species to co-exist and a greater diversity at every trophic level (Triantis et al.,
2003; Williams et al., 2008). Van Buskirk (2005) noted that habitat quality was more
important than competition from other amphibians in determining pond occupancy.
The current data set is simply a snapshot of a relationship that varies across seasons
and decades, it cannot demonstrate if, or how closely, plant and amphibian species
richness would co-vary over time. These data support the use of amphibian richness
as an indicator of plant species richness, but they are not sufficient in isolation. The
current study shows that amphibian species richness has potential as a bioindicator
for the richness of plants, but it needs further testing at more sites, across a broader
time scale and with sufficient survey methods and visits for high cumulative
detection rates (Zanini et al., 2009).
Each amphibian species exhibited a tendency to occupy ponds with a greater number
of wetland plant species. However, the difference in plant species richness between
occupied and non-occupied ponds was only significant for T. cristatus and L.
vulgaris. Other researchers have observed similar effects, notably Gustafson et al.
5. Wetland plants and amphibians
59
(2006) and Sewell (2006) both found that T. cristatus tends to occupy ponds with
higher plant richness, but have more recently concluded from field data and a review
of published papers that T. cristatus alone is not an adequate indicator of biodiversity
(Sewell and Griffiths, 2009). In the present study sample sizes were relatively small
and additional data could have increased the likelihood of detecting significant
differences for other amphibian species in the comparison.
In individual tests of association between plant and amphibian species a number of
relationships were observed, but they did not stand up to the stricter Bonferroni-
corrected alpha levels. Only the positive relationship between A. obstetricans and P.
natans is fully supported. This is perhaps more due to these species having similar
habitat preferences rather than a direct dependence on one another, for example both
thrive in permanent ponds (Bosch et al., 2008; Della Bella et al., 2008).
Plants that can negatively impact a pond‟s suitability for amphibians could have been
expected to show a greater effect on amphibians. For example common duckweed,
Lemna minor, can dominate and create eutrophic conditions (Gee et al., 1997; Janse
and Van Puijenbroek, 1998; Scheffer et al., 2003), but it only interacted with Bufo
bufo presence before the Bonferroni correction was applied. No non-native plants
were observed during the current study, however non-natives (e.g. Crassula helmsii)
can also degrade breeding conditions in amphibian ponds if they crowd out native
plants or choke the water column (Langdon et al., 2004).
Amphibians exhibit many model indicator group qualities, such as wide geographic
ranges, varied population sizes, complex life cycles and sensitivity to modern
environmental threats (Collins and Storfer, 2003). However, in common with similar
studies, here it is observed that single amphibian species do not make good
indicators of plant richness; they show greater indicator potential when used as an
assemblage (Sewell and Griffiths, 2009). There are more amphibian species in
Luxembourg than in the UK where Sewell and Griffiths‟ work was conducted, but
after removing the very common and very rare species, the remaining „suite‟ of four
or seven species is still quite small. A suite of indicators is better able to represent a
5. Wetland plants and amphibians
60
broad gradient, across both environmental conditions and geographical area (Dale
and Beyeler, 2001).
That there is a relationship between plants and amphibians is intuitive: their ecology
is intricately intertwined (e.g. Braz and Joly, 1994; Miaud, 1994; Marco et al.,
2001). Although few comparative studies of amphibians and plants in Western
Europe have been published and indicator studies have largely focussed on water
quality measures. The results here suggest that amphibian assemblages have
potential to be indicative of other taxa within their breeding ponds. However, the
dataset is relatively small and represents a restricted geographical area, so
generalisations should be made with caution. Further work should include plants and
features surrounding the study ponds, because terrestrial habitat is also important in
dictating amphibian species presence (Cushman, 2006).
6. Macroinvertebrates and amphibians
61
6. Amphibian species richness as an indicator of
macroinvertebrate family richness
Macroinvertebrates and amphibians
Summary: Invertebrates are a key food source for amphibians. In turn,
amphibians – especially eggs and larvae – are often preyed on by
invertebrates. Their interdependence makes them ideal candidates for use
as indicators of each other‟s presence or richness within an amphibian
breeding pond. Invertebrates have often been used as indicators of
general aquatic habitat health, but little has been published on their
relationships with amphibians. Although there was a positive trend in the
relationship between macroinvertebrate richness and amphibian species
richness, this was not supported statistically. One specific relationship
stood out: Lissotriton vulgaris and the Glossiphoniidae leech family
occurred in the same ponds and never separately. In the current study
only a small sample of ponds were surveyed for macroinvertebrates and
the time frame was short. A larger, longer-term dataset may have been
able to substantiate the observed trend.
6.1 Introduction
In the previous chapter amphibians were examined as potential indicators for
wetland plant richness. Correlations between amphibians and wetland plants were
found on several measures and it was concluded that there is potential to use
amphibians as indicators. However, more extensive testing of the relationship is
necessary – for example at a broader geographical scale and across a longer time
period. In this chapter the relationships between macroinvertebrates and amphibians
and plants are tested.
Most adult amphibians feed on live prey comprised principally of invertebrates and
their diets reflect niches within the habitat (Dolmen and Koksvik, 1983). They also
frequently prey on their own larvae and those of sympatric species (Duellman and
Trueb, 1986) and any live animal small enough to be ingested. However, the
relationship is not asymmetric and some invertebrate species also predate on
amphibians. For example, the larvae and adults of the great water beetle (Dytiscus
marginalis) and dragonfly naiads (order Odonata, Anisoptera) are formidable
predators that take amphibians in all life stages and some leeches (family
6. Macroinvertebrates and amphibians
62
Haemopidae) feed on amphibians without necessarily killing them (Duellman and
Trueb, 1986).
As a group, invertebrates occupy a broader range of environmental niches than
amphibians, and have regularly been used as indicators of habitat health, particularly
to monitor water quality. Many schemes have been developed for use in rivers, all
known by their acronyms, including the BMWP (Biological Monitoring Working
Party, Hawkes, 1998) and RIVPACS (River Invertebrate Prediction and
Classification System, Centre for Ecology & Hydrology, 2009); with variations on
both used extensively in Europe.
Similar schemes have been developed to monitor and classify standing waters, both
in the UK and continental Europe. For example, PSYM (Predictive System for
Multimetrics, Howard, 2002) and NPS (National Pond Survey, UK, Biggs et al.,
1998) in the UK, and IBEM (Indice de Biodiversité des Etangs et Mares,
Indermuehle et al., 2008) and PLOCH (Plans [PL] d‟eau [O] suisses [CH], Oertli et
al., 2005) in Switzerland. These methodologies require a number of standardised
measurements of the biological features of waterbodies, weighting families or
species according to their rarity or sensitivity to environmental conditions, to
produce a single score. The scores can be used to infer water quality or to measure
the „conservation value‟ of a site.
Some studies have found invertebrates to be ideal surrogate measures of plant
richness (e.g. PSYM, BMWP; Hawkes, 1998; Howard, 2002), which is logical
where one taxon is largely dependent on the other, because an increase in the
availability and variety of habitat can influence organisms further up the food chain.
However, in their large scale study Oertli et al. (2005) found very low correlations
between the plant, invertebrate and amphibian taxa tested. Further, they based the
PLOCH scheme on the sum of scores assigned to individual species rather than using
any taxa as an indicator of another.
The current chapter follows on from the previous one on wetland plants and
amphibians. Macroinvertebrates fill a range of ecological niches within ponds, from
6. Macroinvertebrates and amphibians
63
prey items for amphibians to predators of amphibians. In this chapter the following
questions are addressed:
Is there a correlation in species richness between macroinvertebrates and
wetland plants?
Is amphibian diversity a suitable surrogate measure of macroinvertebrate
diversity?
Do amphibians favour high or low macroinvertebrate diversity ponds?
Are there any strong associations between particular macroinvertebrate
families – especially predatory invertebrates – and amphibian species?
6.2 Methodology
6.2.1 Study area
Table 6.1 The amphibian species present in ponds surveyed for macroinvertebrates.
Pond Amphibian species present
Number of
species Rt P.spp Ha Bb Bc Bv Ao Ia Tc Lh Lv
2 Weiden 3&4 6
5 Laaschtert 2 6
10 Biergwiss 7
15 Werwelslach 1&2 5
17 Giele Botter 6
18 Féitzemuer 5
22 Dréisch 6
25 Neiländigfeld 6
27 Haard 6
31 Weiergewan 7
34 Steinfort 8
See Appendix 1 for species name abbreviations.
Eleven of the ponds surveyed for amphibians and plants in 2007 and 2008 were
chosen for macroinvertebrate surveys (Table 6.1). They were selected to include a
good spread across the study area of south and central Luxembourg, medium to high
amphibian species richness and a variety of amphibian assemblages and pond types.
Ponds 2 and 15 both constituted two pools in very close proximity to each other and
here are considered as a single pond each, as are the big permanent pool and small
ephemeral pools at pond 34.
6. Macroinvertebrates and amphibians
64
6.2.2 Macroinvertebrate survey method
Two rounds of macroinvertebrate surveys were conducted during late spring and
summer, 6-22 May 2008 and 27 June - 16 July 2008 respectively. Surveys were
usually completed within one morning or afternoon, but on one occasion a survey
started in the late afternoon had to be finished the following morning (pond 34,
survey 1).
Macroinvertebrates were considered to be those which could not easily fit through a
2 mm gauge pond net. Microinvertebrates were often retained in the net within
vegetation and substrate, but they were not included because their populations are
subject to large and rapid fluctuations in number. Common and scientific names
follow those given on the taxonomic nomenclature website
www.zipcodezoo.com/Animals (ZipcodeZoo.com, 2009). Non-aquatic invertebrates
were excluded, assumed to have fallen in from overhanging vegetation.
Macroinvertebrate survey methodology followed the UK‟s National Pond Survey
(NPS) (Biggs et al., 1998):
1. Mesohabitats within the pond were identified (e.g. open water, stands of
emergent vegetation, shallow and deep areas).
2. Three minutes of netting time was divided equally between the mesohabitats.
Netting time includes only the time that the net is in the water: typically this
would constitute multiple sweeps of 10-20 seconds each. In small ponds 100%
coverage was achieved in fewer sweeps and less time. Occasionally in larger
ponds, or where there was thick mud or leaf litter, it was not possible to net for
3 minutes because the sorting and identification time was too great (> 7 hours).
3. A 2 mm gauge net, measuring 26 x 25 cm across the opening and 50 cm deep,
was swept in front of the wading surveyor in a continuous motion through the
water for the required netting time. The net was moved throughout the water
column, including across the surface.
4. Samples were removed from the net to a white sorting tray, from which
invertebrates were identified on site using standard field guides (Unwin, 1984;
Croft, 1986; Chinery, 1993; Tachet et al., 2000; Olsen et al., 2001; Wallace,
6. Macroinvertebrates and amphibians
65
2006) to family level (genus or species where possible) and counted. Counts for
very numerous specimens present were estimated, based on an accurate count
of a sub-sample. Unidentified invertebrates were compared via photographs
and field notes between ponds to ensure that the total species number is
accurate.
5. All samples were returned to the pond as soon as possible after identification.
6.2.3 Amphibian data
The survey methods for amphibians are described in the General Methodology
(Chapter 2). Here amphibians are simply described as present or absent (1 / 0
respectively). As for the plant analysis (Chapter 5), the cumulative amphibian
detection rates (Pellet and Schmidt, 2005) of multiple methods and survey visits are
high and it is unlikely that there were false negative occupancy errors for amphibian
species in the sub-set of ponds used in the current chapter.
6.2.4 Analysis
A species weighting system for aquatic invertebrates, such as BMWP or PLOCH,
has not been applied as none has been developed for still water in the Luxembourg
area of Europe. Therefore all analyses used basic counts or presence / absence.
Species and family richness data were not adjusted for different sampling times,
counts were extrapolated to give a full three minutes of survey effort in ponds 10
(survey 1) and 27 (survey 2) where the full netting time or 100% of pond could not
be completed. All analyses reported here are of total catch sizes or family richness,
which is generally a good predictor of species richness (Williams and Gaston, 1994).
The same analyses with the less accurate species-level data produced very similar
results and are not reported, except in comparisons of predatory invertebrates and
amphibians.
The Bray-Curtis Dissimilarity Index (BS) (Bloom, 1981) indicated a low level of
dissimilarity in species composition between surveys (all ponds BS ≤ 0.4) and there
was no difference between surveys in terms of numbers caught (t10 = 0.13, p > 0.05).
6. Macroinvertebrates and amphibians
66
Therefore there was no need to weight the surveys for analysis and data were pooled
by simply adding the surveys together.
The data were described by the percentage identified to family, genus and species
levels and the average and range of catch sizes and family diversity. Outliers (> 2
standard deviations from the mean) were identified with z-scores.
Pearson product-moment correlations were applied to examine the relationships
between all invertebrates – or just predatory macroinvertebrates – and pond size,
amphibian richness, wetland plant diversity and vegetation extent. Only those
invertebrates well known to be predators of amphibians were included here:
dragonfly naiads (Odonata, Anisoptera), great diving beetle (Dytiscidae) larvae and
adults, horse leeches (Haemopidae) and water scorpions (Nepidae). Other aquatic
invertebrates (e.g. other beetles and caddisfly larvae) could predate amphibian eggs
and small larvae, but are not considered here due to limited published information on
exact species interactions.
General macroinvertebrate diversity and predator diversity were compared between
ponds where each moderately common amphibian species was present to those it
was absent from with one-way ANOVA. A priori power analyses in G*POWER
(Erdfelder et al., 1996) indicated sufficient power for statistical analysis.
Chi-square tests were applied to the interactions between presence / absence of pairs
of all moderately common invertebrate (n = 38) and amphibian (n = 4) species.
Sequential Bonferroni corrections (Rice, 1989) were made within amphibian species
to adjust the accepted alpha value for repeated tests.
6. Macroinvertebrates and amphibians
67
6.3 Results
6.3.1 Macroinvertebrate descriptives
Overall, 88.4% of invertebrate specimens were positively identified to family level.
A total of 69 macroinvertebrate families were recorded in the 11 ponds surveyed,
with 60.0% of specimens in these families confidently identified to genus level, but
only 25.8% identified to species level. Ponds held a mean of 19.6 macroinvertebrate
species (s.d. ± 6.1, range 8 to 32; not all identified). The number of genera per site
cannot be calculated because it is not known how many genera the unidentified
species comprised.
The total count of macroinvertebrates netted in a single pond (the sum of surveys 1
and 2) ranged from 167 to 3833 individuals (mean 1623.9 ± s.d. 1123.0) and the
number of families ranged from 14 to 34 (mean 24.5 ± s.d. 5.5). Pond 10, which was
heavily shaded and thick with leaf litter, held the fewest macroinvertebrates in both
total catch size and family count, while pond 31 held the highest total catch of
macroinvertebrates and pond 34 had the most families.
In terms of macroinvertebrate counts and richness, pond 17 is an outlier (z = 2.19).
Among the ponds surveyed for macroinvertebrates pond 17 was unusual in that it
was very large, the only pond with a hard rock substrate and little vegetation or leaf
litter in the water. Most of the following analyses are run both with and without pond
17.
Table 6.2 The five most numerous macroinvertebrate families across all 11 ponds
surveyed.
Type Order Family Total number
caught Percentage of total catch (%)
Mayfly Ephemeroptera Baetidae 2880 16.1
Snail Basommatophora Planorbidae 2424 13.6
Snail Basommatophora Lymnaeidae 2330 13.0
Caddisfly Tricoptera Limnephilidae 1872 10.5
True fly Diptera Chaoboridae 1777 10.0
6. Macroinvertebrates and amphibians
68
Table 6.3 The most commonly occurring macroinvertebrate families across all 11
ponds surveyed.
Type Order Family nponds Percentage of
ponds (%)
Mayfly Ephemeroptera Baetidae 11 100.0
Bug Hemiptera Gerridae 11 100.0
True fly Diptera Chaoboridae 10 90.9
Beetle Coleoptera Dytiscidae 10 90.9
Caddisfly Tricoptera Limnephilidae 10 90.9
Oligochaete worm Haplotaxida Naididae 10 90.9
Bug Hemiptera Notonectidae 10 90.9
Many macroinvertebrate families were found in only one pond (Figure 6.1); 27
families occurred in just one pond and only two occurred in all 11 ponds.
Macroinvertebrate assemblages were generally dominated by the mayfly family
Baetidae (Table 6.2 and Table 6.3), which was present in all 11 ponds surveyed and
very numerous, accounting for 16.1% of the total number of macroinvertebrates
caught during both surveys. The pond skater family Gerridae, comprising only one
species in the current study, Gerris lacustris, was equally widespread (11/11 ponds)
but caught in lower numbers (2.4% of total catch). Snails of the Planorbidae
(ramshorn snail) and Lymnaeidae (freshwater snail) families were also very
numerous, representing 13.6% and 13.0% of the total catch respectively and both at
8/11 ponds. See Appendix 7 for a full list of the invertebrates encountered.
Fish were not found on any survey in any of the ponds surveyed for
macroinvertebrates.
Figure 6.1 The number of ponds that macroinvertebrate families occurred in.
0
5
10
15
20
25
30
1 2 3 4 5 6 7 8 9 10 11
Num
ber
of
macro
invert
ebra
te fam
ilie
s
Number of ponds
6. Macroinvertebrates and amphibians
69
6.3.2 Pond size and macroinvertebrate numbers and family richness
With all ponds included, a Pearson product-moment correlation indicates that there is
no correlation between pond size and the number of macroinvertebrates counted (r =
0.29, n = 11, p > 0.05) or the number of macroinvertebrate families present (r = 0.01,
n = 11, p > 0.05). However, with the outlying pond 17 removed, pond size does
correlate positively with the number of macroinvertebrates caught (r = 0.86, n = 10,
p < 0.01), but still not with the number of families (r = 0.46, n = 10, p > 0.05)
(Figure 6.2).
Figure 6.2 The relationship between pond size and macroinvertebrate family
richness (y = 0.0033x + 3.1533) and the total count of macroinvertebrates (y =
0.1261x + 6.2569) (logarithmic scales; including pond 17).
= number of families; = total count of macroinvertebrates.
6.3.3 Macroinvertebrates and wetland plant richness and extent
No relationship was observed between the richnesses of wetland vegetation and
aquatic macroinvertebrates (Table 6.4 and Figure 6.3). Excluding the outlying pond
17 did not reveal any significant relationships either.
1
10
1 10
Num
ber
of m
acro
invert
ebra
tefa
mili
es o
r to
tal count (l
ogs)
Pond size (log)
6. Macroinvertebrates and amphibians
70
Table 6.4 Pearson's product-moment correlation results for wetland vegetation and
macroinvertebrates, including all ponds surveyed for invertebrates (n = 11).
Vegetation Macroinvertebrates r p
Extent Total count 0.28 0.414
Family richness 0.37 0.268
Species richness Total count 0.53 0.096
Family richness 0.24 0.471
Num
ber
of
macro
invert
ebra
te f
am
ilies
Figure 6.3 The number of macroinvertebrate families in surveyed ponds by (a) the
number of wetland plant species present and (b) the extent of wetland plant cover.
6.3.4 Amphibians as indicators of macroinvertebrate diversity
There is no correlation between the number of amphibian species present and the
number of macroinvertebrate families (r = 0.18, n = 11, p > 0.05) or the total count
of macroinvertebrates (r = 0.487, n = 11, p > 0.05), although Figure 6.4 suggests a
trend. A relationship would be difficult to detect statistically with the small sample
size of 11 ponds, especially with the narrow range of amphibian species richnesses
(5 – 8). Removing the outlying pond 17 does not make either of these comparisons
significant (r = 0.171, n = 10, p > 0.05; and r = 0.497, n = 10, p > 0.05 respectively).
pond 17
0
10
20
30
40
0 5 10 15 20
Number of wetland plant species
(a)
pond 17
0
10
20
30
40
0 20 40 60 80 100
Extent of wetland plant cover (%)
(b)
6. Macroinvertebrates and amphibians
71
Mean c
ount
of
macro
invert
ebra
tes (
±s.e
.)
Mean n
um
ber
of
macro
invert
ebra
te f
am
ilies (
±s.e
.)
Number of amphibian species present
Figure 6.4 (a) The mean count of macroinvertebrates and (b) the mean number of
macroinvertebrate families found in ponds according to the number of amphibian
species present.
6.3.5 Macroinvertebrate richness and amphibian species occupancy
The number of macroinvertebrate families found in ponds occupied / not occupied
by moderately common amphibian species was compared using one-way ANOVA:
there were no significant differences between occupancy states (Figure 6.5).
Although sample sizes were low, a priori power analyses (run in G*POWER,
Erdfelder et al., 1996) suggest that there was sufficient power ( > 0.8) to detect an
effect if it had existed. False absences of the amphibian species in these data are
unlikely (as explained above, section 6.2.3), but it is very likely that invertebrate
families were missed, which would have affected the chance of detecting
differences.
The count of predatory invertebrates varied widely between ponds (mean 30, range
6-59), with the greatest number being caught in the pond with the most amphibian
species (Figure 6.6a). There was no relationship between the number of amphibian
species present and the number of predators caught during surveys (r = 0.105, n =
11, p > 0.05), and the removal of outlying pond 17 did not substantially change the
result (r = 0.084, n = 10, p > 0.05).
0
500
1000
1500
2000
2500
3000
3500
4000
5 (n=2) 6 (n=6) 7 (n=2) 8 (n=1)
(a)
0
10
20
30
40
5 (n=2) 6 (n=6) 7 (n=2) 8 (n=1)
(b)
6. Macroinvertebrates and amphibians
72
Figure 6.5 Mean number of macroinvertebrate families in ponds occupied / not
occupied by moderately common amphibian species.
= occupied, = not occupied; see Appendix 1 for amphibian species name abbreviations.
A priori power analyses (in G*POWER, Erdfelder et al., 1996) indicated sufficient
power ( > 0.8) for ANOVAs comparing amphibian occupancy and the total number
of predatory invertebrates and family and species richness (Figure 6.6b). There were
no differences in the count of predatory invertebrates between ponds occupied or not
occupied by any of the four amphibian species tested. However, there were just
significantly more predatory invertebrate species in ponds occupied by Bufo bufo
than not occupied, and fewer predatory invertebrate families in ponds occupied by A.
obstetricans than not occupied (Table 6.5). The same two relationships are
significant with outlying pond 17 removed.
Table 6.5 One-way ANOVA values comparing the invertebrate predator total count
and the number of families and species in ponds occupied / not occupied by
moderately common amphibian species.
Amphibian species
nponds Number of predatory
invertebrates Direction of relationship Count
F1,9 Families
F3,7 Species
F4,6 Occupied Not
occupied
Bufo bufo 5 6 0.53 0.39 4.64 * +
A. obstetricans 2 9 0.06 5.30 * 0.60 -
T. cristatus 6 5 1.73 0.85 2.01
L. vulgaris 9 2 0.00 0.53 2.18
* = significant at p < 0.05; no comparison was significant after sequential Bonferroni
correction within species; direction of relationship: „+‟ = positive, „-‟ = negative.
n=5 n=2 n=6 n=9n=6 n=9 n=5 n=20
5
10
15
20
25
30
35
Bb Ao Tc Lv
Mean n
um
ber
of m
acro
invert
ebra
te
fam
ilies (
±s.e
.)
Amphibian species
6. Macroinvertebrates and amphibians
73
Figure 6.6 The mean number of predatory invertebrate families with (a) the number
of amphibian species present and (b) amphibian species‟ occupancy.
= occupied, = not occupied; see Appendix 1 for amphibian species name abbreviations.
6.3.6 Specific relationships between macroinvertebrate and amphibian species
Chi-square analyses revealed eight moderately common macroinvertebrate families
to be associated with one or more moderately common amphibian species: Bufo
bufo, A. obstetricans, T. cristatus and L. vulgaris (Table 6.6). However, at 5% error,
these may mostly have been found significant by chance. Just one relationship stands
up to a sequential Bonferroni correction within amphibian species: the leech family
Glossiphoniidae is positively associated with L. vulgaris, always occurring together,
never singly.
0
1
2
3
4
5
6
5 (n=2) 6 (n=6) 7 (n=2) 8 (n=1)
Mean c
ount of
pre
dato
ry invert
ebra
te
fam
ilies in p
ond (
±s.e
.)
Number of amphibian species present
(a)
0
1
2
3
4
5
Bb Ao Tc Lv
Mean n
um
ber
of
pre
dato
ry invert
ebra
te
fam
ilies (
±s.e
.)
Amphibian species
(b)
6. Macroinvertebrates and amphibians
74
Table 6.6 Moderately common macroinvertebrates exhibiting a significant
relationship with amphibian species‟ occupancy (Nponds = 11) before and after
sequential Bonferroni correction.
Macroinvertebrate
(df = 1) Direction of relationship Order Family Genus / species
Bufo bufo
Coleoptera Dytiscidae Hyphydrus ovatus 4.95 * +
Diptera Stratiomyidae Stratiomys sp. 4.95 * -
Odonata Libellulidae † - 4.41 * -
Hydracarina - - 4.44 * - (excl. pond 17)
Odonata Aeshnidae † - 4.44 * + (excl. pond 17)
A. obstetricans
Coleoptera Dytiscidae Hyphydrus ovatus 6.52 * +
Diptera Stratiomyidae Stratiomys sp. 6.52 * -
T. cristatus
Coleoptera Dytiscidae Hyphydrus ovatus 4.95 * -
Odonata Aeshnidae † - 7.64 ** -
L. vulgaris
Diptera Chironomidae Chironomus sp. 6.52 * +
Odonata Lestidae † - 4.28 * -
Rhynchobdellida Glossiphoniidae - 11.00 **
+
† = predatory invertebrates; 2 significance levels: * p < 0.05, ** p < 0.01; = significant
after sequential Bonferroni correction; pond 17 is included except where noted – excluding
pond 17 made no difference to significance levels in other comparisons; direction of
relationship: „+‟ = positive, „-‟ = negative.
6.4 Discussion
The data collected on invertebrates were limited by the amount of time available and
the time-consuming process of identifying and counting invertebrates. In the present
chapter few relationships that have been observed and tested elsewhere are
substantiated, possibly due to the relatively small dataset.
A weak positive relationship was observed between pond size and the count of
macroinvertebrates, but as with other small scale studies no significant correlation
between macroinvertebrate richness and pond size was found (Sewell, 2006). The
higher count in bigger ponds may have been due to unintentionally covering a
greater area, i.e. netting faster, when space was not limited. Although efforts were
made to standardise the survey technique, the results of a short study cannot be
subjected to an extrapolation to find true richness rather than sample richness (Foggo
et al., 2003).
6. Macroinvertebrates and amphibians
75
Other studies have found aquatic macroinvertebrates to correlate with measures of
wetland plant richness and abundance (e.g. Gee et al., 1997; Hawkes, 1998; Howard,
2002). In the present study there appears to be a positive trend, although no statistical
support was found. In terms of biodiversity, habitat heterogeneity is more important
than size (Triantis et al., 2003), therefore the greater heterogeneity afforded by
higher plant richness and quantity could be expected to support a higher number of
macroinvertebrates (Williams et al., 2008). A larger sample size may have allowed
detection of such an effect, but even in much bigger samples of 60 or 80 ponds Oertli
et al. (2002) found only very weak correlations between the plant, invertebrate and
amphibian taxa they tested. Alternatively, measures of microinvertebrate diversity
and productivity might more accurately reflect amphibians within a habitat, but they
would need more intensive monitoring to smooth over the population fluctuations.
Amphibians as a taxon and as separate species have not been demonstrated to be
appropriate indicators of invertebrate diversity. There was very little difference
between the number of macroinvertebrate families in ponds where any single
amphibian species was present or absent. Again, a larger sample size would have
increased the likelihood of detecting an effect and given a greater variety of
amphibian assemblages to test – amongst 11 ponds seven of eleven amphibian
species were either too rare or too common for comparison.
Seven of the macroinvertebrate families recorded, comprising 16 species, are capable
of predating adult or larval amphibians. The presence of several predatory
invertebrates may have been negatively related to that of certain amphibian species,
although this could not be confirmed with the present data. These relationships
require further data to confirm, as they may have occurred by chance or be due to
other co-varying variables not considered here. Bufo bufo, T. cristatus and L.
vulgaris were each negatively associated with a single dragonfly (Odonata,
Anisoptera) family, Libellulidae, Aeshnidae and Lestidae respectively, which may
directly influence each other or are perhaps an indirect effect of relationships with
other organisms. Anuran larvae, such as Bufo bufo, which are generally unpalatable
to vertebrate predators, may be readily taken by insects such as Odonata naiads and
6. Macroinvertebrates and amphibians
76
Dytiscidae larvae (Gunzburger and Travis, 2005) with mouthparts adapted to pierce
the body cavity to avoid skin toxins. The aggregating behaviour of adults and
concentrated masses of eggs and larvae make amphibians an easy prey item for all
classes of predator (Duellman and Trueb, 1986).
No robust relationship between macroinvertebrates and amphibians was observed in
the present study. It is remarkable that little has been published from western /
central Europe looking specifically at amphibians in relation to other pond taxa; the
large scale studies in Switzerland and the UK focussed mainly on measuring water
quality. Further work to build a larger dataset could focus on fewer invertebrates –
perhaps just those identified as key predators or prey of amphibians and would need
a much larger sample of ponds, most usefully over a wider geographic area.
7. Occupancy modelling
77
7. Absent or not detected? Development of an occupancy
modelling approach to amphibian survey design
Occupancy modelling
Summary: When species detection is imperfect, it is difficult to
distinguish between true absences and non-detections. Surveying for
cryptic animals such as amphibians can be time-consuming and
expensive, so it is important to find methods that work, preferably with a
high detection rate. Traditional analyses generally do not account for
imperfect detectability and survey protocols have tended to be designed
by common sense, but without statistical support (e.g. for the number of
repeat surveys necessary). The maximum-likelihood based occupancy
models allow rates of occupancy and detectability to be taken into
account, as well as the site and survey covariates that might influence
them. The detection rates of standard amphibian survey methods alone
were examined: netting returned very poor detection rates for all species,
aural survey was good for anurans with an extended breeding season
(Pelophylax spp. and Alytes obstetricans), visual survey was best for
frogs (Rana temporaria and Pelophylax spp.) and trapping was best for
all newt species. In models including covariates of detection, water
temperature and day of year were most frequently selected by the final
model. Covariates of occupancy were successfully modelled for two
moderately common species, Triturus cristatus and Lissotriton vulgaris:
pond size, length of paths and tracks in the area and elevation appeared
most prominently in the final models, although no clear pattern emerged
for either species. Occupancy modelling is a rapidly developing approach
to analysing survey data of species with low detection. It has great
potential to transform the traditional methodologies in amphibian
surveying and monitoring.
7.1 Introduction
7.1.1 Comparing survey methodologies
In amphibian monitoring it is common to use a range of survey methods to increase
the chances of detecting a species if it is present. Different survey techniques have
varying success rates between habitat types, species and life stages (Ryan et al.,
2002; de Solla et al., 2005), yet inefficient survey protocols may be accepted as
standard practice due to limited money and time, or simply out of habit. The
resources available often dictate the methods used (Kaiser, 2008) and the results
achieved with limited resources can be greatly improved by careful assessment of the
7. Occupancy modelling
78
value of each method in different habitats and for different species (Gooch et al.,
2006).
In Western Europe a number of survey methods are regularly used to sample pond-
breeding amphibians, which can be tailored to certain species or applied more
broadly to sample all of the potentially occurring species. Survey techniques used in
this region often include dip-netting, head-counts (or torch surveys), funnel / bottle
traps, egg searches, refugia searches and aural surveys (Griffiths and Raper, 1994;
Pellet and Schmidt, 2005). These methods all have their own inherent advantages
and disadvantages (see review in Griffiths and Raper, 1994), but when used in
combination would ideally be able to detect all species, at all life stages, in a variety
of habitat conditions.
Many studies have compared the effectiveness of survey methods between species.
For example, „trappability‟ appears to vary more between newt species of different
body sizes than between more similarly-sized species due to their niche use within
the breeding pond (Griffiths, 1987; Griffiths and Mylotte, 1987; Jehle et al., 2000).
Knowing how efficiently each available method detects the study species is essential
to allocating survey effort appropriately (Pierce and Gutzwiller, 2004).
7.1.2 Detectability
Unless a species is detected with 100% accuracy (a detection probability [p] of 1), its
absence from a site cannot be inferred from non-detection (e.g. MacKenzie, 2005;
MacKenzie et al., 2005; Gooch et al., 2006). Particularly high non-detection errors
are found with rare species, especially those that are cryptic in appearance or
behaviour and that occur in difficult-to-survey habitats (Gu and Swihart, 2004).
Many herpetological studies are based on count data that have not been corrected for
detection probability, thereby violating an assumption of conventional statistical
analyses that the rate of detection is perfect or at least constant (Mazerolle et al.,
2007; Altwegg et al., 2008). The traditional approach to lessen the confounding
effects of imperfect detection is to standardise sampling methods and survey
conditions (Heyer et al., 1994). However, this cannot possibly identify or deal with
7. Occupancy modelling
79
all of the potential influences on detection – false absences can have a very powerful
effect on model performance (Moilanen, 2002).
It is well established that data obtained by sampling methods such as trapping and
point counts do not correlate with those of exhaustive surveys (e.g. Quinn et al.,
2007), resulting in two types of prediction error: false positives (due to
misidentifications) and false negatives (due to non-detections; Fielding and Bell,
1997). Therefore a more advanced approach than simple presence / absence or
count-based data modelling is required.
There are two ways to ensure that surveys adequately represent the study species: (1)
always apply more survey effort than the minimum necessary to accurately describe
the community, and (2) adjust the results statistically using detection probabilities
(de Solla et al., 2005). New statistical techniques to model survey methods,
incorporating covariates of species occurrence and detectability, are rapidly
improving our ability to assess approaches to monitoring and to employ methods in
the most effective way.
7.1.3 Occupancy modelling
Occupancy modelling is a relatively new and rapidly developing maximum
likelihood-based approach to species distribution modelling introduced by
MacKenzie et al. (2002), with the important advantage over traditional models that it
accounts for imperfect detection. It differentiates between detection probability and
occupancy by modelling both as functions of site or survey covariates, thereby
allowing true and false absences to be distinguished (Altwegg et al., 2008).
Software developed specifically for running occupancy models has been developed
by Hines (2006): Program PRESENCE is freely available and is widely used among
researchers developing this approach to species distribution modelling. A number of
assumptions about the quality of the data are made (see Section 7.2.3), however the
ad hoc methods (i.e. methods that do not account for detectability) make even
stronger assumptions. Moreover, the occupancy modelling method is reasonably
robust to minor violations of its assumptions and also accounts for missing
7. Occupancy modelling
80
observations by evaluating model likelihood separately for each site (MacKenzie et
al., 2002).
7.1.4 Covariates
Many factors influence the rate of detection – a species‟ detectability will differ
between studies or geographical areas (Mazerolle et al., 2007). Covariates affecting
the detectability of the study species may vary both between sites and over time
(between surveys). For example, anuran breeding behaviour may be triggered by
factors such as rainfall, air and water temperatures, humidity and lunar cycle (e.g.
Banks and Beebee, 1986; Brooke et al., 2000; Marsh, 2000; Friedl and Klump, 2002;
Oseen and Wassersug, 2002; Weir et al., 2005; Kirlin et al., 2006), which, if a
species is present at a breeding pond, will cause its detectability to vary throughout
the season and across sites.
Occupancy is related only to site characteristics which are constant across the season
but vary between sites, for example nearby forest cover, developed land, pond size,
vegetation density, altitude and distance to other ponds (e.g. Halley et al., 1996; Joly
et al., 2001; Knutson et al., 2004; Maletzky et al., 2007; Rinehart et al., 2009).
In the current chapter the detection and occupancy rates of common amphibian
species are examined using presence / absence data in Program PRESENCE
(MacKenzie et al., 2002; Hines, 2006). Initially, basic models were run for each
species and survey method within seasons, with occupancy and detectability as the
only parameters. Finally, full models incorporating covariates detailing the site and
survey conditions are explored.
The objectives of the current chapter are to test the following questions:
Are the commonly used amphibian survey methods appropriate and effective?
How do detection rates vary between survey methods?
How do detection rates vary between species?
What site and survey covariates most affect occupancy and detection rates?
7. Occupancy modelling
81
7.2 Methods
7.2.1 Study area and amphibian data
The current chapter uses data collected specifically for this study at amphibian
breeding ponds in Luxembourg; the study area and survey methods are described in
the General methodology (Chapter 2). All ponds were included in these analyses
(Appendix 3): ponds within 200 m of each other were grouped due to sharing some
of the habitat covariates and in order to maintain independent replicates. This leaves
24 ponds in 2007 and 22 ponds in 2008 for all species, except Alytes obstetricans for
which the first year of surveys („year 1‟) at each pond were used, comprising 40
ponds in total (including the additional ponds surveyed in 2009).
To ensure that a species is available to be detected, only surveys between the first
and last detection dates by any method at any pond are included in the current study
(Appendix 4). The rarest amphibian species (present at only one pond each: Bufo
calamita, Bombina variegata and Hyla arborea) could not be modelled and were
excluded from analysis.
All appropriate survey methods are analysed for each species and life stages are
merged. The life stages recorded by each survey method are detailed in Table 2.2.
Visual encounter surveys (VES), bottle-trapping and dip-netting were modelled for
the four newt species and Rana temporaria. The same methods with the addition of
aural surveys were considered for Pelophylax spp. and Bufo bufo – although most
Bufo bufo calling occurs during a short „explosive‟ breeding period, they were heard
throughout the whole survey period in both 2007 and 2008. Aural survey was the
only methodology considered for A. obstetricans, due to failure of the other methods
to detect them more than once.
7.2.2 Covariate measurements
Thirty-three site-specific covariates (varying between sites, constant within season)
were measured only once per pond per season (Table 7.1 and Table 7.2, detailed in
Appendix 8). Thirteen survey-specific covariates (varying across time and between
sites) were measured during each survey visit to the study ponds (Table 7.3, with
detail in Appendix 8). Measurements of water conductivity and pH varied only
7. Occupancy modelling
82
slightly over the season, therefore here they are included as site-specific and were
measured once a season in as short a period as possible to minimise any inter-pond
differences due to seasonal changes.
Site-specific covariates describing the landscape within a 100 m buffer around study
ponds (Beebee, 1985) were extracted from the maps described in Chapter 2, section
2.1.1 (with ArcMap v.9.3.1.1850). All covariates were transformed to minimise
skew (Fowler et al., 1998) and to create smaller values which are handled better by
Program PRESENCE.
Table 7.1 Site-specific covariates concerning the occupancy of other species and the
method of normalisation.
Covariate Explanation Norm.
Method Unit
waterfowl Presence / non-detection of any water bird (e.g. duck, moorhen, heron)
- 0/1 (not-detected
/ present)
fish Presence / non-detection of any fish; groups of ponds indicated as fish absent if at least one pond never had fish detected
- 0/1
Rt R. temporaria presence / non-detection - 0/1
P.spp Pelophylax spp. Presence / non-detection - 0/1
Bb Bufo bufo presence / non-detection - 0/1
Ao A. obstetricans presence / non-detection - 0/1
Tc T. cristatus presence / non-detection - 0/1
Lh L. helveticus presence / non-detection - 0/1
Lv L. vulgaris presence / non-detection - 0/1
#_amphib_sps Total number of amphibian species ever detected in pond /mean count
Counts given in Appendix 8.
7. Occupancy modelling
83
Table 7.2 Site-specific habitat covariates and the method of normalisation.
Covariate Explanation Norm.
method Unit
pond_size Surface area of the pond (m2) /mean m
2
pH Yearly mean of three pH measurements taken in each study pond; all ponds tested within 2 days in all years; pH meter: Hanna HI991002
/10 pH
cond Yearly mean of three water conductivity measurements taken in each study pond; all ponds tested within 2 days in all years; Conductivity meter: Orion 125
/mean µs
shade_int Percentage of pond surface area shaded by internal, aquatic vegetation (estimates made during vegetation surveys, for dates see section 5.2.1)
/mean %
shade_ext Percentage of pond surface area shaded by external, non-aquatic vegetation
/mean %
open_water Estimate of the percentage of pond surface area with no vegetation
/mean %
elevation Metres above sea level (m.a.s.l.) /mean m.a.s.l.
#_invert_sps Number of invertebrate species counted during invertebrate surveys (11 ponds only, see Chapter 6)
/mean Count
plant_h' Wetland plant species richness, Shannon-Weiner index (H’), measured during plant surveys (Chapter 5).
/mean H'
paths_tracks Length of paths and tracks within 100 m /1000 m
road_<5.5 Length of small roads (<5.5 m wide) within 100 m /1000 m
road_6.5-7.5 Length of medium-sized roads (6.5 - 7.5 m wide) within 100 m
/1000 m
road_all_widths Length of all roads (any width) within 100 m /1000 m
hedgerow Length of hedgerow within 100 m /1000 m
treerow Length of tree rows within 100 m /1000 m
water_stream_ surface
Length of small surface water courses within 100 m /1000 m
undef_building Area of unclassified buildings (mostly houses) within 100 m
/1000 m2
buildings_undef_ ag_ind
Total area of all buildings (houses, agricultural and industrial) within 100 m
/1000 m2
forest_conif Area of coniferous forest within 100 m /100000 m2
forest_brdlvd Area of broadleaved forest within 100 m /100000 m2
forest_mixed Area of mixed forest within 100 m /100000 m2
orchard Area of orchard forest within 100 m /100000 m2
forest_bwood_ orchard
Total area of all forest, brushwood and orchard within 100 m
/100000 m2
Measurements taken in study pond or within a 100 m radius of pond; summary values given
in Appendix 8.
7. Occupancy modelling
84
Table 7.3 Survey-specific covariates and the method of normalisation.
Covariate Explanation Norm.
method Unit
day_of_yr Day of the survey /100 days (day 1 = 1st Jan.)
full_moon_closest Number of days to/from the closest full moon /10 days
pm_timetaken Total time spent at pond for evening surveys - hours
pm_windspeed Evening wind speed estimate /10 0 - none 1 - breeze 2 - wind 3 - strong
pm_wind_dir Evening wind direction - compass points, clockwise 0.1-0.8
pm_rain Evening rainfall /10 0 - none 1 - light 2 - moderate 3 - heavy
pm_gen Local conditions during evening survey /10 0 - dry 1 - damp 2 - wet 3 - snow
clear_shore Shoreline accessible on foot /100 %
turbidity Water turbidity in torchlight /10 0 - clear 1 - cloudy 2 - very cloudy 3 - no visibility
temp_min Minimum water temperature recorded in pond over night of survey
/10 ˚C
temp_max Maximum water temperature recorded in pond over night of survey
/10 ˚C
#traps Number of traps set /100 count
#nets Number of sweeps made with net /100 count
Summary values given in Appendix 8.
7.2.3 Program PRESENCE analyses
Single season occupancy models make a number of assumptions about the data
quality and independence (MacKenzie et al., 2002): (1) sites are closed to changes in
occupancy for the duration of the survey period (i.e. no colonisation or extinction);
(2) occupancy and detection probabilities are the same for all sites, or are functions
of covariates; (3) sites are independent; and (4) no false detections are made.
Violating these assumptions biases the model estimates or prevents models being
fitted at all. Single season models are used throughout this chapter because multi-
season models require a very large dataset to accommodate the number of
parameters (Donovan and Hines, 2007).
Models were constructed in Program PRESENCE (version 2.3, Hines, 2006) for the
eight amphibian species that occurred at more than one pond. For each species all
7. Occupancy modelling
85
life stages and both sexes were modelled together; only Lissotriton helveticus and L.
vulgaris larvae were excluded from all surveys and females from VES due to their
similarity preventing certain identification. Survey years 2007 and 2008 were
considered separately, except for A. obstetricans, where „year 1‟ data from each
pond were analysed together and only one year‟s data was considered for each pond,
to uphold the first assumption.
Survey rounds where too few ponds were visited were excluded to minimise missing
data points and achieve well-fitted models. For each model the bootstrap goodness of
fit procedure was applied (500 iterations) to approximate the standard error (s.e.) of
occupancy (ψ) – this non-parametric method produces a good estimate unless
detectability (p) is very low (MacKenzie et al., 2002).
7.2.4 Comparing survey methods
Initially, basic models were run to compare the detection probabilities of survey
methods for each of the eight modelled species. In these preliminary models
maximum overnight temperature was the only covariate added (ψ(.)
p(survey_method, max_temp)), because survey method alone could not account for
sufficient variation to make well-fitting models. Data from all appropriate survey
methods (Table 2.2) were entered onto a single spreadsheet with dummy covariates
to code for the method represented by each column.
Combining the methods this way could be considered a violation of the assumption
of independence – for example, if a newt is trapped and released, its chances of
being netted during the same visit may be affected. However, it overcomes the
alternative problem of repeated analyses on the same data to compare methods.
Years must still be modelled separately, to meet the assumption of continuous
occupancy. Because a continuous survey-specific covariate was included, p and
s.e.(p) had to be averaged to find a single score for each method. The probability of
detection was compared between methods within years using one-way ANOVAs.
7. Occupancy modelling
86
7.2.5 Predictors of occupancy and detection
Fully parameterised models were run for the common amphibian species: the four
very common species (R. temporaria, Pelophylax spp., Ichthyosaura alpestris and L.
helveticus) were run with detection covariates only and the moderately common
species (Bufo bufo, A. obstetricans, Triturus cristatus and L. vulgaris) were run with
both occupancy and detection covariates. Again, „year 1‟ data from each pond were
used in A. obstetricans models and survey years 2007 and 2008 were modelled
separately for the other species to maintain independence.
Site-specific covariates were tested as predictors of occupancy and survey-specific
covariates as predictors of detection (MacKenzie et al., 2002). Covariates were
chosen a priori, based on published research and assessment of the features and
conditions expected to influence amphibian presence-absence. Covariates (section
7.2.2 and Appendix 8) judged to be appropriate for each species were run
individually, and then covariates in top-ranked models were combined in pairs to
explore whether any better-fitting models could be found. Poorly ranked models (see
next section, 7.2.6) were discarded. The number of covariates was limited to two per
model (in addition to occupancy, ψ, and probability of detection, p) to minimise the
effects of over-parameterisation. A generally accepted rule of thumb is to have no
fewer than ten sites per parameter (Burnham and Anderson, 2004). For example, a
robust model with three parameters should have at least 30 independent samples /
sites.
7.2.6 Model selection
Program PRESENCE ranks models by the Akaike Information Criterion (AIC),
which is a measure of the goodness of fit for estimated models. It is automatically
calculated by Program PRESENCE, but can be expressed as
- Equation 1
where „K‟ is the number of estimated parameters (Mazerolle, 2006). To minimise the
bias caused by small sample sizes the AIC was adjusted to AICc, calculated by
7. Occupancy modelling
87
Equation 2
where „n‟ is the sample size (Sugiura, 1978, in Mazerolle, 2006). AICc converges
with AIC as the number of samples increases and it gives similar results but is more
conservative for small samples (Burnham and Anderson, 1998).
Over- or under-dispersion of the data is indicated by the ĉ score (variance inflation
factor) given by Program PRESENCE, estimated from the chi-square goodness-of-fit
statistic and the degrees of freedom:
Equation 3
ĉ = 1 is a good fit, > 1 shows over-dispersion and < 1 under-dispersion. Only
models with ĉ close to one were accepted. When the most parsimonious model was
slightly over-dispersed (common in count data, often caused by a lack of true
independence, Anderson and Burnham, 2002) estimates were adjusted by the ĉ score
of the most parsimonious model to give a quasi-AICc (QAICc) score. QAIC is
calculated by Program PRESENCE, but can be represented by the following
equations
-
Equation 4
Equation 5
(Burnham and Anderson, 1998).
Delta AICc or delta QAICc (ΔAICc and ΔQAICc respectively) values are more useful
than simple AICc or QAICc because they describe the difference between candidate
models (Connor et al., 2004). They are calculated as follows:
Equation 6
7. Occupancy modelling
88
Equation 7
where „i‟ is model number (in rank order by AIC score) and „min QAIC‟ or „min
AIC‟ the QAIC or AIC value of the top-ranked model (Mazerolle, 2006). The most
parsimonious model is always denoted as ΔAICc or ΔQAICc = 0.00. Any ΔQAICc <
2 indicates good support for the model (Zanini et al., 2008).
An additional measure of the strength of support for a model is the AIC (or QAIC)
weights, which indicate the model‟s strength relative to other candidate models. For
example a model with an AIC weight of 0.4 has twice the support of a model with
AIC weight 0.2 (Mazerolle et al., 2007).
Final models were selected if they met the following criteria: ĉ close to 1, AICc or
QAICc < 2, and a test statistic probability > 0.05. If more than one model was
strongly supported the model estimates of occupancy (ψ) and detection probability
(p) were averaged, following Mazerolle et al. (2007) and Sewell et al. (2010).
Detection rates of the most parsimonious models were plotted against their
associated continuous covariates using the following formula:
Equation 8
where „p‟ is the detection rate and β0 and β1 are Beta estimates given in the Program
PRESENCE output, representing the intercept and slope of a regression equation
(Sewell et al., 2010).
7.2.7 Number of survey visits required
Following Pellet and Schmidt (2005) and Sewell et al. (2010) the minimum number
of survey visits necessary (Nmin) to be 95% certain that a non-detection was a true
absence was calculated with Equation 9, where P is the detection rate derived from
Equation 8.
7. Occupancy modelling
89
Equation 9
7.3 Results
Between five and eleven surveys were conducted at each study pond in 2007, 2008
and „year 1‟; Table 10.2 in Appendix 3 details the ponds included in each year‟s
models. Where more than one life stage was detected all data were modelled
together.
7.3.1 Detectability by different survey methods
ψ(.) p(survey_method, max_temp) models could not be fitted for all species in both
years: R. temporaria from 2008 and Bufo bufo data from both years could not
generate models with a good fit to the data, even when the earliest and latest surveys
were excluded, and including or excluding aural surveys. Unusual detection histories
at some ponds may have prevented models being fitted, for example where there is
perfect detection (p = 1.0) at one pond and low detection rates at other ponds,
particularly with the small sample size in the current study (22 ponds in 2008). AIC
values are not reported individually for these models, because only a single model
was run for each species, therefore ΔAICc was nominally 0.0 and the AIC weight
1.0.
For the two anuran species successfully modelled with more than one survey
method, R. temporaria (2007) and Pelophylax spp. (2007 and 2008), VES had the
highest detection rate, while trapping and netting tended to be less successful (Figure
7.1a-b).
Aural survey was almost as successful as VES for Pelophylax spp., with detection
rates (p) of 0.47 ±0.04 (s.e.) in 2007 and 0.39 ±0.05 in 2008, compared to VES in
2007 0.50 ±0.04 and 2008 0.45 ±0.05. For Pelophylax spp. all survey methods had
slightly higher detection rates in 2007 than 2008. Trapping attained a marginally
higher detection rate for R. temporaria and Pelophylax spp. than netting.
7. Occupancy modelling
90
Only one survey method was tested for A. obstetricans: aural survey for this species
achieved a detection rate of 0.45 ±0.09 (Figure 7.1c).
D
ete
ction r
ate
(p)
(±s.e
.)
Dete
ction r
ate
(p)
(±s.e
.)
Dete
ction r
ate
(p)
(±s.e
.)
Figure 7.1 „Detectability‟ (±s.e.) of anurans by survey method, ψ(.)
p(survey_method, max_temp).
= 2007, = 2008; VES = visual encounter survey.
For all newt species, trapping returned the highest rate of detection overall, making it
the most successful survey method. Netting was the poorest method, while VES
achieved intermediate detection rates (Figure 7.2a-d).
Contrary to the Pelophylax spp. results, detection rates for newts were generally
lower in 2007 than in 2008. Netting is a relatively poor method for newts, achieving
only very low rates of detection; the highest net detection rate for any newt was for
0.0
0.2
0.4
0.6
0.8
1.0
VES Trap Net
(a) R. temporaria
0.0
0.2
0.4
0.6
0.8
1.0
VES Aural Trap Net
(b) Pelophylax spp.
0.0
0.2
0.4
0.6
0.8
1.0
Year 1 - Aural
(c) A. obstetricans
7. Occupancy modelling
91
T. cristatus in 2007 (0.24 ±0.06 s.e.). However, in 2008 no T. cristatus were caught
by net.
D
ete
ction r
ate
(p)
(±s.e
.)
Dete
ction r
ate
(p)
(±s.e
.)
Figure 7.2 'Detectability' (±s.e.) of newts by survey method, ψ(.) p(survey_method,
max_temp).
= 2007, = 2008; VES = visual encounter survey.
7.3.2 Predictors of occupancy and detection
The most frequently occurring covariate of detection among the modelled species
was temperature, which occurred in 21/35 of the successfully run year-method
models (7 „temp_min‟ and 14 „temp_max‟). The „day_of_yr‟ was also an important
covariate, in 16/35 accepted models, indicating that detectability does change over
the season. Only covariates of detection were examined for the four most common
species, which were present in too many study ponds for meaningful analysis of
occupancy covariates.
Bufo bufo data could still not be modelled with the addition of site- and survey-
specific covariates. Occupancy modelling may be unsuitable for some explosive
breeders, particularly in this case where surveys were spread across months rather
than days, because these species are highly detectable for a short period only (de
0.0
0.2
0.4
0.6
0.8
1.0
VES Trap Net
(a) I. alpestris
0.0
0.2
0.4
0.6
0.8
1.0
VES Trap Net
(c) L. helveticus
0.0
0.2
0.4
0.6
0.8
1.0
VES Trap Net
(b) T. cristatus
0.0
0.2
0.4
0.6
0.8
1.0
VES Trap Net
(d) L. vulgaris
7. Occupancy modelling
92
Solla et al., 2005) – larvae tend to be more numerous and more detectable as the
season progresses (Figure 7.3), however they are less conspicuous than breeding
adults. In addition to the skewed detectability over spring and summer, Bufo bufo do
not always use all of the suitable waterbodies available; they congregate to breed and
therefore occupancy is difficult to predict, except perhaps by whether or not a site is
already occupied by that species.
Figure 7.3 The mean number of adult and larval B. bufo seen or captured during
surveys, by day of year.
Based on 2007 and 2008 data from all survey methods; = adults, = larvae.
R. temporaria
Occupancy covariates were not examined for R. temporaria because it was very
common, detected by one or more methods in 87.5% and 90.9% of study ponds in
2007 and 2008 respectively. Although VES and trapping had similar detection rates,
VES found R. temporaria at a greater number of ponds, with naïve and estimated
occupancy rates greater than the other methods (Table 7.4). The final netting models
were under-dispersed and in both years netting detected R. temporaria in fewer
ponds than VES or trapping.
Day of year and temperature emerge as important covariates of detection for R.
temporaria, although across the full range of temperatures encountered there is little
change in detection rate for any method (Figure 7.5a-b), varying by < 0.1. Day of
year had a greater influence on detectability by VES and netting (Figure 7.7), with
the rate of detection by VES decreasing as the year progressed, but increasing for
netting. This may be due to the change in the balance of numbers from largely adults
to largely larvae available to detect (Figure 7.4). Detection rate was positively
0
5
10
15
20
0 50 100 150 200
Mean n
um
ber
seen /
captu
red
Day of year
7. Occupancy modelling
93
affected by the percentage of clear shore for trapping and netting in 2007 (Figure
7.6) and the amount of time taken during trapping surveys in 2008 (Figure 7.8).
Table 7.4 Most parsimonious models for R. temporaria surveys in 2007 and 2008 by
three survey methods.
Models K -2*LL QAICc QAIC
weight
Est. ψ
Est. p (±s.e.)
Visits (±s.e.)
2007 VES (naïve ψ = 0.83)
ψ(.) p(day_of_yr, temp_max) 4 143.65 0.00 0.64 † 0.85 0.46 (±0.063) 5 (±0.9)
ψ(.) p(day_of_yr) 3 146.78 0.22 0.36 † 0.85 0.46 (±0.052) 5 (±0.8)
ĉ = 0.51 Model averaged: 0.85 0.46 (±0.059) 5 (±0.9)
2008 VES (naïve ψ = 0.86)
ψ(.) p(day_of_yr) 3 95.51 0.00 0.68 0.87 0.45 (±0.053) 5 (±0.8)
ψ(.) p(day_of_yr, temp_max) 4 95.03 2.64 0.32 0.87 0.47 (±0.066) 5 (±0.9)
ĉ = 1.25 Model averaged: 0.87 0.46 (±0.057) 5 (±0.9)
2007 Trap (naïve ψ = 0.58)
ψ(.) p(clear_shore, temp_max) 4 126.93 0.00 1.00 0.63 0.46 (±0.080) 5 (±1.2)
ĉ = 1.73
2008 Trap (naïve ψ = 0.59)
ψ(.) p(pm_timetaken) 3 103.76 0.00 0.73 0.67 0.44 (±0.094) 6 (±1.5)
ψ(.) p(clear_shore) 3 105.76 1.80 0.27 0.62 0.49 (±0.081) 5 (±1.1)
ĉ = 1.11 Model averaged: 0.66 0.45 (±0.090) 5 (±1.4)
2007 Net (naïve ψ = 0.54)
ψ(.) p(clear_shore, temp_max) 4 112.13 0.00 1.00 † 0.74 0.28 (±0.066) 10 (±2.6)
ĉ = 0.87
2008 Net (naïve ψ = 0.50)
ψ(.) p(day_of_yr) 3 93.94 0.00 1.00 † 0.52 0.45 (±0.092) 5 (±1.4)
ĉ = 0.95
Covariates are described in Appendix 8; K = number of model parameters; -2*LL = -2* Log
likelihood (residual deviance); ĉ = TestStat/AvgTestStat of the most parsimonious model; †
= AICc and AIC weight used rather than QAICc and QAIC weight (not adjusted for over-
dispersion); Visits = number of visits required for 95% confidence that non-detection
indicates true absence; all life stages modelled together.
Figure 7.4 The mean number of adult and larval R. temporaria seen or captured
during surveys, by day of year.
Based on 2007 and 2008 data from all survey methods; = adults, = larvae.
0
2
4
6
8
10
12
14
16
0 50 100 150 200
Mean n
um
ber
seen /
captu
red
Day of year
7. Occupancy modelling
94
Dete
ction r
ate
(p)
Figure 7.5 Detection rates (±s.e.) of R. temporaria by the most parsimonious
models, where water temperature was a covariate of detection.
Based on (a) 2007 VES data: ψ(.) p(day_of_yr, temp_max), β0 = 5.13, β1 = -0.81; (b) 2007 trap data: ψ(.) p(clearshore, temp_max), β0 = -6.03, β1 = 0.85; (c) 2007 net data: ψ(.)
p(clearshore, temp_max), β0 = -8.91, β1 = 1.47; where β0 and β1 are equivalent, respectively,
to the intercept and slope of a regression equation.
Dete
ction r
ate
(p)
Figure 7.6 Detection rates (±s.e.) of R. temporaria by the most parsimonious
models, where „clear shore‟ was a predictor of detection rate.
Based on (a) 2007 trap data: ψ(.) p(clearshore, temp_max), β0 = -6.03, β1 = 5.77; (b) 2007
net data: ψ(.) p(clearshore, temp_max), β0 = -8.91, β1 = 6.81; where β0 and β1 are equivalent,
respectively, to the intercept and slope of a regression equation.
0.94
0.96
0.98
1.00
(a) Rt VES 2007
-0.02
0.00
0.02
0.04
0.06(b) Rt Trap 2007
-0.02
0.00
0.02
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Maximum overnight water temperature (C)
(c) Rt Net 2007
-0.25
0.00
0.25
0.50
0.75(a) Rt 2007 Trap
-0.25
0.00
0.25
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Clearshore (%)
(b) Rt 2007 Net
7. Occupancy modelling
95
Dete
ction r
ate
(p)
Figure 7.7 Detection rates (±s.e.) of R. temporaria by top models, where „day of
year‟ was a covariate of detection.
Based on (a) 2007 VES data: ψ(.) p(day_of_yr, temp_max), β0 = 5.13, β1 = -3.48; (b) 2008 VES data: ψ(.) p(day_of_yr), β0 = 7.32, β1 = -7.04; (c) 2008 net data: ψ(.) p(day_of_yr), β0 =
-3.52, β1 = 3.06; where β0 and β1 are equivalent, respectively, to the intercept and slope of a
regression equation.
Figure 7.8 Detection rates (±s.e.) of R. temporaria by trapping in 2008, where time
taken was a covariate of detection.
Based on 2008 trap data: ψ(.) p(pm_timetaken), β0 = -0.96, β1 = 1.33; where β0 and β1 are
equivalent, respectively, to the intercept and slope of a regression equation.
-0.2
0.0
0.2
0.4
0.6
0.8
1.0(a) Rt 2007 VES
-0.2
0.0
0.2
0.4
0.6
0.8
1.0(b) Rt 2008 VES
0.0
0.2
0.4
0.6
0.8
1.0
50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
Day of year
(c) Rt 2008 Net
0.0
0.2
0.4
0.6
0.8
1.0
0 10 20 30 40 50 60 70 80 90 100
Dete
ction r
ate
(p)
Time taken in evening(minutes)
Rt 2008 Trap
7. Occupancy modelling
96
Pelophylax spp.
Table 7.5 Most parsimonious models for Pelophylax spp. surveys in 2007 and 2008
by four survey methods.
Models K -2*LL QAICc QAIC
weight
Est. ψ
Est. p (±s.e.)
Visits (±s.e.)
2007 VES (naïve ψ = 0.79)
ψ(.) p(temp_max) 3 153.59 0.00 0.72 † 0.82 0.52 (±0.055) 4 (±0.6)
ψ(.) p(day_of_yr, temp_max) 4 153.52 2.84 0.28 † 0.81 0.53 (±0.067) 4 (±0.8)
ĉ = 0.92 Model averaged: 0.82 0.53 (±0.058) 4 (±0.7)
2008 VES (naïve ψ = 0.73)
ψ(.) p(temp_max) 3 90.67 0.00 0.71 † 0.79 0.39 (±0.061) 6 (±1.2)
ψ(.) p(day_of_yr, temp_max) 4 90.44 2.79 0.29 † 0.79 0.41 (±0.076) 6 (±1.4)
ĉ = 0.65 Model averaged: 0.79 0.40 (±0.066) 6 (±1.3)
2007 Trap (naïve ψ = 0.58)
ψ(.) p(temp_max) 3 125.16 0.00 1.00 0.73 0.25 (±0.062) 11 (±3.1)
ĉ = 1.16
2007 Net (naïve ψ = 0.50)
ψ(.) p(day_of_yr) 3 82.68 0.00 0.59 0.54 0.32 (±0.054) 8 (±1.6)
ψ(.) p(day_of_yr, temp_min) 4 81.39 1.96 0.41 0.54 0.31 (±0.063) 9 (±2.1)
ĉ = 1.37 Model averaged: 0.54 0.31 (±0.058) 8 (±1.8)
2007 Aural (naïve ψ = 0.67)
ψ(.) p(day_of_yr, temp_min) 4 109.78 0.00 0.72 † 0.69 0.51 (±0.061) 5 (±0.7)
ψ(.) p(temp_min) 3 113.63 0.94 0.28 † 0.69 0.51 (±0.050) 5 (±0.6)
ĉ = 0.87 Model averaged: 0.69 0.51 (±0.057) 5 (±0.7)
2008 Aural (naïve ψ = 0.59)
ψ(.) p(temp_max) 3 76.28 0.00 0.39 0.65 0.39 (±0.066) 6 (±1.3)
ψ(.) p(day_of_yr, temp_max) 4 74.53 1.95 0.34 0.62 0.42 (±0.082) 6 (±1.4)
ψ(.) p(day_of_yr) 3 77.06 0.48 0.26 0.60 0.46 (±0.060) 5 (±0.9)
ĉ = 1.64 Model averaged: 0.63 0.42 (±0.070) 6 (±1.2)
Covariates are described in Appendix 8; K = number of model parameters; -2*LL = -2* Log
likelihood (residual deviance); ĉ = TestStat/AvgTestStat of the most parsimonious model; †
= AICc and AIC weight used rather than QAICc and QAIC weight (not adjusted for over-
dispersion); Visits = number of visits required for 95% confidence that non-detection
indicates true absence; all life stages modelled together.
Figure 7.9 The mean number of adult and larval Pelophylax spp. seen or captured
during surveys by day of year.
Based on 2007 and 2008 data from all survey methods; = adults, = larvae.
0
20
40
60
80
100
120
0 20 40 60 80 100 120 140 160 180 200
Mean n
um
ber
seen
/ captu
red
Day of year
7. Occupancy modelling
97
Like R. temporaria, Pelophylax spp. were too common to assess their covariates of
occupancy. No model could be fitted to the 2008 trap data for Pelophylax spp. and
few detections were made by netting in 2008, resulting in a poorly fitted model that
is not reported here (Table 7.5). All of the successful models run for Pelophylax spp.
contained temperature and / or the day of year covariate(s), no other covariates
featured in the most parsimonious models. Detectability increased with temperature
for all survey methods (Figure 7.11). By net and aural survey in 2007 detectability
also increased as the season progressed (with „day of year‟; Figure 7.10) and the
mean number of individuals counted increased throughout the season (Figure 7.9).
Dete
ction r
ate
(p)
Figure 7.10 Detection rates (±s.e.) of Pelophylax spp. by the most parsimonious
models, where „day of year‟ was a covariate of detection rate.
Based on (a) 2007 net data: ψ(.) p(day_of_yr), β0 = -10.97, β1 = 7.67; (b) 2007 aural data: ψ(.) p(day_of_yr, temp_min), β0 = -6.78, β1 = 3.68, under-dispersed; β0 and β1 are equivalent,
respectively, to the intercept and slope of a regression equation.
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
(a) P.spp 2007 Net
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
Day of year
(b) P.spp 2007 Aural
7. Occupancy modelling
98
Dete
ction r
ate
(p)
Figure 7.11 Detection rates (±s.e.) of Pelophylax spp. by the most parsimonious
models, where water temperature was a covariate of detection.
= minimum temperature, = maximum temperature; based on (a) 2007 VES data: ψ(.) p(temp_max), β0 = -3.27, β1 = 2.45, under-dispersed; (b) 2008 VES data: ψ(.)
p(temp_max), β0 = -5.24, β1 = 3.99, under-dispersed; (c) 2007 trap data: ψ(.) p(temp_max), β0 = -3.51, β1 = 1.59; (d) 2007 aural data: ψ(.) p(temp_min, day_of_yr), β0 = -6.78, β1 = 2.19,
under-dispersed; (e) 2008 aural data: ψ(.) p(temp_max), β0 = -5.83, β1 = 4.46; β0 and β1 are
equivalent, respectively, to the intercept and slope of a regression equation.
A. obstetricans
The most parsimonious occupancy model for A. obstetricans was „ψ(.) p(temp_max)‟
(Table 7.6) – none of the additional covariates strengthened the model reported in
section 7.3.1 and models with more covariates could not be fitted. However the
0.0
0.2
0.4
0.6
0.8
1.0
(a) P.spp 2007 VES
0.0
0.2
0.4
0.6
0.8
1.0
(b) P.spp 2008 VES
0.0
0.2
0.4
0.6
0.8
(c) P.spp 2007 Trap
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0 (d) P.spp 2007 Aural
0.0
0.2
0.4
0.6
0.8
1.0
0 5 10 15 20 25
Overnight water temperature (c)
(e) P.spp 2008 Aural
7. Occupancy modelling
99
single final model is under-dispersed (ĉ = 0.08) and therefore is not reliable. The
naïve A. obstetricans occupancy rate (ψ) of „year 1‟ study ponds was 0.33, but the
model estimates occupancy at just 0.16 (±0.074 s.e.), indicating either that false
detections were made or again that the model is unreliable. Further interpretation is
not made due to the poor fit of the model.
Table 7.6 Most parsimonious models for A. obstetricans 'year 1' data using aural
surveys.
Models K -2*LL AICc AIC
weight Est. ψ
Est. p (±s.e.)
Visits (±s.e.)
‘Year 1’ Aural (naïve ψ = 0.33)
ψ(.) p(temp_max) 3 38.33 0.00 1.00 0.16 0.45 (±0.093) 5 (±1.4)
ĉ = 0.08
Covariates are described in Appendix 8; K = number of model parameters; -2*LL = -2* Log likelihood (residual deviance); ĉ = TestStat/AvgTestStat of the most parsimonious model;
Visits = number of visits required for 95% confidence that non-detection indicates true
absence.
I. alpestris
I. alpestris was too common in the study ponds to examine its covariates of
occupancy, and no model could be fitted for the 2008 trap data. The most frequently
occurring detection rate covariates were day of year, time taken and temperature
(Table 7.7). Although „temp_min‟ and „temp_max‟ occurred in three of the top
models, they had only a very slight effect on detection rate: a positive relationship
existed between VES detection rate in 2007 and minimum temperature (β0 = 2.34, β1
= 0.29) and between net detection rate in 2008 and maximum temperature (β0 = -
0.48, β1 = 2.14), while minimum temperature negatively affected detection by net in
2007 (β0 = -5.97, β1 = -1.03).
Detection rate of I. alpestris by VES and trap decreased during the season, while
detection rate by netting increased (Figure 7.14), due to more larvae being available
to net (Figure 7.12). Overall netting was a poor method, requiring many additional
visits to achieve confidence in non-detections (Table 7.7). The general evening
conditions („pm_gen‟: dry / damp / wet / snow) exhibited a negative relationship
with I. alpestris detection by trapping in 2007, however only as the less important of
two covariates in both models. There was a positive relationship between detection
7. Occupancy modelling
100
rate by VES and the time taken to complete evening surveys for I. alpestris in 2008
(Figure 7.13).
Figure 7.12 The mean number of adult and larval I. alpestris seen or captured during
surveys by day of year.
Based on 2007 and 2008 data from all survey methods; = adults, = larvae.
Table 7.7 Most parsimonious models for I. alpestris surveys in 2007 and 2008 by
three survey methods.
Models K -2*LL QAICc QAIC
weight
Est. ψ
Est. p (±s.e.)
Visits (±s.e.)
2007 VES (naïve ψ = 0.96)
ψ(.) p(day_of_yr, temp_min) 4 187.27 0.00 0.60 0.99 0.46 (±0.068) 5 (±1.0)
ψ(.) p(day_of_yr) 3 190.04 -0.33 0.40 0.99 0.45 (±0.057) 5 (±0.9)
ĉ = 1.08 Model averaged: 0.99 0.46 (±0.064) 5 (±0.9)
2008 VES (naïve ψ = 0.91)
ψ(.) p(pm_timetaken) 3 116.66 0.00 0.40 † 0.91 0.71 (±0.063) 3 (±0.4)
ψ(.) p(turbidity, pm_timetaken) 4 114.67 1.03 0.39 † 0.91 0.70 (±0.078) 3 (±0.5)
ψ(.) p(pm_rain, pm_timetaken) 4 115.92 2.28 0.21 † 0.91 0.70 (±0.078) 3 (±0.5)
ĉ = 0.82 Model averaged: 0.91 0.70 (±0.072) 3 (±0.5)
2007 Trap (naïve ψ = 0.92)
ψ(.) p(day_of_yr, pm_gen) 4 167.93 0.00 0.59 0.92 0.69 (±0.062) 3 (±0.4)
ψ(.) p(pm_gen, temp_max) 4 168.66 0.49 0.41 0.92 0.69 (±0.063) 3 (±0.5)
ĉ = 1.48 Model averaged: 0.92 0.69 (±0.063) 3 (±0.4)
2007 Net (naïve ψ = 0.75)
ψ(.) p(day_of_yr, temp_min) 4 125.81 0.00 1.00 † 1.00 0.20 (±0.051) 14 (±3.8)
ĉ = 0.79
2008 Net (naïve ψ = 0.36)
ψ(.) p(temp_max) 3 62.1 0.00 1.00 † 0.65 0.16 (±0.079) 18 (±9.8)
ĉ = 0.71
Covariates are described in Appendix 8; K = number of model parameters; -2*LL = -2* Log
likelihood (residual deviance); ĉ = TestStat/AvgTestStat of the most parsimonious model; †
= AICc and AIC weight used rather than QAICc and QAIC weight (not adjusted for over-dispersion); Visits = number of visits required for 95% confidence that non-detection
indicates true absence; all life stages modelled together.
0
10
20
30
40
50
60
70
0 50 100 150 200
Mean n
um
ber
seen /
captu
red
Day of year
7. Occupancy modelling
101
Dete
ction r
ate
(p)
Figure 7.13 The detection rate (±s.e.) of VES for I. alpestris in 2008, where
„pm_timetaken‟ was a covariate of detection.
Based on 2008 I. alpestris VES data, most parsimonious model: ψ(.) p(pm_timetaken); β0 = -0.14, β1 = 2.13, under-dispersed; β0 and β1 are equivalent, respectively, to the intercept and
slope of a regression equation.
Dete
ction r
ate
(p)
Figure 7.14 Detection rates (±s.e.) of I. alpestris by top models where the day of
year was a covariate of the detection rate.
Based on (a) 2007 VES data: ψ(.) p(day_of_yr, temp_min), β0 = 2.34, β1 = -2.36; (b) 2007
trap data: ψ(.) p(day_of_yr, pm_gen), β0 = 3.14, β1 = -1.54; (c) 2007 netting data: ψ(.) p(day_of_yr, temp_min), β0 = -5.97, β1 = 4.52, under-dispersed; β0 and β1 are equivalent,
respectively, to the intercept and slope of a regression equation.
T. cristatus
A variety of covariates occurred in the top models for T. cristatus (Table 7.8). Pond
size was a predictor of occupancy in the most parsimonious models of VES data in
both years: on average, ponds occupied by T. cristatus were larger (2187.27 m2
0.0
0.2
0.4
0.6
0.8
1.0
0 10 20 30 40 50 60 70 80 90 100
Time taken in evening (minutes)
Ia 2008 VES
-0.2
0.0
0.2
0.4
0.6
0.8
1.0(a) Ia 2007 VES
0.0
0.2
0.4
0.6
0.8
1.0(b) Ia 2007 Trap
0.0
0.2
0.4
0.6
0.8
1.0
50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200Day of year
(c) Ia 2007 Net
7. Occupancy modelling
102
±299.50 s.e.) than non-occupied ponds (1029.60 m2 ±790.61), but the difference is
not significant (t24 = 1.53, p > 0.05). The amount of clear shore (i.e. percentage of
shoreline accessible for survey) was a predictor of T. cristatus detection rate by VES
in three of four well-fitted models in 2007. However, contrary to expectation,
according to Equation 8 it is a negative relationship: detection rate decreases with
increasing shoreline accessibility (Figure 7.15). This was not anticipated because it is
more logical that a species is more likely to be detected where a habitat can be more
completely searched.
Table 7.8 Most parsimonious models for T. cristatus surveys in 2007 and 2008 by
three survey methods.
Models K -2*LL QAICc QAIC
weight
Est. ψ
Est. p (±s.e.)
Visits (±s.e.)
2007 VES (naïve ψ = 0.29)
ψ(pond_size) p(clear_shore) 4 79.55 0.00 0.36 0.29 0.64 (±0.091) 3 (±0.7)
ψ(tracks) p(clear_shore) 4 80.06 0.41 0.30 0.30 0.63 (±0.092) 3 (±0.8)
ψ(pond_size) p(turbidity) 4 81.22 1.36 0.18 0.30 0.54 (±0.105) 4 (±1.1)
ψ(pH) p(clear_shore) 4 81.65 1.71 0.16 0.30 0.63 (±0.092) 3 (±0.8)
ĉ = 1.23 Model averaged: 0.30 0.62 (±0.094) 4 (±0.8)
2008 VES (naïve ψ = 0.27)
ψ(pond_size) p(pm_timetaken) 4 43.92 0.00 1.00 † 0.28 0.80 (±0.102) 2 (±0.6)
ĉ = -0.70
2007 Trap (naïve ψ = 0.29)
ψ(tracks) p(temp_max) 4 59.95 0.00 0.58 0.29 0.75 (±0.081) 3 (±0.5)
ψ(Bb) p(temp_max) 4 61.48 0.88 0.27 0.29 0.75 (±0.082) 3 (±0.5)
ψ(tracks) p(day_of_yr) 4 62.65 1.55 0.15 0.29 0.77 (±0.080) 2 (±0.5)
ĉ = 1.74 Model averaged: 0.25 0.59 (±0.063) 3 (±0.5)
2008 Trap (naïve ψ = 0.41)
ψ(.) p(turbidity, pm_timetaken) 4 67.68 0.00 0.69 0.41 0.62 (±0.150) 4 (±1.3)
ψ(.) p(turbidity) 3 74.44 0.60 0.31 0.42 0.63 (±0.081) 3 (±0.7)
ĉ = 1.87 Model averaged: 0.41 0.62 (±0.129) 4 (±1.1)
2007 Net (naïve ψ = 0.25)
ψ(#invert_sps) p(temp_min) 4 28.54 0.00 1.00 † 0.64 0.34 (±0.079) 8 (±2.1)
ĉ = 0.08
Covariates are described in Appendix 8; K = number of model parameters; -2*LL = -2* Log
likelihood (residual deviance); ĉ = TestStat/AvgTestStat of the most parsimonious model; †
= AICc and AIC weight used rather than QAICc and QAIC weight (not adjusted for over-dispersion); Visits = number of visits required for 95% confidence that non-detection
indicates true absence; all life stages modelled together.
7. Occupancy modelling
103
Figure 7.15 The detection rate (±s.e.) of VES for T. cristatus in 2007, where the
percentage of clear shore was a covariate of detection.
Based on 2007 T. cristatus VES data, most parsimonious model: ψ(pond_size) p(clear_shore); β0 = 3.56, β1 = -3.83; β0 and β1 are equivalent, respectively, to the intercept
and slope of a regression equation.
Figure 7.16 The detection rate (±s.e.) of T. cristatus in the top 2007 trapping model,
where „max_temp‟ was a covariate of detection.
Based on 2007 T. cristatus trap data, most parsimonious model: ψ(tracks) p(temp_max); β0 = -2.29, β1 = 2.70; β0 and β1 are equivalent, respectively, to the intercept and slope of a
regression equation.
Turbidity occurs in one of the top T. cristatus models for VES in 2007 and in the top
2008 trapping models (Table 7.8) – it is logical that turbidity affects detectability by
VES, but less so for trapping. In 2007 detectability by trapping for T. cristatus was
predicted by maximum overnight water temperature, there is a positive relationship
between them (Figure 7.16).
The length of tracks and paths in the 100 m buffer surrounding study ponds is a
predictor of T. cristatus occupancy for both VES in 2007 and trapping in 2008.
There is a significant difference between the ponds where T. cristatus were and were
not recorded (t22 = 2.55, p < 0.05), with more tracks near occupied ponds (not
recorded mean = 96.82 m ±29.73 s.e.; occupied mean = 494.14 m ±239.43).
0.0
0.2
0.4
0.6
0.8
1.0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Dete
ction r
ate
(p)
Clear shore (%)
Tc VES 2007
0.0
0.2
0.4
0.6
0.8
1.0
0 5 10 15 20 25
Dete
ction r
ate
(p)
Maximum overnight water temperature (C)
Tc Trap 2007
7. Occupancy modelling
104
However, this effect is probably an artefact of the data collection or analysis
methods.
No T. cristatus were netted in 2008, giving a detection rate of 0.0, and the best 2007
model is under-dispersed. As shown in the previous section (7.3.1), netting was not a
good method compared to VES and trapping. Achieving 95% confidence that non-
detections of T. cristatus were genuine absences would take 2-4 repeat surveys with
VES or trapping, whereas approximately 8 surveys would be required to gain
confidence in the 2007 netting model with a much poorer detectability rate (Table
7.8).
Figure 7.17 shows VES and trapping to estimate occupancy at similar levels in 2007,
but the under-dispersed model-estimated value for netting in 2007 demonstrates the
potential of a small dataset to distort results. During 2007 seven study ponds were
found to be occupied by T. cristatus by ≥ 1 survey method; in two ponds T. cristatus
were only detected by a single method (one each by trapping and VES). VES and
trapping detected T. cristatus on 53.8% and 67.3% of occupied pond surveys
respectively in 2007, while netting made positive detections on only 26.9% of
surveys in ponds found to be occupied by other methods. In 2008 T. cristatus
occupancy was determined at three ponds by trapping where VES failed, but overall
the rate of positive detections by VES and trapping in 2008 at T. cristatus-occupied
ponds were similar to 2007: 50.0% of VESs and 68.9% of trap surveys were
successful in 2008. Except for the under-dispersed netting model in 2007, there is
not much difference between the naïve and model-estimated occupancy rates (Figure
7.17).
7. Occupancy modelling
105
Occupancy r
ate
(ψ
)
Figure 7.17 Naïve and model-estimated occupancy (ψ) rates (±s.e.) of T. cristatus
by survey method and year.
= naïve ψ, = estimated ψ (uses averaged values from accepted models, Table 7.8);
dotted bars = under-dispersed model; VES = visual encounter survey.
L. helveticus
Very few detections of L. helveticus were made by netting, giving low naïve rates of
occupancy, estimated occupancy and estimated detection rates compared to VES and
trapping (excluding the under-dispersed 2007 VES model; Table 7.9). L. helveticus
was detected in > 90% of study ponds by one or more methods in both survey years
and therefore potential covariates of occupancy were not included in models.
Temperature covariates occurred as predictors of detection rate in 3/6 of the final,
most parsimonious, models for L. helveticus (Table 7.9 and Figure 7.18a-c), both
minimum and maximum overnight temperatures were tested separately for all
models, although they are clearly strongly correlated with each other (2007: r146 =
0.95, p < 0.001; 2008: r107 = 0.96, p < 0.001).
0.0
0.5
1.0
VES Trap Net
(a) 2007
0.0
0.5
1.0
VES Trap Net
(b) 2008
7. Occupancy modelling
106
Table 7.9 Most parsimonious models for L. helveticus surveys in 2007 and 2008 by
three survey methods.
Models K -2*LL QAICc QAIC
weight
Est. ψ
Est. p (±s.e.)
Visits (±s.e.)
2007 VES (naïve ψ = 0.71)
ψ(.) p(pm_gen, temp_max) 4 134.26 0.00 1.00 † 0.75 0.39 (±0.078) 7 (±1.6)
ĉ = 0.73
2008 VES (naïve ψ = 0.73)
ψ(.) p(pm_ timetaken) 3 119.04 0.00 0.59 0.76 0.59 (±0.077) 4 (±0.7)
ψ(.) p(clear_shore) 3 120.47 0.73 0.41 0.75 0.57 (±0.074) 4 (±0.7)
ĉ = 1.96 Model averaged: 0.76 0.58 (±0.076) 4 (±0.7)
2007 Trap (naïve ψ = 0.83)
ψ(.) p(temp_min, pm_rain) 4 147.89 0.00 1.00 0.84 0.58 (±0.053) 4 (±0.5)
ĉ = 1.20
2008 Trap (naïve ψ = 0.86)
ψ(.) p(pm_timetaken) 3 96.68 0.00 0.54 0.85 0.75 (±0.061) 3 (±0.4)
ψ(.) p(pm_timetaken, #traps) 4 95.03 1.84 0.46 0.85 0.76 (±0.073) 3 (±0.5)
ĉ = 1.40 Model averaged: 0.85 0.75 (±0.066) 3 (±0.4)
2007 Net (naïve ψ = 0.42)
ψ(.) p(clearshore) 3 103.59 0.00 1.00 0.43 0.41 (±0.085) 6 (±1.5)
ĉ = 1.10
2008 Net (naïve ψ = 0.36)
ψ(.) p(temp_min) 3 74.29 0.00 0.63 0.38 0.49 (±0.110) 5 (±1.4)
ψ(.) p(temp_max) 3 75.36 0.86 0.37 0.38 0.47 (±0.112) 5 (±1.6)
ĉ = 1.24 Model averaged: 0.38 0.48 (±0.110) 5 (±1.5)
Covariates are described in Appendix 8; K = number of model parameters; -2*LL = -2* Log
likelihood (residual deviance); ĉ = TestStat/AvgTestStat of the most parsimonious model; †
= AICc and AIC weight used rather than QAICc and QAIC weight (not adjusted for over-dispersion); Visits = number of visits required for 95% confidence that non-detection
indicates true absence; all life stages modelled together.
In contrast to T. cristatus, L. helveticus showed a negative relationship between
detection rate and water temperature for VES and trapping in 2007: at the lowest
temperatures encountered detectability by VES and trap for L. helveticus was very
high (Figure 7.18a-b). However, the two accepted models for netting in 2008 also
included water temperature covariates as predictors of detectability, but here there
was a positive relationship (Figure 7.18c). There is no clear reason why temperature
would have an opposite effect on netting to VES and trapping, although again, it
may be due to the small sample size.
7. Occupancy modelling
107
Dete
ction r
ate
(p)
Figure 7.18 Detection rates (±s.e.) of L. helveticus by top models where water
temperature was a covariate of detection rate.
= minimum temperature, = maximum temperature; based on (a) 2007 VES data: ψ(.) p(pm_gen, temp_max), β0 = 2.08, β1 = -1.52, under-dispersed; (b) 2007 trap data: ψ(.) p(temp_min, pm_rain), β0 = 2.54, β1 = -1.62; (c) 2008 netting data: ψ(.) p(temp_min), β0 = -
1.97, β1 = 2.15; β0 and β1 are equivalent, respectively, to the intercept and slope of a
regression equation.
In 2008 the amount of time spent at study ponds was a predictor of L. helveticus
detectability. Detection rate increased slightly with time taken for VES and it
decreased for trap surveys (Figure 7.19). Ponds that took longer to survey generally
had more complex structure (e.g. scrubby or steep banks), which is perhaps reflected
in the diminishing returns from trapping, whereby newts could be less likely to
encounter and enter traps in complex ponds.
Three or four trapping surveys would be needed to be 95% certain that the L.
helveticus-unoccupied ponds were truly unoccupied (Table 7.9). The well-fit 2008
VES model also suggests a minimum of four visits, although the under-dispersed
2007 VES model indicates seven. To achieve the lower occupancy estimate using
netting only, five to six repeat surveys would be required.
0.0
0.2
0.4
0.6
0.8
1.0 (a) Lh VES 2007
0.0
0.2
0.4
0.6
0.8
1.0 (b) Lh Trap 2007
0.0
0.2
0.4
0.6
0.8
1.0
0 5 10 15 20 25
Overnight water temperature (C)
(c) Lh Net 2008
7. Occupancy modelling
108
L. helveticus were present in nearly all of the ponds surveyed in 2007 and 2008:
91.7% and 95.5% respectively with all the survey methods taken together, although
none of the individual methods quite achieved naïve occupancy rates this high
(Figure 7.20). The naïve and estimated occupancy rates match each other quite
closely within methods, even for the under-dispersed VES model. As with T.
cristatus, trapping revealed L. helveticus occupancy at more ponds than the other
methods: trap surveys found L. helveticus in four 2007 and five 2008 ponds where
they were never detected by VES or netting. Netting performed very poorly, failing
to detect L. helveticus on any survey in 12 and 13 ponds (2007 and 2008
respectively) that were proven to be occupied by VES and/or traps. A contributory
factor to the weakness of netting as a survey method is that it tends to catch larvae
more easily than adults and the larvae of L. helveticus and L. vulgaris cannot be
distinguished in the field, although this limitation also applies to the other methods.
Dete
ction r
ate
(p)
Figure 7.19 Detection rates (±s.e.) of L. helveticus by top models, where „time
taken‟ was a covariate of detection.
Based on (a) 2008 VES data: ψ(.) p(pm_timetaken), β0 = 0.20, β1 = 0.35; (b) 2008 trap data:
ψ(.) p(pm_timetaken), β0 = 2.51, β1 = -2.47; β0 and β1 are equivalent, respectively, to the
intercept and slope of a regression equation.
0.0
0.2
0.4
0.6
0.8
1.0 (a) Lh VES 2008
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
0 10 20 30 40 50 60 70 80 90 100
Time taken in evening (mins)
(b) Lh Trap 2008
7. Occupancy modelling
109
Occupancy r
ate
(ψ
)
Figure 7.20 Naïve and model-estimated occupancy (ψ) rates (±s.e.) of L. helveticus
by survey method and year.
= naïve ψ, = estimated ψ (uses averaged values from accepted models, Table 7.9);
dotted bars = under-dispersed model; VES = visual encounter survey; where β0 and β1 are
equivalent, respectively, to the intercept and slope of a regression equation.
L. vulgaris
Well-fitted models were found for L. vulgaris with all three survey methods in both
years and only the 2008 netting model being under-dispersed. The detection rate of
L. vulgaris was strongly associated with the „day of year‟ covariate – for VES and
trap surveys in 2007 the detection rate decreased as the year progressed (Table 7.10
and Figure 7.22a-b), whereas for netting in 2008 detectability was minimal until
approximately day 120 (end of April, Figure 7.22c). However, this model was under-
dispersed and therefore should be interpreted with caution.
L. vulgaris 2008 trap data show the opposite relationship with „time taken‟ to L.
helveticus. Detectability of L. vulgaris by trap in 2008 increased with time spent
(Figure 7.21), which is more intuitive than the decrease shown by L. helveticus
(Figure 7.19b).
0.0
0.2
0.4
0.6
0.8
1.0
VES Trap Net
(a) 2007
0.0
0.2
0.4
0.6
0.8
1.0
VES Trap Net
(b) 2008
7. Occupancy modelling
110
Table 7.10 Most parsimonious models for L. vulgaris surveys in 2007 and 2008 by
three survey methods.
Models K -2*LL QAICc QAIC
weight
Est. ψ
Est. p (±s.e.)
Visits (±s.e.)
2007 VES (naïve ψ = 0.38)
ψ(Tc) p(day_of_yr) 4 75.55 0.00 1.00 0.55 0.17 (±0.068) 17 (±7.1)
ĉ = 1.19
2008 VES (naïve ψ = 0.41)
ψ(.) p(turbidity) 3 73.72 0.00 1.00 0.61 0.21 (±0.080) 13 (±5.5)
ĉ = 1.12
2007 Trap (naïve ψ = 0.71)
ψ(hedgerow) p(day_of_yr) 4 146.48 0.00 1.00 0.71 0.49 (±0.062) 5 (±0.8)
ĉ = 1.42
2008 Trap (naïve ψ = 0.68)
ψ(elevation) p(pm_timetaken) 4 94.24 0.00 1.00 0.77 0.58 (±0.074) 4 (±0.7)
ĉ = 1.77
2007 Net (naïve ψ = 0.13)
ψ(forest_brdlvd) p(.) 3 30.59 0.00 0.42 0.27 0.09 (±0.084) 32 (±31.1)
ψ(roads_all) p(.) 3 30.83 0.22 0.37 0.19 0.13 (±0.083) 22 (±14.7)
ψ(roads_<5.5m) p(.) 3 31.93 1.24 0.21 0.22 0.11 (±0.083) 26 (±20.6)
ĉ = 1.08 Model averaged: 0.23 0.11 (±0.083) 26 (±21.0)
2008 Net (naïve ψ = 0.09)
ψ(elevation) p(day_of_yr) 3 11.05 0.00 0.68 † 0.27 0.09 (±0.073) 32 (±27.0)
ψ(elevation) p(.) 3 14.56 1.51 0.32 † 0.27 0.07 (±0.047) 42 (±28.8)
ĉ = 0.52 Model averaged: 0.27 0.08 (±0.065) 35 (±27.7)
Covariates are described in Appendix 8; K = number of model parameters; -2*LL = -2* Log
likelihood (residual deviance); ĉ = TestStat/AvgTestStat of the most parsimonious model; †
= AICc and AIC weight used rather than QAICc and QAIC weight (not adjusted for over-dispersion); Visits = number of visits required for 95% confidence that non-detection
indicates true absence; all life stages modelled together.
Dete
ction r
ate
(p)
Figure 7.21 Detection rate (±s.e.) of L. vulgaris by the top trapping model in 2008,
where „pm_timetaken‟ was a covariate of detectability.
Based on 2008 Trap data: ψ(elevation) p(pm_timetaken), β0 = -0.84, β1 = 2.34; β0 and β1 are
equivalent, respectively, to the intercept and slope of a regression equation.
0.0
0.2
0.4
0.6
0.8
1.0
0 10 20 30 40 50 60 70 80 90 100
Time taken in evening(minutes)
Lv Trap 2008
7. Occupancy modelling
111
Dete
ction r
ate
(p)
Figure 7.22 Detection rates (±s.e.) of L. vulgaris by top models, where „day of year‟
was a covariate of detection.
Based on (a) 2007 VES data: ψ(.) p(day_of_yr), β0 = 1.73, β1 = -2.96; (b) 2007 trap data: ψ(hedgerow) p(day_of_yr), β0 = 3.24, β1 = -2.70; (c) 2008 netting data ψ(elevation)
p(day_of_yr), β0 = -12.63, β1 = 8.11, under-dispersed; β0 and β1 are equivalent, respectively,
to the intercept and slope of a regression equation.
Covariates predicting L. vulgaris occupancy of study ponds varied between models
and survey methods, because the ponds found to be occupied varied between
methods. In the most parsimonious 2007 trap model, hedgerow length within 100 m
of the pond was selected – the unoccupied ponds all had no hedgerow nearby, while
the occupied ponds averaged 163.9 m (±49.89 s.e.; t22 = 2.08, p = 0.049). Both the
net and trap models of the 2008 season selected elevation as predictive of
occupancy: on average occupied ponds (detected by any method) were at lower
elevations than unoccupied ponds (mean = 284.0 m.a.s.l. ±6.51 and 350.5 m.a.s.l.
±19.96 respectively; t20 = 4.06, p < 0.005). The 2007 netting model has broadleaved
forest as a predictor of occupancy, but this is strongly biased by the method‟s failure
to detect occupancy at most sites – when occupied sites detected by any method are
compared to unoccupied sites there is no difference in broadleaved forest coverage
(t22 = 0.34, p > 0.05). The presence of T. cristatus in ponds was retained as a
predictor of L. vulgaris occupancy in the most parsimonious VES model in 2007;
they occurred together more frequently than individually (Figure 7.23).
-0.2
0.0
0.2
0.4
0.6
0.8(a) Lv VES 2007
0.0
0.2
0.4
0.6
0.8
1.0(b) Lv Trap 2007
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
Day of year
(c) Lv Net 2008
7. Occupancy modelling
112
Again, netting achieved very low detection rates and consequently low occupancy
rates for L. vulgaris. VES also performed relatively badly with naïve and estimated
occupancy rates far lower than those for trapping and many surveys required for
95% confidence in non-detections (Table 7.10).
L. vulgaris were found in 70.8% and 68.2% (2007 and 2008) of study ponds by ≥ 1
survey method. Traps detected L. vulgaris at least once per season in each of the
occupied ponds, while VES failed in 47.1% and 40.0% and netting failed in 82.4%
and 86.3%. Neither VES nor netting detected L. vulgaris in ponds where they were
not also detected by trapping (Figure 7.24).
Figure 7.23 Occupancy of 2007 study ponds by T. cristatus and L. vulgaris.
Occupancy determined by ≥ 1 detection by any method within th 2007 season.
Occup
an
cy r
ate
(ψ
)
Figure 7.24 Naïve and model-estimated occupancy (ψ) rates (±s.e.) of L. vulgaris by
survey method and year.
= naïve ψ, = estimated ψ (uses averaged values from accepted models, Table 7.10);
dotted bars = under-dispersed model; VES = visual encounter survey.
0
2
4
6
8
10
12
Tc & Lv Tc only Lv only No Tc or Lv
Nu
mb
er
of p
on
ds
(de
tecte
d b
y ≥
1
me
tho
d)
0.0
0.2
0.4
0.6
0.8
1.0
VES Trap Net
(a) 2007
0.0
0.2
0.4
0.6
0.8
1.0
VES Trap Net
(b) 2008
7. Occupancy modelling
113
7.4 Discussion
The probability of detection of any of the methods and species studied here is far
from perfect, ranging from p = 0.00 for T. cristatus by net in 2008 to 0.75 for L.
helveticus by trap in 2008, often with variation between years. Comparison of the
survey methods appropriate for each amphibian species highlighted that some
methods are far more effective than others in terms of achieving higher detection
rates and requiring fewer survey visits for an acceptable level of confidence that non-
detections represent true absences.
For the frogs, R. temporaria and Pelophylax spp., VES had the highest detection
rate: larvae of these species are easily identifiable and adults are quite conspicuous in
their behaviour, especially the Pelophylax spp.. Among the newts, trapping was
clearly the best method and netting was of very limited use, while VES had mostly
intermediate detection rates but was poor for L. vulgaris. Netting is routinely used in
surveys for European newt species (Griffiths and Raper, 1994; Arntzen, 2002),
therefore its inefficiency is surprising. In the current study netting was restricted to
avoid creating too much disturbance or damage to vegetation. It is a relatively quick,
simple method, which is perhaps explains its popularity, but here it has not proven
worthwhile for any species. Netting was never the sole method of detection of a
species on any survey visit and made far fewer detections than VES or trapping, or
aural for anurans.
Aural survey was only modelled for Pelophylax spp. and A. obstetricans.
Unfortunately the A. obstetricans model is unreliable, although it indicates a low-
medium detection rate of 0.45. The population size of A. obstetricans at some study
sites may have been so low that survey conditions could not predict when they
would be detected (Mazerolle et al., 2007), for example at ponds 9-10 just two
detections were made in ten surveys in year 1 and the maximum number of
individuals heard calling was estimated to be just four. For Pelophylax spp. aural
survey had medium success; it achieved higher detection rates than netting, but was
not as effective as VES or trapping. Both Pelophylax spp. and A. obstetricans exhibit
an extended breeding season, so are more likely to be repeatedly detected by aural
7. Occupancy modelling
114
survey, although A. obstetricans mating activity is mostly concentrated into several
peaks across a long season (Marquez, 1992).
7.4.1 Covariates of detection
The survey-specific covariates expected to influence detection, such as turbidity for
VES and dense vegetation for netting, did not often make well-fitted models and
therefore are not reported in the model tables. The most frequently occurring
covariates of detection for the fully modelled species were temperature (maximum or
minimum) and day of year. „Time taken‟ and „clear shore‟ were also in a number of
top models and overnight water temperature generally had a positive effect in the
well-fitted models. Lower oxygen content in warmer water (Steen, 1958)
necessitates more frequent surface movements to breathe and previous studies have
found greater amphibian activity at higher temperatures (e.g. Oseen and Wassersug,
2002; Blair, 1960 in Kirlin et al., 2006). Therefore there is a greater chance of
entering a trap or being seen moving in the torch beam. Maximum temperature was
the only detection covariate fitted to A. obstetricans data, supporting the idea that
anuran call intensity increases with temperature (Marquez, 1992; Navas, 1996;
Navas and Bevier, 2001). For VES and trapping detection rates of all species
decreased as the season progressed, while for netting it increased. Although netting
was generally a poor method, this suggests that it was more effective later in the
season, when VES and traps were less useful. This is probably due to the numbers of
adults available to detect being lower as they leave the water post-breeding and the
number of larvae being higher (Figures 7.3, 7.5, 7.9 and 7.12).
In reality the effects of survey-specific covariates are likely to vary across the
season. For example, early in the year there is a minimum threshold temperature
above which Bufo bufo begin migration to breeding ponds (Reading, 1998), but after
the threshold has been reached and the temperature continues to rise, other covariates
may become more important (Kirlin et al., 2006).
7.4.2 Covariates of occupancy
Only two species, T. cristatus and L. vulgaris, were modelled with covariates of
occupancy. The other moderately common species either did not fit any site-specific
7. Occupancy modelling
115
covariates (A. obstetricans) or could not form acceptable models (Bufo bufo). In
some of the most parsimonious models for T. cristatus, pond size and the length of
tracks in the vicinity were both positively related to the likelihood of occupancy, and
elevation occurred in three of the top models for L. vulgaris. However, overall no
strong patterns emerged and no covariates stood out for either species. With a small
sample size the models can easily be disproportionately influenced by a covariate,
particularly here where each survey method produced slightly different sets of
„occupied‟ and „unoccupied‟ ponds.
7.4.3 Occupancy modelling
Occupancy modelling had variable success with the amphibian species surveyed in
Luxembourg. The rarest species were excluded because they occurred at too few
ponds and the most common species could only be modelled for detection rate (not
for occupancy). Moreover, the sample size in both study years (2007 and 2008) was
only just sufficient to run models with additional parameters – using AICc addresses
the problem of small sample sizes, although a larger dataset would allow
simultaneous modelling of more covariates and produce better-fitting models.
Occupancy modelling is a very „data-hungry‟ analysis technique, it cannot be done
with the low resolution, presence-only data typically collected by record centres. To
collect a more adequately sized dataset within a season the current study would have
required more than one person, and an even greater survey effort would have been
needed for sufficient data for multi-season models.
The present set of models is a preliminary exploration of the application of
occupancy modelling and the types of covariates to be used for species‟ occupancy
and detectability. Occupancy modelling represents a considerable advance from the
traditional logistic regression and discriminant analysis approaches, which tend to
ignore detection rates (Mazerolle et al., 2007) and lend false confidence to
management practices (MacKenzie, 2005). Data collection needs to be undertaken
with the demands of occupancy modelling in mind to ensure that the appropriate
type and standard of data are collected. Typical record centre data are currently not
suitable, but as current revisions of the National Amphibian and Reptile Recording
7. Occupancy modelling
116
Scheme (NARRS) in the UK demonstrate, volunteer schemes can be designed to
collect the required data (Sewell et al., 2010).
The cryptic behaviour of many amphibian species means that a single survey method
is unlikely to detect all species in all sites (Ryan et al., 2002). Variation in the cost
and time required to implement a method add to the considerations necessary when
designing a survey protocol (O‟Donnell et al., 2007). It is important to know the
efficiency of each method and the sampling effort required (de Solla et al., 2005).
Further work should include pairs of survey methods modelled together to find the
most efficient method combinations, however this would also benefit from a larger
dataset.
8. General discussion
117
8. General discussion
General discussion
The analyses in Chapters 3 – 7 employed two different types of data, firstly broad
scale, presence-only data, collected by a large number of people using mixed survey
methods, which provided an overview of the status of amphibian species in
Luxembourg. Secondly a detailed dataset on a subset of sites used to examine
relationships between amphibians and biotic and abiotic variables in and around
breeding ponds. In view of the population declines and extinctions reported among
amphibians globally (e.g. Barinaga, 1990; Alford et al., 2001; Houlahan et al., 2001;
Collins and Storfer, 2003; Beebee and Griffiths, 2005), the record centre data
showed quite a positive outlook for most amphibian species in Luxembourg.
Common species‟ ranges have generally expanded over recent decades and, despite
the identification of L. helveticus as endangered in the Benelux countries by the
European herpetological atlas (Gasc et al., 1997), it was the most widespread newt
species according to the MNHN data. Even the locally threatened A. obstetricans
appears from these data to have increased its range. Range sizes of H. arborea, T.
cristatus and L. vulgaris have shrunk, but while the latter two remain moderately
common, H. arborea has become very rare and more recently is thought to have
gone extinct at all but one pond. National species action plans for H. arborea, T.
cristatus and Bufo calamita are already in place to address specific threats and
ensure their long-term survival in Luxembourg (MNHNL - Groupe herpétologique,
2009a, b; Proess, 2009). These conservation measures will also benefit the non-
targeted species such as L. vulgaris, which was found in association with T.
cristatus.
In Chapters 3 and 4 the problems associated with low resolution data were discussed,
however Chapter 7 introduced an additional consideration: detection rates. The
highest detection rate for any species by any method was 0.75, although most were
much lower. Imperfect detection raises a key issue with record centre data, which is
that some or all species may have been routinely under-detected, and therefore
under-represented, throughout the historical records. As argued by Telfer (2002),
although detection rates are not perfect, survey methods have not changed greatly
therefore detection should have remained fairly constant over time, so the broad
8. General discussion
118
trends demonstrated are probably true. However, for the rare species the margin of
error could have a greater effect, especially when coupled with the unknown bias in
recording (Dennis et al., 1999) – for example there may be many populations of
common species such as L. helveticus that are known by individual surveyors but not
reported to the MNHN, and there are probably no known unrecorded populations of
Bufo calamita. For the latter species one additional population would represent a
50% increase, but it would make little difference to the common species.
Accounting for a species‟ rate of detection could increase estimates of its range size.
However, in the present study the combination of imprecise historical data and weak
occupancy models make for dubious extrapolations. For example, R. temporaria was
found in 54.89% of the grid squares surveyed (Figure 3.3) and its rate of detection
(p) was approximately 0.47 by all survey methods: ideally it would be possible to
deduce that it was likely to occupy 53% more squares than it had been detected in
(83.98%), but the type of surveys and the survey effort were not reliably recorded in
the MNHN data. Further development of the models and a higher quality of record
centre data are needed before such inferences can be made.
Findings from Chapters 5 and 6 suggested that there is potential to use amphibians as
a surrogate measure of general biodiversity in their breeding ponds, but no strong
relationships were established here and surrogates ideally have high detection rates.
The detection rates for macrophytes and macroinvertebrates will also have been
below 100% – potentially very low for invertebrates, but also for small,
inconspicuous or submerged plants. Where conservation or management planning is
required for a range of taxa, there may be no substitute for conducting thorough
surveys appropriate to each group. Although the conservation measures intended for
a specific species, for example the removal of invasives and limiting local farming
activity prescribed for H. arborea (MNHNL - Groupe herpétologique, 2009a),
probably benefit other native species too. Occupancy modelling could be applied to
the species of interest to help design more efficient surveys, but it would be too time-
consuming to gather the initial data to apply it to every species within a large
taxonomic group.
8. General discussion
119
On the whole PAs appear to perform well for amphibians in Luxembourg, with the
common species found both inside and outside. The three very rare species, H.
arborea, Bufo calamita and Bombina variegata, occupied very few sites, all of
which were within PAs. These species are afforded a high level of protection, but
currently have little chance of colonising new areas because the PAs they inhabit are
surrounded by roads, towns and intensive agriculture (Bennett, 2003; Hannah et al.,
2007). The modern protected areas approach promotes connectivity between sites
(Haslett et al., 2010) and new action plans for these species address their isolation,
with new ponds planned for the area around the sole remaining H. arborea site and
new scrapes as potential breeding ponds already made in the large park where
Bombina variegata occupied the only waterbody. According to the MNHN data the
rarest amphibian species in Luxembourg have always been rare, so their general
status has not changed, but the threats have perhaps intensified and at least H.
arborea is now much rarer than a decade ago. Unfortunately the rarest species were
unsuitable for statistical analysis by any of the methods used here, but it is clear that
they are dependent on conservation action to avoid becoming nationally extinct.
Identification of low levels of the fungal disease Batrachochytrium dendrobatidis in
amphibian populations in Luxembourg (see Appendix 9; Wood et al., 2009) adds
further threat to the amphibian population, and is worthy of close monitoring over
future years.
The data collected by the MNHN are a very useful resource both for identifying
populations for further study and to monitor species‟ distributions; similar databases
are kept for all other plant and animal taxa found in Luxembourg. However, for the
purposes of the current study finer details would have increased the strength of
analysis and inference greatly. For example, full grid references would have allowed
every record to be pin-pointed and the surrounding landscape variables to be
measured precisely (rather than simply sharing a grid square) and records of non-
detections (or „null‟ records) would allow exploration of local extinctions more
accurately than the abrupt ending of records. Such details could easily be supplied by
database contributors if the data submission process was simpler, which could now
be done via a website rather than the current unwieldy spreadsheet.
8. General discussion
120
There was some conflicting information between the various datasets employed in
the chapters – for example the MNHN data showed an increase in A. obstetricans
range, but at several sites selected for intense study where it had previously been
recorded it was not found, suggesting local extinctions (and / or low detection rates)
had occurred. Also the map data could not be ground-truthed, and may have
contained inaccuracies, or have had low resolution. For example some of the smaller,
non-permanent study ponds were not included in the ACT maps, and therefore there
may be other small waterbodies, such as ditches, that are suitable amphibian
breeding habitat, but do not show up on the maps.
8.1.1 Future data collection
Natural habitats are rapidly changing under increasingly intense anthropogenic
pressures. The effect on amphibians and other taxa has been largely negative and is
mostly attributable to loss of habitat (Cushman, 2006). Climate change projections
present an uncertain picture for the future of amphibian populations (Araújo et al.,
2006), but heightened monitoring activity – both of amphibians and the factors
threatening their survival – is fast improving knowledge of the ecosystems and
effective conservation measures. Imperfect detection has typically been inadequately
addressed in previous studies by use of multiple methods and ample surveys in the
field, but occupancy modelling represents a big step forward in survey design by
properly accounting for detectability.
The main barrier to the adoption of occupancy modelling for designing all fauna and
flora survey protocols is that it needs high quality data (e.g. repeated visits,
independent / discrete sites) and it requires records of non-detections as well as
presences, and therefore cannot be used on the historical data typically held by
record centres. In countries where data collection capacity is limited occupancy
modelling would be difficult to implement. However, as Luxembourg and the
neighbouring countries have many amateur and professional contributors to their
natural history records, it would be possible to collect a sufficiently large database to
build much stronger models than are presented in the current thesis, from which
standard survey protocols could be developed. Ongoing monitoring is very important
to inform conservation management decisions: Luxembourg has a high level of
8. General discussion
121
monitoring and good information on the highly threatened amphibians. Building on
the existing database and recording in more detail could greatly strengthen the
information that can be drawn from it.
9. References
122
9. References
References
Alford, R. A., Dixon, P. M. & Pechmann, J. H. K. (2001) Global amphibian
population declines. Nature, 412, 499-500.
Alford, R. A. & Richards, S. J. (1999) Global amphibian declines: a problem in
applied ecology. Annual Review of Ecology and Systematics, 30, 133-165.
Altwegg, R., Wheeler, M. & Erni, B. (2008) Climate and the range dynamics of
species with imperfect detection. Biology Letters, 4, 581-584.
AmphibiaWeb (2010) Information on amphibian biology and conservation [web
application]. Berkeley, California: AmphibiaWeb. Electronic database
available at: www.amphibiaweb.org/ (accessed on 6th January 2010).
Anderson, B. J., Akcakaya, H. R., Araujo, M. B., Fordham, D. A., Martinez-Meyer,
E., Thuiller, W. & Brook, B. W. (2009) Dynamics of range margins for
metapopulations under climate change. Proceedings of the Royal Society B-
Biological Sciences, 276, 1415-1420.
Anderson, D. R. & Burnham, K. P. (2002) Avoiding pitfalls when using
information-theoretic methods. Journal of Wildlife Management, 66, 912-
918.
Araújo, M. B. (1999) Distribution patterns of biodiversity and the design of a
representative reserve network in Portugal. Diversity and Distributions, 5,
151-163.
Araújo, M. B., Thuiller, W. & Pearson, R. G. (2006) Climate warming and the
decline of amphibians and reptiles in Europe. Journal of Biogeography, 33,
1712-1728.
Arntzen, J. W. (2002) Testing for equal catchability of Triturus newts by dip netting.
Journal of Herpetology, 36, 272-276.
Arrhenius, O. (1921) Species and area. Journal of Ecology, 9.
Auderset Joye, D., Oertli, B., Lehmann, A., Juge, R. & Lachavanne, J. B. (2002) The
prediction of macrophyte species occurrence in Swiss ponds. 11th
International Symposium on Aquatic Weeds, Moliets et Maa, France. pp.175-
182.
Babbitt, K. J. & Jordan, F. (1996) Predation on Bufo terrestris tadpoles: effects of
cover and predator identity. Copeia, 485-488.
9. References
123
Babbitt, K. J. & Tanner, G. W. (1997) Effects of cover and predator identity on
predation of Hyla squirella tadpoles. Journal of Herpetology, 31, 128-130.
Balmford, A. & Gaston, K. J. (1999) Why biodiversity surveys are good value.
Nature, 398, 204-205.
Balmford, A., Green, M. J. B. & Murray, M. G. (1996) Using higher-taxon richness
as a surrogate for species richness .1. Regional tests. Proceedings of the
Royal Society of London Series B-Biological Sciences, 263, 1267-1274.
Banks, B. & Beebee, T. J. C. (1986) Climatic effects on calling and spawning of the
natterjack toad Bufo calamita: discriminant analyses and applications for
conservation monitoring. Biological Conservation, 36, 339-350.
Barinaga, M. (1990) Where have all the froggies gone? Science, 247, 1033-1034.
Beebee, T. J. C. (1977) Habitats of British amphibians (1): Chalk uplands. Biological
Conservation, 12, 279-293.
Beebee, T. J. C. (1979) Habitats of British amphibians (2): Suburban parks and
gardens. Biological Conservation, 15, 241-257.
Beebee, T. J. C. (1980) Habitats of British amphibians (3): River valley marshes.
Biological Conservation, 18, 281-287.
Beebee, T. J. C. (1981) Habitats of British amphibians (4): Agricultural lowlands
and a general discussion of requirements. Biological Conservation, 21, 127-
139.
Beebee, T. J. C. (1985) Discriminant analysis of amphibian habitat determinants in
south-east England. Amphibia-Reptilia, 6, 35-43.
Beebee, T. J. C., Flower, R. J., Stevenson, A. C., Patrick, S. T., Appleby, P. G.,
Fletcher, C., Marsh, C., Natkanski, J., Rippey, B. & Battarbee, R. W. (1990)
Decline of the natterjack toad Bufo calamita in Britain: paleoecological,
documentary and experimental evidence for breeding site acidification.
Biological Conservation, 53, 1-20.
Beebee, T. J. C. & Griffiths, R. A. (2005) The amphibian decline crisis: a watershed
for conservation biology? Biological Conservation, 125, 271-285.
Benelux Economic Union (1982) Benelux Convention on Nature Conservation and
Landscape Protection. Available at:
http://sedac.ciesin.org/entri/texts/benelux.landscape.protection.1982.html
(accessed on 10th August 2010).
9. References
124
Bennett, A. F. (2003) Linkages in the Landscape: the Role of Corridors and
Connectivity in Wildlife Conservation. IUCN, Gland, Switzerland and
Cambridge, UK.
Biggs, J., Fox, G., Nicolet, P., Walker, D., Whitfield, M. & Williams, P. (1998) A
guide to the methods of the National Pond Survey. Pond Action, Oxford.
Available at: www.pondnetwork.org.uk/Downloads/NPSMethods.pdf
(accessed on 15th February 2008).
Bilton, D. T., McAbendroth, L., Bedford, A. & Ramsay, P. M. (2006) How wide to
cast the net? Cross-taxon congruence of species richness, community
similarity and indicator taxa in ponds. Freshwater Biology, 51, 578-590.
Blaustein, A. R., Romansic, J. M., Kiesecker, J. M. & Hatch, A. C. (2003)
Ultraviolet radiation, toxic chemicals and amphibian population declines.
Diversity and Distributions, 9, 123-140.
Bloom, S. A. (1981) Similarity indices in community studies: potential pitfalls.
Marine Ecology - Progress Series, 5, 125-128.
Bosch, J., Beebee, T., Schmidt, B., Tejedo, M., Martínez-Solano, I., Salvador, A.,
García-París, M., Recuero Gil, E., Arntzen, J. W., Diaz Paniagua, C. &
Marquez, R. (2008) Alytes obstetricans. IUCN 2009. IUCN Red List of
Threatened Species. Version 2009.2. Electronic article available at:
www.iucnredlist.org (accessed on 17th January 2010).
Boulinier, T., Nichols, J. D., Sauer, J. R., Hines, J. E. & Pollock, K. H. (1998)
Estimating species richness: the importance of heterogeneity in species
detectability. Ecology, 79, 1018-1028.
Bousbouras, D. & Ioannidis, Y. (1997) The distribution and habitat preferences of
the amphibians of Prespa National Park. Hydrobiologia, 351, 127-133.
Braz, E. & Joly, P. (1994) Microhabitat use, resource partitioning and ecological
succession in a size-structured guild of newt larvae (g. Triturus, Caudata,
Amphibia). Archiv Fur Hydrobiologie, 131, 129-139.
Brooke, P. N., Alford, R. A. & Schwarzkopf, L. (2000) Environmental and social
factors influence chorusing behaviour in a tropical frog: examining various
temporal and spatial scales. Behavioral Ecology and Sociobiology, 49, 79-87.
9. References
125
Buley, K. & Garcia, G. (1997) The recovery programme for the Mallorcan midwife
toad Alytes muletensis: an update. Dodo. Journal of the Wildlife Preservation
Trusts, 33, 80-90.
Burnham, K. P. & Anderson, D. R. (1998) Model selection and inference: a
practical information-theoretic approach. Springer-Verlag, New York.
Burnham, K. P. & Anderson, D. R. (2004) Multimodel inference - understanding
AIC and BIC in model selection. Sociological Methods & Research, 33, 261-
304.
Carey, C., Cohen, N. & Rollins-Smith, L. (1999) Amphibian declines: an
immunological perspective. Developmental and Comparative Immunology,
23, 459-472.
Caro, T. M. & O'Doherty, G. (1999) On the use of surrogate species in conservation
biology. Conservation Biology, 13, 805-814.
Centre for Ecology & Hydrology (2009) RIVPACS - how RIVPACS works. Centre
for Ecology & Hydrology. Available at:
www.ceh.ac.uk/products/software/HowRIVPACSWorks.html (accessed on
22nd October 2009).
Chamber of Commerce of the Grand Duchy of Luxembourg (2009) Your access to
European markets: Luxembourg: where else? Electronic article available at:
www.pwc.com/lu/en/invest-in-luxembourg/docs/pwc-publ-lux-where-
else.pdf (accessed on 25th July 2010).
Chinery, M. (1993) Insects of Britain and Northern Europe. 3rd edition. Harper
Collins Publishers, London.
Clemons, J. (1998) Provisional atlas of the amphibians and reptiles of
Warwickshire. Warwickshire Museum Service.
Collen, B., Ram, M., Zamin, T. & McRae, L. (2008) The tropical biodiversity data
gap: addressing disparity in global monitoring. Tropical Conservation
Science, 1, 75-88.
Collins, J. & Halliday, T. (2005) Forecasting changes in amphibian biodiversity:
aiming at a moving target. Philosophical Transactions of the Royal Society
B: Biological Sciences, 360, 309.
Collins, J. P. & Storfer, A. (2003) Global amphibian declines: sorting the
hypotheses. Diversity and Distributions, 9, 89-98.
9. References
126
Connor, J. S., Seborg, D. E. & Larimore, W. E. (2004) Analysis of the ΔAIC statistic
for optimal detection of small changes in dynamic systems. The 2004
American Control Conference, Boston, Massachusetts.
Conseil d‟Etat [Luxembourg] (2009) Règlement grand-ducal du 9 janvier 2009
concernant la protection intégrale et partielle de certaines espèces animales
de la faune sauvage. Available at:
www.legilux.public.lu/leg/a/archives/2009/0004/a004.pdf#page=8 (accessed
on 5th August 2010).
Convention on Biological Diversity (2002) Strategic Plan for the Convention on
Biological Diversity (COP 6 Decision VI/26). Sixth Meeting of the
Conference of the Parties to the Convention on Biological Diversity. The
Hague, Netherlands, 7-19th April 2002. Available at:
www.cbd.int/decision/cop/?id=7200 (accessed on 7th April 2010).
Corser, J. D. (2001) Decline of disjunct green salamander (Aneides aeneus)
populations in the southern Appalachians. Biological Conservation, 97, 119-
126.
Council Directive 92/43/EEC of 21 May 1992 on the Conservation of natural
habitats and of wild fauna and flora. Available at: www.eur-
lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONSLEG:1992L0043:20070
101:EN:PDF (accessed on 11th April 2010).
Council Directive 2009/147/EC of 30 November 2009 on the Conservation of wild
birds (codified version). Available at: www.eur-
lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONSLEG:1992L0043:20070
101:EN:PDF (accessed on 11th April 2010).
Council of Europe (1979) Convention on the Conservation of European Wildlife and
Natural Habitats. Bern, 19.IX.1979. Available at:
http://conventions.coe.int/Treaty/en/Treaties/Html/104.htm (accessed on 10th
May 2010).
Council of Europe (1992) Council Directive 92/43/EEC of 21 May 1992 on the
conservation of natural habitats and of wild fauna and flora. Available at:
http://eur-
lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:31992L0043:EN:HT
ML (accessed on 10th May 2010).
9. References
127
Croft, P. S. (1986) A key to the major groups of British freshwater invertebrates.
Field Studies Council, Shrewsbury.
Crouch, W. B. & Paton, P. W. C. (2002) Assessing the use of call surveys to monitor
breeding anurans in Rhode Island. Journal of Herpetology, 36, 185-192.
Cushman, S. A. (2006) Effects of habitat loss and fragmentation on amphibians: a
review and prospectus. Biological Conservation, 128, 231-240.
Dale, V. H. & Beyeler, S. C. (2001) Challenges in the development and use of
ecological indicators. Ecological indicators, 1, 3-10.
Daszak, P., Berger, L., Cunningham, A. A., Hyatt, A. D., Green, D. E. & Speare, R.
(1999) Emerging infectious diseases and amphibian population declines.
Emerging Infectious Diseases, 5, 735-748.
Davies, B., Biggs, J., Williams, P., Whitfield, M., Nicolet, P., Sear, D., Bray, S. &
Maund, S. (2008) Comparative biodiversity of aquatic habitats in the
European agricultural landscape. Agriculture Ecosystems & Environment,
125, 1-8.
de Solla, S. R., Shirose, L. J., Fernie, K. J., Barrett, G. C., Brousseau, C. S. &
Bishop, C. A. (2005) Effect of sampling effort and species detectability on
volunteer based anuran monitoring programs. Biological Conservation, 121,
585-594.
Della Bella, V., Bazzanti, M., Dowgiallo, M. G. & Iberite, M. (2008) Macrophyte
diversity and physico-chemical characteristics of Tyrrhenian coast ponds in
central Italy: implications for conservation. Hydrobiologia, 597, 85-95.
Dennis, R. L. H., Sparks, T. H. & Hardy, P. B. (1999) Bias in butterfly distribution
maps: the effects of sampling effort. Journal of Insect Conservation, 3, 33-
42.
Dennis, R. L. H. & Thomas, C. D. (2000) Bias in butterfly distribution maps: the
influence of hot spots and recorder's home range. Journal of Insect
Conservation, 4, 73-77.
Denton, J. S. (1991) The distribution and breeding site characteristics of newts in
Cumbria, England. Herpetological Journal, 1, 549-554.
Devictor, V., Godet, L., Julliard, R., Couvet, D. & Jiguet, F. (2007) Can common
species benefit from protected areas? Biological Conservation, 139, 29-36.
9. References
128
Dolmen, D. (1980) Distribution and habitat of the smooth newt, Triturus vulgaris
(L.), and the warty newt, T. cristatus (Laurenti), in Norway. Proceedings of
the European Herpetological Symposium, Cotswold Wildlife Park, UK.
pp.127-137.
Dolmen, D. & Koksvik, J. I. (1983) Food and feeding habits of Triturus vulgaris (L.)
and T. cristatus (LAURENTI) (Amphibia) in two bog tarns in central
Norway. Amphibia-Reptilia, 4, 17-24.
Donald, P. F. & Fuller, R. J. (1998) Ornithological atlas data: a review of uses and
limitations. Bird Study, 45, 129-145.
Donovan, T. M. & Hines, J. (2007) Exercises in occupancy modeling and estimation.
Electronic article available at:
www.uvm.edu/envnr/vtcfwru/spreadsheets/occupancy/occupancy.htm
(accessed on 7th June 2009).
Dormann, C. F., McPherson, J. M., Araújo, M. B., Bivand, R., Bolliger, J., Carl, G.,
Davies, R. G., Hirzel, A., Jetz, W., Kissling, W. D., Kühn, I., Ohlemüller, R.,
Peres-Neto, P. R., Reineking, B., Schröder, B., Schurr, F. M. & Wilson, R.
(2007) Methods to account for spatial autocorrelation in the analysis of
species distributional data: a review. Ecography, 30, 609-628.
Duellman, W. E. & Trueb, L. (1986) Biology of Amphibians. The John Hopkins
University Press, Baltimore.
Dvorak, J. & Gvozdik, L. (2009) Oviposition preferences in newts: does temperature
matter? Ethology, 115, 533-539.
Egan, R. S. & Paton, P. W. C. (2004) Within-pond parameters affecting oviposition
by wood frogs and spotted salamanders. Wetlands, 24, 1-13.
Elith, J., Graham, C. H., Anderson, R. P., Dudik, M., Ferrier, S., Guisan, A.,
Hijmans, R. J., Huettmann, F., Leathwick, J. R., Lehmann, A., Li, J.,
Lohmann, L. G., Loiselle, B. A., Manion, G., Moritz, C., Nakamura, M.,
Nakazawa, Y., Overton, J. M., Peterson, A. T., Phillips, S. J., Richardson, K.,
Scachetti-Pereira, R., Schapire, R. E., Soberon, J., Williams, S., Wisz, M. S.
& Zimmermann, N. E. (2006) Novel methods improve prediction of species'
distributions from occurrence data. Ecography, 29, 129-151.
Erdfelder, E., Faul, F. & Buchner, A. (1996) GPOWER: a general power analysis
program. Behavior Research Methods Instruments & Computers, 28, 1-11.
9. References
129
Everitt, B. S. & Dunn, G. (2001) Applied Multivariate Data Analysis. Arnold,
London.
Field, A. (2005) Discovering Statistics Using SPSS. SAGE Publications Ltd.,
London.
Field, S. A., Tyre, A. J. & Possingham, H. P. (2005) Optimizing allocation of
monitoring effort under economic and observational constraints. Journal
Information, 69.
Fielding, A. H. & Bell, J. F. (1997) A review of methods for the assessment of
prediction errors in conservation presence/absence models. Environmental
Conservation, 24, 38-49.
Foggo, A., Rundle, S. D. & Bilton, D. T. (2003) The net result: evaluating species
richness extrapolation techniques for littoral pond invertebrates. Freshwater
Biology, 48, 1756-1764.
Fowler, J., Cohen, L. & Jarvis, P. (1998) Practical Statistics for Field Biology. 2nd
edition. John Wiley & Sons Ltd, Chichester.
Frantz, A. C., Proess, R., Burke, T. & Schley, L. (2009) A genetic assessment of the
two remnant populations of the natterjack toad (Bufo calamita) in
Luxembourg. The Herpetological Journal, 19, 53-59.
Friedl, T. W. P. & Klump, G. M. (2002) The vocal behaviour of male European
treefrogs (Hyla arborea): implications for inter- and intrasexual selection.
Behaviour, 139, 113-136.
Gasc, J.-P., Cabela, A., Crnobrnja-Isailovic, J., Dolmen, D., Grossenbacher, K.,
Haffner, P., Lescure, J., Martens, H., Martínez Rica, J. P., Maurin, H.,
Oliviera, M. E., Sofianidou, T. S., Veith, M. & Zuiderwijk, A. (1997) Atlas of
amphibians and reptiles in Europe. Societas Europaea Herpetologica and
Muséum National d'Histoire Naturelle, Paris.
Gaston, K. J. (1996) Species-range-size distributions: patterns, mechanisms and
implications. Trends in Ecology & Evolution, 11, 197-201.
Gaston, K. J. (2000) Global patterns in biodiversity. Nature, 405, 220-227.
Gaston, K. J., Charman, K., Jackson, S. F., Armsworth, P. R., Bonn, A., Briers, R.
A., Callaghan, C. S. Q., Catchpole, R., Hopkins, J., Kunin, W. E., Latham, J.,
Opdam, P., Stoneman, R., Stroud, D. A. & Tratt, R. (2006) The ecological
9. References
130
effectiveness of protected areas: the United Kingdom. Biological
Conservation, 132, 76-87.
Gaston, K. J., Jackson, S. E., Nagy, A., Cantú-Salazar, L. & Johnson, M. (2008)
Protected areas in Europe - principle and practice. Annals of the New York
Academy of Sciences, 1134, 97-119.
Gee, J. H. R., Smith, B. D., Lee, K. M. & Griffiths, S. W. (1997) The ecological
basis of freshwater pond management for biodiversity. Pond Conservation
Symposium, 91-104.
Gollmann, G., Szymura, J. M., Arntzen, J. W. & Piálek, J. (1997) Bombina variegata
(Linneaus, 1758). In: Atlas of amphibians and reptiles in Europe. Eds.: J.-P.
Gasc, A. Cabela, J. Crnobrnja-Isailovic, D. Dolmen, K. Grossenbacher, P.
Haffner, J. Lescure, H. Martens, Martínez Rica, J. P., H. Maurin, M. E.
Oliviera, T. S. Sofianidou, M. Veith & A. Zuiderwijk, pp. 98-99. Societas
Europaea Herpetologica and Muséum National d'Histoire Naturelle, Paris.
Gooch, M. M., Heupel, A. M., Price, S. J. & Dorcas, M. E. (2006) The effects of
survey protocol on detection probabilities and site occupancy estimates of
summer breeding anurans. Applied Herpetology, 3, 129-142.
Goodman, D. (1987) The demography of chance extinction. In: Viable populations
for conservation. Ed.: M. E. Soulé, pp. 11-34. Cambridge University Press,
Cambridge.
Grafe, T. U. & Meuche, I. (2005) Chorus tenure and estimates of population size of
male European tree frogs Hyla arborea: implications for conservation.
Amphibia-Reptilia, 26, 437-444.
Grand-Duché de Luxembourg (2004) Mémorial A n° 10 du 29.01.2004. Protection
de la Nature et des Ressources Naturelles. Available at:
www.legilux.public.lu/leg/a/archives/2004/0010/a010.pdf#page=2 (accessed
on 29th September 2009).
Green, D. M. (2003) The ecology of extinction: population fluctuation and decline in
amphibians. Biological Conservation, 111, 331-343.
Grelle, C. E. V. (2002) Is higher-taxon analysis an useful surrogate of species
richness in studies of Neotropical mammal diversity? Biological
Conservation, 108, 101-106.
9. References
131
Griffiths, R. A. (1985) A simple funnel trap for studying newt populations and an
evaluation of trap behavior in smooth and palmate newts, Triturus vulgaris
and Triturus helveticus. Herpetological Journal, 1, 5-10.
Griffiths, R. A. (1987) Microhabitat and seasonal niche dynamics of smooth and
palmate newts, Triturus vulgaris and Triturus helveticus, at a pond in mid-
Wales. Journal of Animal Ecology, 56, 441-451.
Griffiths, R. A. & Mylotte, V. J. (1987) Microhabitat selection and feeding relations
of smooth and warty newts, Triturus vulgaris and Triturus cristatus, at an
upland pond in mid-Wales. Holarctic Ecology, 10, 1-7.
Griffiths, R. A. & Pavajeau, L. (2008) Captive breeding, reintroduction, and the
conservation of amphibians. Conservation Biology, 22, 852-861.
Griffiths, R. A. & Raper, S. J. (1994) A review of current techniques for sampling
amphibian communities. JNCC Report No.210. pp. 26. Joint Nature
Conservation Committee, Peterborough.
Gu, W. D. & Swihart, R. K. (2004) Absent or undetected? Effects of non-detection
of species occurrence on wildlife-habitat models. Biological Conservation,
116, 195-203.
Guest, J. & Harmer, A. (2006) Atlas of the Amphibians of Cheshire and Wirral.
Nepa Books.
Gunzburger, M. S. & Travis, J. (2005) Critical literature review of the evidence for
unpalatability of amphibian eggs and larvae. Journal of Herpetology, 39,
547–571.
Gustafson, D. H., Pettersson, C. J. & Malmgren, J. C. (2006) Great crested newts
(Triturus cristatus) as indicators of aquatic plant diversity. Herpetological
Journal, 16, 347-352.
Halley, J. M., Oldham, R. S. & Arntzen, J. W. (1996) Predicting the persistence of
amphibian populations with the help of a spatial model. Journal of Applied
Ecology, 33, 455-470.
Hamer, A. J. & McDonnell, M. J. (2008) Amphibian ecology and conservation in the
urbanising world: a review. Biological Conservation, 141, 2432-2449.
Hannah, L., Midgley, G., Andelman, S., Araujo, M., Hughes, G., Martinez-Meyer,
E., Pearson, R. & Williams, P. (2007) Protected area needs in a changing
climate. Frontiers in Ecology and the Environment, 5, 131-138.
9. References
132
Hansen, A. J. & DeFries, R. (2007) Ecological mechanisms linking protected areas
to surrounding lands. Ecological Applications, 17, 974-988.
Harrop, S. R. (1999) Conservation regulation: a backward step for biodiversity?
Biodiversity and Conservation, 8, 679-707.
Hartel, T., Nemes, S., Cogalniceanu, D., Ollerer, K., Schweiger, O., Moga, C. I. &
Demeter, L. (2007) The effect of fish and aquatic habitat complexity on
amphibians. Hydrobiologia, 583, 173-182.
Haslam, S. M., Sinker, C. A. & Wolseley, P. A. (1982) British Water Plants. Field
Studies Council. Reprinted from Field Studies 4(2): 243-351, 1975,
Shrewsbury.
Haslett, J. R., Berry, P. M., Bela, G., Jongman, R. H. G., Pataki, G., Samways, M. J.
& Zobel, M. (2010) Changing conservation strategies in Europe: a
framework integrating ecosystem services and dynamics. Biodiversity and
Conservation, 19, 2963-2977.
Haslett, J. R., Berry, P. M. & Zobel, M. (2008) European habitat management
strategies for conservation: current regulations and practices with reference
to dynamic ecosystems and ecosystem service provision. The RUBICODE
Project - Rationalising Biodiversity Conservation in Dynamic Ecosystems.
Electronic article available at:
www.rubicode.net/rubicode/RUBICODE_Review_on_Habitat_Management.
pdf (accessed on 15th August 2010).
Hawkes, H. A. (1998) Origin and development of the Biological Monitoring
Working Party score system. Water Research, 32, 964-968.
Hels, T. & Buchwald, E. (2001) The effect of road kills on amphibian populations.
Biological Conservation, 99, 331-340.
Heyer, W. R., Donnelly, M. A., McDiarmid, R. W., Hayak, L.-A. C. & Foster, M. S.
(1994) Measuring and Monitoring Biological Diversity: Standard Methods
for Amphibians. Smithsonian Institute Press, Washington.
Hines, J. E. (2006) PRESENCE2- Software to estimate patch occupancy and related
parameters. USGS-PWRC. Available at: www.mbr-
pwrc.usgs.gov/software/presence.html.
9. References
133
Holenweg Peter, A.-K. (2001) Dispersal rates and distances in adult water frogs,
Rana lessonae, R. ridibunda and their hybridogenetic associate R. esculenta.
Herpetologica, 57, 449-460.
Houlahan, J. E. & Findlay, C. S. (2003) The effects of adjacent land use on wetland
amphibian species richness and community composition. Canadian Journal
of Fisheries and Aquatic Sciences, 60, 1078-1094.
Houlahan, J. E., Findlay, C. S., Meyer, A. H., Kuzmin, S. L. & Schmidt, B. R.
(2001) Global amphibian population declines. Nature, 412, 500-500.
Houlahan, J. E., Findlay, C. S., Schmidt, B. R., Meyer, A. H. & Kuzmin, S. L.
(2000) Quantitative evidence for global amphibian population declines.
Nature, 404, 752-755.
Howard, S. (2002) A guide to monitoring the ecological quality of ponds and canals
using PSYM. Environment Agency and Pond Action, Solihull and Oxford.
Hubbard, C. E. & Hubbard, J. C. E. (1984) Grasses: a guide to their structure,
identification, uses and distribution in the British Isles. 3rd edition. Penguin
Books Ltd., London.
Indermuehle, N., Angélibert, S. & Oertli, B. (2008) IBEM: Indice de Biodiversité
des Etangs et Mares. Manuel d‟utilisation. pp. 33. Ecole d‟Ingénieurs HES de
Lullier, Genève.
IUCN (2001) IUCN Red list categories and criteria: version 3.1. IUCN Species
Survival Commission. IUCN, Gland and Cambridge.
IUCN (2006) Mediterranean Red List of Threatened Species - Reptiles and
Amphibians. IUCN. Available at:
www.uicnmed.org/web2007/cd_rep_amp/special_list_amp.html (accessed on
25th July 2010).
IUCN (2009) IUCN Red List of Threatened Species. Version 2009.1. IUCN.
Electronic database available at: www.iucnredlist.org (accessed on 5th May
2010).
IUCN, Conservation International & NatureServe (2008) An analysis of amphibians
on the 2008 IUCN Red List. Electronic database available at:
www.iucnredlist.org/amphibians (accessed on 5th January 2010).
9. References
134
IUCN and UNEP (2009) The World Database on Protected Areas (WDPA). UNEP-
WCMC, Cambridge, UK. Available at:
www.wdpa.org/MultiResult.aspx?Country=128 (accessed on 9th April 2010).
Jackson, S. F. & Gaston, K. J. (2008) Land use change and the dependence of
national priority species on protected areas. Global Change Biology, 14,
2132-2138.
Janse, J. H. & Van Puijenbroek, P. J. T. M. (1998) Effects of eutrophication in
drainage ditches. Environmental Pollution, 102, 547-552.
Jehle, R., Bouma, P., Sztatecsny, M. & Arntzen, J. W. (2000) High aquatic niche
overlap in the newts Triturus cristatus and T. marmoratus (Amphibia,
Urodela). Hydrobiologia, 437, 149-155.
Joint Nature Conservation Committee (2010) Council Directive 92/43/EEC on the
Conservation of natural habitats and of wild fauna and flora. Available at:
www.jncc.gov.uk/page-1374 (accessed on 10th June 2010).
Joly, P., Miaud, C., Lehmann, A. & Grolet, O. (2001) Habitat matrix effects on pond
occupancy in newts. Conservation Biology, 15, 239-248.
Jongman, R. H. G., Bela, G., Pataki, G., Scholten, L., Mérő, Á. & Mertens, C. (2008)
The effectiveness and appropriateness of existing conservation policies and
their integration into other policy sectors. The RUBICODE Project -
Rationalising Biodiversity Conservation in Dynamic Ecosystems. Electronic
article available at:
www.rubicode.net/rubicode/RUBICODE_Report_on_Conservation_Policy.p
df (accessed on 15th August 2010).
Junck, C., Schoos, F. & Proess, R. (2003) Bufo calamita, Laurenti, 1768. In:
Ferrantia 37: Verbreitungsatlas der Amphibien des Großherzogtums
Luxemburg. Ed.: R. Proess. Musée National d'Histoire Naturelle,
Luxembourg.
Kaiser, K. (2008) Evaluation of a long-term amphibian monitoring protocol in
Central America. Journal of Herpetology, 42, 104-110.
KARCH (2002) Cartes de répartition des amphibiens. Available at:
www.karch.ch/karch/f/amp/verbka/fs3.html (accessed on 5th June 2010).
9. References
135
Kats, L. B. & Ferrer, R. P. (2003) Alien predators and amphibian declines: review of
two decades of science and the transition to conservation. Diversity and
Distributions, 9, 99-110.
Kiesecker, J., Blaustein, A. & Belden, L. (2001) Complex causes of amphibian
population declines. Nature, 410, 681-684.
Kiesecker, J. M. & Blaustein, A. R. (1995) Synergism between UV-B radiation and a
pathogen magnifies amphibian embryo mortality in nature. Proceedings of
the National Academy of Sciences of the United States of America, 92, 11049.
Kirlin, M. S., Gooch, M. M., Price, S. J. & Dorcas, M. E. (2006) Predictors of winter
anuran calling activity in the north Carolina Piedmont. Journal of the North
Carolina Academy of Science, 122, 10-18.
Knutson, M. G., Richardson, W. B., Reineke, D. M., Gray, B. R., Parmelee, J. R. &
Weick, S. E. (2004) Agricultural ponds support amphibian populations.
Ecological Applications, 14, 669-684.
Kwet, A. (2009) European Reptile and Amphibian Guide. New Holland, London.
Landres, P. B., Verner, J. & Thomas, J. W. (1988) Ecological uses of vertebrate
indicator species - a critique. Conservation Biology, 2, 316-328.
Langdon, S. J., Marrs, R. H., Hosie, C. A., McAllister, H. A., Norris, K. M. & Potter,
J. A. (2004) Crassula helmsii in UK ponds: effects on plant biodiversity and
implications for newt conservation. Weed Technology, 18, 1349-1352.
Lebboroni, M., Ricchiardino, G., Bellavita, M. & Chelazzi, G. (2006) Potential use
of anurans as indicators of biological quality in upstrearns of central Italy.
Amphibia-Reptilia, 27, 73-79.
Lips, K. R. (1998) Decline of a tropical montane amphibian fauna. Conservation
Biology., 12, 106-117.
Lips, K. R., Brem, F., Brenes, R., Reeve, J. D., Alford, R. A., Voyles, J., Carey, C.,
Livo, L., Pessier, A. P. & Collins, J. P. (2006) Emerging infectious disease
and the loss of biodiversity in a Neotropical amphibian community.
Proceedings of the National Academy of Science of USA, 102, 3165-3170.
Long, L., Saylor, L. & Soulé, M. (1995) A pH/UV-B synergism in amphibians.
Conservation Biology, 9, 1301-1303.
9. References
136
Loyola, R. D., Kubota, U. & Lewinsohn, T. M. (2007) Endemic vertebrates are the
most effective surrogates for identifying conservation priorities among
Brazilian ecoregions. Diversity and Distributions, 13, 389-396.
Luoto, M., Virkkala, R. & Heikkinen, R. K. (2007) The role of land cover in
bioclimatic models depends on spatial resolution. Global Ecology and
Biogeography, 16, 34-42.
MacKenzie, D. I. (2005) What are the issues with presence-absence data for wildlife
managers? Journal of Wildlife Management, 69, 849-860.
MacKenzie, D. I., Nichols, J. D., Lachman, G. B., Droege, S., Royle, J. A. &
Langtimm, C. A. (2002) Estimating site occupancy rates when detection
probabilities are less than one. Ecology, 83, 2248-2255.
MacKenzie, D. I., Nichols, J. D., Sutton, N., Kawanishi, K. & Bailey, L. L. (2005)
Improving inferences in population studies of rare species that are detected
imperfectly. Ecology, 86, 1101-1113.
Maletzky, A., Kyek, M. & Goldschmid, A. (2007) Monitoring status, habitat features
and amphibian species richness of Crested newt (Triturus cristatus
superspecies) ponds at the edge of the species range (Salzburg, Austria).
Annales De Limnologie-International Journal of Limnology, 43, 107-115.
Marco, A., Lizana, M., Alvarez, A. & Blaustein, A. R. (2001) Egg-wrapping
behaviour protects newt embryos from UV radiation. Animal Behaviour, 61,
639-644.
Margules, C. R. & Pressey, R. L. (2000) Systematic conservation planning. Nature,
405, 243-253.
Marnell, F. (1998) Discriminant analysis of the terrestrial and aquatic habitat
determinants of the smooth newt (Triturus vulgaris) and the common frog
(Rana temporaria) in Ireland. Journal of Zoology, 244, 1-6.
Marquez, R. (1992) Terrestrial paternal care and short breeding seasons:
reproductive phenology of the midwife toads Alytes obstetricans and A.
cisternasii. Ecography, 15, 279-288.
Marsh, D. M. (2000) Variable responses to rainfall by breeding Tungara frogs.
Copeia, 2000, 1104-1108.
Marsh, D. M. & Trenham, P. C. (2001) Metapopulation dynamics and amphibian
conservation. Conservation Biology, 15, 40-49.
9. References
137
Martínez Rica, J. P. (1997) General features of European climate and vegetation.
Consequences on the distribution of herpetofauna. In: Atlas of amphibians
and reptiles in Europe. Eds.: J.-P. Gasc, A. Cabela, J. Crnobrnja-Isailovic, D.
Dolmen, K. Grossenbacher, P. Haffner, J. Lescure, H. Martens, Martínez
Rica, J. P., H. Maurin, M. E. Oliviera, T. S. Sofianidou, M. Veith & A.
Zuiderwijk, pp. 17-22. Societas Europaea Herpetologica and Muséum
National d'Histoire Naturelle, Paris.
Mazerolle, M. J. (2006) Improving data analysis in herpetology: using Akaike's
Information Criterion (AIC) to assess the strength of biological hypotheses.
Amphibia-Reptilia, 27, 169-180.
Mazerolle, M. J., Bailey, L. L., Kendall, W. L., Royle, J. A., Converse, S. J. &
Nichols, J. D. (2007) Making great leaps forward: accounting for
detectability in herpetological field studies. Journal of Herpetology, 41, 672-
689.
Memmott, J., Cadotte, M., Hulme, P. E., Kerby, G., Milner-Gulland, E. J. &
Whittingham, M. J. (2010) Putting applied ecology into practice. Journal of
Applied Ecology, 47, 1-4.
Menetrey, N., Sager, L., Oertli, B. & Lachavanne, J. B. (2004) Looking for metrics
to assess the trophic state of ponds. Macroinvertebrates and amphibians.
Meeting on Conservation and Monitoring of Pond Biodiversity, Lullier
Geneva, Swizterland. pp.653-664.
Miaud, C. (1993) Predation on newt eggs (Triturus alpestris and T. helveticus):
identification of predators and protective role of oviposition behaviour.
Journal of Zoology, 231, 575-582.
Miaud, C. (1994) Role of wrapping behavior on egg survival in three species of
Triturus (Amphibia, Urodela). Copeia, 535-537.
Milner-Gulland, E. J., Fisher, M., Browne, S., Redford, K. H., Spencer, M. &
Sutherland, W. J. (2010) Do we need to develop a more relevant conservation
literature? Oryx, 44, 1-2.
Ministère de l'Environnement (2003) Le réseau écologique "Natura 2000" : sa
préparation au Luxembourg. Le Gouvernement du Grand-Duché du
Luxembourg. Available at:
9. References
138
www.environnement.public.lu/conserv_nature/travaux/natura2000/index.htm
l (accessed on 9th April 2010).
Ministère de l'Environnement (2007) Plan National pour la Protection de la Nature
(PNPN) - rapport final (avant-projet). Ministère de l'Environnement and
Administration des Eaux et Forêts, Luxembourg. Available at:
www.environnement.public.lu/conserv_nature/dossiers/PNPN/PNPNvfinale2
00407-2.pdf (accessed on 10th August 2010).
Ministerio de Medio Ambiente y Medio Rural y Marino (2008) Atlas y Libro Rojo
de los Anfibios y Reptiles de España. Available at:
www.mma.es/portal/secciones/biodiversidad/inventarios/inb/anfibios_reptile
s/indice.htm (accessed on 5th June 2010).
MNHNL - Groupe herpétologique (2009a) Plan d‟action: Rainette arboricole (Hyla
arborea). Plan national pour la protection de la nature (PNPN) - Plans
d‟actions espèces, Le Gouvernement du Grand-Duché de Luxembourg.
Electronic article available at:
www.environnement.public.lu/conserv_nature/dossiers/Plans_d_actions/PAE
_Hyla_arborea.pdf (accessed on 5th August 2010).
MNHNL - Groupe herpétologique (2009b) Plan d‟action: Triton crêté (Triturus
cristatus). Plan national pour la protection de la nature (PNPN) - Plans
d‟actions espèces, Le Gouvernement du Grand-Duché de Luxembourg.
Electronic article available at:
www.environnement.public.lu/conserv_nature/dossiers/Plans_d_actions/PAE
_Triturus_cristatus.pdf (accessed on 5th August 2010).
Moilanen, A. (2002) Implications of empirical data quality to metapopulation model
parameter estimation and application. Oikos, 96, 516-530.
Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B. & Kent, J.
(2000) Biodiversity hotspots for conservation priorities. Nature, 403, 853-
858.
NARRS (2011) National Amphibian Survey. Available at:
www.narrs.org.uk/natamphibsurvey.htm (accessed on 5th January 2011).
Navas, C. A. (1996) The effect of temperature on the vocal activity of tropical
anurans: a comparison of high and low-elevation species. Journal of
Herpetology, 30, 488-497.
9. References
139
Navas, C. A. & Bevier, C. R. (2001) Thermal dependency of calling performance in
the eurythermic frog Colostethus subpunctatus. Herpetologica, 57, 384-395.
Nicolet, P., Biggs, J., Fox, G., Hodson, M. J., Reynolds, C., Whitfield, M. &
Williams, P. (2004) The wetland plant and macroinvertebrate assemblages of
temporary ponds in England and Wales. Biological Conservation, 120, 261-
278.
Norris, K. M. & Hosie, C. A. (2005) A quantified ethogram for oviposition in
Triturus newts: description and comparison of T. helveticus and T. vulgaris.
Ethology, 111, 357-366.
O‟Donnell, R. P., Quinn, T., Hayes, M. P. & Ryding, K. E. (2007) Comparison of
three methods for surveying amphibians in forested seep habitats in
Washington. A report submitted to the Cooperative Monitoring, Evaluation,
and Research Committee, Available at:
www.dnr.wa.gov/Publications/fp_cmer_04_402.pdf (accessed on 28th May
2010).
Oertli, B., Auderset-Joye, D., Castella, E., Juge, R., Lehmann, A. & Lachavanne, J.
B. (2005) PLOCH: a standardized method for sampling and assessing the
biodiversity in ponds. Aquatic Conservation-Marine and Freshwater
Ecosystems, 15, 665-679.
Oertli, B., Auderset Joye, D., Castella, E., Juge, R., Cambin, D. & Lachavanne, J. B.
(2002) Does size matter? The relationship between pond area and
biodiversity. Biological Conservation, 104, 59-70.
Olsen, L.-H., Sunesen, J. & Pedersen, B. V. (2001) Small Freshwater Creatures.
Oxford University Press, New York.
Orizaola, G. & Brana, F. (2003) Oviposition behaviour and vulnerability of eggs to
predation in four newt species (genus Triturus). Herpetological Journal, 13,
121-124.
Orłowski, G. (2007) Spatial distribution and seasonal pattern in road mortality of the
common toad Bufo bufo in an agricultural landscape of south-western Poland.
Amphibia-Reptilia, 28, 25-31.
Oseen, K. L. & Wassersug, R. J. (2002) Environmental factors influencing calling in
sympatric anurans. Oecologia, 133, 616-625.
9. References
140
Papageorgiou, K. & Vogiatzakis, I. N. (2006) Nature protection in Greece: an
appraisal of the factors shaping integrative conservation and policy
effectiveness. Environmental Science & Policy, 9, 476-486.
Parmesan, C. & Yohe, G. (2003) A globally coherent fingerprint of climate change
impacts across natural systems. Nature, 421, 37-42.
Paszkowski, C. A., Scrimgeour, G., Gingras, B. A. & Kendall, S. (2002) A
comparison of techniques for assessing amphibian assemblages on streams in
the western boreal forest. Canadian Field-Naturalist, 116, 116-119.
Pavignano, I., Giacoma, C. & Castellano, S. (1990) A multivariate analysis of
amphibian habitat determinants in north western Italy. Amphibia-Reptilia, 11,
311-324.
Pearson, D. L. (1994) Selecting indicator taxa for the quantitative assessment of
biodiversity. Philosophical Transactions of the Royal Society of London
Series B-Biological Sciences, 345, 75-79.
Pellet, J. & Schmidt, B. R. (2005) Monitoring distributions using call surveys:
estimating site occupancy, detection probabilities and inferring absence.
Biological Conservation, 123, 27-35.
Phillips, S. J., Anderson, R. P. & Schapire, R. E. (2006) Maximum entropy modeling
of species geographic distributions. Ecological Modelling, 190, 231-259.
Picco, A. M. & Collins, J. P. (2008) Amphibian commerce as a likely source of
pathogen pollution. Conservation Biology, 22, 1582-1589.
Pierce, B. A. & Gutzwiller, K. J. (2004) Auditory sampling of frogs: detection
efficiency in relation to survey duration. Journal of Herpetology, 38, 495-
500.
Pimm, S., Raven, P., Peterson, A., Şekercioğlu, Ç. H. & Ehrlich, P. R. (2006)
Human impacts on the rates of recent, present, and future bird extinctions.
Proceedings of the National Academy of Sciences, 103, 10941-10946.
Pimm, S., Russell, G., Gittleman, J. & Brooks, T. (1995) The future of biodiversity.
Science, 269, 347.
Pope, S., Fahrig, L. & Merriam, H. (2000) Landscape complementation and
metapopulation effects on leopard frog populations. Ecology, 81, 2498-2508.
Pounds, J. A. (2001) Climate and amphibian declines. Nature, 410, 639-640.
9. References
141
Pounds, J. A., Fogden, M. P. L. & Campbell, J. H. (1999) Biological response to
climate change on a tropical mountain. Nature, 398, 611-615.
Proess, R. (2003a) Die Amphibienwanderung im oberen Abschnitt der Strae Bridel-
Steinsel (Luxemburg). Bulletin de la Société des naturalistes
luxembourgeois, 103, 93-100.
Proess, R. (2003b) Verbreitungsatlas der Amphibien des Großherzogtums
Luxemburg. Musée National d'Histoire Naturelle, Luxembourg.
Proess, R. (2009) Plan d‟action: Crapaud calamite (Bufo calamita). Plan national
pour la protection de la nature (PNPN) - Plans d‟actions espèces, Le
Gouvernement du Grand-Duché de Luxembourg. Electronic article available
at:
www.environnement.public.lu/conserv_nature/dossiers/Plans_d_actions/PAE
_Bufo_calamita.pdf (accessed on 5th August 2010).
Qian, H. (2007) Relationships between plant and animal species richness at a
regional scale in China. Conservation Biology, 21, 937-944.
Quinn, T., Hayes, M. P., Dugger, D. J., Hicks, T. L. & Hoffmann, A. (2007)
Comparison of two techniques for surveying headwater stream amphibians.
Journal of Wildlife Management, 71, 282-288.
RAVON (2008) Levende Atlas. Available at:
www.ravon.nl/Soorten/Levendeatlas/tabid/325/Default.aspx (accessed on 5th
June 2010).
Reading, C. J. (1998) The effect of winter temperatures on the timing of breeding
activity in the common toad Bufo bufo. Oecologia, 117, 469-475.
Rice, W. R. (1989) Analyzing tables of statistical tests. Evolution, 43, 223-225.
Rinehart, K. A., Donovan, T. M., Mitchell, B. R. & Long, R. A. (2009) Factors
influencing occupancy patterns of Eastern Newts across Vermont. Journal of
Herpetology, 43, 521-531.
Root, T. L., Price, J. T., Hall, K. R., Schneider, S. H., Rosenzweig, C. & Pounds, J.
A. (2003) Fingerprints of global warming on wild animals and plants.
Nature, 421, 57-60.
Rose, F. (1989) Colour identification guide to the grasses, sedges, rushes and ferns
of the British Isles and north-western Europe. Penguin Group, London.
9. References
142
Rose, F. (2006) The Wild Flower Key: how to identify wild flowers, trees and shrubs
in Britain and Ireland. 2nd Edition, revised and updated by C. O‟Reilly.
edition. Frederick Warne, Penguin Books Ltd., London.
Rouse, J., Bishop, C. & Struger, J. (1999) Nitrogen pollution: an assessment of its
threat to amphibian survival. Environmental Health Perspectives, 107, 799-
803.
Ryan, T. J., Philippi, T., Leiden, Y. A., Dorcas, M. E., Wigley, T. B. & Gibbons, J.
W. (2002) Monitoring herpetofauna in a managed forest landscape: effects of
habitat types and census techniques. Forest Ecology and Management, 167,
83-90.
Sahlén, G. & Ekestubbe, K. (2001) Identification of dragonflies (Odonata) as
indicators of general species richness in boreal forest lakes. Biodiversity and
Conservation, 10, 673-690.
Samways, M. J., Bazelet, C. S. & Pryke, J. S. (2010) Provision of ecosystem services
by large scale corridors and ecological networks. Biodiversity and
Conservation, 19, 2949-2962.
Scheffer, M., Szabo, S., Gragnani, A., van Nes, E. H., Rinaldi, S., Kautsky, N.,
Norberg, J., Roijackers, R. M. M. & Franken, R. J. M. (2003) Floating plant
dominance as a stable state. Proceedings of the National Academy of
Sciences of the United States of America, 100, 4040-4045.
Schmidt, B. R. & Zumbach, S. (2008) Amphibian road mortality and how to prevent
it: a review. In: Urban Herpetology. Eds.: J. C. Mitchell, R. E. Jung Brown &
B. Bartolomew, pp. 131-167. Society for the Study of Amphibians and
Reptiles.
Schoorl, J. & Zuiderwijk, A. (1980) Ecological Isolation in Triturus cristatus and
Triturus marmoratus (Amphibia: Salamandridae). Amphibia-Reptilia, 1, 235-
252.
Semlitsch, R. D. (2000) Principles for management of aquatic-breeding amphibians.
Journal of Wildlife Management, 64, 615-631.
Sewell, D. (2006) Great crested newts (Triturus cristatus) as indicators of pond
biodiversity. PhD thesis. The Durrell Institute of Conservation and Ecology,
The University of Kent, Canterbury.
9. References
143
Sewell, D., Beebee, T. J. C. & Griffiths, R. A. (2010) Optimising biodiversity
assessments by volunteers: the application of occupancy modelling to large-
scale amphibian surveys. Biological Conservation, 143, 2102-2110.
Sewell, D. & Griffiths, R. A. (2009) Can a single amphibian species be a good
biodiversity indicator? Diversity, 1, 102-117.
Shirose, L. J., Bishop, C. A., Green, D. M., MacDonald, C. J., Brooks, R. J. &
Helferty, N. J. (1997) Validation tests of an amphibian call count survey
technique in Ontario, Canada. Herpetologica, 53, 312-320.
Sinsch, U. (1984) Thermal influences on the habitat preference and the diurnal
activity in three European Rana species. Oecologia, 64, 125-131.
Skelly, D. K., Yurewicz, K. L., Werner, E. E. & Relyea, R. A. (2003) Estimating
decline and distributional change in amphibians. Conservation Biology, 17,
744-751.
Societas Herpetologica Italica (2009) Atlante degli Anfibi e dei Rettili d'Italia.
Available at: www.3.unipv.it/webshi/atlas/atlas.htm (accessed on 5th June
2010).
Stace, C. (1997) New flora of the British Isles. 2nd edition. Cambridge University
Press, Cambridge.
Stebbins, R. C. & Cohen, N. W. (1995) A Natural History of Amphibians. Princeton
University Press, Chichester.
Steen, H. (1958) Determinations of the solubility of oxygen in pure water.
Limnology and Oceanography, 3, 423-426.
Stuart, S. N., Chanson, J. S., Cox, N. A., Young, B. E., Rodrigues, A. S. L.,
Fischman, D. L. & Waller, R. W. (2004) Status and trends of amphibian
declines and extinctions worldwide. Science, 306, 1783-1786.
Stumpel, A. (1997) Hyla arborea (Linneaus, 1758). In: Atlas of amphibians and
reptiles in Europe. Eds.: J.-P. Gasc, A. Cabela, J. Crnobrnja-Isailovic, D.
Dolmen, K. Grossenbacher, P. Haffner, J. Lescure, H. Martens, Martínez
Rica, J. P., H. Maurin, M. E. Oliviera, T. S. Sofianidou, M. Veith & A.
Zuiderwijk, pp. 124-125. Societas Europaea Herpetologica and Muséum
National d'Histoire Naturelle, Paris.
9. References
144
Swihart, R. K., Gehring, T. M., Kolozsvary, M. B. & Nupp, T. E. (2003) Responses
of 'resistant' vertebrates to habitat loss and fragmentation: the importance of
niche breadth and range boundaries. Diversity and Distributions, 9, 1-18.
Tachet, H., Richoux, P., Bournaud, M. & Usseglio-Polatera, P. (2000) Invertébrés
d'eau douce : Systématique, biologie, écologie. 4th edition. CNRS Editions,
Paris.
Tarr, T. L. & Babbitt, K. J. (2002) Effects of habitat complexity and predator identity
on predation of Rana clamitans larvae. Amphibia-Reptilia, 23, 13-20.
Telfer, M. G., Preston, C. D. & Rothery, P. (2002) A general method for measuring
relative change in range size from biological atlas data. Biological
Conservation, 107, 99-109.
Thomas, C. D., Bulman, C. R. & Wilson, R. J. (2008) Where within a geographical
range do species survive best? A matter of scale. Insect Conservation and
Diversity, 1, 2-8.
Thomas, C. D., Cameron, A., Green, R. E., Bakkenes, M., Beaumont, L. J.,
Collingham, Y. C., Erasmus, B. F. N., de Siqueira, M. F., Grainger, A.,
Hannah, L., Hughes, L., Huntley, B., van Jaarsveld, A. S., Midgley, G. F.,
Miles, L., Ortega-Huerta, M. A., Peterson, A. T., Phillips, O. L. & Williams,
S. E. (2004a) Extinction risk from climate change. Nature, 427, 145-148.
Thomas, J. A., Telfer, M. G., Roy, D. B., Preston, C. D., Greenwood, J. J. D., Asher,
J., Fox, R., Clarke, R. T. & Lawton, J. H. (2004b) Comparative losses of
British butterflies, birds, and plants and the global extinction crisis. Science,
303, 1879-1881.
Tilman, D. (1986) A consumer-resource approach to community structure. American
Zoologist, 26, 5-22.
Triantis, K. A., Mylonas, M., Lika, K. & Vardinoyannis, K. (2003) A model for the
species-area-habitat relationship. Journal of Biogeography, 30, 19-27.
Unwin, D. M. (1984) A key to the families of British beetles (reprinted 1988, with
minor amendments). Field Studies Council, Shrewsbury.
Van Buskirk, J. (2005) Local and landscape influence on amphibian occurrence and
abundance. Ecology, 86, 1936-1947.
Vitt, L. J. & Caldwell, J. P. (2009) Herpetology. 3rd edition. Academic Press,
Elsevier Inc., London.
9. References
145
Wallace, I. (2006) Simple key to Caddis larvae. 1st edition. Field Studies Council,
Shrewsbury.
Warkentin, I. G., Bickford, D., Sodhi, N. S. & Bradshaw, C. J. A. (2009) Eating
frogs to extinction. Conservation Biology, 23, 1056-1059.
Weir, L. A., Royle, J. A., Nanjappa, P. & Jung, R. E. (2005) Modeling anuran
detection and site occupancy on North American Amphibian Monitoring
Program (NAAMP) routes in Maryland. Journal of Herpetology, 39, 627-
639.
Wilkinson, J. (2007) Ecology and conservation of the European common toad (Bufo
bufo) in Jersey, British Channel Islands. PhD thesis. The Durrell Institute of
Conservation and Ecology, University of Kent, Canterbury, UK.
Williams, P., Whitfield, M. & Biggs, J. (2008) How can we make new ponds
biodiverse? A case study monitored over 7 years. Hydrobiologia, 597, 137-
148.
Williams, P., Whitfield, M., Biggs, J., Bray, S., Fox, G., Nicolet, P. & Sear, D.
(2003) Comparative biodiversity of rivers, streams, ditches and ponds in an
agricultural landscape in Southern England. Biological Conservation, 115,
329-341.
Williams, P. H. & Gaston, K. J. (1994) Measuring more of biodiversity - can higher
taxon richness predict wholesale species richness? Biological Conservation,
67, 211-217.
Wolff, F. (2009) Quatrième rapport national de la Convention de la diversité
biologique du Grand-Duché de Luxembourg. Le Gouvernement du Grand-
Duché de Luxembourg, Ministère du Développement durable et des
Infrastructures, Available at: www.cbd.int/doc/world/lu/lu-nr-04-fr.pdf
(accessed on 7th April 2010).
Wood, L. R., Griffiths, R. A. & Schley, L. (2009) Amphibian chytridiomycosis in
Luxembourg. Bulletin de la Société des Naturalistes Luxembourgois, 110,
109-114.
Wood, P. J., Greenwood, M. T. & Agnew, M. D. (2003) Pond biodiversity and
habitat loss in the UK. Area, 35, 206-216.
Zanini, F., Klingemann, A., Schlaepfer, R. & Schmidt, B. R. (2008) Landscape
effects on anuran pond occupancy in an agricultural countryside: barrier-
9. References
146
based buffers predict distributions better than circular buffers. Canadian
Journal of Zoology - Revue Canadienne De Zoologie, 86, 692-699.
Zanini, F., Pellet, J. & Schmidt, B. R. (2009) The transferability of distribution
models across regions: an amphibian case study. Diversity and Distributions,
15, 469-480.
Zhao, S. Q., Fang, J. Y., Peng, C. H. & Tang, Z. Y. (2006) Relationships between
species richness of vascular plants and terrestrial vertebrates in China:
analyses based on data of nature reserves. Diversity and Distributions, 12,
189-194.
ZipcodeZoo.com (2009) ZipcodeZoo Index to Animals. Available at:
www.zipcodezoo.com/Animals/Animalia.asp (accessed on 15th August
2009).
10. Appendices
10. Appendices
Appendix 1 – Amphibian nomenclature ........................................................... 148
Appendix 2 – Contributors to the MNHN database .......................................... 150
Appendix 3 – Study ponds ............................................................................... 151
Appendix 4 – Species dates .............................................................................. 152
Appendix 5 – Species distribution maps ........................................................... 153
Appendix 6 – Plants species surveyed .............................................................. 155
Appendix 7 – Macroinvertebrate families surveyed .......................................... 159
Appendix 8 – Occupancy modelling covariates ................................................ 161
Appendix 9 – Publications ............................................................................... 163
Appendix 1 – Amphibian nomenclature
148
Appendix 1 – Amphibian nomenclature
Twelve amphibian species in Luxembourg were studied here, species names
throughout follow Speybroeck et al. (2010), who are the most recent authority on
European herpetological nomenclature. In figures and tables species are often
referred to by their initials, which are also given below.
Speybroeck et al. (2010) recommend retaining the natterjack toad (Bufo calamita) in
the Bufo genus, concluding that there is not enough evidence yet to re-group it into
Epidalea as suggested by others.
Two green frog species are found in Luxembourg: Pelophylax lessonae and P. kl.
esculentus. P. kl. esculentus originally arose as a hybrid of P. lessonae and P.
ridibundus, but persists in the absence of one parent species by only producing
gametes with the genome of the absent parent and breeding with the available parent
species and (Holenweg Peter, 2001). P. kl. esculentus x P. kl. esculentus offspring
are not viable, necessitating the persistence of sexual parasitism on a parent species.
P. lessonae potentially occupies some sites without P. kl. esculentus, although their
individual distributions have not been studied in Luxembourg. Green frogs can be
distinguished in the field (Mutz, 2009), but this requires morphometric measures of
multiple individuals at every site – a time consuming task outside the scope of the
current study; they were grouped for all analyses and are collectively referred to as
Pelophylax spp. (or abbreviated to P.spp) throughout.
Appendix 1 – Amphibian nomenclature
149
Table 10.1 Nomenclature of study species (following Speybroeck et al., 2010).
Initials Order Family Genus Species
Anura (frogs and toads) Ranidae Rafinesque-Schmaltz, 1814 (true frogs) Rt
Rana Linnaeus, 1758 temporaria Linnaeus, 1758 – common/grass frog
Pelophylax Fitzinger, 1843
P.spp kl. esculentus (Linnaeus, 1758) – edible (green) frog lessonae (Camerano, 1882) – pool (green) frog
Ha
Hylidae Rafinesque, 1815 (tree frogs) Hyla Laurenti, 1768 arborea (Linnaeus, 1758) – common tree frog
Bb
Bufonidae Gray, 1825 (true toads) Bufo Laurenti, 1768 bufo (Linnaeus, 1758) – common toad
Bc calamita (Laurenti, 1768) – natterjack Ao
Alytidae Fitzinger, 1843 (painted frogs and midwife toads) Alytes Wagler, 1829 obstetricans (Laurenti, 1768) – common midwife toad
Bv
Bombinatoridae Gray, 1825 (fire-bellied toads) Bombina Oken, 1816 variegata (Linnaeus, 1758) – yellow-bellied toad
Caudata or Urodela (salamanders and newts) Salamandridae Goldfuss, 1820 (true salamanders and newts)
Ia
Ichthyosaura Sonnini and Latreille, 1801 alpestris (Laurenti, 1768) – Alpine newt
Tc
Triturus Rafinesque, 1815 cristatus (Laurenti, 1768) – great crested newt
Lv Lh
Lissotriton Bell, 1839 vulgaris (Linnaeus, 1758) – smooth newt helveticus (Razoumowsky, 1789) – palmate newt
Ss
Salamandra Garsault, 1764 salamandra (Linnaeus, 1758) – fire salamander
Appendix 1 References
Holenweg Peter, A.-K. (2001) Dispersal rates and distances in adult water frogs,
Rana lessonae, R. ridibunda and their hybridogenetic associate R. esculenta.
Herpetologica, 57, 449-460.
Mutz, T. (2009) Eine einfache Methode zur Bestimmung von Wasserfröschen
(Pelophylax sp.) im Freiland, vorgestellt am Beispiel einer Population im
Naturschutzgebiet >>Heiliges Meer<< bei Hopsten, Nordrhein-Westfalen.
Zeitschrift für Feldherpetologie, 16, 201-218.
Speybroeck, J., Beukema, W. & Crochet, P. A. (2010) A tentative species list of the
European herpetofauna (Amphibia and Reptilia) - an update. Zootaxa, 2492,
1-27.
Appendix 2 – Contributors to the MNHN database
150
Appendix 2 – Contributors to the MNHN database
1955-2005 - Musée national d‟histoire naturelle, 25 rue Münster, L-2160
Luxembourg.
Arendt, Alexandra
Backes, S.
Baghli, A.
Baum
Bechet, Georges
Braunert, C.
Carrières, Evelyne
Casagranda, Béatrice
Conzemius, T.
Cungs, Josy
Diederich, Jules
Dion, M.
Dolisy, D.
Dupont, R.
Engel, Edmée
Erpelding, Andre
Ewert, M.P.
Felten, C.
Felten, P.
Gerend, J.
Gerend, R.
Grégoire, Christian
Grof, Marc
Groh, K.
Haag, Guillaume
Heidt, C.
Hengemühle, M.
Hentgen
Hermes, Jean
Hoffelt, J.
Hoffmann, Jos
Isekim
Junck, Claudine
Jungers, P.
Kalmes, Pierre
Klopp, F.
Kraus, C.
Kraus, M.
Krippel, Yves
LANIUS
Lauff, Max
Manderscheid
Mangen, Jean-Marie
Meyer, Marc (MNHNL)
Origer, G.
Owaller, Marc
Pelles, A.
Pfeiffenschneider, Manou
Philippo, S.
Pir, Jacques
Poncelet, M.
Proess, Roland (Ecotop)
Pütz, F.
Reichling, Léopold
Reuter, C.
Risch, J.-P.
Risch, Stéphane
Roesgen
Scheer, Anne
Schley, Laurent
Schmidt, Gérard
Schmit, J.
Schmit, R.
Schmitz, C.
Schmitz, R.
Schneider, Nico
Schoentgen, F.
Schoos, Fernand
Schoos, R.
Schotel, Jan
Schrankel, Isabelle
Seiler, R.
Sowa, F.
Steichen, Claude
Stomp, N.
Strozyk, Ulrich
Thilmany, B.
Thonon, Philippe
Thorn, Robert
Weimerkirch, S.
Weis, A.
Weiss, J.
Willems, G
Appendix 3 – Study ponds
151
Appendix 3 – Study ponds
Table 10.2 Study ponds in 25 m and 200 m groupings.
Years
Ponds Grid Reference
Pond size (at start)
(m2) 07 08 ‘1’ 25 m 200 m
1 Weiden 1&2
Weiden 1-5
LC 648 926 486
2 Weiden 3&4 LC 648 924 1134
3 Weiden 5 LC 646 923 780
4 Laaschtert 1 Laaschtert 1&2
LC 722 934 128
5 Laaschtert 2 LC 722 932 150
6 Laaschtert 3 Laaschtert 3-5
LC 725 933 323
7 Laaschtert 4&5 LC 726 934 409
8 Pesch 1&2 Pesch 1&2 LC 744 801 3600
9 Flakewiss Flakewiss & Biergwiss
LC 800 870 270
10 Biergwiss LC 798 869 270
11 Jakobsbierg Jakobsbierg LC 955 913 2750
12 Scheedchen Scheedchen LC 964 900 225
13 Kléiwiss Kléiwiss & Briderfeld
LC 681 791 900
14 Briderfeld LC 681 793 200
15 Werwelslach 1&2 Werwelslach 1&2 LC 683 790 234
16 Loorlach Loorlach LC 693 796 221
17 Giele Botter Giele Botter LC 592 674 8000
18 Féitzemuer Féitzemuer LC 684 663 840
19 Laangwiss Laangwiss LC 713 640 1250
20 Lamidden Lamidden LC 715 630 270
21 Looreck Looreck LC 712 632 700
22 Dréisch Dréisch LC 708 635 700
23 Hongerbruch Hongerbruch LC 753 682 250
24 Véierhärebesch Véierhärebesch LC 749 683 108
25 Neiländigfeld Neiländigfeld LC 752 669 80
26 Leiwesdallchen Leiwesdallchen LC 745 668 225
27 Haard Haard LC 725 602 137
28 Bloklapp 1&2
Bloklapp 1-3
LC 754 600 693
29 Bloklapp 3 LC 754 600 1000
30 Stueden Stueden LC 763 650 1232
31 Weiergewan Weiergewan LC 900 670 6000
32 Stuppicht 1
Stuppicht 1&2
LC 800 890 1250
33 Stuppicht 2 LC 799 890 700
34 Steinfort Steinfort LC 610 813 3989
35 Renkert Merscheid Renkert Merscheid LA 754 1140 912
36 Alersang Putscheid Alersang Putscheid LA 781 1141 899
37 Biwelser Béral Biwelser Béral LA 800 1135 160
38 Vianden Teggelbaach Vianden Teggelbaach LA 833 1124 704
39 Hosingen Niklosboesch 1&2
Hosingen Niklosboesch 1-6
LA 764 1178 1496
40 Hosingen Niklosboesch 4-6 LA 750 1180 2029
41 Follebur 1
Follebur 1&2
LA 682 1244 680
42 Follebur 2 LA 680 1243 680
43 Lellingen Ortschaft Lellingen Ortschaft LA 689 1163 834
Plant and invertebrate chapters (5 and 6) group ponds less than 25 m apart; occupancy models (Chapter 7) group ponds within 200 m of each other; Gauss-Luxembourg grid
reference system.
Appendix 4 – Species dates
152
Appendix 4 – Species dates
Table 10.3 Survey dates in 2007.
Species First
detection in pond(s)
Last detection
in pond(s)
R. temporaria 25.02.07 3 13.06.07 11
Pelophylax spp. 07.03.07 1 07.08.07 11
H. arborea 19.04.07 37 25.06.07 37
Bufo bufo 06.03.07 24-27 11.06.07 24-27
Bufo calamita - - - -
A. obstetricans 16.04.07 11 07.08.07 11
Bombina variegata 26.04.07 32 16.07.08 32
I. alpestris 06.03.07 24-27 25.06.07 37
T. cristatus 07.03.07 1-5 25.06.07 37
L. helveticus 13.03.07 28-31 21.06.07 16
L. vulgaris 06.03.07 24-27 11.06.07 25
Table 10.4 Survey dates in 2008.
Species First
detection in pond(s)
Last detection
in pond(s)
R. temporaria 27.02.08 6-9 30.05.08 24-27
Pelophylax spp. 01.04.08 38-39 15.07.08 40
H. arborea 23.04.08 37 02.06.08 37
Bufo bufo 03.03.08 36 30.05.08 24-27
Bufo calamita 07.04.08 40 27.05.08 40
A. obstetricans 24.04.08 40 15.07.08 40
Bombina variegata 05.05.08 32 17.06.08 32
I. alpestris 27.02.08 6-9 16.06.08 37
T. cristatus 29.02.08 1-5 16.06.08 37
L. helveticus 27.02.08 6-9 29.05.08 13-14
L. vulgaris 28.02.08 17-21 30.05.08 24-27
Table 10.5 Survey dates in 2009 – additional ponds used in „year 1‟ data for A.
obstetricans.
Species First
detection in pond(s)
Last detection
in pond(s)
A. obstetricans 12.05.09 10, 32-34 02.06.09 34, 39-40
nb: surveying in 2009 was all within the A. obstetricans breeding season, therefore the first
and last dates reflect survey dates more than in years 2007 and 2008.
Appendix 5 – Species distribution maps
153
Appendix 5 – Species distribution maps
(a) Rana temporaria
(b) Pelophylax spp.
(c) Hyla arborea
(d) Bufo bufo
(e) Alytes obstetricans
(f) Bufo calamita
Figure 10.1 Distributions of amphibian species in Luxembourg in 1 km squares,
1996-2005, data from the MNHN database.
Continued…
20 km
N
Appendix 5 – Species distribution maps
154
…continued from previous page.
(g) Bombina variegata
(h) Ichthyosaura alpestris
(i) Triturus cristatus
(j) Lissotriton helveticus
(k) Lissotriton vulgaris
(l) Salamandra salamandra
20 km
N
Appendix 6 – Plants species surveyed
155
Appendix 6 – Plants species surveyed
Table 10.6 All wetland plants surveyed and the percentage of study ponds in which
each plant occurred (names following Stace, 1997).
Plants marked „*‟ could not be identified to species level. „~‟ indicates plants never
occurring at ≥ 5% cover.
Common name Latin name Ponds
occurring in (%)
* ~ Algal pond weed / Blanket weed - 9.3
Creeping bent Agrostis stolonifera L. 7.0
~ Narrow-leaved water plantain Alisma lanceolatum With. 9.3
Water plantain Alisma plantago-aquatica L. 39.5
Alder Alnus glutinosa (L.) Gaertn. 9.3
~ Marsh foxtail Alopecurus geniculatus L. 2.3
* ~ unID'd Alopecurus sps Alopecurus L. sp. 7.0
Meadow foxtail Alopecurus pratensis L. 11.6
~ Wild angelica Angelica sylvestris L. 16.3
~ Fool's-water-cress Apium nodiflorum (L.) Lag. 2.3
~ Lesser water-parsnip Berula erecta (Huds.) Coville 2.3
~ Autumnal water-starwort Callitriche hermaphroditica L. 4.7
~ Various-leaved water-starwort Callitriche platycarpa Kütz. 4.7
~ Common water-starwort Callitriche stagnalis Scop. 4.7
Marsh-marigold Caltha palustris L. 18.6
~ Hedge bindweed Calystegia sepium (L.) 2.3
Slender tufted-sedge Carex acuta L. 2.3
~ Water sedge Carex aquatilis Wahlenb. 2.3
~ Brown sedge Carex disticha Huds. 2.3
Common sedge Carex nigra (L.) 14.0
Cyperus sedge Carex pseudocyperus L. 9.3
Bottle sedge Carex rostrata Stokes 44.2
* unID'd Carex sps Carex sp. 16.3
~ Bladder-sedge Carex vesicaria L. 7.0
True fox-sedge Carex vulpina L. 4.7
Rigid hornwort Ceratophyllum demersum L. 4.7
Soft hornwort Ceratophyllum submersum L. 2.3
~ Opposite leaved golden saxifrage Chrysoplenium oppositifolium L. 2.3
~ Marsh thistle Cirsium palustre (L.) Scop. 16.3
* ~ unID'd Marsh-orchid sp. Dactylorhiza sp. 2.3
* ~ unID'd hair-grass sp Deschampsia sp. 2.3
Common Spike-rush Eleocharis palustris (L.) Roem. & Schult. 32.6
* unID'd Spike-rush Eleocharis sp. 4.7
~ Canadian Waterweed Elodea canadensis Michx. 2.3
~ Nuttall's pondweed** Elodea nuttallii (Planch.) H. St. John 2.3
~ Great willowherb Epilobium hirsutum L. 11.6
Water horsetail Equisetum fluviatile L. 2.3
Marsh Horsetail Equisetum palustre L. 18.6
~ Hemp-agrimony Eupatorium cannabinum L. 2.3
Alder buckthorn Frangula alnus Mill. 2.3
Common Marsh-bedstraw Galium palustre L. 27.9
Floating sweet-grass Glyceria fluitans (L.) B. Br. 51.2
~ Reed sweet-grass Glyceria maxima (Hartm.) Holmb. 11.6
* ~ unID'd sweetgrass sp Glyceria sp. 2.3
Mare's tail Hippuris vulgaris L. 9.3
~ Square-stalked St John's-wort Hypericum tetrapterum Fr. 2.3
Yellow iris Iris pseudacorus L. 32.6
Continued…
Appendix 6 – Plants species surveyed
156
…continued from previous page
Common name Latin name Ponds
occurring in (%)
Jointed rush Juncus articulatus L. 4.7
~ Round-fruited rush Juncus compressus Jacq. 2.3
~ Compact rush Juncus conglomeratus L. 4.7
Soft-rush Juncus effusus L. 62.8
Hard rush Juncus inflexus L. 34.9
* ~ unID'd Juncus sp Juncus sp. 2.3
Common duckweed Lemna minor L. 60.5
* ~ unID'd Duckweed sp Lemna sp. 2.3
Ivy-leaved duckweed Lemna trisulca L. 16.3
~ Water lobelia Lobelia dortmanna L. 4.7
~ Greater bird's-foot trefoil Lotus pedunculatus Cav. 7.0
~ Ragged robin Lychnis flos-cuculi L. 7.0
Gypsywort Lycopus europaeus L. 39.5
Creeping-jenny Lysimachia nummularia L. 25.6
~ Tufted loosestrife Lysimachia thyrsiflora L. 2.3
Purple-loosestrife Lythrum salicaria L. 41.9
Water mint Mentha aquatica L. 16.3
~ Tufted forget-me-not Myostis laxa Lehm. 2.3
Water forget-me-not Myostis scorpioides L. 25.6
~ Alternate water-milfoil Myriophyllum alternifolium DC. 2.3
Whorled water-milfoil Myriophyllum verticillatum L. 2.3
~ Least water-lily Nuphar pumila (Timm) DC. 2.3
White water-lily Nymphaea alba L. 7.0
~ Fine-leaved water-dropwort Oenanthe aquatica (L.) Poir. 2.3
Amphibious bistort Persicaria amphibia (L.) Gray 16.3
Reed canary-grass Phalaris arundinacea L. 2.3
Common reed Phragmites australis (Cav.) Trin. ex Steud. 9.3
Swamp meadow-grass Poa palustris L. 11.6
Curled pondweed Potamogetan crispus L. 9.3
Broad-leaved pondweed Potamogetan natans L. 27.9
Fennel pondweed Potamogetan pectinatus L. 2.3
* unID'd Potamogetan sps Potamogetan sp. 2.3
~ Common fleabane Pulicaria dysenterica (L.) Bernh. 2.3
Common water-crowfoot Ranunculus aquatilis L. 2.3
~ Lesser spearwort Ranunculus flammula L. 4.7
Greater spearwort Ranunculus lingua L. 11.6
Round-leaved crowfoot Ranunculus omiophyllus Ten. 4.7
Creeping buttercup Ranunculus repens L. 51.2
Aquatic liverwort (a bryophyte) Riccia fluitans 2.3
Narrow-fruited water-cress Rorippa microphylla (Boenn.) Hyl. ex Á. & D. Löve 4.7
~ Water-cress Rorippa nasturtium-aquaticum (L.) Hayek. 11.6
~ Clustered dock Rumex conglomeratus Murray 16.3
~ Curled dock Rumex crispus L. 11.6
~ Marsh Dock Rumex palustris Sm. 9.3
~ Wood dock Rumex sanguineous L. 7.0
White willow Salix alba L. 27.9
Goat willow Salix caprea L. 11.6
Grey willow Salix cinerea L. 23.3
Crack-willow Salix fragilis L. 25.6
Purple willow Salix purpurea L. 2.3
Osier Salix viminalis L. 9.3
Common club-rush Schoenoplectus lacustris (L.) Palla 9.3
Wood club-rush Scirpus sylvaticus L. 9.3
~ Skullcap Scutellaria galericulata L. 16.3
Continued…
Appendix 6 – Plants species surveyed
157
…continued from previous page.
Common name Latin name Ponds
occurring in (%)
~ Woody nightshade Solanum dulcamara L. 16.3
Unbranched bur-reed Sparganium emersum Rehmann 4.7
Branched bur-reed Sparganium erectum L. 25.6
* Aquatic moss Sphagnum sp. 25.6
Greater duckweed Spirodela polyrhiza (L.) Schleid. 9.3
* Algal pond weed Charophyta sp. 2.3
Lesser bulrush Typha angustifolia L. 18.6
Bulrush Typha latifolia L. 41.9
Greater bladderwort Utricularia vulgaris L. 7.0
~ Common valerian Valeriana officinalis L. 2.3
Brooklime Veronica beccabunga L. 11.6
Marsh speedwell Veronica scutellata L. 14.0
Table 10.7 Non-wetland plants surveyed and the percentage of study ponds in which
each plant occurred (names following Stace, 1997).
Plants marked „*‟ could not be identified to species level. „~‟ indicates plants never
occurring at ≥ 5% cover.
Common name Latin name Ponds
occurring in (%)
* ~ unID'd Grasses - 2.3
~ Field maple Acer campestre L. 2.3
~ Garlic mustard Alliaria petiolata (M. Bieb.) Cavara & Grande 2.3
~ Silver birch Betula pendula Roth. 4.7
~ Hornbeam Carpinus betulus L. 2.3
~ Common knapweed Centaurea nigra L. 2.3
~ Rosebay willowherb Chamerion angustifolium (L.) Holub 18.6
~ Enchanter's-nightshade Circaea lutetiana L. 2.3
~ Dogwood Cornus sanguinea L. 2.3
~ Hazel Corylus avellana L. 4.7
~ Hawthorn Crataegus monogyna Jacq. 16.3
* ~ Hawthorn hybrid Crataegus sp. 2.3
~ Cock's-foot Dactylis glomerata L. 20.9
~ Broad-leaved willowherb Epilobium montanum L. 4.7
* ~ unID'd Horsetail sp Equisetum sp. 4.7
Beech Fagus sylvatica L. 7.0
~ Ash Fraxinus excelsior L. 7.0
~ Cleavers Galium aparine L. 9.3
~ Hedge bedstraw Galium mollugo L. 2.3
~ Wood avens Geum urbanum L. 4.7
~ Ground-ivy Glechoma hederacea L. 2.3
* ~ unID'd hogweed Heracleum sp. 2.3
~ White dead nettle Lamium album L. 2.3
~ Lesser hawkbit Leontodon saxatilis Lam. 2.3
* ~ unID'd pine tree Pinus sp. 2.3
~ Annual meadow-grass Poa annua L. 7.0
~ Wood meadow-grass Poa nemoralis L. 2.3
Continued…
Appendix 6 – Plants species surveyed
158
…continued from previous page.
Common name Latin name Ponds
occurring in (%)
~ Smooth meadow-grass Poa pratensis L. 2.3
* ~ unID'd meadow-grass sp Poa sp. 2.3
* ~ Poplar sp Populus sp. 14.0
~ Aspen Populus tremula L. 4.7
~ Creeping cinquefoil Potentilla reptans L. 2.3
~ Blackthorn Prunus spinosa L. 7.0
~ Dog-rose Rosa canina L. 2.3
~ Bramble Rubus fruticosus L. agg. 18.6
~ Broad-leaved dock Rumex obtusifolius L. 7.0
~ Eared willow Salix aurita L. 7.0
* unID'd Willow sp Salix sp. 7.0
~ White campion Silene latifolia 2.3
~ Prickly sow-thistle Sonchus asper L. Hill 2.3
~ Dandelion Taraxacum officinale Wigg. group 2.3
~ Red clover Trifolium pratense L. 2.3
~ White clover Trifolium repens L. 2.3
* unID'd Elm tree sp Ulmus sp. 2.3
~ Common nettle Urtica dioica L. 14.0
Appendix 6 References
Stace, C. (1997) New flora of the British Isles. 2nd edition. Cambridge University
Press, Cambridge.
Appendix 7 – Macroinvertebrate families surveyed
159
Appendix 7 – Macroinvertebrate families surveyed
Table 10.8 The macroinvertebrates found and their frequency of occurrence (names
following ZipcodeZoo.com, 2009).
Eleven ponds surveyed; ? = unknown / unidentified.
Phylum Class Type Order Family
Total number of individuals
counted
Number of ponds
Annelida Clitellata Worm Haplotaxida Naididae 339 10
Leech Arhynchobdellida Erpobdellidae 9 2
Haemopidae 23 3
Rhynchobdellida Glossiphoniidae 66 9
Arthropoda Arachnida Watermite Hydracarina ? 102 5
Trombidiformes Hydrodromidae 1 1
Hydryphantidea 14 4
Limnocharidae 4 1
Branchiopoda Waterflea Cladocera ? 140 2
Daphniidae 20 1
Insecta Beetle Coleoptera ? 9 3
Chrysomelidae 8 1
Dytiscidae 230 10
Elmidae 10 1
Gyrinidae 5 3
Haliplidae 37 6
Helophoridae 18 3
Hydrochidae 5 1
Hydrophilidae 71 7
Noteridae 88 8
Scirtidae 5 2
True fly Diptera ? 4 1
Chaoboridae 1777 10
Chironomidae 328 8
Culicidae 9 3
Heleinae 16 1
Psychodidae 2 1
Ptychopteridae 10 4
Sciomyzidae 1 1
Stratiomyidae 83 9
Tabanidae 4 1
Tipulidae 3 2
Mayfly Ephemeroptera Baetidae 2880 11
Caenidae 1 1
Leptophlebiidae 13 1
Potamanthidae 3 1
Siphlonuridae 210 2
Bug Hemiptera Corixidae 113 5
Gerridae 364 11
Hydrometridae 3 1
Naucoridae 105 4
Nepidae 16 4
Notonectidae 862 10
Pleidae 344 6
Continued…
Appendix 7 – Macroinvertebrate families surveyed
160
…continued from previous page.
Phylum Class Type Order Family
Total number of individuals
counted
Number of ponds
Arthropoda Insecta Dragonfly Odonata ? 3 2
Aeshnidae 100 7
Corduliidae 120 9
Gomphidae 1 1
Libellulidae 28 6
Damselfly Odonata Coenagrionidae 391 8
Lestidae 372 6
Platycnemididae 17 1
Stonefly Plecoptera Capniidae 2 1
Caddisfly Tricoptera Glossosomatidae 3 1
Hydropsychidae 1 1
Lepidostomatidae 1 1
Limnephilidae 1872 10
Phryganeidae 15 1
Polycentropodidae 18 1
Sericostomatidae 10 1
Malacostraca Crustacean Amphipoda Gammaridae 8 1
Isopoda Asellidae 10 2
Ostracoda ? ? 1000 1
Mollusca Bivalvia Bivalve Veneroida Corbiculidae 20 2
Pisidiidae 747 7
Gastropoda Snail Basommatophora Lymnaeidae 2330 8
Planorbidae 2424 8
Platyhelminthes Turbellaria Flatworm Tricladida Planariidae 1 1
Number of families identified: 62
Number of families not identified: 6
Appendix 8 – Occupancy modelling covariates
161
Appendix 8 – Occupancy modelling covariates
Table 10.9 Summary of survey-specific covariate values for each study year.
Covariate 2007 ponds 2008 ponds Year 1 ponds
min max mean mode min max mean mode min max mean mode
day_of_yr 56 219 - - 58 197 - - 56 219 123.58 72
full_moon_ closest
0.07 14.72 7.19 - 0.21 14.67 8.38 - 0.07 14.74 7.58 -
pm_timetaken 0.08 2.25 0.57 - 0.08 1.50 0.40 - 0.08 2.25 0.53 -
pm_windspeed 0 2 - 0 0 3 - 0 0 2 - 0
pm_wind_dir - - - 0.3 (E)
- - - 0.7 (NW)
- - - 0.3 (E)
pm_rain 0 3 - 0 0 3 - 0 0 3 - 0
pm_gen 0 2 - 0 0 2 - 0 0 2 - 0
clear_shore 10 100 76.21 - 5 100 74.66 - 10 100 75.55 -
turbidity 0 3 - 1 0 3 - 2 0 3 - 1
temp_min 1 22 10.95 - 1 20 8.76 - 1 22 11.28 -
temp_max 3 25 14.09 - 4 23 11.47 - 3 25 14.03 -
#traps 5 30 19.57 - 0 30 18.87 - 0 30 19.45 -
#nets 5 30 19.57 - 0 30 18.92 - 0 30 19.44 -
See explanations in Chapter 7, Table 7.3.
Appendix 8 – Occupancy modelling covariates
162
Table 10.10 Summary of site-specific covariates for each study year.
Covariate 2007 ponds 2008 ponds 'year 1' ponds
mean s.e._ mean s.e._ mean s.e._
pond_size 1247.33 387.38 1242.64 409.04 1159.34 274.37
pH 7.36
(5.97-8.20)
- 7.32
(6.27-8.20)
- 7.68
(5.91-8.99)
-
cond 423.75 52.82 386.54 53.37 352.26 42.42
shade_int 63.20 5.99 63.72 6.23 59.20 5.36
shade_ext 21.58 4.96 22.41 5.30 21.23 4.11
open_water 30.94 5.32 30.57 5.47 35.29 5.25
elevation 299.64 7.80 305.17 10.06 332.48 11.73
#_invert_sps 29.70 2.20 31.09 2.43 31.09 2.43
plant_h' 4.07 0.60 4.31 0.64 3.71 0.44
paths_tracks 212.71 78.84 268.68 86.39 282.31 69.21
road_<5.5 70.54 30.17 76.95 32.62 90.83 27.67
road_6.5-7.5 16.75 11.63 18.27 12.66 11.49 8.03
road_all_widths 87.29 33.03 95.23 35.60 102.31 28.94
hedgerow 116.08 38.31 120.64 41.51 97.26 27.20
treerow 46.71 27.37 50.95 29.74 37.17 19.10
water_stream_surface 40.71 17.53 44.82 19.11 80.14 21.93
undef_building 5.63 5.58 6.14 6.09 36.49 32.74
buildings_undef_ag_ind 5.63 5.58 6.14 6.09 53.71 49.89
forest_conif 254.54 145.74 335.55 176.96 2442.71 1184.18
forest_brdlvd 8295.67 3818.29 11770.27 4788.81 10076.86 3172.21
forest_mixed 2338.88 1528.61 2398.23 1687.31 3073.83 1442.03
orchard 651.83 524.52 711.09 571.58 446.97 360.98
forest_bwood_orchard 11650.63 4170.52 15334.82 5239.51 16115.60 3741.00
See explanations in Chapter 7, Table 7.2.
Table 10.11 Site-specific variables concerning the presence / non-detection of other
species.
Covariate 2007 2008 ‘year 1’ ponds
1 0 1 0 1 0
waterfowl 12 12 11 11 21 14
fish 4 20 3 19 4 31
Rt 22 2 21 1 24 2
P.spp 22 2 22 0 26 2
Bb 12 12 11 11 18 13
Ao 2 22 3 19 13 22
Tc 11 13 10 12 11 15
Lh 23 1 22 0 31 1
Lv 17 7 17 5 18 8
#_amphib_sps 5.58 0.27 5.91 0.20 5.65 0.27
1 = present, 0 = not detected; see explanations in Chapter 7, Table 7.1 and species name
abbreviations in Appendix 1.
Appendix 9 – Publications
163
Appendix 9 – Publications
Wood, L. R., Griffiths, R. A. Groh, K., Engel, E. & Schley, L. (2008)
Interactions between freshwater mussels and newts: a novel form of
parasitism? Amphibia-Reptilia, 29, 457-462.
Amphibia-Reptilia 29 (2008): 457-462
Interactions between freshwater mussels and newts:a novel form of parasitism?
Laura R. Wood1,*, Richard A. Griffiths1, Klaus Groh2, Edmée Engel3, Laurent Schley4
Abstract. Small freshwater mussels are sometimes found attached to the toes of aquatic phase amphibians, but the ecologicalimplications of this interspecific relationship are unknown. Toe condition and mussel presence were recorded for newts caughtin 37 ponds in Luxembourg between March and June 2007. All four local newt species were affected (Lissotriton helveticus,L. vulgaris, Mesotriton alpestris and Triturus cristatus), primarily by the mussel Sphaerium nucleus but also by Pisidiumsubtruncatum. Mussel attachment was observed in three ponds, with a particularly high occurrence in one pond, where 23%of newts were affected and significantly more toes were damaged than in other ponds. Mussels caused local tissue and bonedamage to their host and may interfere with egg-laying. Twenty-two newts with attached mussels were observed in aquariafor up to 3 days: 13 mussels detached when the newt’s toe fell off and nine remained attached. If the mussels benefit fromthe interaction through, for example, enhanced dispersal then the relationship between the two taxa represents a novel formof parasitism.
Keywords: dispersal, Lissotriton, Mesotriton, Sphaerium nucleus, symbiosis, Triturus.
Introduction
Interactions between organisms of differentspecies may have a mutually positive or nega-tive outcome for both parties, or an overall pos-itive result for one species at the expense of theother. The term ‘symbiosis’ describes a rangeof interspecific associations, the exact type de-pending on the advantage or disadvantage eachparticipant derives from the relationship. Hostorganisms may serve a range of purposes for thepartner, from providing temporary food, shelteror transport, to hosting its entire life (Ricklefsand Miller, 1999).
In a parasitic relationship the host is neg-atively affected while the parasite benefits. Incontrast a commensal relationship is where oneparticipant benefits whilst the other incurs no
1 - The Durrell Institute of Conservation and Ecology, Uni-versity of Kent, Marlowe Building, Canterbury, CT27NR, UK
2 - VHÖ/BBN, Mainzer Str. 25, D-55546 Hackenheim,Germany
3 - Musée national d’histoire naturelle, 25 rue Münster,L-2160 Luxembourg
4 - Service de la Conservation de la Nature, Direction desEaux et Forêts, 16 rue Eugène Ruppert, L-2453 Luxem-bourg∗Corresponding author; e-mail: [email protected]
notable benefit or harm. Phoresy is a specificform of commensalism in which a mobile or-ganism affords provisional transport to a ses-sile one (Kwet, 1995). A phoretic relationshipbetween invertebrates living in bromeliad wa-ter reservoirs and visiting vertebrates has beenwell studied (Lopez et al., 2002; Lopez et al.,2005). Bromeliad annelids and ostracods ori-ent towards and attach to frog or lizard skin,thereby gaining transport to another reservoirand succeeding in dispersal (Lopez et al., 2005).Furthermore, the ostracods can survive pass-ing through the gut of amphibians and mam-mals, potentially achieving dispersal by beingingested while the animal drinks (Lopez et al.,2002).
Seidel (1990) recorded high rates of ostracod‘attacks’ on yellow-bellied toads (Bombina var-iegata) at a pond complex, and instances of indi-vidual toads carrying ostracods between ponds.Furthermore, the rate of association increased asponds dried out, indicating that attachment wasnot accidental.
Some freshwater molluscs need another ani-mal as a host during parts of their life cycle. Forexample freshwater unionid or margaritiferidmussel larvae that parasitise fishes’ gills or fins,
© Koninklijke Brill NV, Leiden, 2008. Also available online - www.brill.nl/amre
458 L.R. Wood et al.
gaining shelter, sustenance and ultimately dis-persal (Howerth and Keller, 2006). A heavy par-asite load is costly to the host, and potentiallyfatal, however this is not always a one-sided re-lationship, some fish also use the mussels to in-cubate their embryos (Reichard et al., 2006).
The present study examines the occurrence offreshwater bivalve mussels found clamped ontonewts’ toes, which might be a form of phoresyor dispersal parasitism. This phenomenon islittle-reported in the literature, although infor-mal communication with professional herpetol-ogists indicates that it has been sporadically ob-served in several European countries. There area few published accounts from the late 19th andearly 20th centuries including two short notes byDarwin (1878, 1882) detailing a mussel foundclipped onto a duck’s foot and, later, on a greatdiving beetle (Dytiscus marginalis) and a frog(species not given). The only recent account, toour knowledge, is that of Kwet (1995) detail-ing mussels found on common toads’ toes (Bufobufo) in Germany, but with only two individualtoads observed with mussels and no analysis.
Freshwater mussels were observed on newts’toes at three ponds in Luxembourg, but in par-ticularly high numbers at one pond, affectingall four local newt species (Lissotriton helveti-cus, L. vulgaris, Mesotriton alpestris and Tritu-rus cristatus). Mussels were firmly shut aroundthe tip of the newt’s digit (see fig. 1), often withapparent tissue and bone damage to the newt.The data collected so far allow simple compar-isons between the newt sexes and species, anddescribe the local incidence of the phenomenon.
Materials and methods
Ponds in central and southern Luxembourg were surveyedby bottle trapping up to six times between March and June2007. In total 37 ponds at 17 sites were surveyed. A pondwas defined as “a body of water, of man-made or naturalorigin, between 1 m2 and 2 ha, which usually holds waterfor at least four months of the year” (National Pond Survey,Biggs et al., 1998). Sites were considered distinct if studyponds were separated by >1 km. Sites were selected fromhistorical records of amphibian presence in Luxembourgand from topographical maps showing waterbodies.
Bottle traps, constructed from 2 litre drinks bottles withthe top section cut off and inverted (see Griffiths, 1985),were set underwater resting on the substrate of the pond,held in place with a cane and with an air bubble retainedinside. Up to 25 traps were set per pond, or a maximum of 30between immediately adjacent ponds, at 2 m intervals alongthe margin in multiples of 5 or in a continuous line. Trapswere removed and emptied the following morning, within12 hours of setting. All captured newts were identified tospecies, sexed and counted, and their feet and toes wereexamined for damage or attached mussels. The conditionof each toe was recorded as intact, damaged or with musselattached.
In pond ‘Dréisch’, where the incidence of mussel at-tachment to newts was highest, six visits were made dur-ing the fieldwork period. From five of these visits 22 newts(14 Mesotriton alpestris, 4 Triturus cristatus, and 4 Lissotri-ton vulgaris) caught with mussels clamped to their toes wereobserved for 24 or 36 hours. They were kept in a 30 litre co-vered plastic box, with between 1 and 10 other newts, innatural shade on site or indoors in a cool day-lit room with20 cm depth of pond water and some vegetation. The timeelapsed between capturing the newt and the mussel becom-ing detached was measured to the following 12 hour intervaland the affected toe was checked for damage.
Mussel sampling
On 16th March 2007 mussels in Dréisch were sampled insix 10 × 10 cm quadrats. Quadrats were spaced along theaccessible shallow pond margin, random placing was notpossible due to dense vegetation and deep water. Substrateand plant matter from the pond bottom were lifted by handinto a bucket and then sorted and sieved carefully to collectall the freshwater mussels.
Mussels in each sample were counted and preserved inethanol for later identification. They were identified accord-ing to Glöer and Meier-Brook (2000) and Killeen, Aldridgeand Oliver (2004).
Analysis
Pearson’s chi-square test was used to compare the incidenceof mussel attachment between sexes and species of newtand fore/hind legs at Dréisch, the rate of damaged toes be-tween Dréisch and all other ponds, and the difference be-tween middle toes and outer toes. Pseudoreplication mighthave occurred as newts were not individually recognisable;however recapture rates would be expected to be relativelylow and not to adversely affect analyses.
Results
Mussels were found attached to newts in threeponds at two sites: Dréisch (Gauss-Luxembourggrid reference: LC 708 636) and Laaschtert3 & 4 (LC 725 934 & LC 727 935). Two
Interactions between freshwater mussels and newts 459
Figure 1. Male M. alpestris with a mussel attached to the fourth toe of the hind left foot.
species of freshwater mussel were identified atDréisch: Sphaerium nucleus and Pisidium ob-tusale, although only S. nucleus was seen at-tached to newts. Only one mussel species, P.subtruncatum, was found at Laaschtert, how-ever mussels at this site were not surveyed or
collected with quadrats. The density of mus-sels at Dréisch was very high – S. nucleuswas recorded at a mean density of 2216.7/m2
(standard error: ±371.0/m2, range: 1100-3300,n = 6) and P. obtusale at 716.7/m2 (standard er-ror: ±293.7/m2, range: 0-1900, n = 6) with
460 L.R. Wood et al.
Figure 2. The condition of the toes of newts caught in pond ‘Dréisch’ compared to those caught in 34 other ponds. Newtsrecorded as ‘Intact’ had all their toes present uninjured, ‘Damaged’ newts had some shortened or missing toes, and newts‘With mussels’ had one or more mussels attached and possibly some additional damage; n indicates the total number of newtsin each category.
an overall mean mussel density of 2933.3/m2
(standard error: ±315.9/m2, range: 1700-4000,n = 6).
Individuals of all four newt species presentin Luxembourg were observed with musselsattached to their toes: Mesotriton alpestris (n =23), Triturus cristatus (n = 10), Lissotritonhelveticus (n = 2) and L. vulgaris (n = 7).Of all newts captured in Dréisch 23% (totaln = 161) were carrying mussels (see fig. 2).The occurrence was less frequent at Laaschtert3 and 4 where mussel attachment was recordedon 3.6% (n = 1, total n = 28) and 7.0%(n = 4, total n = 57) of newts respectively.Laaschtert 3 and 4 data were excluded fromstatistical analysis of the frequency of musselattachment due to low overall capture rates andthe corresponding lack of data.
Individuals of all three newt species presentin Dréisch were found with one or more mus-sels attached (fig. 2); there was no significantdifference in frequency of mussel attachmentbetween the species (χ2 = 3.21, df = 2, P =0.20). T. cristatus males were significantly more
likely to have a mussel attached than females(χ2 = 5.98, df = 1, P < 0.05), but there wasno significant difference between sexes of theother species (M. alpestris: χ2 = 1.75, df = 1,P = 0.18; L. vulgaris: χ2 = 1.18, df = 1,P = 0.28). One female M. alpestris was foundwith a mussel on each hind foot and her owneggs adhered to the mussels, showing that mus-sels can interfere with oviposition.
Newts in Dréisch (n = 161) were nearlyfive times more likely to have damaged, miss-ing or split toes than newts from other ponds(excluding Laaschtert, n = 1982, see fig. 2;χ2 = 155.73, df = 1, p < 0.001). Middle toeswere significantly more likely to have musselsattached than the shorter outer toes on each foot(χ2 = 7.27, df = 1, P < 0.001). There was nodifference in rate of attachment between frontand hind legs (χ2 = 3.04, df = 1, P = 0.08).In captivity newts only lost mussels when the af-fected toe fell off with the mussel still clampedonto it (n = 13). Nine out of 22 mussels re-mained attached for longer than the captivity pe-
Interactions between freshwater mussels and newts 461
riod of 24 or 36 hours (24 hrs: n = 3; 36 hrs:n = 6).
Discussion
During amphibian surveys in Luxembourg,newts were found with small freshwater musselsattached to their toes – in one pond, ‘Dréisch’,23% of the newts captured were carrying one ormore mussels. The current data set is limited toa single season and three ponds at two sites; toofew newts were captured at one site, so analyseswere restricted to pond Dréisch. In addition, theoccurrence of damaged digits was significantlyhigher in newts from Dréisch, possibly as a re-sult of previous mussel attachment. Of the 22newts observed in captivity for up to three days,the 13 mussels that became detached only did sobecause the toe fell off. Therefore attachment ofmussels does have a negative impact on newts.The ability to regenerate digits ensures that theimpact is only temporary, although full regener-ation takes at least 1.5 years (Arntzen, Smithsonand Oldham, 1999). A different kind of impactwas observed on a single occasion: the femaleMesotriton alpestris with a mussel attached toeach hind foot and eggs adhered to the mussels.Clearly her ability to wrap eggs was reduced, atleast in the short-term, indicating that perhapsmussels regularly prevent or hinder reproduc-tive behaviours in which the feet play a part,such as egg-laying and courtship.
Although a greater percentage of the largerbodied newts carried mussels, this trend wasnot significant. If the inter-species differencesare real, they may be due to differences in thespecies’ behaviour and habitat use, or simplydue to toe size and structure. For example, Tritu-rus cristatus spends more time on the pond bot-tom than Lissotriton vulgaris (Dolmen, 1988)and feeds predominantly on benthic prey (Jolyand Giacoma, 1992), so would be more likelyto come into contact with mussels.
Why mussels attach to newts, and why theydo not release, is open to speculation. Newtsresting and travelling on the pond bottom will
regularly contact freshwater mussels, especiallywhen the mussels are at a very high density asin Dréisch. An open, filter feeding mussel willclamp shut defensively when prodded, whichwould probably occur if a newt steps on it. Fol-lowing such accidental attachment, the musselmay be unable to release due to constant stimu-lation from the trapped toe. Alternatively, Kwet(1995) hypothesised that attachment to amphib-ians may be a dispersal mechanism for musselsfrom overcrowded ponds, although presumablystill with an element of accidental attachment inthe first place.
Dréisch is a closed pond (no water flowing inor out) and is relatively isolated. Isolation andideal conditions for Sphaerium nucleus preventeasy dispersion and create the extraordinarilyhigh mussel density. Although most Sphaeriumand Pisidium species are able to climb plantstems underwater, they cannot traverse dry landwithout the assistance of some kind of vector(Boycott, 1936). There are only two waterbod-ies within 500 m: the nearest pond (400 m) isan inaccessible motorway run-off pool and thesecond (480 m) is separated from Dréisch by arailway track and embankment. In this case, am-phibians may provide transport within Dréisch,but are very unlikely to be useful vectors fordispersal to other ponds. However if there wereother ponds in close proximity, movement ofnewts between ponds during the breeding sea-son (e.g. Hagström, 1979; Dolmen, 1981; Grif-fiths, 1984) could facilitate very local musseldispersal. Over greater distances passive mus-sel dispersal could occur through attachmentto larger animals: mussels might become stuckin the mud on other animals’ feet (Pip, 1986;Bilton, Freeland and Okamura, 2001), or even“nip” onto ducks’ toes (Darwin, 1882; Boycott,1936).
Based on the present observations it is notpossible to determine whether freshwater mus-sel attachment to newts represents an interactionthat has evolved by natural selection or is simplyaccidental. If dispersal is facilitated in this way,the relationship appears to be parasitic, because
462 L.R. Wood et al.
the mussel benefits while the newt is harmed.On land, mussels seem to impede a newt’s lo-comotion, but an experimental approach wouldbe necessary to determine whether newts are ca-pable of transporting mussels a sufficient dis-tance on land for such ‘dispersal parasitism’ tosucceed. Subject to environmental conditions,Sphaerid mussels can survive extended periodsin anoxic conditions: approximately 20 days at16◦C, increasing to more than 100 days below7◦C (Holopainen, 1987), and avoid desiccationby closing tightly. Therefore if the host newtmoves to another pond during the same season,and before its toe falls off, a mussel may sur-vive the transfer. Further work should build onthe present study, allowing greater descriptionof the phenomenon and developing an under-standing of where and under what conditionsfreshwater mussels attach to amphibians.
Acknowledgements. This project was funded by a re-search scholarship (‘bourse de formation-recherche’) fromthe Luxembourg Ministère de la Culture, de l’EnseignementSupérieur et de la Recherche to LW (BRF06/056), with ad-ditional funding for fieldwork equipment from The Am-phibian Conservation and Research Trust, UK. A licence totrap amphibians was granted by the Luxembourg Ministèrede l’Environnement (ref. 64512). We thank L. Glesener andM. Sylla for fieldwork assistance and two referees for usefulcomments on the manuscript.
References
Arntzen, J.W., Smithson, A., Oldham, R.S. (1999): Mark-ing and tissue sampling effects on body condition andsurvival in the newt Triturus cristatus. J. Herpetol. 33:567-576.
Biggs, J., Fox, G., Nicolet, P., Walker, D., Whitfield, M.,Williams, P. (1998): A guide to the methods of theNational Pond Survey. Oxford, Pond Action.
Bilton, D.T., Freeland, J.R., Okamura, B. (2001): Dispersalin freshwater invertebrates. Ann. Rev. Ecol. Syst. 32:159-181.
Boycott, A.E. (1936): The habitats of fresh-water molluscain Britain. J. Anim. Ecol. 5: 116-186.
Darwin, C. (1878): Transplantation of shells. Nature 18:120-121.
Darwin, C. (1882): On the dispersal of freshwater bivalves.Nature 25: 529-530.
Dolmen, D. (1981): Local migration, rheotaxis, andphilopatry by Triturus vulgaris within a locality in cen-tral Norway. Br. J. Herpetol. 6: 151-158.
Dolmen, D. (1988): Coexistence and niche segregation inthe newts Triturus vulgaris (L.) and T. cristatus (Lau-renti). Amphibia-Reptilia 9: 365-374.
Glöer, P., Meier-Brook, C. (2000): Süsswassermollusken.Ein Bestimmungsschlüssel für die BundesrepublikDeutschland, 13th Edition. Hamburg, Deutscher Jugend-bund für Naturbeobachtung.
Griffiths, R.A. (1984): Seasonal behaviour and intrahabitatmovements in an urban population of Smooth newts,Triturus vulgaris (Amphibia: Salamandridae). J. Zool.Lond. 203: 241-251.
Griffiths, R.A. (1985): A simple funnel trap for studyingnewt populations and an evaluation of trap behaviourin smooth and palmate newts, Triturus vulgaris andTriturus helveticus. Herpetol. J. 1: 5-10.
Hagström, T. (1979): Population ecology of Triturus crista-tus and T. vulgaris (Urodela) in SW Sweden. HolarcticEcol. 2: 108-114.
Holopainen, I.J. (1987): Seasonal variation of survival timein anoxic water and the glycogen content of Sphaeriumcorneum and Pisidium amnicum (Bivalvia, Pisidiidae).Amer. Malac. Bull. 5: 41-48.
Howerth, E.W., Keller, A.E. (2006): Experimentally in-duced glochidiosis in smallmouth bass (Micropterusdolomieu). Vet. Pathol. 43: 1004-1008.
Joly, P., Giacoma, C. (1992): Limitation of similarity andfeeding habits in three syntopic species of newts (Tritu-rus, Amphibia). Ecography 15: 401-411.
Killeen, I., Aldridge, D., Oliver, G. (2004): Freshwaterbivalves of Britain and Ireland. Cardiff and Cambridge,Field Studies Council.
Kwet, A. (1995): Erdkröten (Bufo bufo) als Transportwirtevon Kugelmuscheln (Sphaerium corneum). Salamandra31: 61-64.
Lopez, L.C.S., Filizola, B., Deiss, I., Rios, R.I. (2005):Phoretic behaviour of bromeliad annelids (Dero) andostracods (Elpidium) using frogs and lizards as dispersalvectors. Hydrobiologia 549: 15-22.
Lopez, L.C.S., Gonçalves, D.A., Mantovani, A., Rios, R.I.(2002): Bromeliad ostracods pass through amphibian(Scinaxax perpusillus) and mammalian guts alive. Hy-drobiologia 485: 209-211.
Pip, E. (1986): A study of pond colonization by freshwatermolluscs. J. Mollus. Stud. 52: 214-224.
Reichard, M., Ondrackova, M., Przybylski, M., Liu, H.,Smith, C. (2006): The costs and benefits in an un-usual symbiosis: experimental evidence that bitterlingfish (Rhodeus sericeus) are parasites of unionid musselsin Europe. J. Evol. Biol. 19: 788-796.
Rickleffs, R.E., Miller, G.L. (1999): Ecology. Basingstoke,W.H. Freeman.
Seidel, B. (1990): Amphibien als Transporteure limnis-cher Muschelkrebse: Ein Parameter zur Analyse derVerteilung von Bombina variegata. Amphibia-Reptilia11: 253-261.
Received: January 25, 2008. Accepted: April 25, 2008.
Appendix 9 - Publications
170
Wood, L. R., Griffiths, R. A. & Schley, L. (2009) Amphibian
chytridiomycosis in Luxembourg. Bulletin de la Société des
Naturalistes Luxembourgois, 110, 109-114.
Bull. Soc. Nat. luxemb. 110 (2009) 109
Amphibian chytridiomycosis in Luxembourg
Laura R. Wood1, Richard A. Griffiths1 & Laurent Schley2
1 Durrell Institute of Conservation & Ecology, University of Kent, Marlowe Building, Canterbury, CT2 7NR, UK ([email protected]; [email protected])2 Administration de la nature et des forêts, 16 rue Eugène Ruppert, L-2453 Luxembourg ([email protected])
Wood, L.R., R.A. Griffiths & L. Schley, 2009. Amphibian chytridiomycosis in Luxembourg. Bulletin de la Société des naturalistes luxembourgeois 110: 109-114.
Abstract. Aquatic-phase amphibians at eight sites in Luxembourg were tested for the fungus Batrachochytrium dendrobatidis, which causes a disease responsible for population declines among many amphibian species worldwide. Infected amphibians were found at two of the sites tested; two further sites also showed marginally positive results. The ecology of B. den-drobatidis and the necessity for biosecurity protocol implementation by fieldworkers are discussed.
Key words. Batrachochytrium dendrobatidis, biosecurity, conservation, fieldwork, mass mortality, population decline.
1. IntroductionIn recent years much attention has been focused on the apparent global decline of amphibians. According to the IUCN Red List many more amphibian species are threatened than birds or mammals, with 32.5% amphibians, 23% mammals and 12% birds assessed as ‘globally threatened’ (Stuart et al. 2004). The threats facing amphibians are complex and mostly caused by humans, including non-native species introductions, exploitation, habitat loss, climate change, pollution and disease (Collins & Storfer 2003).Following its relatively recent discovery almost simultaneously in wild and captive populations (Berger et al. 1998, Pessier et al. 1999), the chytridiomycete fungus Bat-rachochytrium dendrobatidis (hereafter Bd) has been implicated in amphibian declines and extinction events worldwide. The reason for the rapid global emergence of Bd is the subject of much debate, but likely due to a suite of factors including introduction by human agency into wild populations, for example release of non-native species, espe-cially the North American bullfrog (Rana catesbeiana) (Garner et al. 2006) and the cumulative effect of environmental changes
modifying an already present host-parasite relationship (Pounds et al. 1999).Prior to the identification of Bd, chytridi-omycete fungi had not been described in parasitic relationships with vertebrates, they were thought to be either free-living or parasitic on other fungi, algae, plants or invertebrates (Pessier et al. 1999). There-fore Bd is the first member of the phylum known to parasitise vertebrates and in a very short space of time has been associated with amphibian declines on every continent they inhabit (see www.spatialepidemiology.net/bd/). Bd infection has been observed to travel very rapidly, described by Lips et al. (2006) as an ‘epidemic wave’ and its interac-tions with amphibians and the environment appear to vary geographically (Morehouse et al. 2003).Infection with Bd causes a disease called chytridiomycosis. The fungus only occurs in keratinised tissue and may cause death by interfering with the respiration and osmoregulation properties of amphibian skin (Daszak et al. 1999). However not all amphibian species are susceptible. Out of 28 European species tested by Garner et al. (2005) 20 were found to be infected in some area or at some stage of their life history. According to the latest data available on the
110 Bull. Soc. Nat. luxemb. 110 (2009)
website dedicated to sharing information about Bd, 29 species have tested positive in nine European countries (http://www.spa-tialepidemiology.net/bd/).Bd has been detected in both France and Germany and at very high prevalence in Switzerland. However this does not mean that it is not present in other nearby coun-tries as sampling effort has been patchily distributed and not all data are in the online database. All but two of the amphibian species present in Luxembourg have been found to be susceptible to the disease else-where in Europe: only Triturus cristatus and Bombina variegata have not tested positive so far, but the extent of survey effort has not been quantified.Increasing knowledge of the global distribu-tion of Bd infections is vital. Accordingly, we tested amphibians in Luxembourg for the disease in a simple presence/absence study. The resulting data are useful from both local and global perspectives, informing manage-ment decisions on site and contributing to wider knowledge of the disease.
2. Materials and methodsAmphibians from 14 ponds in eight differ-ent locations in Luxembourg were tested for Bd infection (see Table 1). Test sites were selected mainly (but not exclusively) for having an historical record of the Common
midwife toad, Alytes obstetricans, a threat-ened species known to be particularly sus-ceptible to the disease. Up to 34 individuals were tested at each site between May and July 2008.During all fieldwork, appropriate biosecu-rity measures were applied. Footwear and survey equipment were disinfected between sites and clean disposable vinyl gloves were worn for handling amphibians.Amphibians were caught in bottle traps, by hand or by net. Only species known to be susceptible to Bd were tested; see Table 1. Keratinised skin parts were targeted with a cotton swab that was swept three times across the drink patch, feet and thighs of adult animals or rotated on the mouthparts of anuran larvae.Swabs were stored in individual airtight tubes in a cool, dry environment. The real-time polymerase chain reaction (PCR) Taqman procedure of Boyle et al. (2004) was used to extract and amplify Bd DNA from the swabs. Samples were compared to ampli-fication standards of 100, 10, 1 and 0.1 Bd zoospore equivalents. This method is widely used and is specific to Bd.Each sample was replicated twice. Samples that amplified ≥1.0 Bd zoospore equiva-lent in both replications were taken to be true positives, single amplification samples were retested, and samples that did not
Nearest town or village Pond name Grid reference
Number of swabs by species Total swabsA.o B.c R.kl.e M.a L.h L.v
Putscheid Alersang LA 78152 14160 17 6 2 25Bissen Laaschtert 2 LC 72266 93271 1 1 2 4
Laaschtert 3 LC 72520 93396 1 1 2Laaschtert 4&5 LC 72679 93499 1 1 2 4
Useldange Weiden 3&4 LC 64855 92458 30 30Schoos Stuppicht 1&2 LC 80024 89074 3 27 30Hunnebour Mersch LC 74102 88067 25 25Blaschette Flakewiss2 LC 80068 87043 1 1 2
Biergwiss LC 79823 86966 1 4 14 19Steinfort Steinfort Carrières LC 61072 81361 4 6 5 6 13 34Niedercorn Giele Botter LC 59207 67483 19 19
Table 1. Sites of Bd testing and the species tested at each site. A.o = Alytes obstetricans; B.c = Bufo calamita; R.kl.e = Rana klepton. esculenta (green frog complex); M.a = Mesotriton alpestris; L.h = Lissotriton helveticus; L.v = L. vulgaris.
Bull. Soc. Nat. luxemb. 110 (2009) 111
amplify in either replicate were taken to be negative. Due to the very small quanti-ties of DNA involved, and the potential for contamination between samples, those that amplified only once and at very low levels were ignored. The quantity of Bd zoospores detected on positive animals is taken from the mean value of Bd zoospore equivalents detected in the replicates of each sample.This study aimed only to establish whether Bd is present in Luxembourg, therefore inferences about the prevalence of disease can not be made.
3. ResultsAmphibians tested positive for Bd zoospores at two of the eight sites surveyed. Two fur-ther sites may also be positive, however the number of zoospores detected was very low – if greater confidence in results was required these samples would need further testing. The sites tested and the results are shown in Fig. 1.No infection was found at Alersang, Laas-chtert 2 or 3, Weiden 3&4, Mersch or Biergwiss and Flakewiss2.A single green frog (R. kl. esculenta) tested positive at Laaschtert 4&5; this is a clear positive because Bd DNA amplified in both sample replicates on the first PCR run (mean 8.3 zoospores). A clear positive result was also found in one Alpine newt (Mesotriton alpestris) at the forest ponds Stuppicht 1&2 (mean 2.5 zoospores).At Giele Botter pond a palmate newt (Lissot-riton helveticus) returned a marginal positive result on both replicates, only one of which was ≥0.5 zoospore equivalents (0.07 and 0.6). Screening was unable to give conclu-sive results for two newts swabbed at Stein-fort Carrières: a smooth newt (L. vulgaris) sample that was tested twice (4 replicates) amplified only a single replicate each time (0.4 and 2.2 zoospores); one palmate newt (L. helveticus) gave a marginally positive test on both replicates (mean 0.7 zoospores).The amplification standards, with a known quantity of Bd zoospores, all amplified as expected, indicating that the tests had run successfully.
4. Discussion4.1. Bd in LuxembourgBd was found at half of the sites tested, although two of the positive results are not clear. The highest infection rate among the animals randomly sampled was 10% at the Laaschtert ponds near Bissen (n = 10): how-ever finding true prevalence rates would require more extensive testing.The sites tested were selected for their amphibian fauna, with at least four of the eight sites presently hosting the Common midwife toad (A. obstetricans) and two other sites from where it has recently become extinct. According to the Luxem-bourg Musée national d’histoire naturelle (MNHN) records A. obstetricans was present at the Laaschtert pond complex until at least 1997 (F. Schoos, ‘Amphibia Data’ record 466, MNHN, 2007) but it was not detected
Fig. 1. The locations of ponds tested for Bd infection and test results (1 = Alersang; 2 = Laaschtert 2; 3 = Laaschtert 3; 4 = Laaschtert 4+5; 5 = Weiden 3+4; 6 = Stuppicht 1+2; 7 = Mersch; 8 = Flakewiss; 9 = Biergwiss; 10 = Steinfort Carrières; 11 = Giele Botter).
112 Bull. Soc. Nat. luxemb. 110 (2009)
during two seasons of intensive survey effort by LRW in 2007 and 2008.
4.2. Biology of BdBd was only discovered recently (Berger et al. 1998, Pessier et al. 1999), so relatively little is known about its ecology, and its modes of transmission are still not fully understood. Long-distance transmission seems likely to have occurred principally via infected ani-mals in the global amphibian trade (Garner et al. 2006) for the pet, science and food industries (Kriger & Hero 2007), and then subsequent releases or escapes into the wild. Besides its own motility in water, on a local level there are many possible vectors of Bd; for example animals travelling between sites (e.g. on bird feathers) and human vectors such as vehicles and hiking boots (Daszak et al. 2003, Johnson & Speare 2005). It can survive, and remain viable, away from a host amphibian for up to 7 weeks in lake water (Johnson & Speare 2003) and can survive up to three hours out of the water (Johnson & Speare 2005). Therefore it could easily be transported locally.The effects of Bd infection vary widely. In some places mass-mortality events have been observed, for example the large num-bers of dead and dying A. obstetricans found around ponds in Peñalara National Park in central Spain (Bosch et al. 2001). Yet some-times there may be a very high prevalence of Bd, but no pathological symptoms or any apparent population decline (Kriger & Hero 2007). Furthermore, many of the so-called ‘enigmatic declines’ (Stuart et al. 2004) could have been caused by undetected disease events. To the authors’ knowledge no mass-mortality events of amphibians have been reported in Luxembourg. However, because of the patchy nature of amphibian recording, it is possible that they may have occurred without being witnessed. Alternatively Bd may simply be present but not causing ani-mals to die.Although several treatments or cures for Bd infection have been found (Woodhams et al. 2003, Bishop et al. 2008) these are imprac-ticable in the field, therefore Bd has not yet been mitigated for in the wild (Garner 2008).
Ascertaining the presence of Bd in Luxem-bourg, and neighbouring countries (Garner et al. 2005), is a key step to understanding the pressures faced by declining amphib-ian species in the region. The actual rate of infection, whilst regulating the impact of the disease, does not alter the biosecurity pro-tocol that should be implemented by field-workers. The disease is present, and care should be taken to prevent further spread by human agency.
4.3. Biosecurity issuesSimple disinfection procedures, such as washing boots in commercially available disinfectants, can be used by fieldworkers to lessen the risk of carrying Bd between study sites. At first this is time-consuming, how-ever it rapidly becomes part of the survey routine. It can be argued that such biose-curity measures are ‘too little, too late’ and even futile because wild animals move freely and frequently between ponds. Despite the counter-opinions, it is generally consid-ered that precautionary principles should be applied because of the potentially severe consequences of spreading Bd.A guideline leaflet produced by UK herpeto-logical conservation organisations provides up-to-date, simple advice on how to mini-mise the risk of transferring Bd between sites and the recommended disinfectants. The leaflet ‘Amphibian disease precautions: a guide for UK fieldworkers’ (Version 1, Feb-ruary 2008) is free to download from www.arg-uk.org.uk/Publications.htm. Anybody working in or near to amphibian aquatic habitat is strongly recommended to follow the guidelines, to report unusual die-off events and to be aware of the symptoms of disease.A great deal of research is being conducted on Bd around the world. The data from Lux-embourg contribute to knowledge of the extent of the disease. Further work could include retesting the samples that gave bor-derline positive results and making detailed local investigations to determine more pre-cisely where Bd infection occurs and its prevalence.
Bull. Soc. Nat. luxemb. 110 (2009) 113
AcknowledgementsThe disease testing was funded by the Administration de la nature et des forêts, during research work under LRW’s student bursary from the Ministère de la culture, de l´enseignement supérieur et de la recherche (Luxembourg, BFR 06/056). Many thanks to fieldwork assistants Liza Glesener and Véronique Ludwig who sped up the swabbing process greatly, and to Edmée Engel (Musée national d’histoire naturelle) for her collaboration and support throughout the project. Trent Garner and Frankie Clare from the Zoological Society of London (ZSL) kindly instructed LRW in sample collection and processing and provided laboratory facilities. We also thank Jennifer Sears, of ZSL and the Durrell Institute of Conservation and Ecology, whose knowledge of Bd was very helpful in writing this paper.
LiteratureBd database: www.spatialepidemiology.net/bd/
Global surveillance data for Batra-chochytrium dendrobatidis. Imperial College London. Accessed on the World Wide Web on 26/11/2008.
Berger, L., R. Speare, P. Daszak, D. E. Green, A. A. Cunningham, C. L. Goggin, R. Slocombe, M. A. Ragan, A. D. Hyatt, K. R. McDonald, H. B. Hines, K. R. Lips, G. Marantelli & H. Parkes, 1998. Chytridiomycosis causes amphibian mortality associated with population declines in the rain forests of Australia and Central America. Proceedings of the National Acad-emy of Sciences USA 95: 9031-9036.
Bishop, P. J., R. T. M. Poulter, J. N. Busby, P. J. Bishop, M. I. Butler, P. Harlow & R. Speare, 2008. Chloramphenicol is active against chytridiomycosis in frogs. Paper presented at the 6th World Congress of Herpetology, Manaus, Brazil. 17-22 August 2008.
Bosch, J., I. Martínez-Solano & M. García-París, 2001. Evidence of a chytrid fungus infection involved in the decline of the common mid-wife toad (Alytes obstetricans) in protected areas of central Spain. Biological Conserva-tion 97: 331–337.
Boyle, D. G., D. B. Boyle, V. Olsen, J. A. Morgan & A. D. Hyatt, 2004. Rapid quantitative detec-tion of chytridiomycosis (Batrachochytrium dendrobatidis) in amphibian samples using real-time Taqman PCR assay. Diseases of Aquatic Organisms 60:141-148.
Collins, J. P. & A. Storfer, 2003. Global amphibian declines: Sorting the hypotheses. Diversity and Distributions 9: 89-98.
Daszak, P., L. Berger, A. A. Cunningham, A. D. Hyatt, D. E. Green & R. Speare, 1999. Emerg-ing infectious diseases and amphibian popu-lation declines. Emerging Infectious Diseases 5: 735-748.
Daszak, P., A. A. Cunningham, A. D. Hyatt, 2003. Infectious disease and amphibian population declines. Diversity and Distributions 9: 141-150.
Garner, T. W. J., M. W. Perkins, P. Govindarajulu, D. Seglie, S. Walker, A. A. Cunningham & M. C. Fisher, 2006. The emerging amphibian pathogen Batrachochytrium dendrobatidis globally infects introduced populations of the North American bullfrog, Rana catesbeiana. Biology Letters 2: 455-459.
Garner, T. W. J., S. Walker, J. Bosch, A. D. Hyatt, A. A. Cunningham & M. C. Fisher, 2005. Chytrid fungus in Europe. Emerging Infec-tious Diseases 11: 1639-1641.
Garner, T. W. J., 2008. Mitigating Batra-chochytrium dendrobatidis in the wild: where we are now and where are we going? Paper pre-sented at the Zoological Society of London symposium: Halting the global decline in amphibians: Research and practice. London, UK. 20-21 November 2008.
Johnson, M. & R. Speare, 2003. Survival of Batra-chochytrium dendrobatidis in water: quaran-tine and disease control implications. Emerg-ing Infectious Diseases 9: 922-925.
Johnson, M. & R. Speare, 2005. Possible modes of dissemination of the amphibian chytrid Batrachochytrium dendrobatidis in the envi-ronment. Diseases of Aquatic Organisms 65: 181-186.
Kriger, K. M. & J.-M. Hero, 2007. Large-scale sea-sonal variation in the prevalence and severity of chytridiomycosis. Journal of Zoology 271: 352-359.
Lips, K. R., F. Brem, R. Brenes, J. D. Reeve, R. A. Alford, J. Voyles, C. Carey, L. Livo, A. P. Pessier & J. P. Collins, 2006. Emerging infec-tious disease and the loss of biodiversity in a Neotropical amphibian community. Proceed-ings of the National Academy of Sciences USA 103: 3165-3170.
Morehouse, E. A., T. Y. James, A. R. D. Ganley, R. Vilgaly, L. Berger, P. J. Murphy & J. E. Long-core, 2003. Multilocus sequence typing sug-gests the chytrid pathogen of amphibians is a recently emerged clone. Molecular Ecology 12: 395-403.
Pessier, A. P., D. K. Nichols, J. E. Longcore & M. S. Fuller, 1999. Cutaneous chytridiomycosis in poison dart frogs (Dendrobates spp.) and
114 Bull. Soc. Nat. luxemb. 110 (2009)
White’s tree frogs (Litoria caerulea). Journal of Veterinary Diagnostic Investigation 11: 194-199.
Pounds, J. A., M. P. L. Fogden & J. H. Campbell, 1999. Biological response to climate change on a tropical mountain. Nature 398: 611-615.
Stuart, S. N., J. S. Chanson, N. A. Cox, B. E. Young, A. S. Rodrigues, L. Fischman & R. W.
Waller, 2004. Status and trends of amphibian declines and extinctions worldwide. Science 306: 1783-1786.
Woodhams, D. C., R. A. Alford & G. Marantelli, 2003. Emerging disease of amphibians cured by elevated body temperature. Diseases of Aquatic Organisms 55: 65-67.