Diversity and Distributions, (Diversity Distrib.) (2008) 
BIODIVERSITY 
RESEARCH 
Bat assemblages on Neotropical land- 
bridge islands: nested subsets and null 
model analyses of species co-occurrence 
patterns 
Christoph F. J. Meyer1* and Elisabeth K. V. Kalko1,2 
^Institute of Experimental Ecology, University of 
Ulm, Albert-Einstein-Allee 11, 89069 Ulm, 
Germany, 2Smithsonian Tropical Research 
Institute, PO Box 0843-03092, Balboa, Panama 
^Correspondence: Christoph R I. Meyer, Institute 
of Experimental Ecology, University of Ulm, 
Albert-Einstein-Allee 11, 89069 Ulm, Germany. 
Tel.: +49-731-502-2675; 
Fax:+49-731-502-2683; 
E-mail: christoph.meyer@uni-ulm.de 
ABSTRACT 
A fundamental goal of ecology is to understand whether ecological communities 
are structured according to general assembly rules or are essentially dictated by 
random processes. In the context of fragmentation, understanding assembly patterns 
and their mechanistic basis also has important implications for conservation. 
Using distribution data of 20 bat species collected on 11 islands in Gatun Lake, 
Panama, we tested for non-randomness in presence-absence matrices with respect to 
nestedness and negative species co-occurrence. We examined the causal basis for the 
observed patterns and conducted separate analyses for the entire assemblage and 
for various species submatrices reflecting differences in species' trophic position 
and mobility. Furthermore, we explored the influence of weighting factors (area, 
isolation, abundance) on co-occurrence analyses. Unweighted analyses revealed 
a significant negative co-occurrence pattern for the entire assemblage and for 
phytophagous bats alone. Weighting analyses by isolation retained a pattern of species 
segregation for the whole assemblage but nullified the non-random structure 
for phytophagous bats and suggested negative associations for animalivores and 
species with low mobility. Area- and abundance-weighted analyses always 
indicated random structuring. Bat distributions followed a nested subset structure 
across islands, regardless of whether all species or different submatrices were analysed. 
Nestedness was in all cases unrelated to island area but weakly correlated with island 
isolation for incidence matrices of all species, phytophagous bats, and mobile species. 
Overall, evidence for negative interspecific interactions indicative of competitive 
effects was weak, corroborating previous studies based on ecomorphological analyses. 
Our findings indicate that bat assemblages on our study islands are most strongly 
shaped by isolation effects and species' differential movement and colonization 
ability. From a conservation viewpoint this suggests that even in systems with high 
fragment-matrix contrast, a purely area-based approach may be inadequate, and 
structural and functional connectivity among patches are important to consider 
in reserve planning. 
Keywords 
Assembly rules, Chiroptera, habitat fragmentation, land-bridge islands, nestedness, 
species co-occurrence. 
INTRODUCTION 
The question whether communities are shaped predominantly 
by biotic interactions such as competition or are randomly 
assembled from species pools has been pondered by ecologists 
for decades (Weiher & Keddy, 1999). Ever since Diamond (1975) 
in his seminal paper on assembly of insular communities 
posited that faunal assemblages are competitively structured, 
the identification and explanation of non-random patterns in 
assemblage composition have been a central theme in community 
ecology. Moreover, knowledge of patterns and causes of species 
distribution in insular biotas has been central in providing 
guidelines to biodiversity conservation, e.g. concerning reserve 
design (Whittaker, 1998). Diamond's controversial assembly 
rules model has been the focus of a long-standing and intense 
debate centred mainly around theoretical and statistical aspects 
? 2008 The Authors 
Journal compilation ? 2008 Blackwell Publishing Ltd 
DOI: 10.1111/j.1472-4642.2007.00462.x 
www.blackwellpublishing.com/ddi 
C. F. J. Meyer and E. K. V. Kalko 
(Connor & Simberloff, 1979; Strong etal, 1984). In a recent 
meta-analysis, Gotelli & McCabe (2002) demonstrated for a 
variety of taxa that assemblages with fewer co-occurring species 
than expected by chance are common, in line with Diamond's 
proposition that competitive interactions play a generally 
important role in structuring many species assemblages. Most 
recently, tests for species co-occurrence patterns have also been 
extended to include neutral models (Ulrich, 2004; Bell, 2005; 
Gotelli & McGill, 2006), which posit ecological equivalence among 
species and argue for random processes shaping local and 
regional community structure (Hubbell, 2001). Apart from 
Diamond's assembly rules a range of other patterns of community 
non-randomness has been suggested and examined, including 
Fox's favoured states model (Fox & Brown, 1993), core-satellite 
organization (Hanski, 1982; Ulrich & Zalewski, 2006), and nested 
subsets of species (Patterson & Atmar, 1986; Wright et ah, 1998). 
Nestedness is a pattern frequently reported for faunal 
assemblages in natural or anthropogenically fragmented 
systems and has been documented for a broad range of taxa 
(cf. review by Wright et ah, 1998). Assemblages exhibit a nested 
distributional pattern when the species present at species-poor 
sites constitute subsets of those from progressively species-rich 
sites, rather than a random draw of those present in the entire 
regional species pool (Patterson & Atmar, 1986). The concept 
of nestedness has been useful in interpreting community 
structure and function across a range of ecological studies, 
e.g. in situations where species assemblages are characterized 
by hierarchical niche relationships among species (Patterson 
& Brown, 1991) or, more recently, in studies of mutualistic 
species interactions (e.g. Bascompte et ah, 2003). Nested subset 
theory has also received considerable attention regarding its 
relevance to biodiversity management and conservation, 
concerning its potential to identify fragmentation-sensitive species, 
but particularly as it relates to the SLOSS debate (single large 
or several small) regarding reserve design where its utility, 
however, appears to be limited (Boecklen, 1997; Fischer & 
Lindenmayer, 2005; Martinez-Morales, 2005). 
In contrast to Diamond's assembly rules model, nested 
subset theory does not invoke competition as the structuring 
mechanism underlying community assembly. Instead, nested 
patterns in species assemblages are thought to arise as a result 
of species' differential colonization or extinction, nested habitat 
structure, passive sampling, as well as distance or area effects 
(Wright et ah, 1998). Nestedness appears to be particularly 
characteristic of extinction-dominated systems such as habitat 
fragments or land-bridge islands undergoing faunal relaxation, 
where species loss has been observed to occur in a predictable 
order based on species' differential extinction vulnerability 
(Patterson & Atmar, 1986; Wright et ah, 1998). 
In this study, we used null model analyses (Gotelli & Graves, 
1996) to test for patterns of species co-occurrence and nested- 
ness in presence-absence matrices of phyllostomid bats 
sampled as part of a comprehensive project investigating 
fragmentation effects on Neotropical bats within a landscape 
of small land-bridge islands in Gatun Lake, Panama (Meyer & 
Kalko, in press). Compared to other taxa, few studies have 
assessed bat assemblages with respect to species co-occurrence 
patterns and nested subset structure, and analyses to date have 
largely been restricted to bats on oceanic islands (Connor & Sim- 
berloff, 1979; Wright et ah, 1998; Gotelli & McCabe, 2002; 
Watling & Donnelly, 2006). Moreover, to our knowledge no 
study has so far investigated patterns of nestedness and species 
associations for bats in the context of habitat fragmentation. In 
view of their high functional significance in Neotropical ecosys- 
tems as pollinators, seed dispersers, and arthropod predators 
(Kalko, 1998; Patterson et ah, 2003), it is important to evaluate 
whether and how fragmentation affects the structure of 
phyllostomid bat assemblages. 
We explored possible mechanisms underlying the observed 
distribution patterns of bats on islands and examined the use 
of a series of weighting factors (island area, island isolation, 
species abundance) on the outcome of co-occurrence analyses. 
In addition to conducting analyses for the whole species 
assemblage, we were particularly interested in investigating 
differences in the outcome of co-occurrence and nestedness 
analyses for different subsets of species. In the case of analyses 
testing for negative species associations, this was predicated on 
the assumption that non-random co-occurrence patterns are less 
likely to be detected at the assemblage level, i.e. in datasets that 
include a large number of ecologically disparate species that 
obviously do not compete for resources. In contrast, at the level 
of ensembles (Fauth et ah, 1996) interactive effects among species 
can be expected to be stronger, a point that has recently been 
demonstrated for phyllostomid bats based on ecomorphological 
analyses (Moreno et ah, 2006). We therefore contrasted incidence 
matrices of phytophagous and animalivorous phyllostomid bats 
in our analyses. It is conceivable that in the context of habitat 
fragmentation, alterations in resources (food, roost sites) as a 
result of fragmentation may disrupt effects of species ecological 
interactions even at the ensemble level, making the detection of 
deterministic structure less likely. 
Dispersal or mobility is another factor that may shape patterns 
of species co-occurrence and nestedness (Cutler, 1991; Zalewski & 
Ulrich, 2006), but so far has rarely been explicitly considered in 
bat studies (but see Arita, 1997). If species differ in their ability to 
colonize new patches, then they will be differentially affected by 
fragmentation, suggesting that differential mobility should 
influence patterns of species distribution. In addition to conducting 
separate analyses for phytophagous and animalivorous bats, 
we therefore divided species based on differences in mobility 
(high vs. low mobility species). High species mobility can be 
expected to randomize faunal composition and should hence 
lead to random co-occurrence patterns. In contrast, we predicted 
an underdispersed pattern of species co-occurrence, i.e. negative 
species associations for incidence matrices of less vagile species 
(cf. Zalewski & Ulrich, 2006). 
Like many patterns and processes in ecology, the outcome of 
nestedness and species co-occurrence analyses may be contingent 
on spatial scale (Wright et ah, 1998; Jenkins, 2006), and this scale 
dependency has been explicitly addressed in various studies 
(e.g. Patterson & Brown, 1991; Gotelli & Ellison, 2002; Fischer & 
Lindenmayer, 2005). Here our analyses focus on the spatial scale of 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 
Nestedness and co-occurrence patterns in bat assemblages 
Table 1   Characteristics of the study islands in Gatun Lake, Panama. Island isolation is given as distance to the nearest mainland. For each island 
the number of phyllostomid bat species in each of the four species submatrices used in the co-occurrence and nestedness analyses is given. 
Island Area (ha) Isolation (km) 
Number of species 
ID no. Phytophagous Animalivorous High mobility Low mobility 
1 Guava 2.5 1.93 5 1 3 3 
2 Chicha 2.8 0.51 10 0 4 6 
3 Tres Almendras 3.4 0.15 12 3 5 10 
4 Pina* 4.4 0.02 11 3 5 9 
5 Mona Grita 5.9 0.25 9 3 7 5 
6 Guanabano 7.2 2.25 4 1 3 2 
7 Pato Horqueta 11.4 3.40 10 3 5 8 
8 Cacao 12.8 0.16 11 1 5 7 
9 Guacha 16.3 1.42 6 1 3 4 
10 Trinidad 17.3 2.02 6 1 3 4 
11 Leon 50 1.55 11 2 4 9 
?Island no. 8 in Adler & Seamon (1991) 
individual islands. Consequently, inferences based on our findings 
should be viewed as being most relevant at this local scale and 
may not necessarily apply to the same degree at larger spatial scales. 
METHODS 
Study area 
Lake Gatun is a large artificial reservoir, which was created 
in 1914 following the damming of the Chagres River as part of 
the construction of the Panama Canal. Lake formation isolated 
numerous former hilltops, resulting in a large number of 
forested islands ranging in size from < 1 ha to the 1560 ha Barro 
Colorado Island (BCI) (Adler & Seamon, 1991). We surveyed 
the bat fauna of a total of 11 islands that ranged in size from 
2.5 to 50 ha and that were located between 0.02 and 3.4 km 
from the mainland (Table 1). Islands in the lake are covered with 
semideciduous lowland tropical moist forest (Holdridge, 1967), 
which is typically shorter in stature and less diverse in tree species 
composition than on the adjacent mainland (Leigh et ah, 1993). 
The climate is highly seasonal with a long rainy season punctuated 
by a severe 4-month dry season (Windsor, 1990). 
Bat sampling 
On each island, bats were sampled in a standardized manner 
using at each site six 6-m mist nets set at ground level and one net 
wall consisting of four stacked 6-m nets, reaching subcanopy or 
canopy level depending on the height of the forest. Each island 
was sampled for seven entire nights between October 2003 and 
October 2005 with a minimum time interval of 30 days between 
netting nights to reduce possible net-shyness. The bats were 
marked and standard measurements and demographic data were 
collected following Handley et al. (1991). For a full account of 
sampling procedures see Meyer & Kalko (in press). Species 
accumulation curves and species richness estimators indicated a 
high level (> 95%) of inventory completeness for all study islands. 
In total, we obtained 8447 captures of 39 species belonging to five 
families (Meyer & Kalko, in press). 
In the context of our analyses it is important to consider 
whether the observed species distributions represent primarily 
patterns of occupancy rather than patterns of differential habitat 
use or foraging behaviour. To assess this, we calculated species 
turnover between successive sampling years as T =(/ + ?)/ 
(Sj + S2), where / is the number of species recorded in the second 
but not in the first year, E the number of species found in the 
first but not in the second year, and S1 and S2 the total number 
of species during both years (Aguirre et ah, 2003). Turnover rates 
can vary between 0 (no turnover) and 1 (complete turnover). 
Overall species turnover between sampling years was very low 
(0.08) and mean turnover rates across sites were similarly low 
(0.16 ?0.10 SD). Over the temporal scale of our study, we 
therefore believe that it is reasonable to consider the islands as 
sufficiently isolated for most species and to regard the presence- 
absence data used in our analyses as primarily representing 
species occurrence. It has to be noted, however, that our study 
sites cannot be considered completely insular for all species as some 
highly vagile species (mostlyArtibeusjamaicensis,A. Hturatus,and 
Uroderma bilobatum) may regularly move among neighbouring 
islands on a nightly basis (Meyer & Kalko, in press). But even for 
these highly mobile species, recapture data (Meyer & Kalko, in 
press) indicate that long-distance cross-water movements are 
rare. In any case, the water matrix seems to be a movement barrier 
significant enough that many species apparently do not regularly 
cross it; otherwise turnover rates should have been much higher. 
Data analysis 
Species matrices 
We excluded all non-phyllostomid species, since they cannot be 
adequately sampled with mist nets (Kalko, 1998) as well as one 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 
C. F. J. Meyer and E. K. V. Kalko 
Table 2 Phyllostomid bat species captured on Gatun Lake islands, 
Panama, and used in nestedness and species co-occurrence analyses. 
Nomenclature follows Simmons (2005). 
Number of 
islands 
ID occupied Trophic 
no. Species (abbreviation) (N??=ll) level* Mobility 
1 Artibeus jamaicensis (Ajam) 11 P High 
2 Artibeus lituratus (Alit) 11 P High 
3 Artibeus phaeotis (Apha) 4 P Low 
4 Artibeus watsotii (Awat) 6 P Low 
5 Carollia castanea (Ccas) 5 P Low 
6 Carollia perspicillata (Cper) 10 P Low 
7 Chiroderma villosum (Cvil) 7 P High 
8 Glossophaga soricina (Gsor) 10 P Low 
9 Lophostoma silvicolum (Lsil) 2 A Low 
10 Micronycteris hirsuta (Mhir) 3 A Low 
11 Micronycteris microtis (Mmic) 9 A Low 
12 Phyllostomus discolor (Pdis) 1 P High 
13 Phyllostomus hastatus (Phas) 2 A High 
14 Platyrrhinus helleri (Phel) 5 P Low 
15 Trinycteris nicefori (Tnic) 1 A Low 
16 Totiatia saurophila (Tsau) 2 A Low 
17 Uroderma bilobatum (Ubil) 11 P High 
18 Vampyrodes caraccioli (Vcar) 4 P High 
19 Vampyressa nymphaea (Vnym) 1 P Low 
20 Vampyressa pusilla (Vpus) 9 P Low 
*P, phytophagous; A, animalivorous (Giannini & Kalko, 2004, 2005). 
species, Centurio senex, that was present in the study area only 
during part of the rainy season at low numbers and for which 
sampling may thus also not have been sufficient. This resulted in 
a total of 20 phyllostomid species for analysis (Table 2) for 
which we constructed a species (rows) by site (columns) matrix. 
We additionally generated presence-absence matrices for four 
different species subsets as follows: first, we classified bats into 
phytophagous and animalivorous species based on dietary analyses 
(Giannini & Kalko, 2004, 2005). Second, we employed principal 
components analysis (PCA) to divide species according to their 
mobility (Table 3, Fig. 1). Variables included in the PCA were mean 
Table 3 Results of a principal components analysis on four 
variables related to species mobility, showing the loadings for each 
variable and the proportion of variance explained by the first two 
components (PCI andPC2). 
Variable PCI PC2 
Maximum recapture distance (dmax) -0.48 0.56 
Mean recapture distance (dmea]1) -0.58 0.04 
Aspect ratio -0.37 -0.84 
Wing loading -0.55 0.05 
Variance explained (%) 67.0 21.1 
and maximum recapture distances as well as two morphological 
variables, aspect ratio and wing loading (body weight divided 
by wing area), which are linked to mobility in bats. Species 
characterized by high wing loading and long and narrow wings 
(high aspect ratio) are fast and energy-efficient flyers, whereas 
those with short and broad wings have higher manoeuvrability 
in cluttered habitats but increased commuting costs over larger 
distances (Norberg & Rayner, 1987). 
Since we obtained reliable mark-recapture data only for a 
limited set of species during our study period, we calculated 
movement distances based on a large dataset collected as part of 
the BCI long-term bat project with > 40,000 capture/recapture 
data (Kalko et ah, 1996; unpublished data), which covered all 
the species included in our analyses. Although these data mostly 
constitute movement distances within a rather large patch of 
unfragmented forest rather than cross-water movements, we 
believe that they reflect the general movement abilities of the 
different species reasonably well. Values for aspect ratio and wing 
loading (NnT2) were taken from Meyer et al. (2008) and von 
Staden (2002). Based on the ordination diagram two clusters of 
species characterized by high and low mobility, respectively, 
could be distinguished (Fig. 1). This categorization (Table 2) 
was subsequently used for the construction of separate incidence 
matrices for nestedness and co-occurrence analyses. 
Co-occurrence patterns 
Co-occurrence indices. We calculated two metrics of co-occurrence, 
the checkerboard score (C-score) introduced by Stone & Roberts 
(1990) and the variance ratio (V-ratio) popularized by Schluter 
(1984). The number of checkerboards and the number of species 
combinations, two other commonly employed co-occurrence 
measures, have been shown to be prone to type II errors and may 
not reveal significant patterns in noisy datasets. Conversely, the 
C-score and the V-ratio are based on the average co-occurrence 
and covariance, respectively, of all species pairs and are therefore 
relatively insensitive to noise in the data (Gotelli, 2000). The 
C-score in particular has been demonstrated to be superior to the 
other indices with respect to type I and II error rates. For a 
detailed account of the statistical properties and performance of 
these indices, see Gotelli (2000). The C-score measures the extent 
to which species are segregated across sites but does not require 
perfect checkerboard distributions (Gotelli, 2000). For any two 
species, the number of 'checkerboard units' (CU) is defined as 
CU= (_Rj-S)(J?rS), where _R; and _Rj are the number of occurrences 
(= row totals) for species i and j, respectively, and S denotes the 
number of co-occurrences. The C-score is the average number of 
checkerboard units over all possible pairs of species in the matrix. 
The V-ratio measures the variability in the number of species 
per site and represents the ratio between the variance in species 
richness per site (= variance of column sums), and the sum of the 
variances of species occurrence over sites (= sum of row variances) 
(Schluter, 1984; Gotelli, 2000). The ratio equals 1 if the average 
covariance between species pairs is 0. In the case of positive or 
negative covariance between species pairs, the V-ratio is smaller 
or greater than 1, respectively (Gotelli, 2000). In an assemblage 
structured by negative species interactions, the C-score should be 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 
Nestedness and co-occurrence patterns in bat assemblages 
CO 
CM 
< O 
CL 
o     g 
2   "D 
O CM 
a> 1 
Q. 
to 
< 
?-l 
1 
-1 
I               I 
0              1 
PCA1 (67.0%) 
I 
2 
dmax, dmean, aspect ratio, wing loading 
Figure 1   Ordination of 20 phyllostomid bat 
species in the principal components space of 
a set of variables related to species mobility. 
For full species names see Table 2; dmean and 
dmaK = mean and maximum recapture 
distance, respectively. 
significantly larger than expected by chance, while the opposite 
should be true for the V-ratio (Gotelli, 2000; Gotelli & McCabe, 
2002). 
Null models and randomization algorithms. We calculated the 
above indices and compared them to those of 5000 randomly 
assembled communities using the software EcoSim 7.72 (Gotelli 
& Entsminger, 2006). The outcome of species co-occurrence 
analyses is sensitive to the selection of appropriate null models 
and choice of the randomization algorithm. We used the 
sequential-swap algorithm to generate random null matrices 
(Gotelli, 2000; Gotelli & Entsminger, 2001). Although use of the 
swap algorithm has been subject to criticism (Sanderson et al., 
1998; Manly & Sanderson, 2002), re-evaluation (Gotelli & 
Entsminger, 2003) confirmed its overall good statistical properties 
and performance. We compared three alternative null models, 
differing in the way row and column totals are treated: 
1 Fixed-fixed (FF). With this algorithm, both the row and the 
column totals of the original matrix are fixed, thus preserving 
differences in occurrence frequencies among species (row sums) 
and differences in species richness among sites (column sums). 
Gotelli (2000) advocated these constraints particularly for island 
datasets. As the V^-ratio is exclusively determined by the row and 
column totals of the matrix, it cannot be derived for this null 
model (Gotelli, 2000). 
2 Fixed-weighted (FW). Results of co-occurrence analyses can 
be strongly affected by the use of weighting factors, although this 
approach so far has only infrequently been applied (but see 
Gotelli & Ellison, 2002; Jenkins, 2006). Column weighting adjusts 
the probability of a species occurring at a particular site during 
randomization using factors thought to contribute to intersite 
differences in community composition. Here, we explored the 
influence of two weighting factors, island area and island isolation 
(distance from the mainland), on the outcome of null model 
analyses. 
3 Weighted-fixed (WF). Co-occurrence scores may be affected 
by sampling artefacts due to differences in species abundances or 
detection probabilities (MacKenzie et al, 2004; Peres-Neto, 2004; 
Ulrich & Zalewski, 2006). We explored the possible influence of 
such sampling errors by using a null model with fixed column 
totals but weighting the row totals by setting the occurrence 
frequency of each species proportional to its total relative 
abundance across all sites. 
To allow for meaningful comparisons of our results with those 
from other studies, we calculated a standardized effect size (SES) 
as ([observed score - mean simulated score] /standard deviation 
of simulated scores), which indicates the number of standard 
deviations that the observed index is above or below the mean 
index of simulated matrices (Gotelli & McCabe, 2002; Gotelli & 
Entsminger, 2006). 
Quantification ofnestedness 
A variety of different methods are available for calculating the 
nestedness of presence-absence matrices (reviewed by Wright et ah, 
1998). Of these, Atmar & Patterson's Nestedness Temperature 
Calculator (NTC; Atmar & Patterson, 1995) is the most widely used 
in biogeographical studies and analyses of habitat fragmentation. 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 
C. F. J. Meyer and E. K. V. Kalko 
Table 4 Summary of species co-occurrence analyses of phyllostomid bats with observed and simulated co-occurrence metrics, standardized 
effect sizes [SES], and associated tail probabilities (in parentheses) for each of the different species presence?absence matrices and null model 
algorithms. For details see text. Significant and marginally significant results are highlighted in bold, na, not applicable, as the V-ratio cannot be 
tested with the fixed-fixed null model. 
Species matrix 
Score 
Index obs simFf SiniFW-area SlrnFW-isolatioii simvw 
C-score [SES] All species 1.368 1.245 (0.026) 1.417(0.554) 0.915 (0.033) 1.513 (0.634) 
[2.28] [-0.16] [1.93] [-0.38] 
Phytophagous 0.890 0.764(0.041) 1.311 (0.878) 0.710 (0.256) 1.366 (0.869) 
[2.14] [-1.15] [0.64] [-1.13] 
Animalivorous 2.267 2.399 (0.679) 1.693 (0.222) 1.493 (0.076) 3.686 (0.902) 
[-0.49] [0.83] [1.59] [-1.23] 
High mobility 0.143 0.143 (1.000) 0.613 (0.923) 0.453 (0.876) 0.456 (0.818) 
[0.00] [-1.29] [-1.04] [-0.82] 
Low mobility 1.821 1.733 (0.188) 1.863 (0.535) 1.166(0.048) 2.288 (0.761) 
[0.88] [-0.09] [1.79] [-0.74] 
V-ratio [SES] All species 4.053 na 3.746 (0.661) 
[0.37] 
5.021 (0.083) 
[-1.42] 
4.128 (0.432) 
[-0.27] 
Phytophagous 3.819 na 2.808 (0.946) 
[1.55] 
4.005 (0.406) 
[-0.32] 
3.624 (0.814) 
[0.82] 
Animalivorous 1.264 na 1.785 (0.253) 
[-0.91] 
1.811 (0.151) 
[-1.21] 
1.201 (0.753) 
[0.45] 
High mobility 2.119 na 1.529 (0.956) 
[1.40] 
1.694(0.927) 
[1.15] 
1.954 (0.839) 
[0.82] 
Low mobility 3.157 na 3.022 (0.609) 
[0.19] 
4.066 (0.070) 
[-1.55] 
3.054 (0.712) 
[0.46] 
The matrix temperature T calculated by the NTC is a measure of 
unexpected species presences and absences or system 'disorder' 
where 0? corresponds to a perfectly nested matrix and 100? 
indicates a random species distribution pattern. However, 
recent research cautions against use of the NTC as it has been 
demonstrated to have a number of important shortcomings 
relating to the definition of the isocline of perfect order, the way of 
matrix reorganization, the robustness of the packing algorithm, 
and choice of an appropriate null model (Fischer & Lindenmayer, 
2002; Rodriguez-Girones & Santamaria, 2006; Ulrich & Gotelli, 
2007). 
Here we used the binary matrix nestedness temperature 
calculator (BINMATNEST) recently developed by Rodriguez- 
Girones & Santamaria (2006), which is the latest in a growing line 
of algorithms to quantify nestedness which overcome these 
difficulties. BINMATNEST implements an isocline of perfect 
order that is unambiguously defined, is based on robust genetic 
algorithms to determine the reordering of rows and columns 
that leads to minimum matrix temperature, and provides a set of 
three alternative null models to assess the statistical significance 
of matrix temperature (Rodriguez-Girones & Santamaria, 2006). 
Following the authors' recommendations, we used the null 
model 3 in evaluating statistical significance as it has been shown 
to be associated with the smallest type I error. Reported f-values 
are based on 5000 random matrices. To evaluate the causal role of 
colonization and extinction in shaping community structure, 
we used a Spearman rank correlation between the matrix 
reorganization vectors, i.e. island rank order in the maximally 
packed matrix, and island isolation and area, respectively (Pat- 
terson & Atmar, 2000; Rodriguez-Girones & Santamaria, 2006). 
RESULTS 
Co-occurrence patterns 
Results of analyses depended considerably on the type of null 
model algorithm employed (Table 4). For the FP-model, the 
observed C-score for the incidence matrix of the entire assemblage 
and for the submatrix of phytophagous bats was significantly 
higher than expected by chance, suggesting a negative pattern 
of species co-occurrence. Conversely, for the sub matrices of 
animalivorous bats and species of both high and low mobility, 
the C-score did not deviate from null model expectations, 
indicating random species co-occurrence. Using island isolation 
as a column constraint in the analysis {FW-isolation) produced 
partly contrasting results. This model detected significant non- 
randomness in species co-occurrence again for all species but 
also for the subsets of animalivorous bats and less mobile species, 
whereas the pattern for phytophagous and highly mobile species 
was not significant (Table 4). The use of island area as weighting 
factor (FW-area), on the other hand, suggested random co- 
occurrence patterns as the null hypothesis was never rejected 
for any of the species matrices. This was similarly true for the 
abundance-weighted (WF) model (Table 4). 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 
Nestedness and co-occurrence patterns in bat assemblages 
Table 5 Results of nestedness analyses conducted on the species by site matrix for all phyllostomid bats and for different subsets of phyllostomid 
species caught on 11 islands in Gatun Lake, Panama. Given are observed matrix temperatures (Tobs), expected nestedness temperatures (Texp), as 
well as Monte Carlo-derived probabilities that the matrix was randomly generated. Also indicated are the results of Spearman rank correlations 
of island rank order in the maximally nested matrix with the rank order of island area and isolation. 
Rank correlation with island 
Area Isolation 
Species matrix 7L, (SD) 
All 14.80 37.40 (5.36) < 0.0001 0.109 0.745 0.618 0.046 
Phytophagous 10.47 28.71(6.45) 0.001 -0.073 0.839 0.582 0.063 
Animalivorous 9.16 21.48 (8.63) 0.062 0.100 0.765 0.473 0.141 
High mobility 0.83 19.63 (7.38) < 0.0001 0.100 0.765 0.536 0.091 
Low mobility 14.04 33.78 (6.38) 0.0004 -0.064 0.860 0.509 0.110 
The V-ratio detected marginally significant deviations from 
null expectations only for the PW-model weighted by island 
isolation and when either all species or bats characterized by 
low mobility were considered. In the remainder of the cases, 
the V-ratio did not differ significantly from random expectations 
(Table 4). 
Nestedness 
Phyllostomid bat assemblages on Gatun Lake islands were highly 
significantly nested when all species were considered 
(P < 0.0001, Table 5, Fig. 2). The bat distribution across islands 
remained more significantly nested (P < 0.001) than expected by 
chance using incidence matrices of phytophagous species alone 
or based on the data sets comprising species of both mobility 
classes. For gleaning animalivores, the difference between observed 
and expected nestedness temperature was marginally significant 
(P = 0.062) (Table 5). Spearman rank correlations between row 
order in the maximally nested matrix with causal factors suggested 
that island isolation was an important determinant of nestedness 
in phyllostomid bat distributions for incidence matrices of the whole 
phyllostomid assemblage (rs = 0.62, P = 0.046), phytophagous bats 
(r, = 0.58, P = 0.063), and mobile species (r, = 0.54, P = 0.091) 
but not for animalivorous (rs = 0.47, P = 0.141) or less vagile 
species (rs = 0.51, P = 0.110). In contrast, the nested order of 
islands was unrelated to the rank order of island areas for all five 
species matrices (Table 5), indicating that island area is not 
causally linked to nested structure in our study system. 
ID 
1 
17 
2 
8 
6 
11 
20 
7 
4 
18 
5 
14 
3 
16 
10 
9 
13 
12 
15 
19 
Sites 
11       8 
0 0 1 
1 0 0 
0 0 1 
0 1 0 
0 1 1 
0 
0 
0 
0 
0 1 
1 0 
1 0 0 
10 
Figure 2 Maximally nested presence?absence matrix of 
phyllostomid bat distributions on Gatun Lake islands, Panama. For 
island and species ID numbers refer to Tables 1 and 2, respectively. 
DISCUSSION 
Co-occurrence patterns 
Overall, co-occurrence analyses did not provide strong evidence 
that species composition of phyllostomid bats on Gatun Lake 
islands is highly structured by negative interspecific interactions. 
Corroborating recent findings by Jenkins (2006), the outcome 
of analyses was sensitive to weighting factors, adding further 
support to the notion that, whenever possible, co-occurrence 
analyses should incorporate weights for important factors likely 
to contribute to the observed patterns, such as in our case, island 
isolation. Moreover, as expected, different results emerged depending 
on whether the whole assemblage or particular species subsets were 
considered. Unweighted analyses based on the C-score showed a 
negative pattern of co-occurrence indicating mutually exclusive 
species distributions for the entire assemblage as well as for 
phytophagous species, whereas a random pattern was suggested 
for all other species matrices examined. Area- and abundance- 
weighted analyses always indicated random assemblage structure 
for both co-occurrence indices and irrespective of the species 
matrix analysed. Weighting analyses by island isolation, however, 
retained a non-random pattern for the whole species set but 
rendered the result non-significant for phytophagous bats. 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 
C. F. J. Meyer and E. K. V. Kalko 
By contrast, we found that less vagile bats, and based on the 
C-score also animalivorous species, tended to occur together less 
often than expected by chance, whereas our results suggest 
random structuring for species with high mobility. This finding is 
consistent with our initial hypothesis that high species mobility 
should lead to random assemblage composition while incidence 
matrices of less mobile species should be more likely to exhibit 
non-random structure. This is because colonization rates compared 
to local persistence should be higher in mobile species, whereas 
those with limited mobility should be more affected by local 
extinction processes (Zalewski & Ulrich, 2006). Our results 
generally support Zalewski & Ulrich's (2006) call for taking 
species' differential dispersal abilities into account when analysing 
patterns of community assembly. Moreover, our results indicate 
that island isolation but not area had to some degree confounded 
unweighted analyses, a finding in line with a marked species-distance 
effect and absence of a significant species-area relationship at 
local scales (Meyer & Kalko, in press). 
A meta-analysis by Gotelli & McCabe (2002) of nearly one 
hundred species presence-absence matrices of a diverse array of 
invertebrate and vertebrate taxa revealed that assemblages of 
plants, ants, birds, bats, and non-volant mammals exhibited 
non-random patterns of species co-occurrence consistent 
with Diamond's (1975) assembly rules model stressing the 
importance of competitive interactions. Non-random matrices 
are characterized by having SES for the C-score > |2.0| (Gotelli 
& McCabe, 2002). These authors reported strong deviations 
from randomness as based on average SES values for birds 
(3.65) and non-volant mammals (3.10). Particularly strong 
effects of species segregation were also indicated for three 
presence-absence matrices of bats (average SES > 4.0). In the 
present study, incidence matrices that showed significant or 
marginally significant deviation from randomness had absolute 
SES values between 1.42 and 2.28 (Table 4), i.e. values closer to 
those found for herps and most invertebrate assemblages 
(average SES < 1.5) (Gotelli & McCabe, 2002). Spatial scale of the 
study may have a strong effect on C-scores and SES obtained in 
co-occurrence analyses (Jenkins, 2006). As Gotelli & McCabe's 
(2002) analysis for bats was exclusively based on presence- 
absence matrices of bats on oceanic islands, the discrepancy 
between studies likely indicates different structuring mechanisms 
prevailing on old oceanic islands vs. recently isolated land- 
bridge islands. More specifically, it possibly reflects fundamental 
differences between these systems regarding geographical scale, 
size, age, habitat diversity, and resource abundance and echoes 
the importance of evolutionary processes and historical events 
in shaping patterns of species distribution on oceanic islands 
(Drake etal., 2002). 
Ulrich (2004) and Bell (2005) recently evaluated whether 
patterns of species co-occurrence could be accounted for by 
invoking a neutral community model in which species are 
regarded as being ecologically equivalent and local assemblage 
structure is determined by random colonization, migration, 
and extinction (Hubbell, 2001). Interestingly, they found that 
non-random patterns of species segregation may indeed be 
generated just as well by neutral ecological drift models as by 
traditional null models. By analogy with genetic drift, zero-sum 
ecological drift models imply that relative abundances of species 
that are ecologically equivalent should change only owing to 
chance events (Hubbell, 2001). However, in Ulrich's (2004) analysis 
SES values generated by the neutral model for the C-score were 
comparatively low (c. 0.5). Therefore, it has been argued that the 
strong negative co-occurrence patterns found by Gotelli & 
McCabe (2002) for some taxa cannot be accounted for solely on 
the basis of a neutral model (Gotelli & McGill, 2006). This may 
also be because neutral theory is mainly concerned with resident 
organisms and neutral models may therefore have limited 
applicability for mobile animals (Chave, 2004). Moreover, 
because many of the important parameters in neutral models can 
rarely be measured directly, this greatly limits their utility as a null 
hypothesis for testing empirical patterns (Gotelli & McGill, 2006). 
Finally, in a more applied context, it has been contended that 
neutral theory cannot adequately address the question of how 
fragmentation will differentially alter the composition of species 
and their interactions with other species in the community 
(Chase, 2005). This is because the neutral theory explicitly dis- 
regards differences in species traits even though species are known 
to be differentially affected by fragmentation based on their traits 
(Henle etal, 2004). Mobility in our case apparently plays an 
important role in determining assemblage structure, which 
would be generally congruent with a neutral model. However, 
whether the non-random patterns observed to some degree in 
our analyses reflect competitive interactions or are attributable 
mainly to stochastic processes, remains an open question. 
Competitive interactions could result in ecomorphological size 
divergence via character displacement. Alternatively, competition 
may not be sufficiently strong to effect the local extinction of 
species but may reduce the abundance of those species that 
experience more competitive pressure, a phenomenon known as 
density compensation (Patterson etal., 2003). Meta-analyses 
that evaluated the degree to which each of five ensembles 
(aerial insectivores, frugivores, gleaning animalivores, high-flying 
insectivores, and nectarivores) from 15 bat assemblages throughout 
the New World show signs of competitive species interactions, have 
provided little support for pervasive and consistent deterministic 
structuring across most locations based on these two independent 
lines of evidence, although non-randomness was detected in a few 
cases (Stevens & Willig, 1999,2000). This may reflect the fact that, 
in studies conducted over large areas, high heterogeneity and 
variability in environmental conditions may prevent competitive 
interactions from inducing deterministic structure in a ubiquitous 
fashion (Stevens & Willig, 2000; Moreno et al., 2006). 
Our results suggest that this probably applies equally well to 
heterogeneous fragmented landscapes. Nevertheless, certain 
evidence for effects of ecological interactions on bat community 
assembly mechanisms comes from a recent study by Moreno 
etal. (2006). Also following an ecomorphological approach 
and focusing on the local habitat scale, they detected significant 
non-random patterns at the ensemble level for frugivorous 
phyllostomid bats, indicating that when environmental conditions 
are sufficiently homogeneous, interspecific interactions may to 
some degree structure local bat assemblages. 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 
Nestedness and co-occurrence patterns in bat assemblages 
Nested ness 
According to our analysis, assemblages of phyllostomid bats on 
Gatun Lake islands exhibit a highly nested structure such that 
species that occur on depauperate islands are also found on 
larger, more species-rich islands. While nested subset patterns seem 
to be common in ecological systems, comparative assessments 
indicate that nestedness is particularly prevalent in systems that 
are mainly shaped by extinction processes mediated through area 
effects (Patterson & Atmar, 1986; Wright et al, 1998; Patterson 
& Atmar, 2000; Feeley, 2003; Watling & Donnelly, 2006). 
Land-bridge islands or habitat fragments are typically uniformly 
colonized patches where species loss occurs in most cases selec- 
tively and in a predictable order based on species' differential 
extinction vulnerability, e.g. due to differences in area require- 
ments, resulting in a nested subset structure. Such a mechanism 
of area-related extinction during faunal relaxation has for 
instance been reported for resident bird assemblages on islands 
in Lake Guri, an artificial reservoir in Venezuela (Feeley, 2003), 
where islands were isolated only 20 years ago. 
By contrast, we found that island nested rank order was sig- 
nificantly or marginally significantly correlated with the rank 
order of island isolation but not island area for the entire assem- 
blage as well as for phytophagous and highly mobile species. This 
mainly reflects, and is in agreement with the finding of a strong 
effect of island isolation on species richness and lack of a significant 
species-area relationship at the local scale (Meyer & Kalko, 
in press). Thus, isolation-dependent, selective colonization, 
enhanced by the rescue effect (Brown & Kodric-Brown, 1977) by 
lowering local extinction rates of mobile species (Wright et ah, 
1998), appears to be the likely cause of nestedness and dominant 
process structuring phyllostomid bat assemblages on our study 
islands. The reason for this may also be that bat assemblages on 
Gatun Lake islands, which are > 90 years old, have likely reached a 
quite stable species richness and composition, which is suggested 
by the very low turnover rates observed across sampling years. 
Conversely, avian assemblages on the much more recently isolated 
Lake Guri islands are probably still undergoing relaxation, 
whereby local extinction continues to be the dominant structuring 
process (Feeley, 2003). This may explain, at least in part, the 
contrasting patterns observed with respect to the causal factors 
underlying nestedness in both systems. 
Differential colonization can produce nested subset patterns if 
highly mobile species are present even on the most isolated islands 
and less vagile species occupy only the closer islands. Our findings 
support this notion insofar as this pattern seems to be mainly 
driven by differential mobility between generally more mobile 
phytophagous bats and comparatively less vagile animalivorous 
species. This suggests that factors related to species' fragmentation 
sensitivity can be important in determining patterns of faunal 
nestedness and assemblage composition in fragmented landscapes. 
Even though we found nestedness to be unrelated to area on 
the scale of individual islands, i.e. at the spatial scale at which 
sampling was performed, area-dependent local extinction may 
well be the dominant structuring process shaping assemblages 
at the landscape scale. At this larger spatial scale the amount of 
forest cover was the chief determinant of species richness and 
composition (Meyer & Kalko, in press). This suggests that our 
findings regarding the underlying causes of nestedness in our 
study system may be scale-dependent, as has been found in other 
studies that have explicitly addressed the scale-dependency of 
nestedness patterns (e.g. Fischer & Lindenmayer, 2005). 
Martinez-Morales (2005) advocated the use of nestedness 
analyses as a potentially valuable tool for conservation to identify 
species sensitive to fragmentation. He found that certain groups of 
tropical cloud forest birds exhibited a nested structure significantly 
correlated with fragment area. This contrasts with our findings as 
phytophagous bats, which showed a nested arrangement correlated 
with island isolation, can generally be considered relatively 
fragmentation-tolerant compared to animalivores (Meyer et ah, 
2008) and suggests that such an approach may not be generally 
applicable and needs to be explored further in future studies. 
Also, responses to fragmentation are often species-specific, with 
some species being negatively affected and others benefiting from 
fragmentation. Therefore, because of their focus on a unidirectional 
change in species composition, nestedness analyses may not be 
an ideal tool to identify fragmentation-sensitive species (Fischer 
& Lindenmayer, 2005). 
Nested habitat distributions may also produce nested subsets 
if many species are habitat specialists, however, this is unlikely to 
contribute to the observed pattern of nestedness in our case as 
habitat heterogeneity is relatively low across the study islands 
(Meyer & Kalko, in press). Recently, Higgins et al. (2006) showed 
that stochastic processes, such as the random placement of 
individuals according to different species-abundance and island- 
size distributions, can by itself result in nested subset structure. 
Their study demonstrated that in interactive systems in which 
the species-abundance distribution of each island is determined 
largely by colonization dynamics rather than in situ dynamics, 
individual-based processes become important in generating 
non-random patterns of species composition such as nestedness. 
For bats, our study islands can clearly be regarded as such an 
interactive system suggesting that random processes may, at least 
in part, account for the strong degree of nestedness detected in 
our analyses. 
Conclusions 
In summary, there was limited evidence for negative species 
associations congruent with niche-based community assembly 
invoking competitive interspecific interactions. In line with previous 
findings our results suggest that deterministic structuring may be 
hard to detect in situations with high heterogeneity in environ- 
mental conditions such as in fragmented landscapes. Patterns 
of nestedness and species co-occurrence indicate that assemblage 
composition of phyllostomid bats on Gatun Lake islands is in 
large part determined by isolation-dependent, differential 
colonization reflecting species-specific differences in mobility. 
In an applied context, our results imply that community-level 
nestedness indices, when applied at a spatial scale that better 
matches patch-utilization than birth-death processes, may be 
inadequate as conservation planning tools (see also Fischer & 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 
C. F. J. Meyer and E. K. V. Kalko 
Lindenmayer, 2005). Even if assemblages exhibit significant 
nestedness, particularly in fragmented landscapes with a high patch 
to matrix contrast, conservation strategies should entail measures 
to incorporate the importance of structural and functional 
connectivity among remnant patches into reserve planning. 
ACKNOWLEDGEMENTS 
We thank the Smithsonian Tropical Research Institute for logistical 
support, the Autoridad del Canal de Panama for permission to 
work on the islands in Gatun Lake, and the following people 
for help with fieldwork: A. Bravo, K. Burger, S. Estrada Villegas, 
J. Frund, I. Geipel, N. Herdina, A. Lang, J. Nagel, A. Reside, 
R. Rodriguez, A. Sjollema, C. Stubenrauch, and C. Weise. We are 
grateful to Bruce Patterson, Luis Aguirre, and two anonymous 
reviewers for helpful comments on an earlier version of the 
manuscript. Miguel Rodriguez-Girones kindly provided a copy 
of the BINMATNEST program. Financial support was provided by 
grants from the German Academic Exchange Service (DAAD) to 
C. Meyer and the German Science Foundation (DFG) to E. Kalko. 
REFERENCES 
Adler, G.H. & Seamon, J.O. (1991) Distribution and abundance 
of a tropical rodent, the spiny rat, on islands in Panama. Journal 
of Tropical Ecology, 7, 349-360. 
Aguirre, L.F., Lens, L., van Damme, R. & Matthysen, E. (2003) 
Consistency and variation in the bat assemblages inhabiting 
two forest islands within a neotropical savanna in Bolivia. 
Journal of Tropical Ecology, 19, 367-374. 
Arita,  H.T.   (1997)   Species  composition  and  morphological 
structure of the bat fauna of Yucatan, Mexico. Journal of Animal 
Ecology, 66, 83-97. 
Atmar, W & Patterson, B.D. (1995) The nestedness temperature 
calculator: a visual basic program, including 294 presence- 
absence matrices. AICS Research, Inc., University Park, NM 
and The Field Museum, Chicago, IE. 
Bascompte, I., Jordano, P., Melian, C.J. & Olesen, J.M. (2003) 
The nested assembly of plant-animal mutualistic networks. 
Proceedings of the National Academy of Sciences USA, 100, 
9383-9387. 
Bell, G. (2005) The co-distribution of species in relation to the 
neutral theory of community ecology. Ecology, 86, 1757-1770. 
Boecklen, W.J. (1997) Nestedness, biogeographic theory, and the 
design of nature reserves. Oecologia, 112, 123-142. 
Brown, J.H. & Kodric-Brown, A. (1977) Turnover rates in insular 
biogeography: effect of immigration on extinction. Ecology, 
58,445-449. 
Chase, J.M. (2005) Towards a really unified theory of meta- 
communities. Functional Ecology, 19, 182-186. 
Chave, J. (2004) Neutral theory and community ecology. Ecology 
Letters, 7,241-253. 
Connor, E.F. & Simberloff, D. (1979) The assembly of species 
communities: chance or competition? Ecology, 60,1132-1140. 
Cutler, A.H. (1991) Nested faunas and extinctions in fragmented 
habitats. Conservation Biology, 5, 496-505. 
Diamond, J.M. (1975) Assembly of species communities. Ecology 
and evolution of communities (ed. by M.L. Cody and J.M. Dia- 
mond), pp. 342-444. Harvard University Press, Cambridge, MA. 
Drake, D.R., Mulder, C.P.H., Towns, DR. & Daugherty, C.H. 
(2002) The biology of insularity: an introduction. Journal of 
Biogeography, 29, 563-569. 
Fauth, J.E., Bernardo, J., Camara, M., Resetarits Jr, W.J., Van 
Buskirk, J. & McCollum, S.A. (1996) Simplifying the jargon of 
community ecology: a conceptual approach. The American 
Naturalist, 147, 282-286. 
Feeley, K. (2003) Analysis of avian communities in Lake Guri, 
Venezuela, using multiple assembly rule models. Oecologia, 
137,104-113. 
Fischer, J. & Lindenmayer, 0.B. (2002) Treating the nestedness 
temperature calculator as a 'black box' can lead to false conclu- 
sions. Oikos, 99, 193-199. 
Fischer, J. & Lindenmayer, D.B. (2005) Perfectly nested or sig- 
nificantly nested - an important difference for conservation 
management. Oikos, 109,485-495. 
Fox, B.J. & Brown, J.H. (1993) Assembly rules for functional 
groups in North American desert rodent communities. Oikos, 
67, 358-370. 
Giannini, N.P. & Kalko, E.K.V. (2004) Trophic structure in a large 
assemblage of phyllostomid bats in Panama. Oikos, 105,209 -220. 
Giannini, N.P. & Kalko, E.K.V. (2005) The guild structure of 
animalivorous leaf-nosed bats of Barro Colorado Island, 
Panama, revisited. Acta Chiropterologica, 7, 131-146. 
Gotelli, N.J. (2000) Null model analysis of species co-occurrence 
patterns. Ecology, 81,2606-26021. 
Gotelli, N.J. & Ellison, A.M. (2002) Assembly rules for New 
England ant assemblages. Oikos, 99, 591-599. 
Gotelli, N.J. & Entsminger, G.L. (2001) Swap and fill algorithms 
in null model analysis: rethinking the knight's tour. Oecologia, 
129,281-291. 
Gotelli, N.J. & Entsminger, G.L. (2003) Swap algorithms in null 
model analysis. Ecology, 84, 532-535. 
Gotelli, N.J. & Entsminger, G.L. (2006) EcoSim: null models soft- 
ware for ecology. Version 7. Acquired Intelligence Inc. & Kesey- 
Bear., Jericho, VT 05465. http://www.garyentsminger.com/ 
ecosim/index.htm. 
Gotelli, N.J. & Graves, G.R. (1996) Null models in ecology. Smith- 
sonian Institution Press, Washington, D.C. 
Gotelli, N.J. & McCabe, DJ. (2002) Species co-occurrence: a 
meta-analysis of J. M. Diamond's assembly rules model. Ecology, 
83,2091-2096. 
Gotelli, N.J. & McGill, B.J. (2006) Null versus neutral models: 
what's the difference? Ecography, 29, 793-800. 
Handley Jr, CO., Wilson, DE. & Gardner, AL. (1991) Demography 
and natural history of the common fruit bat Artibeus jamaicensis 
on Barro Colorado Island, Panama. Smithsonian Institution 
Press, Washington, D.C. 
Hanski, I. (1982) Communities of bumblebees: testing the core- 
satellite hypothesis. Annales Zoologici Fennici, 19, 65-73. 
Henle, K., Davies, K.F., Kleyer, M., Margules, C. & Settele, J. (2004) 
Predictors of species sensitivity to fragmentation. Biodiversity 
and Conservation, 13,207-251. 
10 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 
Nestedness and co-occurrence patterns in bat assemblages 
Higgins, C.L., Willig, M.R. & Strauss, R.E. (2006) The role of 
stochastic processes in producing nested patterns of species 
distributions. Oikos, 114, 159-167. 
Holdridge, L.R. (1967) Life zone ecology. Occasional Papers of the 
Tropical Science Center, San lose, Costa Rica. 
Hubbell, S.P. (2001) The unified neutral theory of biodiversity 
and biogeography. Princeton monographs in population biology, 
p. 375. Princeton University Press, Princeton, NJ. 
lenkins, D.G. (2006) In search of quorum effects in meta- 
community structure: species co-occurrence analyses. Ecology, 
87, 1523-1531. 
Kalko, E.K.V. (1998) Organisation and diversity of tropical bat 
communities through space and time. Zoology, 101, 281-297. 
Kalko, E.K.V., Handley Jr, CO. & Handley, D. (1996) Organiza- 
tion, diversity and long-term dynamics of a Neotropical bat 
community. Long-term studies in vertebrate communities 
(ed. by M.L. Cody and J. Smallwood), pp. 503-553. Academic 
Press, Los Angeles, CA. 
Leigh Jr, E.G., Wright, S.J. & Herre, E.A. (1993) The decline of 
tree diversity on newly isolated tropical islands: a test of a null 
hypothesis and some implications. Evolutionary Ecology, 7, 
76-102. 
MacKenzie, D.I., Bailey, L.L. & Nichols, J.D. (2004) Investigating 
species co-occurrence patterns when species are detected 
imperfectly. Journal of Animal Ecology, 73, 546-555. 
Manly, B.F.J. & Sanderson, J.G. (2002) A note on null models: 
justifying the methodology. Ecology, 83, 580-582. 
Martinez-Morales, M.A. (2005) Nested species assemblages as a 
tool to detect sensitivity to forest fragmentation: the case of 
cloud forest birds. Oikos, 108, 634-642. 
Meyer, C.F.J., Frund, J., Pineda Lizano, W. & Kalko, E.K.V. (2008) 
Ecological correlates of vulnerability to fragmentation in Neo- 
tropical bats. Journal of Applied Ecology, doi: 10.1111/j.1365- 
2664.2007.01389.x. 
Meyer, C.F.J. & Kalko, E.K.V. (in press) Assemblage-level 
responses of phyllostomid bats to tropical forest fragmentation: 
land-bridge islands as a model system. Journal of Biogeography. 
Moreno, C.E., Arita, H.T. & Solis, L. (2006) Morphological 
assembly mechanisms in Neotropical bat assemblages and 
ensembles within a landscape. Oecologia, 149, 133-140. 
Norberg, U.M. & Rayner, J.M.V. (1987) Ecological morphology 
and flight in bats (Mammalia; Chiroptera): wing adaptations, 
flight performance, foraging strategy and echolocation. Philo- 
sophical Transactions of the Royal Society of London Series B, 
Biological Sciences, 316, 335^427. 
Patterson, B.D. & Atmar, W. (1986) Nested subsets and the struc- 
ture of insular mammalian faunas and archipelagos. Biological 
Journal of the Linnean Society, 28, 65-82. 
Patterson, B.D. & Atmar, W. (2000) Analyzing species composi- 
tion in fragments. Isolated vertebrate communities in the tropics. 
Proceedings of the 4th International Aymposium (ed. by G. 
Rheinwald), pp. 9-24. Bonner Zoologische Monographen, 
Bonn, Germany. 
Patterson, B.D. & Brown, J.H. (1991) Regionally nested patterns 
of species composition in granivorous rodent assemblages. 
Journal of Biogeography, 18, 395-402. 
Patterson, B.D, Willig, M.R. & Stevens, R.D. (2003) Trophic 
strategies, niche partitioning, and patterns of ecological 
organization. Bat ecology (ed. by TH. Kunz and M.B. Fenton), 
pp. 536-579. University of Chicago Press, Chicago, IL. 
Peres-Neto, P.R. (2004) Patterns in the co-occurrence of fish 
species in streams: the role of site suitability, morphology and 
phylogeny versus species interactions. Oecologia, 140, 352-360. 
Rodriguez-Girones, MA. & Santamaria, L. (2006) A new algo- 
rithm to calculate the nestedness temperature of presence- 
absence matrices. Journal of Biogeography, 33, 924-935. 
Sanderson, J.G., Moulton, M.R & Selfridge, R.G. (1998) Null 
matrices and the analysis of species co-occurrences. Oecologia, 
116,275-283. 
Schluter, D. (1984) A variance test for detecting species associa- 
tions, with some example applications. Ecology, 65, 998-1005. 
Simmons, N.B. (2005) Order Chiroptera. Mammal species of the 
world: a taxonomic and geographic reference (ed. by 0.E. Wilson 
and 0.M. Reeder), pp. 312-529. Johns Hopkins University 
Press, Baltimore, MD. 
von Staden, 0. (2002) Comparison of organization, diversity, 
and structure of two Neotropical bat communities in Panama. 
PhD Thesis, University of Tubingen, Tubingen, Germany. 
Stevens, R.D. & Willig, M.R. (1999) Size assortment in New 
World bat communities. Journal oj Mammalogy; 80, 644?658. 
Stevens, R.D. & Willig, M.R. (2000) Density compensation in 
New World bat communities. Oikos, 89, 367-377. 
Stone, L. & Roberts, A. (1990) The checkerboard score and species 
distributions. Oecologia, 85, 74-79. 
Strong Jr, DR., Simberloff, D., Abele, E.G. & Thistle, A.B. (1984) 
Ecological communities. Princeton University Press, Princeton, 
NJ. 
Ulrich, W (2004) Species co-occurrences and neutral models: 
reassessing J. M. Diamond's assembly rules. Oikos, 107, 603- 
609. 
Ulrich, W & Gotelli, N.J. (2007) Null model analysis of species 
nestedness patterns. Ecology, 88, 1824-1831. 
Ulrich, W & Zalewski, M. (2006) Abundance and co-occurrence 
patterns of core and satellite species of ground beetles on small 
lake islands. Oikos, 114, 338-348. 
Watling, J.I. & Donnelly, M.A. (2006) Fragments as islands: a 
synthesis of faunal responses to habitat patchiness. Conservation 
Biology, 20, 1016-1025. 
Weiher, E. & Keddy, P. (1999) Ecological assembly rules: perspectives, 
advances, retreats. Cambridge University Press, Cambridge, UK. 
Whittaker, R.J. (1998) Island biogeography: ecology, evolution, and 
conservation. Oxford University Press, Oxford, UK. 
Windsor, D.M. (1990) Climate and moisture variability in a 
tropical forest: long-term records from Barro Colorado Island, 
Panama. Smithsonian Contributions to the Earth Sciences, 29, 
1-145. 
Wright, DEL, Patterson, B.D., Mikkelson, G.M., Cutler, A. & 
Atmar, W (1998) A comparative analysis of nested subset 
patterns of species composition. Oecologia, 113, 1-20. 
Zalewski, M. & Ulrich, W. (2006) Dispersal as a key element of 
community structure: the case of ground beetles on lake 
islands. Diversity and Distributions, 12, 767-775. 
? 2008 The Authors 
Diversity and Distributions, Journal compilation ? 2008 Blackwell Publishing Ltd 11