PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINE-S return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

 

 

DATE DUE DATE DUE DATE DUE
——-—;v}
E. ,., f: t’ 3")
_. 1? V a

if

 

 

 

 

 

 

 

 

 

 

 

A

 

 

11/00 c/CIRC/DmDuepes-p.“

 

3.4
‘ tg‘s‘:.l

THE RESILIENCY OF LIZARD COMMUNITIES TO HABITAT
FRAGMENTATION IN DRY FORESTS OF SOUTHWESTERN PUERTO RICO

By

Kristen S. Genet

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Department of Zoology

1 999

ABSTRACT

THE RESILIENCY OF LIZARD COMMUNITIES TO HABITAT
FRAGMENTATION IN DRY FORESTS OF SOUTHWESTERN PUERTO RICO

By

Kristen S. Genet

The. effects of habitat fragmentation on lizard communities were quantified
for a subtropical dry forest ecosystem in southwestern Puerto Rico. Diversity
and abundance of 10 lizard species were determined for small (<1 ha), medium
(1-10 ha), and large (>10 ha) forest fragments and were compared to sites in
Guanica Forest, the largest continuous tract of dry forest remaining in Puerto
Rico. Characteristics of the study sites and surrounding landscape were also
quantified to assess the response of lizards to landscape-level phenomena;
habitat characteristics of anoles were also quantified to assess the ecological
status of the endangered Anolis cooki. Comparisons of lizard community
composition and structure among fragments and continuous forest revealed that
lizards were relatively resilient to the consequences of habitat fragmentation,
although species richness was significantly positively correlated with fragment
area. The ability of lizards to survive and maintain populations in this
fragmented landscape indicated that conservation and/or restoration of small
patches of secondary forest would likely lead to increased lizard populations that

would be able to colonize additional forest patches as habitat becomes available.

ACKNOWLEDGMENTS

The guidance and support provided by my graduate committee members,
Thomas M. Burton, J. Alan Holman, and Peter G. Murphy, were invaluable
throughout the development and completion of this research adventure. I am
also very grateful for the assistance provided in the field by John Genet, Ian
Ramjohn, and Skip Van Bloem. Ariel E. Lugo and the lntemational Institute of
Tropical Forestry (llTF) provided assistance, resources, and access to equipment
and data. Miguel Canals Mora provided valuable logistical support and advice in
Puerto Rico. Permission to conduct scientific research at the field site was
granted by the Puerto Rico Departamento de Recursos Naturales y Ambientales
(DRNA permit numbers 97-lC-77 and 98—lC-81). Olga Ramos and Carlos
Rodriguez (llTF) provided spatial data that contributed to the GIS analyses, and
Brian Noyle contributed valuable assistance in those analyses. Thomas A.
Jenssen provided helpful comments on an earlier draft of chapter three. Thanks
to Renate Snider, Jenifer Fenton, Beth Bardon, John Sorochan, Jason
Henderson, and Cheryl Webster for comments on earlier drafts of this thesis.
Funding for this project was provided by the Department of Zoology and the
Ecology, EvolutionaryBiology and Behavior program at Michigan State

University.

TABLE OF CONTENTS

LIST OF TABLES .
LIST OF FIGURES

INTRODUCTION
Objectives and Hypotheses
The Study Region

CHAPTER 1
The Reference Lizard Community of Guénica Forest
Introduction
Materials and Methods
Study Region
Guanica Commonwealth Forest
Reference Site Selection
Transect Sampling
Plot Sampling
Results
Discussion

CHAPTER 2
Lizard Community Composition and Structure in a Fragmented
Subtropical Dry Forest Landscape
Introduction
Materials and Methods
The Study Region
Dry Forest Fragment Study Sites
Fragment and Landscape Analysis
Lizard Community Sampling
Statistical Analyses
Results
Fragment and Landscape Characteristics
Minimum Fragment Area
Principal Components Analysis
Correspondence Analysis
Discussion
Lizard Community Responses to Fragmentation
Conservation of Dry Forest Fragments

11
11
16
16
17
18
20
23
25
31

37
37
41
41
42
45
48
51
52

57
59
62
66

74

CHAPTER 3
Structural Habitats of Dry Forest Trunk-Ground Ecomorphs:
Anolis cristatellus and Anclis cocki

Introduction

Materials and Methods

Results

Discussion

Current and Projected Future Status of A. cocki

SUMMARY
APPENDIX A
APPENDIX B
APPENDIX C
LITERATURE CITED

77
77
80
82
91
99
1 05
1 06
1 16

119

1.1

1.2

1.3

1.4

1.5

1.6

1.7

2.1

2.2

2.3

2.4

2.5

2.6

LIST OF TABLES

Lizard families and species of lizards represented in the dry forest zone of
southwestern Puerto Rico (compiled from Rivero 1978, Schwartz and
Henderson 1991 ).

Summary of sampling design in Guanica Commonwealth Forest
Lizard species present in study sites within Guénica Forest.
Proportional similarity of sites within Gua'nica Forest.

Lizard community composition in Guénica Forest. Relative abundance of
the ten species indicates dominance of A. cn'statellus and S. nicholsi.
Community diversity and evenness metrics also reflect this pattern.

Variables which differed significantly among sites in Guénica Forest,
KruskaI—Wallis ANOVA.

Densities of lizards in Guéniw Forest. One way ANOVA was used to test
for differences among sites. Numbers represent mean (SE) densities (#
individuals/ha), and values within a row with matching superscripts differ
significantly from one another (T ukey’s HSD, p<0.05). n=total number of
individuals per site.

Characteristics of selected forest fragments and reference site (Gua'nica
Forest) selected as study sites.

Summary of sampling design in forest fragment study sites.

Occurrence of lizard species in the 12 forest fragments and the reference
site (Guénica Forest). X = species detected by quantitative transect and
plot sampling methods. P = present, detected during qualitative search
(site walk).

Pearson correlation coefficients (r) of species abundance (# individuals!
ha) and species richness with fragment characteristics. *p<0.05, **p<0.01.

Pearson correlation coefficients (r) of species abundance (# individuals/
ha) and species richness with immediately adjacent land use types around
fragment perimeter. *p<0.05, **p<0.01.

Pearson correlation coefficients (r) describing associations between lizard,

‘ termite, and vegetation variables. * p<0.05, **p<0.01

vi

2.7
2.8
2.9
2.10
3.1
3.2

3.3

3.4

A1

B1
8.2

8.3

Eigenvalues and eigenvectors from PCA of fragment and landscape
characteristics. Additional principal components accounted for 5 5%
of the variation in the data set.

Correspondence analysis of lizard abundance in dry forest fragments (A.
cocki omitted from analysis). Total inertia was 0.50; each additional
dimension explained <1 % of the total variability.

Partial contributions of row (site) and column (species) points to total
inertia of first four CA dimensions. Anolis cocki omitted from analysis.

Squared cosines of rows (sites) and columns (species) points for the first
four CA dimensions. Cumulative total represent the total quality of the
display of each site or species in four-dimensional space. Anolis cocki
omitted from analysis.

Age/size classes for A. cristatellus and A. cocki based on SVL (Schwartz
and Henderson 1991).

Distribution of A. cristatellus and A. cocki among study sites. 1 = presence
of A. cristatellus, 2 = presence of A. cocki. Refer to text and Figure 2.1 in
Chapter 2 for a complete description of fragment study sites.

Overall perch height (SE) and perch diamter (SE) for A. cristatellus and A.
cocki; all sites and age/size classes combined. Differences in perch
height or diameter were not significantly different between species.

Perch height (SE) and perch diameter (SE) for A. cristatellus and A. cocki
at sites where they are in sympatry; age/size classes combined.
Differences between species in perch height or diameter were not
significant (Mann-Whitney U Test, p>0.05).

Summary of sampling design in Gua'nica Commonwealth Forest.

Densities of dry forest lizards in small fragments (<1 ha). Numbers
represent mean (SE) densities (# individuals/ha), n = total number of
individuals per site.

Densities of dry forest lizards in medium fragments (1 -1 0 ha). Numbers
represent mean (SE) densities (# individuals/ha), n = total number of
individuals per site.

Densities of dry forest lizards in large fragments (>10 ha). Numbers
represent mean (SE) densities (# individuals/ha), n = total number of

. individuals per site.

vii

3.4

8.5

86

C1

Correspondence analysis of lizard abundance in dry forest fragments.
Total inertia was 0.57; each additional dimension explained <5% of the
total variability.

Partial contributions of row (sites) and column (species) points to inertia of
first four dimensions. All 10 species of lizards included in analysis.

Squared cosines of row (sites) and column (species) points for the first
four CA dimensions. Cumulative total represent the total quality of the
display of each site or species in four-dimensional space. All 10 species
of lizards included in analysis.

Location of Anolis cristatellus and A. cocki perches, including frequency

(absolute and rounded to percentage) for each perch type selection. Type
of perch also included ( t = tree, s = shrub, c = cactus, v = vine).

viii

1.1

1.2

1.3

1.4

1.5

2.1

2.2

2.3

2.4

2.5

LIST OF FIGURES

Land use map of Guanica Forest (eastern section) and vicinity. Study
sites within the reference site are labeled as follows: Co=coastal scrub,
Ce=central site of upland forest on predominantly south-facing slopes, N=
northern site of upland forest on predominantly north-facing slopes, R1-
R6=ravines. The relative size and location of fragment sites 4, 7, and 10
are also shown on this map.

Transect sampling design.

Five most predominant species comprising the lizard community in
Guanica Forest. Species composition was not significantly different
among sites (Kruskal-Wallis ANOVA, p=0.454).

Differences in percentage rock cover in 2 x 2 m subplots among sites
(mean t SE). Points with matching letters indicate significant pairwise
differences (Tukey’s HSD, p<0.05).

Difference in litter depth in 2 x 2 m subplots among sites (mean 1 SE).
Points with matching letters indicate significant pairwise differences
(Tukey’s HSD, p<0.05).

Map of southwestern Puerto Rico showing the relative size and location
of the12 forest fragments and the reference site (Guanica Forest).

Number of lizard species (mean i SE) for fragment size classes and
reference site. Differences were not significantly different (ANOVA,
p=0.102; Kruskal-Wallis ANOVA, p=0.095).

Relationship between fragment area (log transformed for analysis and
graphical display) and proportion of reference lizard community supported
by fragments.

Proportion of reference lizard community supported by fragments. Sites 7,
10, and 13 (sampled only in 1998) omitted from display.

PCA results of fragment ordination, see text for fragment identification
numbers; (A) PC 1 vs. PC 2, (B) PC 1 vs. PC 3. Eigenvectors scaled by a
factor of 5 for plotting purposes.

2.6

3.1

3.2

3.3

3.4

3.5

3.6

3.7

Correspondence analysis of study sites and lizard species. Size of plotting
symbol = CA 3 (21.25%), shading pattern of plotting symbol = CA 4
(6.44%). Sites are numbered as in text, species codes as follows: ancr =
A. cristatellus, anst = A. strata/us, anpu = A. pulchellus, anpo = A.
poncensis, amex = A. exsul, amwe = A. wetmorei, spni = S. nicholsi,

spro = S. roosevelti, phwi = P. wirshingi. Anolis cocki omitted from
analysis. 94.72% of total inertia is explained in the first four dimensions.

Age distribution of A. cristatellus (n=2413) and A. cocki (n=76) in 1997-
1998. Sample sizes for age/size classes are given in the text.

(A) Mean perch height (A. cristatellus, n=2413; A. cocki, n=76), and (B)
mean perch diameter (A. cristatellus, n=728; A. cocki, n=31) for A.
cristatellus and A. cocki age/size classes at all study sites. Error bars
represent standard error of the mean. Mean perch size selection (height
and diameter) was significantly different between adult males of the two
species (Mann-Whitney U Test, p<0.05), but females and juveniles did not
differ significantly in structuralhabitat characteristics (Mann-Whitney U
Test, p>0.05). ‘

(A) Mean perch height and (B) mean perch diameter of Ano/is cristatellus
in allopatry (n=1872) and sympatry (n=289) with A. cocki. Error bars
represent standard error of the mean. Perch diameter was significantly
less in sympatry with A.cooki (Mann Whitney U Test, p=0.001), but
perch height was not significantly different (Mann-Whitney U Test,
p=0.632)).

Mean (j; SE) perch height for A. cristatellus (n=289) and A. cocki (n=76)
where they occur in sympatry. Within age/size classes, differences
between species were not significant (Mann-Whitney U Test, p>0.05).

Relationship between SVL and perch height for adult male (A) A.
cristatellus (n=47) and (B) A. cocki (n=39) at sympatric sites. Neither
species showed a significant association between the characters

‘ (Spearrnan rank correlation, p>0.05)

Density (mean 1 SE) of A cristatellus (n=289) and A. cocki (n=76) where
they occur in sympatry. Ana/is cristatellus adult females/sub-adult males ,
and juveniles were significantly more abundant than the same A. cocki
age/size classes (Mann-Whitney U Test, p<0.05).

Structural Habitat (mean :t SE) perch height and perch diameter of
A. cristatellus (n=2413) among study site size classes. Perch height and
diameter are both significantly different among site size classes (Kruskal-

- Wallis ANOVA, p<0.05).

A1

3.1

8.2

B3

B4

B5

Species-area curves for transect sampling at Guénica Forest sites. (A)
Coastal site: A. pulchellus added via qualtitative sampling, 8. nicholsi, S.
roosevelti, and P. wirshingi added via plot sampling; (B) Central site: A.

, pulchellus and A. poncensis added via qualitative sampling, 8. nicholsi

and S. roosevelti added via plot sampling; (C) North site: A. pulchellus
added via qualitative sampling, 8. nicholsi, S. roosevelti, and P. wirshingi
added via plot sampling; (D) Ravines: S. nicholsi and P. wirshingi added

. via plot sampling.

Land use map of site 5 and vicinity, illustrating the delineation of the
1000 m buffer around study sites.

Species-area curves for transect sampling at small fragment sites; no
species were present at site 40. (A) Site 36: no additional species added
by qualitative or plot sampling; (8) Site 30: A. pulchellus added via
qualitative sampling, 8. nicholsi and P. wirshingi added via plot sampling;
(C) Site 27: S. nicholsi and P. wirshingi added via plot sampling.

Species-area curves for transect sampling at medium fragment sites.

(A) Site 28: A. strata/us and A. pulchellus added via qualitative sampling,
8. nicholsi, S. roosevelti, and P. wirshingi added via plot sampling; (B) Site
13: S. nicholsi and P. wirshingi added via plot sampling; (C) Site 16: S.
nicholsi, S.’ roosevelti, and P. wirshingi added via plot sampling, (D) Site
10: A. poncensis added via qualitative sampling, S. nicholsi added via plot
sampling. ’

Species-area curves for transect sampling at large fragment sites.

(A) Site 7: S. nicholsi and P. wirshingi added via plot sampling; (8) Site 5:
P. wirshingi added via plot sampling; (C) Site 4: S. nicholsi, S. roosevelti,
and P. wirshingi added via plot sampling, (D) Site 2: S. nicholsi, S.
roosevelti and P. wirshingi added via plot sampling.

Correspondence analysis of all study sites and all 10 lizard species.

A.) Dimension 1 vs. Dimension 2, B.) Dimension 1 vs. Dimension 3
Species codes: ANOCRl=A.cristate/Ius, ANOCOO=A. cocki, ANOSTR=A.
stratulus, ANOPUL=A. pulchellus, ANOPON=A. poncensis, AMEEXS=A.
exsul, AMEWET=A. wetmorei, SPHNIC=S. nicholsi, SPHROO=S.
roosevelti, PHYWIR=P. wirshingi. Refer to Chapters 1 and 2 for site
identification.

xi

INTRODUCTION

Habitat fragmentation is one of the greatest threats to natural ecosystems
and global biodiversity (Wilcox and Murphy 1985, Turner 1996, Dale and
Pearson 1997). Thus, the causes and consequences of fragmentation are
important considerations for developing conservation and management
strategies for fragmented landscapes. Alteration, degradation, and
fragmentation of tropical forest ecosystems as a result of human impact have
received considerable recognition and attention as a source of loss of both
biodiversity and ecosystem structure and function (i.e., Lugo et al. 1978, Murphy
and Lugo 1986a, Janzen 1988a, D, Murphy et al. 1995), as rates of deforestation
in the tropics and subtropics have increased over the past several decades
(FAO 1993, Whitmore 1997).

Laurance et al. (1997) reviewed the effects of tropical forest fragmentation
on ecosystem structure, diversity, and function. Bierregard et al. (1997)
identified the major priorities for research, conservation, and management in
these systems. Fundamental information on the natural history of species and
knowledge of species-level responses to habitat fragmentation are lacking
(Laurance and Bierregard 1997); this study addressed these issues concerning
lizard community ecology in a highly fragmented subtropical dry forest
landscape in southwestern Puerto Rico.

Tropical and subtropical dry forests are more imperiled than their moist

and wet forest counterparts (Janzen 1988a, b, Redford et al. 1990, Murphy and

Lugo 1990, Lerdau etal. 1991, Murphy and-Lugo 1995). Although it is true that
tropical rain forests harbor a more speciose community, tropical dry- forests
support a rich array of biological interactions and responses to the more
pronounced seasonality of these systems. Rather than limiting conservation
efforts solely to numbers of species, interactions among and between species
are also crucial components of a naturally functioning ecosystem that need to be
protected and preserved (Janzen 1988a).

Subtropical and tropical dry forests are located in frost free regions where
the mean annual biotemperature (air temperature corrected for extremely high or
low temperatures, >30o C'and <0o C, respectively) is greater than 17° C, mean
annual precipitation ranges from 250-2000 mm, and the mean annual ratio of
potential evapotranspiration to precipitation (PET/P) is greater than 1.0
(Holdridge 1967). The majority of forests within this classification are found in
Africa, Central America and tropical islands, where they constitute 70-80% of the
total forested area (Murphy and Lugo 1986b).

Although dry forests constitute approximately 42% of the total forested
area in the tropics and subtropics (Murphy and Lugo 1986b, 1990), they are
underrepresented in terms of research focus and published literature.
Historically, dry forests extended over a much larger area than at present, but as
a result of a long legacy of disturbance and deforestation, their original extent
will probably never be known. It is thought that many of the dry savannas or
thorn woodlands are derived from degraded or disturbed dry forests (Murphy

and LUgo 1986b). Dry forests are perhaps a more endangered ecosystem than

moist or wet forests as -a result of the greater human pressure and disturbances.
In general, people prefer to live in the climate of the dry or moist tropics (sensu
Holdridge 1967), and the forest structure and underlying soils have made these
regions profitable to clear for agriculture and cattle ranching (Murphy and Lugo
1986b).

There are few remaining intact representatives of dry forest in Central
America and the Caribbean (Janzen 1988a, b, Murphy and Lugo 1995). Two
notable examples of dry forest under active management and protection are the
Guanacaste National Park in northwestern Costa Rica (Janzen 1988b) and the
Gua'nica Commonwealth Forest in southwestern Puerto Rico (Murphy and Lugo
1990, Murphy et al. 1995). Janzen (1988a, b) has stated that restoration
ecology and habitat management will be the only solutions to salvaging the
diverse array of species and biological activities in the world’s tropical dry
forests, and has actively pioneered such projects in the dry forests of Costa
Rica. From the biological perspective, the task at hand is to devise a plan for
sustainable land management strategies that meet the needs of the local people
- and contribute to the conservation of biodiversity (Lamb et al. 1997).

Species exhibit tremendous variability in their responses to habitat
fragmentation. Species that are the most susceptible to the consequences of
fragmentation, undoubtedly leading to eventual population declines and
extinction, are those that have one or more of the following characteristics
(modified from Turner 1996 and Laurance et al. 1997): ( 1) extreme sensitivity to

habitat alteration or intolerance ofconditions in the surrounding matrix (i.e.

habitat specialists), (2) large habitat area requirements, (3) vulnerability to any
sort of exploitation, (4) unstable or highly fluctuating population dynamics, or
ecological dependence on species with unstable populations, (5) naturally low
population densities or patchy distributions (i.e., rare species), (6) limited
dispersal abilities, or (7) low fecundity. Studying individual species or ecological
assemblages of species will allow us to gain a successively more
comprehensive understanding of how different groups of organisms respond to
the pressures of habitat fragmentation.

Studying animals’ responses to habitat fragmentation is necessary
because many species are very vulnerable to habitat alteration and respond
quickly to any changes in habitat structure and microclimate. Since dry forest
systems are critical habitat areas for. many species of vertebrates (notably high
avian diversity; Kepler and Kepler 1970, Faaborg and Arendt 1990), assessing
the ecological response of the fauna to habitat fragmentation is a necessary step
in addressing biodiversity and conservation management strategies. Small
sized, ectothermic species with short generation times may exhibit the most
sensitive and rapid responses to habitat modification. Responses of vertebrate
species in particular may have ramifications on the entire community structure
as a result of higher order interactions.

Lizards (suborder Lacertilia) are probably the most common as well as the
most conspicuous terrestrial vertebrates, inhabiting every region of Puerto Rico.
The island supports a very diverse lizard fauna, with species accounts ranging

from ‘33 (Rivero 1978) to 43 (Schwartz and Henderson 1991) species.

Differences in surveys and species accounts are most likely attributable to
unresolved and controversial taxonomic distinctions. The most diverse group
within the lizard fauna is found among the Anolis lizards, with 12 species
ecologically distributed‘throughout the island (Rand 1964, Roughgarden 1995).
Lizards are among the higher order predators on the island. They are primarily
insectivorous, but have a broad diet ranging from garbage scraps to other
smaller lizards, making them important in terms of trophic structure of these
tropical forest communities. Lizards also lend themselves well to ecological
studies, as evidenced by the wealth of literature on this vertebrate group.

The vast majority of lizard studies in Puerto Rico have focused on
communities within the Luquillo Rain Forest (e.g., Andrews 1971, Lister 1981,
Reagan 1986, 1992). Of the diverse lizard fauna of the island, fewer than 15
species are found in the dry forest zone of the southwest (Rivero 1978). Studies
of the lizards in the dry forest zone have investigated growth (Lewis 1986),
activity budgets (Lewis and Saliva 1987), diet selection (Lewis 1989), and
competitive interactions (Ortiz and Jenssen 1982, Jenssen et al. 1984). Ortiz
(1990) reported on the status and distribution of Anolis cocki, and the impacts of
competitive interference and habitat disturbance on this endangered species.
Studies of this nature are of paramount importance in light of extensive human
disturbances to natural habitats, and questions regarding faunal communities
within fragmented and protected landscapes are in dire need of additional

research.

Assessing the response of the dry forest lizard community is the first step
in gaining a better understanding of how faunal assemblages respond to
landscape-level habitat fragmentation phenomena. In order to protect and
preserve biodiversity, it is critical to study at the landscape and regional level in
, addition to whole ecOsystem studies (Noss 1983, Franklin 1993, Stohlgren et al.
1997). Through interpretation and analysis of historical land use and land cover
maps (Vélez Rodriguez 1995a-f, Lugo et al. 1996, Kramer 1997) and current
aerial photographs, temporal variability and changes in landscape structure and
function can be inferred. In addition, abiotic landscape and fragment features
can be combined with biotic attributes to assess landscape structure, function,
and change in this highly fragmented and altered environment. This approach
will provide an indication of the ecological associations of lizards in this region,
which may subsequently provide valuable information concerning the
conservation potential and future management strategies for this ecosystem and

other similarly disturbed systems.

Objectives and Hypotheses
The overall objective of this study was to assess the ability of forest
fragments to support native Puerto Rican dry forest lizard communities and to
serve as refugia in the face of continuous habitat disturbance. Even small forest
remnants have conservation potential and important biological, economic, and
social values. These remnants can support assemblages of endemic species

(Turner and Corlett 1996), increase ecosystem connectedness by allowing

dispersal or migration of some species (Powell and Bjork 1995), provide sources

for recolonization of adjacent patches via metapopulation dynamics (Fahrig and

Merriam 1985) or establishment in the matrix habitat through the “rescue effect’

(sensu Brown and Kodric-Brown 1977) or conserve local vegetation populations

and seed sources (Nason et al. 1997), perform natural “ecosystem services”

such as local soil stabilization and erosion control, and provide natural areas for
recreation purposes.

One hypothesis underlying this work is that the fragments of Puerto Rico’s
dry forest region serve as refugia which contain source populations of native
animal species (Brown and Kodric-Brown 1977, Fahrig and Merriam 1985).
Characteristics such as area, plant community composition, microclimate,
aspect, and disturbance history determine the potential of a fragment to serve as
a refugium for a given species or assemblage of species.

The specific objectives of this study were:

(1) to compare the diversity and abundance of native lizards in a large, relatively
undisturbed, protected dry forest (Guanica Commonwealth Forest) and in
forest fragments of varying sizes in the dry forest life zone. These data will
be used to test the hypothesis that the diversity and abundance of lizards
present in an array of fragments will collectively be comparable to those of
Guénica Forest.

(2) to determine variation in diversity and abundance of lizard species among
fragments relative to fragment size, disturbance history, and characteristics

of the landscape surrounding forest fragments. This objective will allow me

to test the hypothesis that the diversity and abundance of lizards will
increase as total fragment area increases.

(3) to determine (a) the minimum critical fragment size which contains a
community of lizards similar to that found in undisturbed dry forest habitats,
and (b) the minimum critical fragment size required to support 50% and 75%
of the reference (Gua’nica Forest) lizard community.

(4) to determine the influence of landscape-level phenomena on the distribution,
diversity, and abundance of dry forest lizard species. These data will allow
me to test the hypothesis that landscape characteristics such as land use
adjacent to fragments and fragment isolation Will influence lizard community
composition of individual forest fragments.

(5) to determine critical habitat requirements of selected lizard species in
fragmented systems relative to continuous forest. The hypothesis to be
tested is that fragment size and landscape position will affect habitat within
fragments differentially such that habitat suitability for lizard species will be

altered.

The Study Region
Ewel and Whitmore (1973) provided an excellent description of the
subtropical dry forest zone on Puerto Rico and the US Virgin Islands. In the dry
forest life zone of southwestern Puerto Rico, annual precipitation ranges from
600-1100 mm, and Gua’nica Forest, situated in the approximate center of this

regidn, receives approximately 860 mm of rainfall annually with considerable

year to year variability (Murphy et al. 1995). Guénica Forest covers an area of
approximately 4000 hectares, the largest continuous tract of dry forest on the
island. Originally, dry forests covered the majority of southern and western
Puerto Rico (as well as smaller regions of the northeastern coast and
surrounding islands and cays), but only 4% of the historical extent of dry forest
remains today (Murphy and Lugo 1990). A more thorough discussion of land
use history and changes in southwestern Puerto Rico can be found in Chapter 2.
The plant community of Guénica Forest, and the dry forest zone in
general, can be characterized by three major vegetation associations: coastal
scrub, deciduous, and semi-evergreen forest (Murphy and Lugo 1986a),
although along the elevational gradient from coastal to upland sites these
classifications can be refined to six floristically and structurally distinct
categories (Murphy et al. 1995). Tree communities are usually stunted, rarely
exceeding 15 m in height, with the possible exceptions of more humid and
protected environments of ravines and valleys (Famsworth 1993); the canopy is
typically broad and sparsely leafed (Ewel and Whitmore 1973). The forest is
typified by many multiple stemmed trees and a high density of small stems (< 5
cm DBH) reflecting regeneration by coppicing as a result offence post
harvesting or other disturbances (i.e., hurricanes) (Murphy and Lugo 1986a, S.
Van Bloem, pers. comm). In comparison to dry forest in other tropical and
subtropical regions, Gua’nica Forest is structurally smaller in terms of canopy
height, canopy foliage, and total biomass (Murphy and Lugo 1990). Possible

explanations for these differences include profound seasonality with low rainfall

alternating with extreme drought (Murphy et al. 1995), harsh limestone substrate,
and probably most importantly, the disturbance history (Murphy and Lugo 1990).

In spite of its harsh environment and structure, Guénica Forest harbors a
rich diversity of wildlife and biological interactions. Both in terms of species
richness and absolute abundance, Gua'nica Forest supports a more diverse
assemblage of birds than the moist and wet forest zones in Puerto Rico (Kepler
and Kepler 1970). Indeed, birds represent a valuable taxonomic'group for
research, and their responses to habitat fragmentation and alteration have been
studied extensively in comparison to other faunal groups (Turner 1996,
Bierregard et al. 1997). Previous studies of faunal assemblages within Guanica
Forest have focused primarily on the avian communities (Faaborg and Arendt
1990 and references within); studies of other vertebrate and invertebrate

communities are lacking.

10

Chapter 1

The Reference Lizard Community of Guanica Forest

Introduction

Lizards (Reptilia, Lacertilia) are ubiquitous inhabitants of subtropical and
tropical regions. Past surveys of the island of Puerto Rico have recorded 33
(Rivero 1978) to 43 (Scwartz and Henderson 1991) species on the main island
as well as the surrounding smaller islands and cays. These species represent
five families: Gekkonidae, Teiidae, Scincidae, Anguidae, and Polychridae.
Within this diverse assemblage, 13 species (eight of which are endemic to the
Puerto Rican Bank) and four families are represented in the dry forest life zone
(sensu Holdridge 1967), although some species within this subset have
restricted habitat requirements or represent rare and/or endangered species
which are not continuously distributed throughout the dry zone (Table 1.1,
Rivero 1978, Schwartz and Henderson 1991). Along with snakes and some
birds, lizards are predominantly carnivorous and occupy the higher trophic levels
in the dry forest ecosystem. Any perturbations in their community composition or
population density have the potential to subsequently affect the entire ecological
community. '

Anoline lizards (Polychridae, genus Ano/is) have been present in the
Caribbean throughout its geologic history; the lineage extends back at least into
the middle Jurassic, approximately 175 million years ago (Estes 1983,

Roughgarden 1995). Puerto Rican Anolis are derived from the Central

11

Caribbean Complex lineage (Shochat and Dessauer 1981 ), and a single
successful colonization event (likely from Hispaniola; Reagan 1996) and
subsequent radiation are theorized as the origin of the current anole fauna
(Williams 1969). Modern Anolis lizards are veryconspicuous and widespread
vertebrates in Caribbean terrestrial ecosystems (Williams 1969, 1976,
Moerrnond 1979). These lizards are diurnal, primarily insectivorous, and are
ecologically distributed along a vertical gradient throughout every habitat type in
Puerto Rico (Rand 1964). These lizards are fundamentally arboreal, but some
ecomorphs (sensu Williams 1983) are well adapted to grass and bush
environments. Little is known about anole survivorship and longevity, but
preliminary data suggest that some of the smaller species (e.g., A. strata/us)
have a mean population turnover time of 1.4 years and a relatively constant
mortality rate throughout their lifetimes; survivorship for larger species may be

higher, but no estimates are available (Reagan 1996).

Table 1.1 Families and species of lizards represented in the dry forest zone of
southwestern Puerto Rico (compiled from Rivero 1978, Schwartz and
Henderson 1991).

 

 

 

 

Polychridae Teiidae - Gekkonidae Scincidae
Anolis cristatellus Ameiva exsul SphaerOdactyIus nicholsi *Mabuya mabouya
TAnolis cocki IAmeiva wetmorei Sphaerodactylus roosevelti
TA nolis strata/us . *Sphaerodactylus macro/epis
1‘Anolis pulchellus Phyllodactylus wirshingi
Anolis poncensis . . - *Hemidactylus brooki

 

 

*species which are rare or patchily distributed and not encountered during this
study.

I species endemic to the Puerto Rican Island Bank (Puerto Rico, Virgin Islands,
and surrounding islets and cays).

12

Within the dry forest life zone of southwestern Puerto Rico, there are 5
species of Anolis lizards (Table 1.1). Of these, Anolis cristatellus (Duméril and
Bibron), A. pulchellus (Duméril and Bibron), and A. stratulus (Cope) are widely
distributed across the island and found in both xeric and mesic environments,
while A. cocki (Grant) and A. poncensis (Stejneger), are restricted to the drier
southwestern region of the island. Anolis lizards are among the most thoroughly
studied organisms, and this collection of species is no exception. The more
widespread species have been studied more intensively; research-topics have
included interspecific interactions (Hess and Losos 11991), distribution and
abundance (e.g., Philibosian 1975, Gorman and Harwood 1977, Reagan 1992,
Dial and Roughgarden 1994), habitat and resource partitioning (e.g., Schoener
and Schoener 1971a, b, Lister 1981., Goto and Osborne 1989), and foraging
(Reagan 1986). Research in the dry~forest zone has emphasized competitive
interference between A. cristatellus and A. cocki (Ortiz and Jenssen 1982,
Jenssen et al. 1984); this interference has implications for the status and
distribution of the threatened A. cocki, a candidate for the federal endangered
species list (Marcellini et al. 1985, Ortiz 1990).

. Members of the family Teiidae (represented bysthe genus Ameiva in the
dry forest life zone) are considered insectivorous, but stomach contents analysis
has determined that they are opportunistic and generalized omnivores (Rivero
1978, Lewis 1989). These lizards are characterized by extremely long tails and
continual active foraging through the forest substrate (Rivero 1978, Scwartz and

Henderson 1991). They are adapted to digging under rocks or other substrate

13

debris in search of food, and have been found to excavate burrows as deep as
30 cm in sandy soils (Rivero 1978). Ameiva spp. are diurnal, foraging most
vigorously during the warmest midday hours; they can actively thermoregulate
by basking or seeking shelter (Rivero 1978).

Two species of teiid lizards inhabit the dry forest of southwestern Puerto
Rico, Ameiva exsul (Cope) and A. wetmorei (Stejneger). The distribution of A.
exsul is restricted to the warm and sandy lowlands around the entire perimeter of
the island (Heatwole and Torres 1967, Schwartz and Henderson 1991). This
species seems to prefer open forest or roadside conditions, and does very well
in human-altered environments. Ameiva wetmorei, on the other hand, has a
much more restricted distribution, being found only in the arid region of
southwestern Puerto Rico and on Caja de Muertos island (Rivero 1978,
Schwartz and Henderson 1991). This species is found in open, dry scrubby
habitats; its activity patterns and natural history characteristics are very similar to
those of A. exsul (Heatwole and'Torres 1967). Previous studies of these lizards
have focused primarily on the more abundant, larger A. exsul (Lewis 1986, 1989,
Lewis and Saliva 1987);-little is known about the ecology of A. wetmorei.

The remaining Puerto Rican dry forest lizard species belong to the family
Gekkonidae. This large and diverse family in the tropics and subtropics is
characterized by generally immovable eyelids, nocturnal or crepuscular habits
(although some species can be diurnal as well), and expanded toepads which
are covered with lamellae and setae that aid them in navigating smooth vertical

surfaces (Rivero 1978). Many species of geckos have become well established

14

in and around buildings and residences (Rivero 1978). The geckos of the dry
forest zone can be found in leaf litter and under rocks; some populations are
also very successful in littoral situations under accumulations of palm trash
(Rivero 1978).

Of the five species of geckos that are present in the dry forest zone of
southwestern Puerto Rico, two (Sphaerodactylus macro/epis GiJnther and
Hemidactylus brooki Gray) are relatively uncommon in natural forested habitats
(Table 1.1). Sphaerodactylus macro/epis, although it is generally the most
common and widespread gecko in Puerto Rico, is absent from the most xeric
habitats typical of the dry forest (Rivero 1978, Schwartz and Henderson 1991).
Hemidaclylus brooki can be found in the dry forest region, but is associated
primarily with human dwellings and rarely occurs in naturally forested habitat
(Schwartz and Henderson 1991).

The other two Sphaerodactylus species, 8. nicholsi (Grant) and S.
roosevelti (Grant), are restricted to dry coastal areas (Thomas and Schwartz
1966); Phyllodactylus wirshingi (Kerster and Smith) has a similar distributional
* range (Schwartz and Henderson 1991). These geckos are primarily ground-
dwellers, inhabiting the forest floor in piles of leaf litter and underneath rocks
(Rivero 1978), although some individuals are also found under peeling bark near
the base of dead trees such as Leucaena Ieucocephala (pers. obs.) Other than
comments and observations provided in the original descriptions of these
species, very little is knOwn of the habits and ecology of any of the Puerto Rican

gekkonids (Hass1991 ).

15

Although some assemblages or individual species of dry forest lizards
have been studied extensively, additional investigations of community
composition and abundance would provide meaningful contributions to what is
already known about basic lizard biology. The objectives of this chapter are to
determine the baseline conditions and characteristics of a relatively undisturbed
lizard community in southwestern Puerto Rico’s dry forest life zone. This
community will serve as a reference against which other lizard communities

(disturbed or otherwise impacted) can be compared (Chapter 2).

Materials and Methods
Study Region

The dry forest life zone (sensu Holdridge 1967) of southwestern Puerto
Rico encompasses an area of approximately 121,640 hectares (ha) extending
approximately 120 km eastward from Cabo Rojo on the extreme southwestern
corner of the island, and depending upon local topography, the northern
boundary of the dry zone ranges from 3 - 20 km inland (Figure 1.1 inset, Ewel
and Whitmore 1973). The vast majority of Puerto Rico’s dry forest habitat
occurs in this region, although only approximately 4% (roughly 5000 he) remains
of the original forested area (Murphy and Lugo 1990, Murphy et al. 1995). Dry
forests are also found along the northeastern coast of the island and constitute
the predominant life zone classification on all surrounding islands (Culebra,
Vieques, Caja de Muertos, Mona, and Desecheo) (Ewel and Whitmore 1973).

_ The dry zone in the southwestern region of the island lies in the

16

orographic rain shadow of the Cordillera Central, and receives between 600-

1100 mm of precipitation annually, variable with specific location and seasonality
(Ewel and Whitmore 1973). There are two wet and dry periods annually; most of
the rainfall occurs in either the longer wet season from August until November or

the shorter wet season during the month of May.

Guénica Commonwealth Forest

Guénica Forest, one of the best remaining representatives of dry forest
habitat, is situated at the approximate center of the dry zone’s east-west
orientation, where it receives approximately 860 mm of rainfall annually (Ewel
and Whitmore 1973, Murphy and Lugo 1990). The area was designated a
Commonwealth Forest in 1917, and has been protected and managed to varying
degrees since the 1930’s. In 1982, protection and management were
heightened when Guénica Forest became a Biosphere Reserve in the UNESCO
Man in the Biosphere program.

Guénica Forest’s boundaries currently encompass approximately 4000
ha, significantly increased from the original 2079 ha reserve (Lugo et al. 1996).
This area has been subjected to a wide variety of activities and uses throughout
its history, including tree plantations (Haematoxylum campechianium and
Swietenia mahogam), charcoal and fencepost production, agriculture (both
cultivated crops and livestock), and human settlements (Murphy et al. 1995).
Most of these activities ceased by the 1930’s or 1940’s, but fencepost harvesting

continued until the 1970’s (M. Canals Mora, pers. comm.)

17

Guénica Forest is divided into two sections, separated by Guanica Bay
and the town of Ensenada. The western portion represents a smaller, more
recently acquired, and less intensively managed tract than the eastern portion.
The forest supports a wide array of vegetation types and associations, with three
primary categories: coastal scrub, deciduous, and semi-evergreen forest along
the gradient from coastal to upland sites exceeding an elevation of 200 m. An
additional feature of Guanica Forest is the extensive cave system that has
resulted from dissolution of limestone. Collapsed cave systems now form narrow
ravines and canyons along a north-south orientation in the forest, supporting
flora and fauna in a more mesic environment which often floods in the wet

seasons and during large storms (Famsworth 1993).

. " . Reference Site Selection

Four sites Were selected within the continuous eastern portion of Guénica
Forest for this study (Figure 1.1). These sites represent the vegetation types
and topographical features described above, and match the characteristics of
additional study sites (forest fragments, discussed in Chapter 2). Each site
measured 100 ha (with the‘eXception of the ravines), and was delineated using
topographic maps (USGS standard 1224,000 quandrangle maps for Guénica and
Punta Verraco). These areas within Guénica Forest included a coastal scrub
(Co, Figure 1.1) along the southern coast and two upland forest sites in the
central and northern section of the forest. The central site (Ce, Figure 1.1)

included predominantly south-facing slopes, while the slopes in the northern site

18

69: «.5 :0 E505 8.x 2a or new K .v 8% E0868“. .6 8:80. new 9% 9.8.2

9.... .mocssummém .8607. 3.0.3.56: >.Em:_Eouoa so 628 Ema: co 2.» Eoctocuz .moqsm

9.68.538 >=cm£Eouoa co $28 9.63: So can itcoouoo .928 .Smmoouoo 832.2 mm 3.0%. 2m
2% 3:882 05 55.; 3% 63m 2.503 new 226$ 5283 v.20“. 8.530 ho aaE cm: 6cm... 3 2:9”.

 

    
          

Ea. 5:2 665.2
ace. 5th .2332
25.83 3.8.8-52 I
6:263 6283“. I
$22 32:3 :30 g
7.98 .388 608.0 I
as; D

95. 6.3.3:?

new. 93.52 .6 :25 I
cm... 25..

allii I 1J

220522 a F

   

 

 

 

.d 2 25

N 25

    

ocoN ma...
7.20“. >5 .85235

19

(N, Figure 1.1) were primarily north-facing. Six ravines (R1-R6, Figure 1.1) were
selected for sampling from topographic maps and personal observations; this
site was not a continuous 100 he as the others, rather the number of ravines was
selected to match the sampling regime for each of the other sites (Table 1.2).
Due to differences in the biology and activity patterns of the 10 commonly
encountered dry forest lizard species, two distinct survey strategies were
required to obtain accurate estimates of diversity and abundance. Seven
members of the genera Anolis and Ameiva are either arboreal or free-ranging
lizards and abundant and conspicuous diurnal species. These species were
censused along 100 m transects of fixed area. Three members of the genera
Sphaerodactylus and.PhyIIodactylus are primarily nocturnal and/or fossorial,
making transect surveys unrealistic to obtain accurate diversity and abundance
measures. These species were sampled by carefully searching through litter
and ground debris in 2 x 2m plots. The sampling design for the reference site is
summarized in Table 1.2. The number of transects and plots represented the
necessity of efficient and accurate sampling, coupled with time and labor
intensive survey strategies. Species-area curves were constructed as a

measure of sampling efficiency at each site (Appendix A).

Transect Sampling
Other investigators have often used transect surveys to efficiently and
accurately estimate lizard population densities (Rand 1964, Schoener and

Schoener 1971 a, b, Reagan 1992). Thus, this method was chosen for sampling

20

Table 1.2 Summary of sampling design in Guénica Commonwealth Forest

 

 

Total area of Number of Number of Site
Site ID Area (ha) transects (m2) 10m x 10m 2m x 2m Walkt
(# of transects) plots'r plots
Coastal 1 00 252.0 (6) 8 24 yes
(Co)
Central 100 . 2520 (6) 8 24 yes
(Ce)
North 100 2520 (6) 8 24 yes
(N) ~
Ravines * 2520 (6) 6 1 8 no
(R)

 

TEocated at the beginning of each transect and additional randomly established
individual plots

* Presence/absence data only

* Ravines were not continuous 100 ha, rather six individual ravines were
selected to match the sampling design of the other three sites.

visually conspicuous lizards above the litter layer. Lizard surveys were
conducted along six randomly located 100 m transects in each of the study sites
during the summers of 1997 and 1998 (coastal, central, and north sites were
sampled in May - August, 1997; these were resampled and ravines were added
to the design in July-August 1998). In the field, the transect locations were
determined with the aid of topographic maps, a compass, and landmarks. When
site topography allowed, sampling locations were stratified using four aspect
categories (N-, S-, E-, and W-facing slopes). The number of transects
established within each CategOry was determined by the proportion of the total

area-comprised by that category at each site.

21

Each 100 m transect ran perpendicular to the slope, and was subdivided
into sample plots at 5 m intervals. Lizards were censused in each plot by slowly
walking along the fixed-area transect and stopping to identify and count the
individuals within a 2.5 m radius (Figure 1.2). Methods used in this study were
modifications of the Frye strip census method (Overton 1971) utilized by Reagan
(1992). A fixed area transect was used to improve the accuracy of the area
estimation, as visually estimating angles or distances to individuals can be
imprecise (Heckel and Roughgarden 1979). Anolis spp. and Ameiva spp. were
censused with this method because they are visually conspicuous. The mean
number of individuals of each species per m2 was calculated and used to
estimate the density of lizards on a per hectare basis (#Iha); relative abundance
was determined by dividing the density of each species by the total density of all
species combined.

Each transect plot was carefully surveyed visually until all individuals had
been detected (approximately 5 minutes, but duration was highly dependent on
structural diversity and density of the vegetation). All transects were surveyed
by a single observer to eliminate any observer bias. Locations (i.e., perches) of
each lizard were recorded to ensure that no lizards were counted more than
once. Each transect was surveyed twice to ensure that all individuals were
counted. Adult Anolis lizards exhibit remarkable fidelity to their perch and
territory (Rand 1964, Philibosian 1975), and a second survey of the transect was
made within a few hours (to account for diurnal differences in activity and

abundance, Reagan 1986). This method yielded more accurate counts of

22

individuals in the sampling area than from one survey alone. Juvenile anoles
may be more mobile and introduce error into this method, but Andrews and Rand
(1983) suggested that the mean distance between captures of juvenile A.
limifrons at Barro Colorado Island in Panama was 2.9 m, a distance contained

within the area of the circular transect sample plot.

 

Figure 1.2 Transect sampling design

Transect Length = 100m
Total transect area sampled: 420 m2

  
 
 
 

Circular plot:
radius = 2.5m
area = 20 m2

  

 

Plot Sampling
At the beginning of each transect a 10m x 10m plot was established and
three 2m x 2m subplots were randomly selected within each. Because the area
sampled with 10m x 10m plots was substantially less than that sampled with the
transect methods, two additional 10 x 10 m plots were also randomly established
within each site, with the exception of the ravines (refer to Table 1.2 for a

complete summary of sampling design). Plot sampling accounted for less total

23

area sampled than transect sampling because of the greater time commitment
involved in careful searches of subplots.

Each subplot was thoroughly searched for geckos by examining and
clearing away all leaf litter and overturning all rocks to investigate potential
refuges within the substrate. After each plot was searched, all litter and rocks
were replaced to minimize impacts on substrate habitat. The following data were
recorded for each subplot: (1) species identification of each individual, (2)
number of individuals of each species, (3) percent rock cover, (4) average litter
depth (cm), and (5) canopy cover, visually classified as open, partially open, or
closed. Percent rock was visually estimated, and litter depth was determined by
taking the mean of three random measurements within each subplot.

In addition to the quantitative transect and plot sampling methods
described above, qualitative focal searches (site walks) of study sites sampled in
1997 were made during the summer of 1998 to ensure that the species lists for
each site were complete and accurate. A species list was considered complete
if all ten dry forest lizard species were present at any given site. Site walks ,
yielded only qualitative presence/absence data, and were conducted only at
sites with incomplete species lists. Data from the sites which were resampled in
1998 were used to test .for temporal differences in lizard community composition
and abundance. All statistical analyses were performed with Systat (version 5.0,

Evanston, IL).

24

Results

The survey of the reference lizard community of Guénica
Commonwealth Forest yielded 10 species representing three families and four
genera (Table 1.3). The distributions of these lizard species were rather patchy,
even within a continuous forest. Anolis cristatellus, A. exsul, A. wetmorei, and S.
nicholsi were found at all sites. Anolis cocki was found only at the coastal site
and a single individual was found in a ravine in close proximity to the coast.

Lizard communities at all sites were consistently dominated by two
species, A. cristatellus and S. nicholsi (Figure 1.3), which represented at least
75% of the lizard community at all sites. The occurrence of the remaining eight
species differed among sites. Both Ameiva spp. occurred at all sites, but the
remaining species were absent from one or more sites (Table 1.3). Differences
in proportional community similarity (Table 1.4) reflect differential community
composition relative to available microhabitat characteristics. The dominance of
A. cristatellus and S. nicholsi at each site clearly resulted in relatively low
diversity (Shannon Index, H’) and evenness (H’lH’mx) estimates (Table 1.5).
* Lizard community composition did not differ significantly among the four habitat
types (sites) (Kruskal-Wallis ANOVA, p=0.454). Additionally, absolute
abundance of lizards did not differ in Guanica Forest between 1997 and 1998
(paired ttest, p=0.908).

Individual species or groups of species showed significant differences
among sites. Because of low sample size in all habitats, it was very difficult to

determine whether the data were normally distributed. Both parametric and

25

nonparametric analyses of variance (ANOVA and Kruskal-Wallis ANOVA,
respectively) were used to test for statistically significant differences among

sites(a=0.05) and results are reported for both methods.

 

 

 

 

 

0
g Iothers
B IPhylodactyfus wirshing’
s I Sphaerodactylus roosevelti
g D Sphaerodactylus nicholsi

. 3 IAmaiva exsul
& I Anois cristatelus

 

 

 

 

 

 

Sites

Figure 1.3 Five most predominant species comprising the lizard community in
Guanica Forest. Species composition was not significantly
different among sites (Kruskal-Wallis ANOVA, p=0.454).

Table 1.3 Lizard species present in study sites within Guanica Forest

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Coastal Central North Ravines Total
Anolis cristatellus X X X X X
Anolis cooki X . X X
Anolis stratulus X X X X
Anolis pulchellus X X X X
Anolis poncensis X X X
Ameiva exsul X X X X X
Ameiva wetmorei X X X X X
Sphaerodactylus nicholsi X X X X X
Sphaerodactylus roosevelti .. X X X X
Phyllodactylus wirshingi X X X X
Total species richness 9 8 8 7 10

 

 

 

 

 

 

 

26

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 1.4 Proportional Similarity of Sites within Guanica Forest
Site Coastal Central North
Central 0.75
North 0.59 0.65
Ravines 0.81 0.71 0.58

Table 1.5 Lizard community composition in Guanica Forest. Relative

abundance of the ten species indicates dominance of A. cristatellus

and S.’ nicholsi. Community diversity and evenness metrics also

reflect this pattern.
_2.__5 ecies 993M 29mm m 891mg natal
Anolis cristatellus 0.23 0.38 0.55 0.27 0.33
Anolis cooki 0.02 0 0 <0.01 <0.01
Anolis stratulus 0 0.01 0.01 0.01 0.01
Anolis pulchellus <0.01 <0.01 <0.01 0 <0.01
Anolis poncensis <0.01 <0.01 0 0 <0.01
Ameiva exsul 0.04 0.02 0.01 0.01 0.02
Ameiva wetmorei 0.01 <0.01 <0.01 <0.01 <0.01
Sphaerodactylus nicholsi 0.53 0.42 0.26 0.67 0.53
Sphaerodactylus roosevelti 0.09 0.17 <0.01 0 0.05
Phyllodactylus wirshingi 0.09 0 0.18 0.04 0.07
Diversity (H’) 0.59 0.48 0.45 0.35 0.49
Evenness (J’) 0.62 0.57 0.53 0.45 0.49

 

Absolute abundance of lizards sampled along transects and in plots

differed significantly among sites within Gua'nica Forest (Tables 1.6 and 1.7).
Individual species also exhibited statistically significant associations with the
habitat types represented by the four sites in the forest; abundance of A.
cristatellus, A. cocki, and S. nicholsi all differed significantly among sites. Anolis
cocki was significantly more abundant at the coastal site while 8. nicholsi was

more abundant in ravines than any other site.

27

Table 1.6 Variables which differed significantly among sites in Gua'nica
Forest, Kruskal-Wallis ANOVA.

 

Variable p value
Anolis cristatellus abundance 0.003
Anolis cocki abundance 0.013
Sphaerodactylus nicholsi abundance 0.018
Absolute transect abundance 0.005
(# Anolis and Ameiva per transect)
Absolute plot abundance 0.04
(# geckos per plot)

 

Measured habitat variables (% rock cover and litter depth) differed
significantly among sites (ANOVA, p<0.001). Percent rock cover in 2 x 2m
subplots was arcsine-square root transformed prior to analysis so that the data
more closely approached a normal distribution (Zar .1984). Plots sampled in the
central site had significantly lower percentage rock cover and a deeper litter
layer than the other 3 sites (Figures 1.4 and 1.5, respectively). Density of S.
nicholsi, which is a species associated with substrate habitats, is significantly
correlated with litter depth (Spearrnan Rank correlation, rs=0.227, df=90,
p<0.05).

Although slope degree and angle were not measured, there were no
significant differences in the composition of the lizard community or abundances
of any individual species or groups of species with respect to aspect. Although
aspect may be very important in structuring plant communities in the dry forest
(Ramjohn et al., unpubl. data), it did not appear to play an important role in

structuring lizard communities in this region.

28

dozmezmm bacon 9.63.026 .mc._aEmm 9:22:30 9:3 .0: Sn 2.m m .m 8282 22s .9: 3.0ch ..

 

 

 

wflc NHC OHC rHC
.8 F. a? .8 P. mom 0 .3 F. X: .8322 witnessed
OHC NHC F":
o .. .mom. mom .3 P. 3. £288. 8588528
0?": ”NC 0": QHC
team». «mum o.3 .. m 5 a_.... a. .8 .83. m8 520.: 3889.“..ch
NHC FHC NHC N":
.N. N .m. e .v. m .m. a 205.2. 965...
QHC NH: .0": PF":
.5. mm .4. o .‘F F. em .3. we see meme...
OHC OHC
o o . t .. c.3823 9.05..
O":
o .. .. .. 3.20.3 ”.354
QHC NHC NHC OHC.
.mm. mm .m. m .m. m 0 9.32m 20% .
PHC OHC OH: 0H:
“.4. v ..o . .o 2...... ea .208 £8...
Emu: 5.": 3 We mm":
2.3 F. Rm .3. .8 ”.2 F. 5.. $8. EN 3.3320 2.8..
855. £32 ‘ .828 $1.de Bum—w

 

 

 

 

30:22.6 .8 $2 2 now: am? <>Oz< >95 25 .7120“. 8.530 c. 223.. .c 3.550 NF 03¢...

27. .cm 226.29... .0 69:3:
.22 n : ..mo.ova .09.. 3.9.2.... 550cm 26 EB. 3.56586 .mtfi 29622.3 9.529: 5.3
32 m 55.3 mo:_m> ucm _.m£m.m:c.>.cc. 3 83.2% $8 cmcE E8262 22.832 82.». 96:5

29

 

 

100
’6
d1 80—
E
.2
g60‘ a
.9 E c
E 40— E“ i
8
3 be
x 20— P
0
a?
s 0‘
I l l l

 

 

Coastal Central North Ravines

Figure 1.4 Differences in percentage rock cover in 2 x 2 m subplot
among sites (mean 1 SE). Points with matching letters
indicate significant pairwise differences
(T ukey's HSD, p<0.05).

 

10

8-
E6‘ abc
E i
§.4_ EC
is Ea. 5"
a 2—

0-

I I I T

 

 

 

Coastal Central, North Ravines

Figure 1.5 Differences in litter depth in 2 x 2 m subplots
among sites (mean i SE). Points with matching
letters indicate significant pairwise differences
(T ukey’s HSD, p<0.05).

3O

Discussion

Guanica Forest is the most extensive tract of dry forest in Puerto Rico,
and it supports the richest and most complete native dry forest lizard community
on the island. This large and continuous forest offers a wide variety of habitat
types that accommodate many species’ ecological preferences. Of the 13
species that have been recorded in the dry zone of southwestern Puerto Rico
(Table 1.1, Rivero 1978, Schwartz and Henderson 1991), 10 were found within
Guanica Forest during this study.

The three species that are absent from the reference community were
Sphaerodactylus macro/epis, Hemidactylus brooki, and Mabuya mabouya.
Sphaerodactylus macro/epis has been reported to occur in the more humid
ravines within the northern boundaries of the forest (M. Canals Mora, pers.
comm), but Was not encountered during my surveys as a result of its relative
rarity and patchy distribution in the most arid parts of the island. Hemidactylus
brooki is primarily edificarian, and although it is rarely encountered in natural
forested situations, this species is very abundant in and around buildings
(Schwartz and Henderson 1991, pers. obs.).. Mabuya mabouya is also very rare
in Puerto Rico, with fewer than 15 specimens ever collected from the island
(Rivero 1978).

Anolis cocki was significantly more abundant at the coastal site than at
any other site within “Guanica Forest. One individual was recorded from a ravine
in close proximity to the cOast, but this species was conspicuously absent from

inland and upland locations. Previous studies have indicated that A. cocki is

31

restricted to a limited number of discontinuous coastal scrub habitats within 1 km
of the coast in southwestern Puerto Rico (Marcellini et al. 1985, Jenssen 1990).
Over the past two decades, field studies have documented disappearing
populations, reductions in the extent of its previous distribution, and microhabitat
invasions by A. cristatellus where the two species were previously allopatric
(Jenssen et al. 1984, Marcellini et al. 1985, Ortiz 1990). Anolis cocki is of
special concern for local conservation officials, who take its distribution into
consideration in current management and development planning (M. Canals
Mora, pers. comm.) The implications of competitive interference between
these two species is addressed in the third chapter of this thesis.

Although A. cocki was present in the coastal regions of Gua’nica Forest, it
was by no means abundant. Ortiz (1990) reported that A. cocki was locally
abundant where present, but low population densities reported here could be
indicative of increased competitive pressure from A. cristatellus. Another
potential explanation for low A. cocki densities is misidentification, which could
lead to underestimates of. actual abundance. It is very difficult to visually
distinguish between these two species in the field, but following the
recommendations of Marcellini and Jenssen (1983), accurate species
identifications were possible for adult individuals as long as the tail had not been
recently autotomized. Williams (1972) was the first to suggest that A. cocki was
potentially threatened by extinction, and evidence of competitive interference
and displacement into less desirable habitats (both attributable to the more

abundant and widespread A. cristatellus) indicates that A. cocki is under intense

32

ecological pressure and has an uncertain future (Ortiz and Jenssen 1982,
Jenssen et al. 1984, Marcellini et al. 1985, Jenssen 1990).

Another anoline lizard, A. poncensis, is restricted to the arid regions of
southwestern Puerto Rico (Schwartz and Henderson 1991). In this study, it was
found to occur only at the coastal site in Guanica Forest, but was not
encountered often or at all during quantitative transect sampling. This species is
a grass-bush ecomorph, and shares much of its habitat with A. pulchellus, a
much more widely distributed species which is often found at extremely high
population densities in grassy habitats (up to 20,000 individuals/ha, Gorman and
Harwood 1977). Where sympatric, A. poncensis often occupies bushes and
fenceposts while A. pulchellus seems to prefer the more exposed grassy areas
(Rivero 1978). To my knowledge, the potential for competitive interference and
displacement between A. poncensis and A. pulchellus has not yet been
assessed, although these two species may represent a dynamic situation similar
to that discussed above for A. cocki and A. cristatellus. While this was not a
question addressed in this study, it is nonetheless a potentially interesting
avenue for future research.

The fourisites represented different habitat types within a continuous
forest; habitat variables differed significantly among these sites. The site in the
central portion of Guanica Forest had significantly less rock and deeper leaf litter
compared to substrate characteristics of the other three sites. Geckos would be
expected to respond to variations in substrate habitat, as they are found

primarily in and among substrate debris in all habitat types in the dry forest

33

region (Rivero 1978). Sphaerodactylus nicholsi abundance showed a significant
positive correlation with leaf litter depth, but 8. nicholsi was surprisingly much
more abundant in ravines than at the central site where the litter layer was
deepest. While S. nicholsi prefers a dry habitat (Thomas and Schwartz 1966),
their soil and leaf litter arthropod prey may be more abundant in the more humid
substrate habitat of ravines. The distribution and abundance of S. nicholsi in
Gua'nica Forest may represent a tradeoff between their preference for a xeric
habitat and the need for an abundant and stable food supply. Sphaerodactylus
roosevelti and P. wirshingi abundance would also be expected to vary with
substrate characteristics, but their low occurrence in this study precluded the
detection of any significant patterns.

Anolis cristatellus abundance also differed significantly among the four
sites in Guanica Forest. This species was least abundant at the coastal site,
where it is partitioning the habitat and competing with the other trunk-ground
ecomorph, A. cocki. Since anoline lizards are largely arboreal, the density of
individuals may also be influenced by the number and distribution of trees
present in the habitat. Although not specifically addressed in this study, the four
sites very probably differ in the distribution of trees and tree species
composition, leading to differences in A. cristatellus density. The ravines contain
larger mature trees and more abundant vines and lianas than the surrounding
dry forest (Famsworth 1993), which provide a greater density of perches for
these arboreal lizards. Within the reference site, semi-evergreen forest has

higher basal area as compared to deciduous and scrub forest types, but stem

34

density is greatest in deciduous forest (Lugo et al. 1978, Castilleja 1991,
Ramjohn et al. unpubl. data).

The total number of lizards encountered during quantitative sampling
varied greatly among species. While patterns potentially exist to explain
distribution and abundance of other species within this community, their
infrequent occurrence did not allow determination of any noticeable trends.
Increasing the number of transects and plots as well as sampling additional
habitat types and geographic locations within Gua'nica Forest would greatly
improve the reliability of these results. Altering the survey methods would also
lead to more accurate density estimates. Some species (i.e., Anolis stratulus)
are present in great numbers higher in the canopy than can be accurately
detected from ground-level observations, and vertical transect methods yield
dramatically higher density estimates than ground transects (Reagan 1992,
1996). The relative abundance and density estimates reported for A. stratulus
are likely to be underestimates for these reasons.

In conclusion, the reference lizard community of Guanica Forest is
comprised of 10 species, some of which are differentially distributed throughout
the four different forest habitat types sampled in this study. This continuous
forest provides excellent habitat for dry forest lizards, and provides baseline
conditions against which data from other areas can be compared. Remaining
dry forest habitat in southwestern Puerto Rico exists as fragments of varying
size, degree of isolation, and vegetation type; it is important to determine the

impact of large scale habitat fragmentation on the distribution and abundance of

35

dry forest lizards to assess the status and conservation potential of both the
species and the forest fragments. The influences of fragment and landscape

characteristics will be addressed in the following chapter.

36

Chapter 2

Lizard Community Composition and Structure
in a Fragmented Subtropical Dry Forest Landscape

Introduction

Habitat fragmentation, resulting in reduction of total habitat area and
isolation of forest fragments, has been recognized as a major threat to tropical
forest ecosystems (Turner 1996, Dale and Pearson 1997, Laurance and
Bierregard 1997). Typically, fragments are able to support only a subset of the
biological community found in a continuous forest, and the composition and
complexity of the fragment community depends on a variety of fragment
characteristics including size and degree of isolation from other forest patches
(Cutler 1991).

The theory of island biogeography (MacArthur and Wilson 1967) was
developed to describe continental mainlands and true islands surrounded by
ocean; it has since been adopted and tested for habitat ‘islands’ (fragments)
surrounded by an ‘ocean’ of alternative land uses (e.g., Harris 1984, Bierregard
and Dale 1996). The theory of island biogeography predicts that the rates of
immigration and extinction in a habitat ‘island’ (fragment) are regulated by the
size of the island and the distance to the nearest ‘mainland’ source of colonizers
(MacArthur and Wilson 1967, Quinn and Harrison 1988). Understanding the
consequences of habitat fragmentation on fragile tropical forest ecosystems will
aid in the preservation and conservation of biodiversity worldwide. Turner

(1996) emphasized that determining the long-term viability of fragments and their

37

extinction-colonization equilibria will be the greatest contribution to the
development of feasible and effective conservation strategies.

Globally, tropical and subtropical dry forests have been more severely
impacted by human activities than wet tropical forests (Tosi and Voertman
1964). The climate, soils, and topography of dry forest regions make them very
attractive for settlement and conversion to agriculture. Most of the dry forest
habitat in Puerto Rico and elsewhere has been extensively disturbed or
completely eliminated as a result of human activities (Murphy et al. 1995). In
southwestern Puerto Rico, most dry forest habitat exists as islands of remnant
and recovering forests sparsely scattered throughout a highly human-dominated
and disturbed landscape.

These trends of disturbance and deforestation are characteristic of the
entire island; 95% of the island was forested during pre-colonial times, but by
1948 only 5% of the forest remained, mostly in steep and topographically
inaccessible areas (Birdsey and Weaver 1982). Puerto Rico has since
experienced a shift in its economy from a predominantly agricultural to an
industrialized society with greater emphasis on manufactured products
(Morales-Carrion 1983, Dietz 1986). This shift has reduced anthropogenic
impacts on the remaining forested areas, and by 1985 the forested area had
increased to 34% (Birdsey and Weaver 1987). In the vicinity of Guanica Forest,
the amount of forested land has steadily increased since 1936 (Lugo et al.

1996).

38

. Anthropogenic disturbance of forested landscapes has long-term effects,
both in duration and environmental consequences. It is imperative that tracts of
recovering as well as remnant forest be preserved; structure and diversity of
secondary forests are able to support rich ecological communities, although the
species composition-of undisturbed dry forests may not be reached for centuries,
if at all (Aide et al.‘1996, Thomlinson et al. 1996, Rivera and Aide 1998).

The remnant and recovering fragments of dry forest in southwestern
Puerto Rico are scattered throughout a landscape dominated by agriculture and
urban development. Much of the native biota of this region is restricted to these
isolated fragments, and understanding how certain species or groups of species
respond to habitat fragmentation will greatly contribute to the management and
conservation of remaining dry forest patches. Vertebrates (especially birds and
mammals) have been a popular group of study organisms for fragmentation
research (Turner 1996, Bierregard et al. 1997). Previous studies indicated that
species which have large area requirements, poor dispersal capabilities, limited
abilities to cope with disturbance, or low population densities are particularly
susceptible to local extinction in the face of habitat fragmentation (Turner 1996,
Laurance et al. 1997).

The dry forest zone of southwestern Puerto Rico is a critical habitat area
for many species of vertebrates. In this region, lizards are conspicuous and
abundant as well as ecologically important organisms in terms of community
structure and trophic interactions (e.g., Lewis 1989). Habitat fragmentation has

the potential to negatively affect the lizards via reduced habitat suitability and

availability resulting‘from altered microclimatic regimes, or limited dispersal
opportunities resulting from isolation (Saunders et al. 1991). Previous studies
concerning the effects of habitat fragmentation on lizard communities found that
the species composition of fragmented habitats was biased towards generalists
and more related to site ecology than degree of isolation from other fragments
(Kitchener et al. 1980, Kitchener and How 1982, Sarre et al. 1995). Species
found in small woodland remnants where habitat alteration had resulted in
reduced tree density, understory shrub cover, and litter were subsets of those
which occurred there prior to fragmentation (Smith et al. 1996). In other
anthropogenically influenced habitats, the most disturbed areas supported the
least diverse lizard communities (Lenart et al. 1997).

Certain ecological characteristics may allow some species to persist in
disturbed habitats, at least in the short term. The majority of dry forest lizard
species in southwestern Puerto Rico can be classified as ecological generalists
(Schwartz and Henderson 1991), which may confer an advantage in this
fragmented landscape. In areas where forest disturbance and degradation are a
continual threat to the native biota, habitat specialists (with narrow ecological
preferences or restricted to areas of ‘natural’ vegetation) do not persist as well
as generalists that can tolerate a wider range of habitat conditions, including
more degraded habitats (Kitchener 1982, Kitchener and How 1982, Sarre et al.
1995). Lizards have been found to be relatively resilient to habitat
fragmentation. For example, lizards have not suffered any extinctions and have

perSisted even in small remnants in the Australian wheatbelt (Kitchener et al.

40

1980). Overall, generalist species have less specific habitat requirements and
are also better able to migrate between fragments, despite having to cross less
hospitable habitat that most often surrounds isolated forest remnants.
Assessing the response of the dry forest lizard community is a first step in
gaining a better understanding of how faunal assemblages respond to
landscape-level habitat fragmentation phenomena. Provided that the trend of
forest recovery continues in Puerto Rico, native lizards will likely be able to
colonize and establish populations in additional patches of recovering forest as
they become available. The objectives of this study were to: (1) evaluate the
response of the lizard community to fragment and landscape characteristics in
the dry forest zone, (2) compare lizard communities in fragments of varying sizes
to the reference community of Gua’nica Commonwealth Forest, (3) assess the
ability of fragments to support representative lizard communities, and (4)
evaluate conservation potential and management opportunities for forest

fragments in this region.

Materials and Methods
The Study Region
The southwestern region of Puerto Rico contains the majority of the dry
forest habitat on the island, although this habitat exists primarily as isolated
patches interspersed throughout a landscape dominated by agriculture (pasture)
and urban/residential areas. The dry forest life zone (sensu Holdridge 1967) of

soUthwestem Puerto Rico is typified by low rainfall (600-1100 mrn annually),

41

warm temperatures (mean annual biotemperature >17° C), and a ratio of
potential evapotranspiration to precipitation (PET/P) which exceeds unity (Ewel
and Whitmore 1973). Study sites were located throughout the dry forest life
zone (approximately 18° N and 66° 35’ W to 66° 12’ W). Guanica
Commonwealth Forest, a continuous and relatively undisturbed dry forest, is
located in the approximate geographic center of this region (Figure 2.1). It
encompasses roughly 4000 ha and represents one of the best remaining
examples of subtropical dry forest in the world (Murphy and Lugo 1990), making
it a suitable reference site for this study. Lizard habitats and species

composition of Gua’nica Forest were discussed in the previous chapter.

Dry Forest Fragment Study Sites
Previous studies by Ramjohn et al. (unpubl. data) have identified more

than 300 fragments (open and closed forest) representing both remnant and
recovering patches of dry forest habitat in southwestern Puerto Rico; the
vegetation has been surveyed and studied extensively in 41 of these fragments
(Ramjohn et al. unpubl. data). To assess the impact of habitat fragmentation on
the native dry forest lizard communities, a subset of these fragments was
selected for study. The twelve fragments chosen for lizard community sampling
ranged in area from 0.006 ha (a single isolated tree and surrounding shrubs) to
>800 ha (Figure 2.1). Four fragments were selected in each of three size

categories: small (<1 ha), medium (1-10 ha), and large (>10 ha). The 12

42

.620... 8.530. 23 8:222
2... new 25832. 822 NP 2.. .o 5.300. cam 93 9:92 on. 3:5,ch 02m. 95?. €232,538 .o 322 ed 233....

 

 

 

 

        
  
   

838:2: 5 E k ‘ o k
9 or
wm..o moEm:
z u n. . 0 V1

 

 

 

 

 

.wmmm...u.._.w... . mmh<
33.2....MH . noQNEm.
och 2...

T 320“. >5 30.3235

  

43

fragments contained a range of habitat conditions and vegetative associations,
including both coastal scrub and upland deciduous forests. These fragments
differed in a variety of characteristics, providing the basis for comparisons
among fragments and between fragments and the reference site, Guanica
Forest. Sites within Guanica Forest were chosen to match the characteristics of
the fragments (seeChapter 1).

The 12 forest fragments varied in size, shape, topographic location,
vegetation composition and structure, degree of past and present disturbance,
as well as the type of substrate. For instance, one fragment was located in close
proximity to the coast and situated on limestone outcroppings almost completely
lacking soil (Site 16, Figure 2.1). Other fragments were located on abandoned
pastures and had more well-developed soils (Sites 30, 36, 40; Figure 2.1). In
addition, some sites have been under extreme pressure from human activities
for prolonged periods of time (Sites 5, 10, 16; Figure 2.1) while others have
remained virtually untouched following their isolation from adjacent forests (Site
4, Figure 2.1). Most of the fragments were located in relatively inaccessible
areas such as ravines, steep slopes, and hilltops that were not suitable for
agriculture or urban development. Although such a large number of factors
could not be controlled in this study, the range of environmental conditions
present in the fragments allowed comparisons valuable for generating additional
hypotheses about the complex effects of fragmentation on ecological

communities.

Fragment and Landscape Analysis

Aerial photographs (1 20000) taken in February 1998 were obtained
from the Office of Photogrammetry at the Puerto Rico Highway and
Transportation Authority. Photointerpretation and delineation of polygons to
determine the distribution of land cover types surrounding each of the 12
fragments was the first step in constructing a geographic information system
(GIS) database. Land use and cover types were delineated using the
classification scheme of Lugo et al. (1996). Photointerpreted polygons were
digitized into ARC/INFO (ESRI, Redlands, CA) and topology was constructed.
The coverages were then georeferenced and transformed into Universal
Transverse Mercator (UTM) grid coordinates (zone 19). Adjacent coverages
were aligned, edgematched, and then joined into a final useable coverage~ of the
12 fragment study sites, Guanica Forest, and surrounding land use types for
further analyses. Additional spatial data (hypsography, road networks) were
provided by the lntemational Institute of Tropical Forestry (llTF, Rio Piedras,
Puerto Rico) for the GIS analysis.

To evaluate spatial patterns of lizard species richness and abundance,
several fragment and landscape characteristics were determined from the GIS

database. Fragment characteristics investigated in this study included area,

 

perimeter, and compactness (Table 2.1). Compactness (K1= 2‘11tarea/perimeter;
Bosch 1978, Davis 1986) describes the shape of a fragment and provides an
indication of the amount of edge habitat; edge effects can be very important in

structuring some communities (Saunders et al. 1991). Compact forms are

45

Table 2.1 Characteristics of forest fragments and reference site (Guanica

Forest) selected as study sites

 

 

 

Site ID'r Area Perimeter Patch Distance to nearest Primary
(ha) (m) Compactness Forest Fragment (m) Aspect
Small: -
40 0.01 27 0.99 392 S
36 0.06 93 0.90 1 1 0 S
30 0.26 227 0.79 364 E
27 0.69 409 0.72 1 0 W
Medium:
28 2.0 836 0.60 8 E
13 3.1 920 0.68 .20 Flat
' 16 3.4 8320 0.79 278 E + S
10 6.4 p 1723 0.52 15 N‘
Large:
7 34 3517 0.58 10 All
5 97 5258 0.66 30 All
4 137 7563 0.55 10 All
2 854 23277 0.45 20 All
Ref. Site’ 3725 47885 0.45 All

 

 

*from Ramjohn et al. (unpubl. data)

1

46

area and perimeter measurements are for the entire tract of continuous forest
in which the eastern portion of Guanica Forest lies.

effective in conserving resources, while convoluted forms have enhanced
interface and interaction with the surrounding landscape (Portmann 1967, Harris
and Kangas 1979).

To investigate the isolation of fragments and their relationships with the
surrounding landscape, ARC/INFO was used to create a 1000 m buffer around
each of the study sites (Appendix 8, Figure 8.1). The 1000m buffer was used to
characterize the fragment in a landscape context. Buffer coverages were
imported into ArcView 3.1 (ESRI, Redlands, CA), and the following landscape
variables were measured within the buffer zone: (1) number of closed canopy
forest fragments, (2) distance to nearest closed canopy fragment, (3) percentage
of closed canopy forest, (4) percentage of open canopy forest, (5) percentage of
total forest, (6) percentage of agriculture, and (7) percentage of urban land use.

The land use in the area directly adjacent to the perimeter of each study
site may affect the lizards by inhibiting patch-to-patch dispersal. The influence
of adjacent land use was evaluated by calculating the percentage of fragment
perimeter comprised by each major land use category (open forest, closed
forest, agriculture, and urban). If a fragment was separated from other closed
canopy forest only by a road, it was considered adjacent. Other types of land
use and cover (i.e., water, wetland, barren land) were directly adjacent to very

few sites and were not included in the analysis.

47

Lizard Community Sampling

Quantitative surveys of the lizard communities in the 12 fragments and
reference sites were conducted in the summers of 1997 (May-July) and 1998
(July-August). Nine fragments were sampled in 1997; in 1998, five of these were
resampled and three additional fragments were added to the study. Due to
differences in the biology and activity patterns of the 10 commonly encountered
dry forest lizard species, two distinct survey strategies were required to obtain
accurate estimates of diversity and abundance (described in Chapter 1).
Abundant and visually conspicuous arboreal and free-ranging lizards (Anolis
spp. and Ameiva spp.) were censused along 100 m transects of fixed area.

A Three members of the genera Sphaerodactylus and Phyllodactylus are primarily
nocturnal and/or fossorial, making it unrealistic for transect surveys to obtain
accurate diversity and abundance measures. These species were sampled by
exhaustively searching randomly selected 2 x 2m subplots nested within 10 x
10m plots located at the beginning of each transect (see previous chapter for a
detailed description of transect and plot sampling).

The number of transects and plots was scaled to the area of each site
(Table 2.2) and represented considerations of adequate sampling efficiency
coupled with time constraints. When site area and topography allowed (i.e., in
large fragments), selection of sampling locations was stratified using four aspect
categories (N—, S-, E-, and W-facing slopes). Sampling effort in each of these
categories was determined by the proportion of total fragment area comprised by

eaCh aspect category.

48

In addition to the quantitative transect and plot sampling methods
described above, qualitative focal searches (site walks) of study sites sampled in
1997 were made during the summer of 1998 to make certain that the species
lists for each site were complete and accurate. Some of the large fragments
encompassed areas that-were too large to make quantitative sampling of the
entire site feasible. A species list was compiled for, each site and considered
complete if all ten species found in the reference site were present in any given
fragment. Site walks yielded only qualitative presence/absence data, and were
conducted only at sites with incomplete species lists. Data from the five sites
sampled in both 1997 and 1998 were used to test for temporal differences in
lizard community composition and abundance. Sampling efficiency was
assessed for each fragment, and species-area curves are provided in Appendix

B.

49

Table 2.2 Summary of sampling design in forest fragment study sites

 

 

Site Area Total Area of No. of No. of Site

ID #T (ha) Transects (m2) 10m x10m 2m x 2m Walk"
(No. of transects) plotsi plots

Small

40§ 70.01 entire fragment - 4 ‘ No
searched

355 0.06 236(2) - 4 No

30§ 0.26 511 (2) - 5 Yes

27 0.69 825(2) 4 12 No

Medium

28 2.0 825(2) 4 12 Yes

13 3.1 825(2) 4 12 No

16 3.4 825(2) 4 12 Yes

10 6.4 825(2) 4 12 Yes

Large

7 34 2520(6) 8 24 Yes

5 97 3360 (8) 12 36 Yes

4 137 3360(8) 12 36 Yes

2 854 3360(8) 12 36 Yes

 

I From Ramjohn et al. (unpubl. data).
* Located at the beginning of each transect and additional randomly located
plots (see text).
* Presence/absence data only.
§ Sites were not large enough to establish complete 100 m transects; instead
the entire fragment was searched (site 40) or transects were established along
the long and short axes of the fragment (sites 36 and 30).

50

Statistical Analyses

Sites were compared using a one-way analysis of variance (ANOVA) to
determine if the three size classes of fragments differed in lizard community
composition (both species richness and individual species abundances) from the
reference site. As a result of low sample size, it was difficult to determine
whether the data were normally distributed. Both parametric and nonparametric
methods were used (ANOVA and Kruskal-Wallis ANOVA, respectively), and
results are reported for both analyses.

Pearson correlation analysis was used to investigate associations of
lizard diversity and abundance with fragment and landscape characteristics.
Additionally, potential relationships between lizard variables and termite or plant
species richness were also investigated with correlation analysis. Data were
transformed to better approximate the normal distribution (Zar 1984); fragment
area was log transformed, and percentage data (adjacent and surrounding land
use) were arcsine transformed. Multiple regression models were used to identify
relationships between lizard variables (species richness and individual
abundances) and combinations of fragment and landscape variables. All
univariate‘ statistical analyses were performed using Systat (version 5.0,
Evanston, IL).

Principal components analysis (PCA) was used to summarize the
variation in the fragment and landscape variables (Morrison 1990); a reduced
number of independent variables (principal components) were then used to

inveStigate patterns among fragment and landscape characteristics of the 12

51

fragment study sites. Since variables were not commensurable and a single
variable (fragment perimeter) dominated the covariance matrix, PCA was
performed using the correlation matrix (Joliffe 1986).

Correspondence analysis (Greenacre 1993) was performed using lizard
densities (#Iha) at 12 fragments and four sites within Gua'nica Forest to reveal
possible patterns of associations and differences among species and sites that
may be ordered along an underlying environmental gradient. Lizard densities
were used instead of raw counts for two reasons: ( 1) to ameliorate differences in
sampling methods (transect vs. plot), and (2) to weight the fragments in
proportion to their total frequency of lizards per unit surface area (Greenacre
and Vrba 1984). All multivariate statistical analyses were performed using SAS

(version 6.12, Cary, NC).

Results
The survey of lizard communities in dry forest fragments in southwestern
Puerto Rico revealed a total of 10 species representing three families and four
genera (Table 2.3); the same species were also encountered in the reference
site (see Chapter 1, Table 1.2). Three of these species (A. cocki, A. poncensis,
and S. roosevelti ) were not found in thesmallest fragments (<1 ha). All other

species were represented in at least one fragment of each size class.

52

.020”. 02:80

55.3 09.0 :0 .0 .90. 9:0.3E30 .1. ._. .00:.>2 n a ..00000 8.09.5.0: 5.2, .020. 03030.00: .56: n 2 500000
8.09.5300 5.3 .020. 03030.00: ._2.:00 u 00 .0200 .9080 u 00 9.0 00:20.2 05 5020”. 02:80 :.:..>> 095 .

 

 

 

 

 

 

 

 

 

 

 

o. N 0 0 m o. m 0 v v 0 v 0 N 0 N o I 0025: 00.8% 2.0
N. x x x x n. n. n. x x x x x n. 555...: 05.355830
‘0 x n. x x x n. n. x 32008. .0
E x x x x x x x x x x x x x x x .055... 9.8890200
08.:9200 2.80“.
o. x x x x x x x x x x x 05.59 .<
0. x x x x x x x x x x x x x x x x see 2.08...
002.9 3.50“.
0 x n. x x x x n. 0.88:8 .<
0 x n. n. 0 x x x 0 x n. 02.20.50 .<
o. x x x x x x x x n. .x x 03.5.20 <
0 x x x x x .558 .<
0. x x x x x x x x x x x x x x x x 9.0.0.05 0.5:...
08:59.00 >_.E0...
.502: h m z 00 00 N v 0 N o. 0. m. N N on 00 ow 028:0
02.0 0 .20 .0: 2 SA 2 o: 2 . v
2.02. 9.0. :2000 9:9..033 0 0::30 00.00.00
5:002: u n. 00050:. 05.0800 .0... 3:0 .0082. 9:05:80 >0 00.00.00 00.0000 n x
N 032.

.Cmmhou SEW—40V 0:0 00:0:0m0: 0r: 0C0 9:09:00: “00.0% N r 02. E mgomaw U..NN= ho QOCQtDOOO

53

Fragment and Landscape Characteristics

Lizard species richness and abundance of each species did not differ
significantly among fragment size classes or between fragments and the
reference site (ANOVA, p>0.05; Kruskal-Wallis ANOVA, p>0.05; see Appendix
8, Tables 3.1-3 for species densities in small, medium, and large fragments,
respectively), although there appeared to be a trend of increasing species
richness with increasing fragment area (Figure 2.2). Indeed, there were
significant correlations between species richness and fragment characteristics
(Table 2.4). Only 4 of 33 correlations (~12%) with fragment characteristics were
significant (Table 2.4), however, it should be noted that fragment area and
compactness are very strongly correlated (r = 0886, p < 0.001 ), and perhaps
should be treated as a single variable.

Land use immediately adjacent to fragments did not appear to greatly
affect lizard community composition; only four of 55 correlations (~7%) were
significant (Table 2.5). Multiple regression analyses using backward selection
(p<0.05) yielded no models for species richness and density of the most
abundant species within each family (A. cristatellus, A. exsul, and S. nicholsr)
that contained more than one significant fragment or landscape predictor
variable. The number of lizard species in forest fragment study sites was
significantly correlated with the number of termite and tree species, and both of
these variables were significantly correlated with fragment area (Table 2.6).

Additionally, certain individual lizard species were significantly correlated with

54

variables describing vegetation structure and diversity and number of termite

species in each fragment (Table 2.6).

 

 

12
10—
.3
g 8— f i
(I)
‘56- f
g
E
2 2—
0..
l l l l

 

 

<1 ha 1-10 ha >10 ha Ref. Site

Study Site Size Class

Figure 2.2 Number of lizard species (mean t SE) for fragment size classes
and reference site. Differences were not significantly different
(ANOVA, p=0.102; Kruskal-Wallis ANOVA, p=0.095).

55

 

 

Table 2.4 Pearson correlation coefficients (r) of species abundance
(# individuals/ha) and species richness with fragment
characteristics. *p<0.05, **p<0.01

Lizard Variables Area Perimeter Compactness

Anolis cristatellus 0.312 0.039 -0.561*

Anolis cocki 0.099 0.010 0.069

Anolis stratulus 0.035 -0.108 0.146

Anolis pulchellus 0.021 0078 -0.014

Anolis poncensis 0572* 0.401 —0.486

Ameiva exsul -0.130 -0.246 -0.007

Ameiva wetmorei -0.047 -0.094 -0.133

Sphaerodactylus nicholsi 0.134 —0.058 -0.155

Sphaerodactylus roosevelti 0.314 0.368 -0.458

Phyllodactylus wirshingi -0.082 -0. 147 0.081

Number of lizard species 0756“ 0.565 0654* ~

 

 

 

 

Table 2.5 Pearson correlation coefficients (r) of species abundance

(# individuals/ha) and species richness with immediately

adjacent land use types around fragment perimeter

*p<0.05, **p<0.01
Lizard Adjacent Land Use
Variables Open Forest Closed Forest Total Forest Agriculture Urban
A. cristatellus 0.377 0.509 0.552 -0.538 0.242
A. cocki -0.302 -0.239 -0.317 -0.269 0.266
A. stratulus -0.213 -0.099 -0.205 0.011 0.168
A. pulchellus 0.544 0.712“ 0.768“ -0.395 -0.102
A. poncensis 0.060 0.201 0.045 -0.146 0.249
A. exsul -0.320 -0.372 -0.343 0.366 0.181
A. wetmorei 0.515 0.328 0581* -0.476 -0.296
S. nicholsi 0.140 . -0.155 0.048 0.120 0.010
S. roosevelti 0.206 -0.183 0.024 -0.240 -0.21 1
P. wirshingi 0.371 0.248 0.431 -0.485 -0.186

Species richness 0.288 0.266 0.276 -0.650* 0.348

 

 

56

 

 

 

Table 2.6 Pearson correlation coefficients (r) describing associations

between lizard, termite, and vegetation variables.

* p<0.05, **p<0.01
Lizard # Plant| # TreeI # Stems # Stems # Stems Basal #
Variables Species Species 3 1cm 3 2.5cm _>_ 5cm Area‘ Termite’

DBH' DBH' oau' (mzlha) * Species

A. cristatellus 0.346 0.249 0.242 0.447 0.193 0.232 0.505
A. cocki -0.105 -0.029 -0075 -0.310 -0270 -0.356 -0058
A. stratulus 0.167 0.245 -0219 -0373 -0.286 -0044 0.273
A. pulchellus 0.088 0.059 0.333 0624* 0781** 0744” 0.205
A. poncensis 0.523 0634* 0.368 0.254 -0055 0.068 0.248
A. exsul 0.273 0.154 -0041 -0.258 -0.370 0.022 0.133
A. wetmorei -0.158 -0.182 0.284 0693* 0691* 0.386 0.331
S. nicholsi 0.390 0.301 0.247 0.153 0.009 0.272 0676*
S. roosevelti 0.020 0.044 0.264 0.390 0.081 -0146 0.274
P. wirshingi -0326 -0359 0.137 0.266 0.356 0.036 0.079
Species 0.539 0621* 0.408 0.322 -0.023 0.070 0701*
nchness

 

 

t Ramjohn et al. (unpubl. data)

* J. Genet (unpubl. data)

Minimum Fragment Area

The minimum fragment area required to support 50% and 75% of the

lizard community of Guanica Forest was 0.26 ha (site 30) and 2.0 ha (site 28),

respectively (Figure 2.3). The threshold of fragment area required to support

both 50% and 75% of the reference lizard community was 97 ha. When

fragments sampled only in 1998 (sites 7, 10, and 13) were omitted, the threshold

to support 50% and 75% of the reference lizard community was reduced to 0.26

ha and 2.0 ha, respectively (Figure 2.4).

57

 

_A
O
l
.

'5 ' .
g 08 ~ . 00 0 75/°
8 e
C)
O
C _
g 0-6 t- ---------------------- n 50%
‘fig 0
g O.4- .0 O
‘8
§- ,
00 1i l l l l l

 

 

 

-3 -2 -1 0 1 2 3
Fragment Area (log transformed)

Figure 2.3 Relationship between fragment area (log transformed for
analysis and graphical display) and proportion of reference

 

 

 

 

 

lizard community supported by fragments.
1 0 — 0
g e
0 8 a CO 0
g L---—-----.- ------------- 75%
8 O
E, 0.6 _.
“32’ -------- O- ------------- 150%
g 0.4 —
'E
5-
0-0 *1 l l r l i
-3 -2 -1 0 1 2 3
Fragment Area (log transformed)

Figure 2.4 Proportion of reference lizard community supported by fragments.
Sites 7, 10, and 13 (sampled only in 1998) omitted from display.

58

Principal Components Analysis

Principal components analysis (PCA) of fragment and landscape
characteristics was performed twice, once with all fragment and landscape
variables, and subsequently without the variables describing the percentage of
land in the 1000 m buffer comprised of forest, agriculture, and urban land use.
PCA with the three buffer landscape variables indicated that they were
correlated with and very similar to the land use directly adjacent to the
fragments; interpretations were therefore based on the PCA of the reduced data
set containing seven variables (fragment area, fragment perimeter, nearest
fragment distance, number of fragments in 1000 m buffer, and percentage of
land adjacent to fragment perimeter comprised of forest, agriculture, and urban
development). This analysis reduced the dimensionality of the data set to three
principal components which accounted for >88% of the total variation (Table 2.7,
Figure 2.5).

The first principal component separated out the fragments which were
most isolated (in terms of distance to the nearest forest fragment and number of
neighboring closed canopy forest fragments) and surrounded by agriculture
(sites 30, 36, and 40) from all other sites (Figure 2.5). Surrounding land use was
also important for the interpretation of the second principal component;
fragments that were less isolated (sites 13, 27, and 28) were distinguished from
those fragments that are embedded in more urban areas (sites 10 and 16). The
third principal component contrasted those sites which had the greatest

perimeter, were surrounded predominantly by agricultural land yet not very

59

isolated from other forest fragments with sites surrounded by forest and urban

 

 

 

land use.

Table 2.7 Eigenvalues and eigenvectors from PCA of fragment and
landscape characteristics. Additional principal components
accounted for _<_ 5% of the variation in the data set.

PC 1 PC 2 PC 3
Eigenvalue 3.45 1 .46 1 .27
% Variation , 49.24 20.79 18.19
Cumulative % Variation 49.24 70.03 88.22
Eigenvectors:
Fragment Area 0.47 0.28 0.10
Fragment Perimeter 0.34 0.17 0.64
Nearest Fragment Distance -0.46 0.19 0.15
# Fragments in 1. km Buffer 0.32 -0.39 0.47
% Adjacent Forest 0.32 -0.53 —0.33
% Adjacent Agriculture 042 —004 0.43
% Adjacent Urban Land 0.27 0.65 -021

 

60

 

 

 

 

 

 

 

 

4
%adj. urban
3 1
_J
2 016
10 .4 area
A nearestfragrnent '
0\° 1 T distance perimeter
92 ~30
:5 0 %adj.agiculture 7 '5 02
N
0 -1 — (.13
0' 2
_2 _ #fragmentsjn
. buffer
27 %adj.forest
-3 -—
'4 T l l l T l
-4 -3 -2 -1 0 1 2 3 4
4
perimeter
3 " .2
% adj. agriculture # fragments in
2 — buffer
nearest
$3 1 _ 400 fragment
a) istan
{3; O ’28 '
c0 . 7
_ o .5
8 -1 16 1313 /oadj.urban
-2 _ °/o adj.forest
-3 —
-4 r 1 r T r T l

 

 

 

 

Figure 2.5 PCA results of fragment ordination, see text for fragment
identification numbers. (A) PC 1 vs. PC 2, (B) PC 1 vs. PC 3
Eigenvectors scaled by a factor of 5 for plotting purposes.

61

Correspondence Analysis

Correspondence analysis (CA) of the ten dry forest lizard species in 12
fragments and four sites within Guanica Forest produced an ordination in which
~93% of the total inertia was explained in the first four dimensions (Appendix 8,
Figure 8.5, Tables 8.4-6). From this analysis, A. cocki was identified as a rare
species which was highly influential on the analysis; this spades was then
omitted and CA was performed again to identify patterns among the nine more
- common species. More than 94% of the total inertia was accounted for in the
first four dimensions (Table 2.8, Figure 2.6). Sites 30, 16, 36, and 16 contributed
the most to the row ordinations of the first four dimensions, respectively, while 8.
nicholsi, P. wirshingi, S. ' roosevelti, and A. exsul had the greatest respective
contributions to the column ordinations (Table 2.9). All sites are represented
relatively well in these four dimensions, but A. stratulus, A. pulchellus, A.

poncensis, and A. wetmorei are very poorly depicted (Table 2.10).

Table 2.8 Correspondence analysis of lizard abundance in dry forest
fragments (A. cooki omitted from analysis). Total inertia was 0.50;
each additional dimension explained <1 % of the total variability.

 

 

Dimensio Principal Inertia % Total Inertia Cum. % Total

n Inertia
1 0.20 39.24 39.24
2 0.14 27.79 67.03
3 0.11 21.25 88.28
4 0.03 6.44 94.72
5 0.01 2.65 97.37
6 <0.01 1.64 99.01

 

 

62

Table 2.9 Partial contributions of row (sites) and column (species) points
to inertia of first four CA dimensions. Anolis cooki omitted from

 

 

analysis.
Dimension 1 Dimension 2 Dimension 3 Dimension 4
Site 36 0.11 0.14 0.26 0.08
Site 30 0.29 <0.01 <0.01 0.05
Site 27 0.02 0.13 <0.01 0.13
Site 28 0.06 <0.01 0.14 <0.01
Site 13 0.10 0.04 0.02 0.01
Site 16 0.04 0.36 <0.01 0.25
Site 10 0.04 0.09 0.15 <0.01
Site 7 0.11 <0.01 0.02 0.10
Site 5 0.03 <0.01 <0.01 0.09
Site 4 0.01 0.01 0.01 0.14
Site 2 0.05 0.09 0.09 <0.01
GF coastal <0.01 <0.01 0.01 0.06
GF central 0.02 0.09 0.24 0.01
GF north 0.08 0.04 0.01 0.04
GF ravines 0.04 <0.01 <0.01 0.04
A. cristatellus 0.32 0.05 0.06 0.19
A. stratulus 0.01 0.01 <0.01 <0.01
A. pulchellus <0.01 0.01 <0.01 0.15
A. poncensis <0.01 <0.01 <0.01 0.01
A. exsul 0.01 0.15 0.25 0.42
A. wetmorei 0.01 <0.01 0.02 0.02
S. nicholsi 0.42 <0.01 0.01 <0.01
3. roosevelti 0.08 0.13 0.65 0.08
P. wirshingi 0.15 0.64 <0.01 0.13

 

 

63

Table 2.10 Squared cosines of row (sites) and column (species) points

for the first four CA dimensions. Cumulative total represent the
total quality of the display of each site or species in four-
dimensional space. Anolis cocki omitted from analysis.

 

 

 

Dim. 1 Dim. 2 Dim. 3 Dim. 4 Cumulative
Site 36 0.30 0.27 0.38 0.03 0.98
Site 30 0.97 <0.01 <0.01 0.02 0.99
Site 27 0.11 0.55 <0.01 0.13 0.79
Site 28 0.41 0.01 0.51 <0.01 0.93
Site 13 0.67 0.18 0.07 0.01 0.93
Site 16 0.11 0.75 <0.01 0.12 0.98
Site 10 0.22 0.35 0.44 <0.01 1.0
Site 7 0.76 0.01 0.06 0.11 0.94
Site 5 0.51 0.01 0.01 0.26 0.79
Site 4 0.23 0.14 0.09 0.48 0.94
Site. 2 0.29 0.35 0.27 <0.01 0.91
GF coastal 0.04 0.01 0.71 0.21 0.97
GF central 0.10 0.29 0.58 0.01 0.98
GF north 0.61 0.21 0.05 0.05 0.92
GF ravines 0.68 0.05 <0.01 0.11 0.84
A. cristatellus 0.76 0.09 0.07 0.07 0.99
A. stratulus 0.12 0.18 <0.01 <0.01 0.30
A. pulchellus 0.01 0.07 <0.01 0.34 0.42
A. poncensis 0.03 0.1 1 0.08 0.05 0.27
A. exsul 0.03 0.33 0.41 0.21 0.98
A. wetmorei 0.18 0.04 0.23 0.06 0.51
S. nicholsi 0.97 0.01 0.02 <0.01 1.0
S. roosevelti 0.15 0.17 0.66 0.03 1.0
P. wirshingi 0.24 0.72 <0.01 0.03 0.99

 

64

CA 2 (27.79%)

1.5
1.0
0.5

‘ 0.0

—0.5

— 1.0

-‘I.5

 

I I I l '
._phwi
16 .
an 0
‘57
13
anst , N‘
59'“ .amwe
30' 5: 000 a
ii an28
0o -
e @2
amex ' 56
' ‘va "
(Danna
I I I I I

 

 

 

- ‘I .5 - 1 .0 -0.5 0.0 0.5 1 .0 ‘l .5

Figure 2.6

CA 1 (39.24%)

Correspondence analysis of study sites and lizard species.

Size of plotting symbol = CA 3 (21.25%), shading pattern of
plotting symbol = CA4 (6.44%). Sites are numbered as in text,
species codes as follows: ancr = A. cristatellus, anst = A. stratulus,
anpu = A. pulchellus, anpo = A. poncensis, amex = A. exsul, ‘
amwe = A. wetmorei, spni = S. nicholsi, spro = S. roosevelti,

mm = P. wirshingi. Anolis cocki omittedfrom analysis. 94.72% of

total inertia is explained in the first four dimensions.

65

Discussion
Lizard Community Response to Fragmentation

Dry forest lizards of southwestern Puerto Rico appeared to be relatively
resilient to the effects of habitat fragmentation, including reductions in available
habitat and isolation from other forested areas. Species richness and
composition of lizard communities did not significantly differ among fragments of
varying sizes and the continuous forest reference site, although there was a
definite trend of increasing species richness with increasing fragment area
(Figure 2.2). Anolis cristatellus and S. nicholsi were the dominant species in
Guanica Forest, comprising more than 75% of the lizard community. These two
species also comprised the majority of the lizard community in all fragment study
sites (Appendix 8, Tables 8.1-3).

A number of potential mechanisms may explain why lizard communities
were not found to be as susceptible to the consequences of habitat
fragmentation as other vertebrate groups such as birds and large carnivorous
mammals (Soulé et al. 1979, Schaller and Crawshaw 1980, Newmark 1987,
Laurance and Bierregard 1997). Lizards may be better equipped to cope with
altered microhabitats than other vertebrates (e.g., mammals and birds) as a
result of comparatively lower energy and space requirements (Turner et al. 1969,
Nagy 1987). Forest fragments, especially if very small, are likely to have
reduced resources compared to large expanses of forested habitat. Lizards are
able to maintain higher populations in fragmented habitats than animals with

greater energy and space requirements (Smith et al. 1996).

66

Fahrig and Merriam (1994) proposed that spatial structure of the
landscape must be considered in management programs if the goal is species
conservation. The potential importance of biogeographic and habitat variables
for species persistence in small fragments has been considered in many
previous studies (i.e., Lord and Norton 1990, Cutler 1991, Smith et al. 1996).
The landscape of southwestern Puerto Rico has undergone many changes over
the past century, and organisms have been forced to tolerate and adapt to the
resulting landscape structure. In general, the distance to the nearest fragment
and the number of fragments within 1 km of each study site were independent of
urban land use directly adjacent to the fragment. Percentage of forest directly
adjacent to the fragments and the number of nearby closed canopy fragments
were unrelated to fragment area and perimeter; however, as perimeter
increased, more forest fragments were located within the buffer (a function of
greater area contained in the buffer region of a larger fragment).

Very few of the correlations between lizard abundance or species
richness and fragment and landscape variables were significant (Tables 2.4,
2.5). Lizard species richness was significantly correlated with fragment area
(Table 2.4), as were termite and tree species richness (J. Genet, I. Ramjohn,
pers. comm.) Fragment area most likely explains these observed associations
between species richness of lizard, termite, and plant communities. Lizards in
the Australian wheatbelt region are also associated with available habitat area
(Kitchener et al. 1980, Smith et al. 1996), althoughthis relationship is perhaps

confounded with the larger body size and area requirements of some species in

67

these studies. Anolis pulchellus was somewhat suprisingly significantly
positively correlated with the number of stems _>_ 2.5 cm DBH and basal area; this
grass-bush ecomorph was expected to be negatively correlated with stem
density of larger trees, if associated with any woody vegetation variables at all.
However, stems 3 2.5 cm DBH are predominantly comprised of stems skewed
towards the smallest measurements; dry forests of southwestern Puerto Rico are
typified by a high density of small stems (Murphy and Lugo 1986a, Ramjohn et
al. unpubl. data). On the other hand, A. stratulus abundance was expected to be
positively correlated with stem density (especially 3 5 cm DBH) and basal area,
since this trunk-crown ecomorph prefers perching higher on larger mature trees
(Rivero 1978, Schwartz and Henderson 1991, Reagan 1986, 1992).

Distance between fragments (isolation) has implications for dispersal
abilities of some lizards, and previous studies have found that species richness
has a significant negative relationship with isolation, indicating that dispersal
and colonization of other fragments is a limited phenomenon. in some
ecosystems (Sarre 1995, Sarre et al. 1995, Smith et al. 1996). Only A.
cristatellus was significantly (negatively) correlated with fragment isolation,
which implies that its dispersal and colonizing ability is restricted, while other dry
forest lizard species may not be so hindered. This was a somewhat surprising
result, given that this species is the most widespread and common lizard
throughout Puerto Rico, and is observed in many types of habitats, whether
forested, agricultural, or residential (Rivero 1978, Schwartz and Henderson

1991, pers. obs). The lack of significant relationships between lizard and

68

biogeographic variables indicates that these species are predominantly
generalists, or if they have specialist habitat requirements, these are met in the
fragments. Also, the dry forest fragments in this study may not be truly isolated
to the degree that would negatively affect the lizard communities. The 12
fragments in this study were separated by no more than 400 m from. other
patches of closed canopy forest, and many were isolated from other forested
areas only by roads or short expanses of non-forested land. However, the low
occurrence of the less common lizard species may have precluded the detection
of significant associations.

If fragment and landscape characteristics are not very influential, what are
the important factors in structuring dry forest lizard communities? Based on the
correspondence analysis of fragments and species, it appears (that
environmental and habitat characteristics within the fragment were more
influential than those in the surrounding landscape. Ordination along the first
four axes primarily described differences among sites attributable to vegetation
characteristics; sites with denser and more complex vegetation contrasted with
sites that have open and scrubby vegetation. Lizard species ordinated along a
similar gradient; 8. nicholsi and A. stratulus, both of which prefer a shaded and
protected environment, contrasted with other species that are better adapted to
open, xeric, and rocky conditions. Phyllodactylus wirshingi and S. roosevelti
emerged as distinctly different from all other species along the second and third
axes, respectively. These two species, while relatively common to most sites,

were not frequently encountered during quantitative sampling. They were

69

recorded at a number of sites during qualitative searches, but were not well
represented in the density estimates used in the correspondence analysis. A
similar argument could be posed for A. pulchellus and A. poncensis; both
species were frequently encountered during site walks, but density estimates for
all sites at which they were present were also lacking for these species. Anolis
pulchellus and A. poncensis were also quite poorly represented in this analysis
(Table 2.10).

Most lizards were present even in small (< 1 ha) fragments, with the
exception of A. cocki, A. poncensis, and S. roosevelti, which were only
represented in the medium and large fragments. Anolis cocki has a very limited
and discontinuous distribution which is restricted to very few areas, all within 1
km of the coast (Marcellini et al. 1985, Jenssen 1990). Sphaerodactylus
roosevelti, like A. cocki, is restricted to coastal habitats and relatively uncommon
in forested habitat farther inland (Schwartz and Henderson 1991). Since the
four small fragments were located further inland than some of the other sites,
none were within the range of these two species, although I would expect these
species to be able to maintain populations within small fragments that are within
their range. Anolis poncensis is found in exposed grassy or shrubby areas
(Schwartz and Henderson 1991); this species was encountered most frequently
along fragment margins or in grassy clearings within larger fragments. It is likely
that this species is indeed present along the edges of small fragments, but was
not detected during this study. Another species very similar in appearance and

ecology to A. poncensis, A. pulchellus, was represented in small fragments.

70

Additional investigations into the natural history and ecology of individual
species (i.e., Kitchener et al. 1988, Fobes et al. 1992) would be valuable
contributions to understanding species-level responses to habitat fragmentation
and other disturbance phenomena.

It is possible that species not encountered during this study in various
fragments are indeed present, but perhaps at low densities or in atypical (and
therefore not sampled) microhabitats (e.g., old quarry in site 4). Lizards were
more likely to be detected and added to species lists at sites which were both
quantitatively and qualitatively resampled in 1998 (Sites 30, 28, 16, 5, 4, and 2).
Unequal sampling effort at sites sampled only in 1997 or 1998 may mean that
the species richness in these fragments was underestimated, although the
fragments sampled in both years did not show any significant temporal
differences (paired ttest, p>0.05). Sites 13, 10 and 7 were added to the study in
1998, and all three of these sites had lower species richness than expected,
given their size and habitat quality (5 = 4 for all three sites). The trend of
increasing species richness with increasing fragment area (Figure 2.3) was not
as strong when these three sites were added to the study in 1998.

A common tool for characterizing the species composition of isolated
fragments is the nested subset model, which assumes that the species present
in small fragments are subsets of those present in larger areas (Patterson and
Atmar 1986, Cutler 1991). In the Bahamas, the presence of a given lizard
species and total species richness can be predicted from habitat area (Schoener

and Schoener 1983a, b). In southwestern Puerto Rico, A. cristatellus and A.

71

exsul were present in virtually all fragments (except site 40), and as fragment
area increased, 8. nicholsi was added to the community. Occurrences of other
species were determined more by habitat heterogeneity and availability of
suitable microhabitats than fragment area. For example, grass-bush anoles
were only found at edges or in fragments with interior grassy clearings.

Although little is known about dispersal behavior in these groups of
lizards (T.A. Jenssen, pers. comm), existing data suggest that Anolis spp. show
little vagility (Gorman and Harwood 1977, Andrews and Rand 1983). Non-
forested land surrounding fragments may be dispersal barriers for some species,
but this would be ameliorated if lizards were able to live and maintain
populations in these alternative habitats. Anolis cristatellus is very common in
open areas and along roadsides where fence posts provide suitable perches
(pers. obs, Schwartz and Henderson 1991). This species has also become
adapted to take advantage of novel feeding opportunities at night where insects
aggregate around lights in residential areas (Garber 1978). Ameiva exsul is also
well adapted to anthropogenic habitat disturbance, and is frequently seen in
urban areas (pers. obs, Heatwole and Torres 1967). The wide variety of food
types consumed by A. exsul includes items such as garbage scraps, indicating
that this generalized omnivore prospers in the midst of human disturbance
(Grant 1931, Heatwole and Torres 1967, Lewis 1986). Other species, such as
the grass-bush ecomorphs A. pulchellus and A. poncensis, are also able to

maintain populations in non-forested habitats. Gorman and Hanivood (1977)

72

reported extremely high population densities of A. pulchellus along roadsides
and in open grassy areas (up to 20,000 individuals/ha).

Lizard population dynamics are important to consider when assessing the
long term effects of habitat fragmentation. The small-bodied dry forest lizards
are relatively short-lived, and complete population turnover could potentially
occur over short time scales. Lizard populations fluctuate dramatically over
time, and annual changes can be as great as 50% (Turner 1977), however,
although year to year variability may be great, lizard populations remain
relatively constant over ecological time (Schoener 1985, 1994). Survival is
influenced by many factors; the presence of a greater number of bird species in
dry forests as compared to wet forests in Puerto Rico (Kepler and Kepler 1970)
may lead to lower survivorship in lizards (Schoener and Schoener 1978, 1982).
Some dry forest bird species, most notably the Puerto Rican lizard cuckoo
(Saurothera vieillotr) feed primarily on small lizards (Raffaele 1989). Although it
is not known how this particular species responds to habitat fragmentation, most
native bird populations suffer severe reductions following disturbance and
fragmentation (Robbins et al. 1989, Faaborg and Arendt 1990).

After a fragment becomes isolated, there is a lag before a new equilibrium
. of extinction and colonization rates is reached (Diamond 1972, 1973, Corlett and
Turner 1997). The relatively rapid and unpredictable pace of lizard population
dynamics suggests that there has probably been sufficient time for faunal
relaxation in the 12 dry forest fragments studied, given that the majority of the

fragments have been isolated for at least 20 years and others for more than 60

73

life

an

an-
991
ton
pot
pro
true
con

Spe

SUQI
In in
ham

COml

years (Lugo et al. 1996). Another component of lizard population dynamics is
that density fluctuations are correlated with timing and amount of precipitation
(Andrews and Wright 1994). In the harsh and often drought-stricken dry forest
life zone, water limitation may lead to reduced population sizes in fragments that
are even more susceptible to local extinction than under natural conditions.

The primary mechanisms for the maintenance of lizard communities in dry
forest fragments in southwestern Puerto Rico may be related to: (1) lower energy
and space requirements than other vertebrates (e.g., mammals or birds), (2)
generalist habitat requirements, (3) ability to move through intermittent non-
forested habitats, and (4) ability to successfully establish and maintain
populations in non-forested habitats. Any of these mechanisms will increase the
probability of lizard survival and persistence in a fragmented landscape. The
true explanation for lizard persistence in forest fragments is most likely a
combination of the above factors; and may vary with the species or group of

species considered.

Conservation of Dry Forest Fragments
Overall, lizard species richness was positively related to fragment area,
suggesting that larger fragments should receive the highest conservation priority
in management programs. Greater heterogeneity inherent in larger expanses of
habitat increases the probability of supporting a rich and diverse ecological

community. However, fragments as small as 2 he were able to support a diverse

74

lizard community, indicating that these habitats also deserve attention in
management and conservation activities at the landscape level.

Characteristics of the surrounding landscape were generally not related to
fragment lizard community composition and abundance, although it would be
very useful to examine the influence of surrounding land use in buffers of various
sizes (e.g., 100 m, 500 m). If lizards are not hindered by nonforested land uses,
they will likely be resilient to the negative consequences of habitat
fragmentation. If the trend of increasing forest cover throughout Puerto Rico
continues, lizards in existing forest fragments should be able to serve as
colonizers once additional patches of secondary forest become available in the
landscape. This study represents a preliminary investigation into the complex
effects of habitat fragmentation on lizards in a landscape that is dominated by
anthropogenic influences. Although more than 300 patches of forested habitat
are scattered throughout the dry forest life zone (Ramjohn et al. unpubl. data),
only 12 fragments were considered in this study. Because of the number of sites
and lack of variation in the surrounding landscape (three of the fragments were
entirely enclosed by agricultural land), additional studies investigating a greater
number of fragments are warranted to further investigate the implications of the
surrounding landscape for the persistence of ecological communities in
subtropical dry forest fragments. However, conclusions based on this study
point very strongly to the conservation potential of fragments in this region, at

least if lizards are the focal taxa.

75

SU

pie
req
in s
frag
rem.
hioh
slip I
Rico
of an

ameli

This study has shown that even forest fragments as small as 2 ha contain
suitable habitat to support a diverse dry forest lizard community. Although it is
true that some vertebrates (e.g., birds) require substantial areas for habitat
preservation, other vertebrates (e.g., lizards) have much smaller area
requirements. Given the already fragmented nature of the dry forest landscape
in southwestern Puerto Rico, preservation of the remaining small habitat
fragments should reCeive conservation priority. Fragments containing the few
remaining populations of the endangered A. cooki should receive top and
highest priority; if these patches of habitat are lost, this species will inevitably
slip towards extinction. If forest fragments in the dry forest life zone of Puerto
Rico were integrated into local management and conservation efforts, the threats
of anthropogenic disturbance on natural communities may begin to be

ameliorated.

76

inl

PL

C0

Isle

d0<

Chapter 3

Structural Habitats of Dry Forest Trunk-Ground Ecomorphs:
Anolis cristatellus and Anolis cocki

Introduction

Anolis cristatellus and A. cooki are morphologically and ecologically
similar anoles found in the dry forest region of southwestern Puerto Rico.
Where these two species coexist, few characters can be used to distinguish
them in the field (Marcellini and Jenssen 1983); they are osteologically identical
(Pregill 1981) and have only recently been recognized as separate species
based on karyotypic differences (Gorman et al. 1968, 1980, 1983). Both are
medium sized anoles that are classified as trunk-ground ecomorphs (Williams
1983), although they differ in their distribution, species status, and biotic
interactions.

While A. cristatellus is a widespread and abundant anole throughout
Puerto Rico (Heatwole 1976, Rivero 1978, Schwartz and Henderson 1991), A.
cocki is restricted to a limited number of discontinuous coastal scrub habitat
patches within 1 km of the coast in the extreme southwestern portion of the
island (Marcellini et al. 1985, Jenssen 1990). Although previous studies have
documented small allopatric populations of A. cocki (Jenssen et al. 1984,
Marcellini et al. 1985), virtually the entire distribution of this species overlaps

with that of A. cristatellus. Anolis cocki populations may be diminishing towards

77

eventual extinction (Williams 1972), and the current and future status of this
species requires not only immediate, but also continual conservation attention.

Rand (1964) introduced the concept of structural habitat for Anolis lizards;
the spatial niche occupied by an anole can best be described by the height and
diameter of the perch site. Other perch characteristics such as texture, color,
and thermal microhabitat (sun vs. shade) may also be used to describe
structural habitat (Heatwole 1968, Scott et al. 1976), but the most powerful
variables to discriminate between species or age classes within a species are
perch height and diameter (Schoener 1968). Structural habitat represents one
of the primary resource axes which are partitioned by sympatric anoles; thermal
habitat and prey size comprise the remainder of the ecological niche (Schoener
and Schoener 1971 a, b, Williams 1972, Schoener 1977).

The structural habitat of an anole has implications for a multitude of
activities including foraging, courtship, aggression, and therrnoregulatory
behaviors. Perch substrate has also been found to be important in anole
movement through their three-dimensional habitat; perch diameter influences
sprint speed, clinging ability and agility (Losos and Sinervo 1989). Intra- and
interspecific interactions often elicit escape behaviors, which can be predicted
from the relationships among perch diameter and the locomotion characteristics
mentioned above (Losos and Irschick 1996). However, Moemnond (1979)
suggested that distribution of perches, rather than perch characteristics

themselves, were more influential for anole movement through a habitat.

78

In sympatry, similar species must differ from one another along at least
one of the primary resource'axes to coexist in stable equilibrium (Schoener and
Schoener 1971a, b). In response to competitive interactions, Anolis lizards have
been found to alter selection of perches and perch position (e.g., Jenssen 1973,
Schoener 1975, Lister 1976, Campbell 1999). Shifts in structural habitat among
sympatric species have been documented in response to seasonal variation in
food availability and reproductive condition (Lister 1981) and the co-occurrence
of direct competitors (Salzburg 1984, Losos et al. 1993, Losos and Spiller 1999).
Jenssen et al. (1984) found that where A. cristatellus and A. cocki are allopatric,
their structural habitats are not significantly different. Ortiz and Jenssen (1982)
found that intense aggressive encounters between these two species had
implications for ecological displacement; indeed, when they are sympatric, A.
cocki is driven via intense competitive interference with A. cristatellus to less
desirable and less complex microhabitats including small bushes and standing
dead vegetation (Jenssen et al. 1984).

While many studies have approached competitive interactions from the
interspecific competitive interference perspective, these studies have also
shown that intraspecific interactions often. overshadowed interspecific
competition (e.g., Tokarz and Beck 1987, Brown and Echtemacht 1991).
Indeed, anoles of both Jamaica and Puerto Rico partition their structural habitat
in response to intraspecific competition among age and size classes (Schoener
and Schoener 1971a, b). Stamps (1977) found that within a species, males are

more likely to engage in intense competitive interactions for mates, while

79

females compete for food. The implications for intraspecific interactions may
affect the outcome of competition between A. cristatellus and A. cocki, both inter-
' and intraspecifically.

Given the tenuous existence of A. cocki in southwestern Puerto Rico, it is
important to re-evaluate its structural habitat in the presence of A. cristatellus, a
direct and apparently superior competitor. The objectives of this chapter are to
compare the structural habitats of these two species in areas where they co-
occur in the highly fragmented dry forest landscape in order to assess the
current status of A. cocki and conservation implications of both inter- and
intraspecific interactions. Anolis cocki presently has a limited and patchy
distribution, and habitat fragmentation may further increase the chance of local

population extinctions and eventual species extirpation.

Materials and Methods

Anolis cristatellus and A. cocki were surveyed in forest fragments of
varying sizes and a reference site (Guanica Commonwealth Forest) in the dry
forest zone (sensu Holdridge 1967) of southwestern Puerto Rico during the
summers of 1997 and 1998. A complete description of the reference site and
fragments can be found in chapters 1 and 2, respectively. At each site, 100 m
transects of fixed area (see Chapter 1, Figure 1.2) were randomly located
throughout the study area; the number of transects was scaled according to the
size class of each site (see Chapter 2 for a complete summary of the sampling

design).

80

Transect sampling involved surveying lizards at 5 m intervals along the
length of the transect. At each interval, the following data were recorded for all
A. cristatellus and A. cooki individuals within a 2.5 m radius of the transect
sampling interval: (1) species identification, (2) age/size class, (3) perch height,
(4) perch diameter, and (5) perch species (identification of the tree species used
as a perch). Anolis cristatellus and A. cooki were distinguished in the field by
estimating the ratio of tail length (TL) to snout-vent length (SVL) (Marcellini and
Jenssen 1983). Each individual was assigned to an age/size class upon visual
estimation of TL and SVL (Schwartz and Henserson 1991, Table 3.1).
Individuals were periodically captured (every 1015 individuals) and measured to
ensure that visual estimation of TL and SVL was accurate and specimens were
correctly identified and assigned to the proper age/size class.

Perch height was visually estimated to the nearest 0.1 m using reference
points on the observer’s body; during the 1998 field season, perch diameter was
visually estimated to the nearest 0.5 cm, and subsequently measured with
calipers. A single observer conducted all surveys to avoid any bias introduced

by observer error.

Table 3.1 Age/size classes for A. cristatellus and A. cocki based on SVL
(Schwartz and Henderson 1991)

 

 

 

A. cristatellus A. cooki
adult male >70 mm > 60 mm
adult female/sub-adult male 55 — 69 mm 50 - 59 mm
juvenile < 55 mm < 50 mm

 

81

Structural habitats and density estimates were analyzed to determine
whether inter- or intraspecific differences existed between species and/or
age/size classes. Anolis cristatellus structural habitat was evaluated under both
allopatric and sympatric conditions. All statistical analyses were performed using
Systat (version 5.0). Non-parametric statistics were utilized on untransformed
data throughout, as perch height, perch diameter, and density data are not
normally distributed, nor did transformation improve the distribution.
Significance level for all tests was set to a = 0.05. Perch height data from 1997
and 1998 were pooled; there were no significant differences between years
(Wilcoxon paired sample test; p>0.05). Density estimates for both species from
1997 and 1998 were also pooled, as there were no significant differences in

abundance between years (Wilcoxon paired sample test, p>0.05).

Results

The total number of lizards sampled in this study was dramatically biased
towards A. cristatellus, the more common and widespread species (A.
cristatellus, n = 2413; A. cocki, n = 76). Of these individuals, the distribution
among age/size classes exhibited a striking pattern (Figure 3.1). A large
proportion of the A. cristatellus individuals sampled were juveniles (0.44,
n=1132), with the remainder comprised of adult females/sub-adult males and
adult males(0.41, n=936; 0.15, n=345; respectively). The age distribution of A.
cooki seems to be the inverse pattern; adult males comprise the largest

proportion of sampled individuals (0.52, n=40) while adult females/sub-adult

82

males and juveniles represent proportionally smaller age classes for this species
(0.26, n=20 for females/sub-adult males and 0.22, n=16 for juveniles,

respectively).

 

0.70 1-

 

0.00 T
0.50 T
040 ~-
030 Jr
0.20

0.10 I

0.00

 

Proportion of Sample

 

 

 

 

 

A. cristatellus A. cocki

Iadult male Cladult female/sub-adult male I juvenile

Figure 3.1 Age distribution of A. cristatellus (n=2413) and A. cocki (n=76)
in 1997-1998. Sample sizes for age/size classes are given in
the text. ‘

Anolis cristatellus was found at all but one study site, while A. cocki was
found in only 2 fragments and two isolated areas in the reference site (Table
3.2). Over the entire distributions of these two species, the differences in
structural habitat were not statistically significant between A. cristatellus and A.

cocki. (Table 3.3, Mann-Whitney U Test: perch height, p=0.09; perch diameter,

83

p=0239). Adult anoles showed a tendency to perch higher than juveniles in
both species; adult male A. cristatellus perched significantly higher and on larger
stems than adult male A. cocki (Figure 3.2; Kruskal-Wallis ANOVA, p<0.05). .No
other differences among age/size classes or species were statistically significant
when the entire habitat range of each species was considered.

Although allopatric populations of A. cristatellus did not perch significantly
higher, they selected significantly larger stems than when they co-occur with A.
cocki (Mann-Whitney U Test, p<0.01; Figure 3.3). When the two species were
sympatric, they occupy virtually the identical structural habitat; an overall
comparison of perch height and diameter showed that there were no significant
differences between A. cristatellus and A. cocki perch dimensions (Mann-
Whitney U Test, p>0.05; Table 3.4). Intraspecifically, there were no significant
differences in perch height or diameter for A. cocki (Kolmogorov—Smimov two
sample test, p>0.05, Figure 3.4). However, adult males and females
(considered together with sub-adult males) A. cristatellus perch significantly
higher than juveniles (Kolmogorov-Smimov two sample test, p<0.001; Figure
3.4); low sample size precluded intraspecific comparisons of A. cristatellus perch
diameters in sympatry with A. cocki. Regression analysis indicated that neither
A. cristatellus nor A. cocki adult males showed a significant relati0nship between
SVL and perch” height (p>0.05, Figure 3.5). Low sample size precluded testing
for significant relationships between adult male lizard SVL and perch diameter.

At sites where A. cristatellus and A. cocki are sympatric, A. cristatellus is

significantly more abundant than A. cooki (Mann-Whitney U Test, p<0.001),

although abundance of both species did not differ significantly among study sites
(Kruskal-Wallis ANOVA, p>0.05). lntraspecifically, A. cooki density did not differ
significantly among age/size classes (Kruskal-Wallis ANOVA, p=0514), but A.
cristatellus differed significantly among age/size classes (Kruskal-Wallis
ANOVA, p=0.035). Interspecifically, density of adult males did not differ
between species (Mann-Whitney U Test, p=0.613), but A. cristatellus adult
female/sub-adult males and juveniles were significantly more abundant than
their A. cooki counterparts (Mann-Whitney U Test, P < 0.01; Figure 3.6).

Table 3.2 Distribution of A. cristatellus and A. cooki among study sites

1 = presence of A. cristatellus, 2 = presence of A. cooki
Refer to text and Figure 2.1 in Chapter 2 for a complete description
of fragment study sites.

 

 

 

 

 

 

 

Small Fragments Medium Fragments Lar e Fra ments Reference Site
Site 40 - Site 28 1 Site 7 1 Coastal 1, 2
Site 36 . 1 Site 13 1 Site 5 1 Central 1
Site 30 1 Site 16 1, 2 Site 4 1 ‘ North 1
Site 27 1 Site 10 , 1 Site 2 1, 2 Ravines 1, 2*

 

 

 

 

 

* only a single A. cooki was found in a ravine that was in close proximity to the
coast

J

Table 3.3 Overall perch height (SE) and perch diameter (SE) for A.
cristatellus and A. cooki; all sites and age/size classes combined.
Differences in perch height or diameter were not significantly

different between species.

 

 

Stnlctural Habitat A. cristatellus A. cooki

Perch height (m) 0.818 (0.013) 0.774 (0.06)
n=2413 n=76

Perch diameter (cm) 6.234 (0.194) 5.339 (0.577)
n=728 n=31

 

 

85

 

1.4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

P = 0.011 A
.2 g .
DA. cristatellus
E 1 — IA. cooki
:7 ..
g, 0.8 ‘-
"6
I.
.: 0.6
2
0
n. 0.4
0.2
o
10
I P = 0.034
8 l
E
3 -
b
3 6
0
E
.2
c: 4 _,
.C
2
0
n.
2
0 t
adult nale ' adult ferralel pvenie
sub-adult rrale
Figure 3.2 (A) Mean perch height (A. cristatellus, n=2413; A. cooki, n=76) and

(8) mean perch diameter (A. cristatellus, n=728; A. cooki, n=31)
for A. cristatellus and A. cooki age/size classes at all study sites.
Error bars represent standard error of the mean. Mean perch size
selection (height and diameter) was significantly different between
adult males of the two species (Mann-Whitney U Test, p<0.05),
but females and juveniles did not differ significantly in structural
habitat characteristics (Mann-Whitney U Test, p>0.05).

86

Perch Height (m)

Figure 3.3

Table 3.4

 

 

 

 

Perch Diameter (cm)
0 —l N 01 h 01 0’) \l

 

 

perch Ilo at _ perch
height na p m diameter
Isyn'patric

(A) Mean perch height and (B) mean perch diameter of Anolis
cristatellus in allopatry (n=1872) and sympatry (n=289) with A.
cooki. Error bars represent standard error of the mean. Perch
diameter was significantly less in sympatry with A.cooki (Mann
Whitney U Test, p=0.001), but perch height was not significantly
different (Mann-Whitney U Test, p=0.632)).

Perch height (SE) and perch diameter (SE) for A. cristatellus and
A. cooki at sites where they are in sympatry; age/size classes
combined. Differences between species in perch height or
diameter were not significant (Mann-Whitney U Test, p>0.05).

 

 

Structural Habitat A. cristatellus A. cooki

Perch height (m) 0.76 (0.03) 0.77 (0.06)
n=289 n=76

Perch diameter (cm) 4.78 (0.39) 5.34 (0.58)
n=728 n=31

 

 

87

 

_s
b

—l
N
1
I

DA. cristatellus
A. cooki

Perch Height (m)
.0 P
O) on -

P
A

.0
N

 

 

 

 

 

 

 

 

O

 

adult male _ adult juvenile
female/sub-
adult male

Figure 3.4 Mean (i SE) perch height for A. cristatellus (n=289) and A. cooki
(n=76) where they occur in sympatry. Within age/size classes,
differences between species were not significant (Mann-Whitney U
Test, p>0.05).

Structural habitat of A. cristatellus differed significantly among study site
size classes (Kruskal-Wallis ANOVA, P < 0.001; Figure 3.7). Anolis cooki was
found in only 2 of 12 fragments (site 2 and 16) and 2 of 4 reference site habitats
sampled (coastal and ravine), precluding any statistical evaluation of potential
effects of habitat fragmentation on this species. Neither species exhibited a
perch preference for a particular plant speCies. Both A. cristatellus and A. cooki
perched most often on standing dead vegetation (16% and 17%, of perch sites,
respectively). The remainder of perch sites were comprised of many different
species, each comprising less than 10% of the total perch sites. A complete

summary of perch sites for A. cristatellus and A. cooki can be found in Appendix

C.

88

Perch Helght (m)

Perch Height (m)

Figure 3.5

4.5

3.5 ..

2.5 .

1.5 ..

0.5 ..

4.5

3.5

 

 

 

 

 

 

 

 

 

 

4 .t O A
O
R2=0.0207
O
l e
l— . .
8
O
. . C . g...
e I 0 -.——0——
O
8 I ' ' '
o e U . o
e 8
O ' 'r
68 70 72 74 76 78 80
Male SVL (mm)
+ e
.l .
R2=o.0681 '
O
T .
O O C O
1 0 W
I
. . e 8 ' o .
i + : . C . b 3
60 62 64 66 68 70 72 74 76 78

Male SVL (mm),

Relationship between SVL and perch height for adult male (A) A.
cristatellus (n=47) and (B) A. cooki (n=39) at sympatric sites.
Neither species showed a significant association between the
characters (Spearrnan rank correlation, P ~> 0.05)

89

 

Density (it I ha)

Figure 3.6

Figure 3.7

 

 

DA. cristatellus

. P<0.001
'A- 000‘“ . . |P=0.003

 

 

 

 

 

   

 

 

adult male adult female! juvenile
sub-adult male

Density (mean 1 SE) of A cristatellus (n=289) and A. cooki (n=76)
where they occur in sympatry. Anolis cristatellus adult
females/sub-adult males and juveniles were significantly more
abundant than the same A. cooki age/size classes (Mann-Whitney
U Test, p<0.05).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 ‘- El height - a—
.. ri- . E
E Eldiameter o
v o 8 "" 7.5 t
:5, f i + +2
I .e
.5 0.4 a D
5.; ~~ 2.5 g
m 0.2 o
n.

o _, If i 0

Small Medium Large Reference
Site
Study Site Size Classes

Structural Habitat (mean 1- SE) perch height and perch diameter of
A. cristatellus (n=2413) among study site size classes. Perch
height and diameter are both significantly different among site size
classes (Kruskal-Wallis ANOVA, p<0.05).

90

Discussion

Anolis cn'statellus is one of the most conspicuous and widespread anoles
on the island of Puerto Rico (Rivero 1978, Schwartz and Henderson 1991). It
occured in all habitat types, with the possible exception of mesic and montane
forest interior (Rand 1964, Schoener and Schoener 1971b). Anolis cooki was
significantly less common; its extremely restricted and patchy distribution have
made it a species of special concern for management and conservation officials
in southwestern Puerto Rico (Ortiz 1990, M. Canals Mora, pers. comm). The
implications of both inter- and intraspecific competitive interference interactions
between these two species for the survival of A. cooki reinforces the need to
monitor and document any changes in the ecological status of A. cooki in the
interest of developing management and conservation strategies for the
protection of this species.

The age distribution of A. cooki populations sampled in this study
suggested that the future of this species is uncertain. The vast majority of the
individuals sampled were adults and sub-adult males (78%), with very few
juveniles to replace the adults as they mature, senesce, and die. The age
distribution of A. cristatellus was more stable and indicative of adequate
replacement and persistence of populations through time; a sizable proportion of
individuals sampled were juveniles (44%). Another Puerto Rican anole, A.
stratulus, had a healthy population comprised of 57% adults and 43% juveniles
(sex ratio 1 male: 1 female) in the Luquillo rain forest (Reagan 1996). It

appeared that the reproductive rates of these two species were quite different; if

91

A. cooki is not reproducing at a rate to replace the current populations, the
species is likely to be on the trajectory of natural extinction. However, the
possibility also exists that the sex ratio and age distribution of the population
were closer to stability. Anolis males often “hide” in the vicinity of dominant
males' territories, ready to overtake the territory if predation or some other
circumstance renders the territory unoccupied and undefended (T. Campbell,
pers. comm).

When the data for all age/size classes and sites were podled (including
both allopatric and sympatric populations of A. cristatellus, but only sympatric
populations of A. COOKI), there were no significant differences in structural
habitat characteristics between A. cristatellus and A. cooki. This supported the
contention that these speciesoccupy similar niches and are virtual ecological
equivalents. in both species, there was a tendency for adults to perch higher
than juveniles. Within adults, males of both species perched higher than
females; Lister (1981) also 'found that male anoles in the Luquillo rain forest of
northeastern Puerto Rico perched higher than females. Adult male A.
cristatellus also occupied larger stems than females and juveniles, but no
significant trends were apparent for A. cooki. While some studies have found a
general trend of larger lizards occupying perches of greater diameter (Schoener
1968, Schoener and Schoener 1971a), Lister (1981) found that mean perch
diameters were generally similar both between and within species of Puerto
Rican anoles. Differences in perch diameter between and within species may

reflect the degree of potential competitive interference; species and age/size

92

 

classes which are ecologically less similar should show little difference in perch
size preference, thereby limiting competition to lizards of presumably differing
size and energetic requirements. The tendency for larger anoles to be found on
higher and larger perches than smaller lizards results in reduced intraspecific
spatial overlap (Schoener and Schoener 1971a).

Neither A. cristatellus nor A. cooki showed a significant relationship
between adult male SVL and perch height. Such a relationship would indicate
intraspecific competition among males, with larger males predominating in larger
and more complex habitats. Jenssen et al. (1984) found that with the exception
of sympatric A. cooki, adult males in both allopatric and sympatric habitats
showed a significant positive relationship between SVL and the relative
complexity and height of their microhabitat. The lack of a significant relationship
between adult male size and perch height in either species (absence of
intraspecific competition?) as well as the lack of a significant difference in perch
height between the two species indicated that extensive structural habitat
overlap and predominantly interspecific competitive interactions may be the
driving forces shaping A. cristatellus and A. cooki community dynamics.

Although adult males of the two species differed significantly in perch
height and diameter when all populations of A. cn'statellus (allopatric and
sympatric) were considered, there were no significant differences in‘structural
habitat of A. cristatellus in sympatry with A. cooki. In anoline lizards, the greatest
potential for competitionand aggression interactions is between males (Ortiz

and Jenssen 1982), and differences in structural habitat serve to segregate

93

congeneric species spatially and reduce interspecific competition. Competitive
interactions are less frequent in juveniles, and their selection of lower and
smaller structural habitats probably reflects an active choice rather than the
results of an intraspecific aggressive encounter (Kiester et al. 1975).

Allopatric A. cristatellus populations perched at similar heights and on
larger stems than those sympatric with A. cooki. These two species have
virtually identical habitat requirements, and shifts in structural habitat function to
minimize competitive interference where they co-occur. The ability to shift and
modify structural habitats in response to competitive interactions has been
documented for many Anolis species in many areas throughout the West Indies
and the southeastern United States (e.g., Jenssen 1973, Schoener1975, Lister
1976, Campbell 1999, Echtemacht 1999). Jenssen et al. (1984) found that there
were no interspecific differences in structural habitat in allopatric areas of
southwestern Puerto Rico, but in sympatry, A. cooki perched on significantly
lower and thinner vegetation while A. cristatellus maintained the same structural
habitat characteristics in sympatric and allopatric populations. The results of
this study indicated that these two species may have altered their ecological
roles to some extent in these habitats; although A. cristatellus did not
significantly alter perch height in sympatry with A. cooki, they selected
significantly smaller stems for perches. Based on these data, structural habitats
of these two species do not differ significantly under sympatric conditions.
Although A. cooki presently appears to be maintaining relatively high quality

perCh sites in sympatry with A. cristatellus, population density and age

94

distribution of this species do not indicate a state of stable equilibrium with its
congeneric competitor.

At every study site where both species occur, A. cristatellus is present in
significantly higher densities than A. cooki. Low population densities of A. cooki
do not bode well for the persistence of these populations over ecological, much
less evolutionary time. Adult males were present in similar densities at
sympatric sites, but A. cristatellus adult females/sub-adult males and juveniles
were significantly more abundant than A. cooki females and juveniles. Lower
densities of females and juveniles will probably lead to declining populations and
eventual local extinctions, depending on fecundity and recruitment within the
population.

Neither A. cristatellus nor A. cooki showed a perch preference for a
particular vegetation type or species. Standing dead vegetation comprised the
largest percentage of perch sites for both species ( 5 17% ), although the vast
majority of perches were distributed relatively evenly among dozens of dry forest
trees, shrubs, and vines (Appendix C). Standing dead stems comprise 13% of
total stem density in Guanica Forest (Murphy and Lugo 1986a); dead stem
density may be even higher in forest fragments, constituting a sizable proportion
of available perching habitat for anoles (pars. obs). Lister (1976) indicated that
some arboreal anoles may be generalists with respect to structural habitat, and
this seems to be the case for perch substrate selection by A. cristatellus and A.
cooki. Other investigators have also stated that plant species is not an important

criterion for anoles in selecting perch sites, and have characterized the substrate

95

 

component of structural habitat by vegetation physiognomy (Jenssen et al.
1984). These studies have found that open areas comprise a large proportion of
the habitat of A. cristatellus and A. cooki, both in allopatry and sympatry, and
that A. cooki perches were strongly skewed to standing dead vegetation and
small shrubs in the presence of A. cristatellus (Jenssen et al. 1984, Marcellini et
al. 1985). The results of the current study do not strongly support these
previous conclusions, as shrubby and standing dead vegetation constitute
approximately equal proportions of perch sites for both species. However,
explicit vegetation data necessary to evaluate differences in microhabitat
complexity between species and among age/size classes were not available for

this study, limiting the conclusions strictly to structural perch characteristics.

Current and Projected Future Status of A._c_o_o_ki

Past studies have indicated that A. cristatellus is the dominant species
and superior competitor in habitats where it coexists with A. cooki (Jenssen et al.
1984). This study suggests that although A. cooki appears to have altered perch
site selection in areas of sympatry with A. cristatellus over the past two decades,
other natural history characteristics of this species imply that it is far from
coexisting in stable equilibrium. Anolis cooki did not appear to be driven to lower
and thinner perches as found in Jenssen et al. (1984), however, low population
density and potentially low recruitment (few juveniles in populations) indicated
that A. cooki is likely in decline. Earlier assertions that A. cooki may be on the

pathway to natural extinction driven by competitive interactions with A.

96

cristatellus (Williams 1972, Marcellini et al. 1985, Ortiz 1990) cannot be rejected
based on this study; the long-terrn persistence of this species remains
questionable, and continual monitoring of remaining A. cooki populations will be
critical to the long-term survival of the species.

Differences in structural habitat and abundance between these two
species also have implications for food consumption, courtship,
therrnoregulation, and relative risk of predation. Anoles are opportunistic
predators and are able to consume a wide variety of prey (Reagan 1996);
perhaps a perch position lower to the ground confers an advantage to A.
cristatellus when feeding on leaf litter arthropods. The coastal scrub habitats
where A. cooki is found are exposed to harsh wind and sea spray conditions
which account for the open and wind-pruned vegetation in these areas (Ewel
and Whitmore 1973). Perch positions slightly lower on trees and shrubs in
these habitats may provide more protection and control over active
therrnoregulatory behaviors. Anolis cooki may be at a disadvantage on higher
perches both because the microhabitat may be harsher and also more exposed
to predation risks from birds such as red-tailed hawks (Buteo jamaicensis) and
the Puerto Rican lizard cuckoo (Saurothera vieillotl).

Competitive interference and habitat partitioning between A. cristatellus
and A. cooki appears to be a dynamic interaction through time, and the

evolutionary outcome depends on the ability of A. cooki to persist and reproduce
under the inhospitable conditions inherent in the coastal scrub habitats where it

is found, coupled with the ability to compete effectively with A. cristatellus. The

97

few remaining patches within the range of A. cooki need to be preserved and
protected; it is unlikely that this species could successfully cope with the intense
pressures of anthropogenic habitat modification combined with the natural

competitive pressures already present.

98

SUMMARY

In this study, dry forest fragments in southwestern Puerto Rico were able
to support relatively rich and diverse lizard communities. When fragments were
viewed collectively, they supported the same species composition as Guanica
Commonwealth Forest, the largest continuous tract of protected dry forest in
Puerto Rico. Abundances of the 10 lizard species in the fragments and in the
reference site are provided in Appendix B (Tables B.2-4) and Table 1.7,
respectively. The 12 fragments in this study contained a wide variety of habitat
types, and ranged in size from 0.01 to >800 hectares. Some of these fragments
may be sheltered from excessive anthropogenic disturbances due to their harsh
topography and relative inaccessibility. Other fragments were located in areas
where the potential for human disturbance was much greater. For instance, one
of the study sites was previously part of a golf course and is located along a
heavily traveled road which receives a substantial amount of trash from passing
vehicles (pers. obs).

These fragments varied greatly in the type and quality of habitats. Two
species, A. cristatellus and A. exsul, were present in all fragments except the
very smallest one (0.01 ha). Sphaerodactylus nicholsi was present in all
fragments larger than 0.25 ha. The occurrence of the remaining dry forest
species depended on the habitat heterogeneity of the fragment. The diversity of
lizards was hypothesized to increase as the total area of the fragment increased;

the number of lizard species was indeed significantly (positively) correlated with

99

fragment area. There were no patterns of abundance among species and
fragment or landscape characteristics.

The fragment area required to support the lizard community found in the
reference site was >800 ha. Other large fragments would also be able to
support the reference community if suitable habitat were provided and the site
were located within the range of all 10 dry forest lizard species. Nine species
were recorded at site 4 (137 ha); A. cooki was the only species not encountered
because the fragment was outside that species’ distribution. The minimum f,
fragment area required to support 50% and 75% of the lizard community of
Guanica Forest was 0.26 ha (site 30) and 2 ha (site-28), respectively.
Fragments greater than 97 ha consistently supported lizard communities with
75% of the species found in the reference site, although when sites sampled
only one year were excluded, fragments greater than 2 ha consistently
supported >75% of the reference lizard community.

Analysis of landscape level phenomena using GIS combined with
statistical techniques indicated that patterns of land use surrounding the forest
fragments had little, if any, influence on species richness and the distribution of
lizard communities. Although no generalizations could be made from the results
of this study, landscape characteristics may be important for certain individual
species requirements. Further studies of the landscape ecology of the
fragmented dry forest life zone with a greater number of study sites and

increased variability in landscape characteristics (as well as investigating the

100

influence of surrounding land use in buffers of various sizes) are necessary
before any firm conclusions can be reached.

Habitat characteristics were important for the distribution and abundance
of some dry forest lizards. Sphaerodactylus nicholsi was found at higher
densities when the litter layer was relatively deep, providing a mesic
environment. This may represent a tradeoff between preference for a xeric
habitat (Schwartz and Henderson 1991) and a need for the abundant and stable
arthropod food supply found in the moist substrate. Structural habitats of A.
cristatellus and A. cooki were also investigated in this study. While the rarity of
A. cooki precluded comparisons among fragments or size classes, comparisons
of the two species in allopatry and sympatry yielded provocative results.

Overall, there were no significant differences in structural habitat between A.
cristatellus and A. cooki. in general, adults perched higher than juveniles, and
adult males perched higher than adult females. Allopatric A. cristatellus perched
higher than when they are found in sympatry with A. cooki. Jenssen et al. (1984)
found the reverse to be true; A. cooki was forced to lower and poorer quality
perch sites when sympatric with A. cristatellus than when allopatric.

Although the number of fragments was limited and variation in
surrounding landscape characteristics was low in this study, these results allow
for some conservation and management recommendations. Large fragments
which contain a variety of habitat types and support rich and diverse lizard
communities should receive highest conservation priority, but fragments as small

as‘2 he also supported a large proportion of the reference lizard community and

101

warrant management and conservation efforts. Fragments z 2 ha typically
support a rich lizard fauna and should also be integrated into management and
conservation strategies. Characteristics of the vegetation and substrate are
potentially very important to the organisms living in the fragments; quantification
Of these factors is also necessary in preparing recommendations and guidelines
for conservation biologists and natural resource managers. Additionally,
protection of fragments containing populations of A. cooki is also mandatory for
the preservation of the few remaining populations of this endangered species.
The results of this study corroborated previous investigations of lizard
communities of the highly fragmented wheatbelt region of western Australia
(Kitchener et al. 1980, Kitchener and How 1982, Smith et al. 1996). Among the
vertebrates in this ecosystem, lizards have endured the best despite habitat
fragmentation. This can be attributed to a number of life history characteristics
that make lizards in essence preadapted to coping with the consequences of
habitat fragmentation such as low metabolic and area needs and generalist
habitat requirements. These life history traits are presumably more important for
lizards in fragmented habitats than habitat and biogeographic characteristics;
there were few significant relationships between lizard taxa and either habitat or
biogeographic variables (Smith et al. 1996). Dry forest lizards in southwestern
Puerto Rico also appear to be resilient to the consequences of habitat
fragmentation compared to other faunal groups, although additional studies are
needed to assess the ecological status of lizards before broad generalizations

can be made concerning their sensitivity to the effects of fragmentation.

102

 

These investigations concerning lizard community ecology in the
fragmented dry forest life zone of southwestern Puerto Rico will be most robust
when combined with concurrent investigations of termite (J. Genet, unpubl. data)
and plant communities (Ramjohn et al. unpibl. data). Without a doubt,
responses of representative vertebrate (lizards), invertebrate (termites), and
plant communities will provide a thorough indication of the ecological status of
these groups of organisms in a fragmented landscape and lead to conservation
and management recommendations such that natural communities will be able

to persist in the face of inevitable anthropogenic disturbance.

103

APPENDICES

104

APPENDIX A
(A) Coastal (B) Central

 

 

Cumulative Number of Lizard Species

 

 

 

 

 

 

 

 

 

 

 

0 l I I I I l I I I I
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500
Cumulative Area Sampled (m2) Cumulative Area Sampled (m2)
(3) North (D) Ravines
7
U)
.9
8 6 r 7
Q.
a)
U —
a 5 ‘
.L!
.J _-
‘s 4 —
E
E 3 — T
3
z
3 2 ‘
E
E 1 ‘
:3
o
0 I I T I I T I I I 1

 

 

 

 

0 500 1000 1500 2000 2500

Cumulative Area Sampled (m2)

0 500 1000 1500 2000 2500

Cumulative Area Sampled (m2)

Figure A.1 Species-area curves for transect sampling at Guanica Forest sites.
(A) Coastal site: A. pulchellus added via qualtitative sampling, S.
nicholsi, S. roosevelti, and P. wirshingi added via plot sampling;
(B) Central site: A. pulchellus and A. poncensis added via
qualitative sampling, 8. nicholsi and S. roosevelti added via plot
sampling; (C) North site: A. pulchellus added via qualitative
sampling, 8. nicholsi, S. roosevelti, and P. wirshingi added via plot
sampling; (D) Ravines: S. nicholsi and P. wirshingi added via plot

sampling.

105

APPENDIX B

      

I
0.5 0 0.5 1 Kilometers

 

 

 

 

Land-use

- Urban or built-up land
- Agricultural land
:l Water

- Closed canopy forest
T; Open canopy forest
- Forested wetland
- Non-forested wetland
; Natural barren land
E: Artificial barren land

Figure B.1 Land use map of site 5 and vicinity, illustrating the delineation of
the 1000 m buffer around study sites.

106

(A) Site 36 (B) Site 30

 

 

 

 

 

 

 

 

 

 

8 7

'8

e — -

E 5 -— -

.5

_I

“5 4 - “‘

h”:

s s — -

2

m 2 - -

,g

g 1 A d ~

g

0 0 r I I w I I I I I r

0 50 100 150 200 250 300 0 100 200 300 400 500 60
Cumulative Area Sampled (m2) Cumulative Area Sampled (m2)
(C) Site 27

 

Cumulative Number of Lizard Species
A
I

 

 

1 I I I I
0 200 400 600 800 1000

Cumulative Area Sampled (m2)

 

Figure 82 Species-area curves for transect sampling at small fragment sites;
no species were present at site 40. (A) Site 36: no additional
species added by qualitative or plot sampling; (B) Site 30: A.
pulchellus added via qualitative sampling, 8. nicholsi and P.
wirshingi added via plot sampling; (C) Site 27: S. nicholsi and P.
wirshingi added via plot sampling.

107

(A) Site 28 (B) Site 13

 

Cumulative Number of Lizard Species
c:

 

 

 

 

 

 

 

 

 

1 _.
0 I I I I r I I I
0 200 400 600 800 0 200 400 600 800
Cumulative Area Sampled (m2) Cumulative Area Sampled (m2)
(C) Site 16 (D) Site 10
7 7
6 ..
5 J
4 _.
3 .—

 

Cumulative Number of Lizard Species

0

Figure 3.3

 

 

 

 

 

 

I I I I 0 l I n I I
200 400 600 800 O 200 400 600 800
Cumulative Area Sampled (m2) Cumulative Area Sampled (m2)

Species-area curves for transect sampling at medium fragment
Sites. (A) Site 28: A. stratulus and A. pulchellus added via
qualitative sampling, S. nicholsi, S. roosevelti, and P. wirshingi
added via plot sampling; (B) Site 13: S. nicholsi and P. wirshingi
added via plot sampling; (C) Site 16: S. nicholsi, S. roosevelti, and
P. wirshingi added via plot sampling, (D) Site 10: A. poncensis
added via qualitative sampling, 8. nicholsi added via plot sampling.

108

(A) Site 7 (B) Site 5

 

 

 

 

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

(D

.9

§ 6 4

m

g 5 —

.5

_I

“5 4 —

‘23 3 --

3

z

22’ 2 _I

‘66

- J

g 1

3

o 0 I I I I I I I F I I I I

O 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 3000
Cumulative Area Sampled (m2) Cumulative Area Sampled (m2)
(C) Site 4 (D) Site 2

m 7

.9

0

2t 1

a)

“O

a -I

.8

_l

‘5 —I

.8

E

3 .

z

2 2 . -

E

3 _.

E 1 _

:3

O

0 I I I I I n I I I I I I I I
o 500 10001500 2000 2500 3000 3500 o 500 10001500 2000 2500 3000 3500

Cumulative Area Sampled (m2) Cumulative Area Sampled (m2)

Figure 84 Species-area curves for transect sampling at large fragment
sites. (A) Site 7: S. nicholsi and P. wirshingi added via plot
sampling; (B) Site 5: P. wirshingi added via plot sampling; (C) Site
4: S. nicholsi, S. mosevelti, and P. wirshingi added via plot
sampling, (D) Site 2: S. nicholsi, S. roosevelti and P. wirshingi
added via plot sampling.

109

.:o=mE=mo bacon 863.02: 65383 o>=m=Emza 05:6 8: Sn 2% m cm 66282 92> can: 36QO ..

 

 

 

 

 

 

NHC "C ONE

:3 EV . o o Seems; 5:882er
ONE 0": ONE OHC

.o o o o 508on wzebomboemmcqm
0": mflC OHC OH:

:9; 89 Sam: 88 o o .222: 3588528
V": ONE 0": ONE

Amos meme o o o 295552 2.55
v": 0?": "C ONE

6: m as at an 5 RN o 395 SEE
ONE OHC ONE 0":

o o o o macmocoq 965‘
mu: on: on:

em . o o 3583: See:
v": mu: 0": ONE

6 3 we as 8 o o 3598 See:
OHC ONE OH: O":

o o o o .538 255...
DR": MN": VP": 0":

an : mom at EN :8: 8... c 32355.8 5.6%

R 25 en 35 8 35 3. 36 1865mm.

 

 

.26 :3 m_m:u_>_u:_ :0 59:5: .92 n : .Améflmsezus E
$2306 39 :86 .5352 93:52 .8: F vv 35:52: __mEm :_ 35.2. 598 Eu :0 33650 mm 2an

110

.:o=mE_.mo 25:50 35255 55388 5255535 053 «o: 5: 25 m 5 5655: 553 55 55QO ..

 

 

 

 

 

 

OH: NHC ruc FHC
o E: 2 2. some new sou mom 52822 55082222
on: on: _.u:
o . o 892 mom 22589 52859825
mu: vflC r": VHC -
:82 2 s. C E 2 :2 some mom SE 28 2222: 52.889828
0": FHC OHC VHC
o E v o :2 me 290225 2.55:2
QFHC OPHC NHC V":
38 New 82 em £5 em :2 on sea 2.55.
"C OHC OHC
.. o o 0 225252 255».
on: on: on:
o o o . 32223 25%
ONE ”NC "C
o 65 mm o .. 52225 255.
ONE NF": OHC ONE
0 as m: o o .208 255.
won: mvuc wvuc mm":
82 mmm 52 m a 38 m8 8..” 5 com 82.2225 228:
2 25 2 35 2 85 ea 35 528 m

 

 

va :mmE 65555: 83:52 .8: or .. 3 35:55.: 825:: :_ 352. 522 >5 5 525.50 mm 553.

.25 5: 2552265 5 5:83: .92 u : 23553355 a: 5:556

111

 

 

.co.5E_55 35:2... 9.5.55 .mczaEmm 5255535 @555 5: 5: 2.5 m 5 .5555: 552, 55 5.5% ..

 

 

FHC
. . .. .3: 52 52:25 525852.52
NHC ONE ONE
.5. :2 . o o .5582 522558555
OHC 0N": NNHC ON":
.8 2 2m :5. mm: .82 N3 .55. 38 55:2: 5225559525
mu: ONE NH: O":
.8... 3 .52.. 8.5 .52. 85 o 5552... 2,55.
hr": VNNC 0?": V0":
.2. 5 a. mm 6. mm .5. 5m 525 2.52::
NHC P": r": OHC
.55. 8.5 .89 8.5 .85. mm o 25:85: 25:<
N": N": V": O": .
E m a. 2 a. 5 o 35:25: 25::
r": Own: Fr": OH: ..
a. m a s on .o 3 mm 0 255:5 25:<
mrflc OHC OH: OH: .
.m c 5 o o o .558 255.
5 W: Emu: mm W: mm W: .
.5 we: .2. :5 .5. m5. .5. 5.5 2255.25 255.
.~|3_.m. .. 35 5.125... N. elgm 3.8 m

 

 

 

5.2555 Em. 55:: 55552 28:52 .5: or A. 25:52.. 595. c. 352. 522 55 5 5.550

.25 5: 5555.255 .6 555: .22 u : 525535.255 .5

mm 2an .

112

CA 2 (27.56%)

CA 3 (18.71%)

Figure 8.5

 

Correspondence analysis of all study sites and all 10 lizard species.
A.) Dimension 1 vs. Dimension 2, B.) Dimension 1 vs. Dimension 3
Species codes: ANOCRI=A.cn'state/Ius, ANOCOO=A. cooki,
ANOSTR=A. stratulus, ANOPUL=A. pulchellus, ANOPON=A.
exsul, AM EWET=A. wetmorei,
SPHNIC=S. nicholsi, SPHROO=S. roosevelti, PHYWIR=P.
wirshingi. Refer to Chapters 1 and 2 for site identification.

poncensis, AMEEXS=A.

 

o
ANOCOO

16
o

ANOSIR .PHYWIR

ANOPUL
SPHNIC 0.27 AM

0 Q m
30 7.5. . 60 “.13
4 .28
.ANgCRI

0 .5? R
AMEExg 10 “$3,011

 

 

 

 

CANOCOO

A.NOPON
Ce
0

Cg ‘AMEWET

'28
spnzucANOSTR

30' 7,53 zANSCRrG
ANOPUL o , PHYWIR
o 13

. 1o
AMEEXS
’36

o
SPHROO

 

 

-2

l I T

-1 0 1
CA 1 (36.03%)

113

 

 

Table 8.4 Correspondence analysis of lizard abundance in dry forest

fragments. Total inertia was 0.57; each additional dimension
explained <5% of the total variability.

 

 

Cumulative %
Dimension Principal % Variability Variability
Inertia
1 0.21 36.03 36.03
2 0.16 27.56 63.59
3 0.11 18.71 82.3
4 0.06 10.88 93.18

 

 

 

 

Table 85 Partial contributions of row (sites) and column (species) points
to inertia of first four dimensions. All 10 species of lizards included
in analysis.

Dimension 1 Dimension 2 Dimension 3 Dimension 4
Site 36 0.05 0.22 0.17 0.02
Site 30 0.26 0.02 <0.01 _ 0.01
Site 27 0.02 0.03 0.01 <0.01
Site 28 0.04 0.01 0.15 0.11
Site 13 0.09 <0.01 0.04 0.02
Site 16 0.20 0.45 <0.01 <0.01
Site 10 0.01 0.13 0.10 <0.01
Site 7 0.11 <0.01 0.01 0.15
Site 5 0.03 <0.01 <0.01 0.19
Site 4 0.02 0.01 0.01 0.19
Site 2 0.05 0.04 0.14 0.22
GF coastal <0.01 0.01 0.07 0.06
GF central 0.01 0.07 0.29 0.01
GF north 0.07 <0.01 0.03 <0.01
GF ravines 0.04 0.01 <0.01 0.02
A. cristatellus 0.19 0.21 0.05 0.05
A. cooki 0.11 0.18 0.01 0.43
A. stratulus <0.01 0.02 <0.01 0.01
A. pulchellus <0.01 <0.01 <0.01 0.05
A. poncensis <0.01 <0.01 0.01 <0.01
A. exsul <0.01 0.14 0.15 0.36
A. wetmorei 0.01 <0.01 0.01 0.03
S. nicholsi 0.37 0.063 0.01 <0.01
8. roosevelti 0.05 0.08 0.74 0.01
P. wirshingi 0.27 0.31 0.02 0.05

 

 

114

 

 

 

Table 8.6 Squared cosines of row (sites) and column (species) points

for the first four CA dimensions. Cumulative total represent the

total quality of the display of each site or species in four-

dimensional space. All 10 species of lizards included in analysis.

Dim. 1 Dim. 2 Dim. 3 Dim. 4 Cumulative

Site 36 0.14 0.49 0.26 0.10 0.98
Site 30 0.89 0.06 <0.01 0.04 0.99
Site 27 0.15 0.16 0.04 0.45 0.81
Site 28 0.30 0.07 0.45 0.08 0.90
Site 13 0.62 0.01 0.15 0.15 0.92
Site 16 0.32 0.56 <0.01 0.11 0.99
Site 10 0.06 0.58 0.29 0.06 0.99
Site 7 0.75 0.01 0.04 0.07 0.87
Site 5 0.56 <0.01 0.01 0.03 0.59
Site 4 0.38 0.11 0.06 0.05 0.60
Site 2 0.25 0.15 0.36 0.11 0.86
GF coastal 0.08 0.01 0.70 0.07 0.87
GF central 0.04 0.24 0.69 <0.01 0.97
GF north 0.56 <0.01 0.13 0.23 0.92
GF ravines 0.62 0.10 0.01 0.08 0.82
A. cristatellus 0.47 0.40 0.06 0.04 0.97
A. cooki 0.27 0.35 0.01 0.33 0.97
A. stratulus 0.03 0.34 <0.01 0.06 0.43
A. pulchellus 0.01 0.02 0.01 0.24 0.27
A. poncensis 0.02 0.04 0.12 0.02 0.20
A. exsul <0.01 0.34 0.24 0.34 0.93
A. wetmorei 0.17 <0.01 0.15 0.22 0.55
S. nicholsi 0.87 0.11 0.02 <0.01 0.99
S. roosevelti 0.10 0.12 0.76 0.01 0.99
P. wirshirlgi 0.49 0.43 0.02 0.03 0.96

 

115

APPENDIX C

Table C.1 Location of Anolis cristatellus and A. cooki perches, including

frequency (absolute and rounded to percentage) for each perch
type selection. Type of perch also included ( t = tree, 3 = shrub, c =
cactus, v = vine).

 

Frequency
Perch Location flp_e A. cristatellus A. cooki
standing dead tree t 379 (16%) 13 (17%)
Leucaena Ieucocephala t 203 (8%)
on substrate 174 (7%) 4 (5%)
Gymnanthes Iucida t 136 (6%) 1 (1 %)
Pisonia albida t 114 (5%) 6 (8%)
Exostema can'baeum t 94 (4%) 4 (5%)
Thouinia stn'ata t 92 (4%) 1 (1 %)
Bounen'a succulenta t 83 (3%)
Bucida buceras t 78 (3%)
Amyn‘s elemifera t 76 (3%)
Coccoloba diversifolia t 73 (3%) 2 (3%)
Eugenia rhombea t 58 (2%) 1 (1 %)
Pictetia aculeata t 54 (2%) 2(3%)
Eugenia foetida ‘ t 50 (2%) 1 (1%)
Sa via sessiliflora t 48 (2 %)
Pithecellobium unguis-cati t 42 (2%)
Pilosocereus royenii c 39 (2%) 2 (3%)
Erythmxylum rotundifolium t 37 (2%) 6 (8%)
Krugiodendmn ferreum t 37 (2 %)
vine v 36 (1 %)
Bursera simaruba t 34 (1 %) 4 (5%)
Cappan's flexuosa t 34 (1 %) 2 (3%)
unknown t 32 (1 %)
Tabebuia heterophylla t 25 (1 %) 2 (3%)
Guettarda elliptica t 24 (1 %) 1 (1 %)
Cappan's hastata t 22 (1 %)
Coccoloba kmgii-microstach ya t 22 (1 %) 1 (1 %)
Colubn'na elliptica t 18 (1 %)
Calyptranthes pal/ens t 16 (1 %)
Comocladia dodonaea t 16 (1 %) 2 (3%)
Clusia rosea t 14 (1 %)
Prosopis pallida t 14 (1 %)
Erythmxylum areolatum t 13 (1 %) 3 (4%)
Guaiacum sanctum t 13 (1 %) 2 (3%)
Randia aculeata t 13 (1 %) 3 (4%)

116

 

Table 0.1 (cont’d)
Leptocereus quadricostatus
Plumen'a alba
Guaiacum officianale
Hype/ate trifoliata
Jacquinia berten’i
Antimea acutata
Crossopetalum rhacoma
Swietenia mahaogani
Croton Iucidus
Guettarda krugii
Cappan's indica-cynophyllophora
Eugenia spp.

En'thalis fruticosa

Ficus citn'folia

Reynosia uncinata
Cane/Ia winterana
Eugenia xerophytica
Rochefon‘ia acanthophora
Piscidia carthagenensis
Schaefieria frutescens
Zanthoxylum fla vum
Antirhea Iucida

Celtis tn'nervia
Citharexylum fruticosum
Croton discolor

Eugenia Iingustn'na
Guapira fragrans
Meliococcus bijugatus
Picramnia pentandra
Bumelia obo vata
Colubn'na arborescens
Haematoxylum campechianum
Jatmpha hemandifolia
Ziziphus reticulata
Acacia famesiana
Bemardia dichotoma
Cmton humilis

Lantana camera
Lantana involucrata
Polygala cowelli
Reynosia guama
Rondeletia pilosa

Total

U)Fir-FmmMFFFOIr-O-MPO-FQ-mrifle-trimfifififl”flb"fiflfifififififl0’fififififififin

12(<1%)
12(<1%)
11 (<1%)
10(<1%)
10(<1%)
9(<1%)
9(<1%)
9(<1%)
9(<1%)
8(<1%)

7(<1%)-

7(<1%)
6(<1%)
6(<1%)
6(<1%)
5(<1%)
5(<1%)
5(<1%)
4(<1%)
4(<1%)
4(<1%)
3(<1%)
3(<1%)
3(<1%)
3(<1%)
3(<1%)
3(<1%)
3(<1%)
3(<1%)
2(<1%)
2(<1%)
2(<1%)
2(<1%)
2(<1%)
1(<1%)
1(<1%)
1(<1%)
1(<1%)
1(<1%)
1(<1%)
1(<1%)
1(<1%)

241 3

2 (3%)

1 (1%)
2 (3%)

1 (1%)

1 (1%)
2 (3%)

4 (5%)

76

 

117

LITERATURE CITED

118

r

LITERATURE CITED

Aide, T. M., J. K Zimmerman, M. Rosario, and H. Maroano. 1996. Forest
recovery in abandoned cattle pastures along an elevational graident in
Northeastern Puerto Rico. Biotropica 28(4a): 537-548

Andrews, R. M. 1971. Structural habitat and time budget of a tropical Anolis
lizard. Ecology 52: 262-270

Andrews, R. M. and A. S. Rand. 1983. Limited dispersal of juvenile Anolis
limifrons. Copeia 1983(2): 429-434

Andrews, R. M. and S. J. Wright. 1994. Long-term population fluctuations of a
tropical lizard: a test of causality. Pp. 267-285 In: L. J. Vitt and E. R.
Pianka (eds). Lizard ecology: historical and experimental perspectives.
Princeton University Press, Princeton, NJ

Bierregard, R. 0., Jr. and V. H. Dale. 1996. Islands in an ever-changing sea: the
ecological and socioeconomic dynamics of Amazonian rain forest
fragments. Pp. 187-204 In: J. Schelhas and R. Greenberg (eds). Forest
patches in tr0pical landscapes. Island Press, Washington, D. C.

Bierregard, R. 0., Jr., W. F. Laurance, J. W. Sites, Jr., A. J. Lynam, R. K
Didham, M. Andersen, C. Gascon, M. D. Tocher, A. P. Smith, V. M. Viana,
T. E. Lovejoy, K. E. Sieving, E. A. Kramer, C. Restrepo, and C. Moritz.
1997. Key Priorities for the study of fragmented tropical ecosystems.
Pages 515-525 In: W. F. Laurance and R. 0. Bierregard, Jr. (eds).
Tropical forest remnants: ecology, management, and conservation of
fragmented communities. University of Chicago Press, Chicago, IL

Birdsey, R. A. and P. L. Weaver. 1982. The forest resources of Puerto Rico.
Resource Bulletin, 80-85. USDA Forest Service Southern Experiment
Station, New Orleans, LA

Birdsey, R. A. and P. L. Weaver. 1987. Forest area trends in Puerto Rico. USDA
Forest Service Research Note, 80-331. New Orleans, LA

Bosch, W. 1978. A procedure for quantifying certain geomorphological features.
Geographical Analysis 10: 241-247

Brown, P. R. and A. C. Echtemacht. 1991. Interspecific behavioral interaction of

adult male Anolis cam/inensis (Sauria: lguanidae): A preliminary field
study. Anolis Newsletter IV: 21-30

119

 

Brown, J.H. and A. Kodric-Brown. 1977. Turnover rates in insular biogeography:
effect of immigration on extinction. Ecology 56: 987-992

Campbell, T. 1999. Consequences of the Cuban brown anole invasion in
Florida: it’s not easy being green. Anolis Newsletter V: 12-21

Castilleja, G. 1991. Seed germination and early establishment in a subtropical
dry forest. Ph.D. dissertation, Yale University, New Haven, CT

Corlett, R. T. and I. M. Turner. 1997. Long-term survival in tropical forest
remnants in Singapore and Hong Kong. Pages 333-345 In: W. F.
Laurance and R. 0. Bierregard, Jr. (eds). Tropical forest remnants:
ecology, management, and conservation of fragmented communities.
University of Chicago Press, Chicago, IL

 

Cutler, A. 1991. Nested faunas and extinction in fragmented habitats.
Conservation Biology 5(4): 496-505

Dale, V. H. and S. M. Pearson. 1997. Quantifying habitat fragmentation due to
land use change in Amazonia. Pages 400-409 In: W. F. Laurance and R.
0. Bierregard, Jr. (eds). Tropical forest remnants: ecology, management,
and conservation of fragmented communities. University of Chicago
Press, Chicago, IL ‘

Davis, J. C. 1986. Statistics and data analysis in geology, 2"d ed. John Wiley,
New York, NY

 

Dial, R and J. Roughgarden. 1994. NOtes on the absolute abundance of canopy
anoles, Anolis cuvieri, A. stratulus and A. evermanni (Lacertilia:
Polychridae) in the Luquillo Forest, Puerto Rico. Caribbean Journal of
Science 30(3-4): 278-279

Diamond, J. 1972. Biogeographic kinetics: estimation of relaxation times for
Avifaunas of southwest Pacific islands. Proceedings of the National
Academy of Science, USA. 69: 3199-3202

Diamond, J. 1973. Distributional ecology of New Guinea birds. Science 179:
759-769

Dietz, J. L. 1986. Economic History of Puerto Rico: Institutional Change and
Capitalist Development. Princeton University Press, Princeton, NJ

Echtemacht, A. C. 1999. Possible causes for the rapid decline in population
‘ density of green anoles,1AnoIis carolinensis (Sauria: Polychrotidae)
following invasion by the brown anole, Anolis sagrei, in the southeastern
United States. Anolis Newsletter V: 22-27

120

Estes, R. 1983. The fossil record and early distribution of lizards. Pp. 365-398
in: A. G. J. Rhodin and K. Miyata (eds) Advances in Herpetology and
Evolutionary Biology. Harvard University Press, Cambridge, MA

Ewel, J. J. and J. L. Whitmore. 1973. The ecological life zones of Puerto Rico
and the U. S. Virgin Islands. USDA, Forest Service. lntemational Institute
of Tropical Forestry, Rio Piedras, PR. Forest Service Research Paper
lTF-1 8

FAQ (Food and Agriculture Organization of the United Nations)/UNESCO. 1993.
Forest resource assessment 1990 - tropical countries. Forestry paper no. i.
112. Food and Agriculture Organization of the United Nations, Rome.

Faaborg, J. and W. J. Arendt. 1990. Long-term studies of Guanica Forest birds.
Acta Cientifica 4(1-3): 69-80

I‘ _

Fahrig, L. and G. Merriam. 1985. Habitat patch connectivity and population
survival. Ecology 66: 1762-1768

Fahrig, L. and G. Merriam. 1994. Conservation of fragmented populations.
Conservation Biology 8(1): 50-59

 

Famsworth, E. J. 1993. Ecology of semi-evergreen plant assemblages in the
Gua’nica Dry Forest, Puerto Rico. Caribbean Journal of Science 29(1-2):
106-123

Fobes, T. M., R. Powell, J. S. Parmerlee, Jr., A. Lathrop, and D. D. Smith. 1992.
Natural history of Anolis cybotes (Sauria: Polychridae) from an altered
habitat in Barahona, Dominican Republic. Caribbean Journal of Science
28(3-4): 200-207 -

Franklin, J. F. 1993. Preserving biodiversity: species, ecosystems, or
landscapes? Ecological Applications 3(2): 202-205

Garber, S. D. 1978. Opportunistic feeding behavior of Anolis cristatellus
(Iguanidae: Reptilia) in Puerto Rico. Transactions of the Kansas Academy
of Science 82(1): 79-80

Gorman, G. C., D. G. Buth, M. Soulé, and S. Y Yang. 1980. The relationships of
Anolis cristatellus species group: electrophoretic analysis. Journal of
Herpetology 14(3). 269-278

Gorman, G. c, o. G. Buth, M. Soulé, and s. Y. Yang. 1983. The relationships of

the Puerto Rican Anolis: electrophoretic and karyotypic studies. Pp. 626
642 In: G. J. Anders, Rhodin, and K. Miyata (eds). Advances in

121

herpetology and evolutionary biology: essays in honor of E. E. Williams.
Museum of Comparative Zoology, Harvard University, Cambridge, MA

Gorman, G. C. and R. HanIvood. 1977. Notes on population density, vagility, and
activity patterns of the Puerto Rican grass lizard, Anolis pulchellus
(Reptilia, Lacertilia, lguanidae). Journal of Herpetology 11(3): 363-368

Gorman, G. C., R. Thomas, and L. Atkins. 1968. lntra- and interspecific
chromosome variation in the lizard Anolis cristatellus and its close
relatives. Breviora 293: 1-13

Goto, M. M. and M. A. Osborne. 1989. Nocturnal microhabitats of two Puerto
Rican grass lizards, Anolis pulchellus and Anolis krugi. Journal of
Hematology 23(1): 79-81

Grant, C. 1931. Notes on the ameivas of Puerto Rico. Copeia 1931: 62

Greenacre, M. J. 1993. Correspondence analysis in practice. Academic Press,
Harcourt, Brace and Company. New York, NY

Greenacre, M. J. and E. S. Vrba. 1984. Graphical display and interpretation of
antelope census data in African wildlife areas, using correspondence
analysis. Ecology 65(3): 984-997

Harris, L. D. 1984. The fragmented forest. Island biogeographic theory and the
preservation of biotic diversity. University of Chicago Press, Chicago, IL

Harris, L. D. and P. Kangas. 1979. Designing future landscapes from principles
of form and function. Pp. 725-729 In: G. H. Pilsner and R. C. Smardon
(eds). Our national landscape: applied techniques for analysis and
management of the visual resource. U. 8. Forest Service General
Technical Report PSW-34. Washington, D. C.

Hass, C. A. 1991. Evolution and biogeography of West Indian Sphaerodactylus
(Sauria: Gekkonidae): a molecular approach. Journal of Zoology 225:
525-561

Heatwole, H. 1968. Relationship of escape behavior and camouflage in anoline
lizards. Copeia 1968: 108-113

Heatwole, H. 1976. Herpetolgeography of Puerto Rico. VII. Geographic variation
in the Anolis cristatellus complex in Puerto Rico and the Virgin Islands.
Occasional Paper for the Museum of Natural History at the University of
Kansas, No. 46

122

 

 

 

Heatwole, H. and F. Torres. 1967. Distribution and geographic variation of the
Ameivas of Puerto Rico and the Virgin Islands. Studies on the Fauna of
Curagao and other Caribbean Islands 24(92): 63-111

Heckel, D. G. and J. Roughgarden. 1979. A technique for estimating the size of
lizard populations. Ecology 60(5): 966-975 -

Hess, N. E. and J. B. Losos. 1991. lnterspecific aggression between Anolis
cristatellus and A. gundlachi: comparison of sympatric and allopatric
populations. Journal of Herpetology 25(2): 256-259

Holdridge, L. R. 1967. Life zone ecology. Tropical Science Center, San Jose,
Costa Rica

Janzen, D. H. 1988a. Tropical dry forests, the most endangered major tropical
ecosystem. Pp. 130-137 In: E.O. Wilson (ed.). Biodiversity. National
Academy Press, Washington, DC.

Janzen, D. H. 1988b. Management of habitat fragments in a tropical dry forest:
growth. Annals of the Missouri Botanical Gardens 75: 105-116

Jenssen, T. A. 1973. Shifts in the structural habitat of Anolis opalinus due to
congeneric competition. Ecology 54: 863-869

Jenssen, T. A. 1990. Anolis cooki Grant: Cook’s anole. Catalogue of American
Amphibians and Reptiles No. 488:1-2

Jenssen, T. A., D. L. Marcellini, C. A. Pague, and L. A. Jenssen. 1984. -
Competitive interference between the Puerto Rican lizards, Anolis cooki
and A. cristatellus. Copeia 1984(4): 853-862

Joliffe, l. T. 1986. Principal Component Analysis. Springer-Verlag, New York, NY

Kepler, C. B. and A. K. Kepler. 1970. Preliminary comparison of bird species
diversity in Luquillo and Guanica Forests. Chapter E-14; 183-191 In: H. T.
Odum (ed.). A tropical rain forest. USAEC, Oak Ridge, TN

Kiester, A. R., G. C. Gorman, and D. Colon Arroyo. 1975. Habitat selection
behavior of three species of Anolis lizards. Ecology 56(1): 220-225

Kitchener, D. J. 1982. Predictors of vertebrate species richness in nature

reserves in the Western Australian wheatbelt. Australian Wildlife
Research 9: 1-7

123

 

 

Kitchener, D. J., A. Chapman, J. Dell, B. G. Muir and M. Palmer. 1980. Lizard
assemblage and reserve size and structure in the Western Australian
wheatbelt - some implications for conservation. Biological Conservation
17: 25-62

Kitchener, D. J. and R. A. How. 1982.-Lizard species in small mainland habitat
isolates and islands off southwestern Western Australia. Australian
Wildlife Research 9: 357-363

Kitchener, D. J., R. A. How, and J. Dell. 1988. Biology of Oedura reticulata and
Gehyra van’egata (Gekkonidae) in an isolated woodland of Western
Australia. Journal of Herpetology 22(4): 401 -412

 

Kramer, E. A. 1997. Measuring landscape changes in remnant tropical dry
forests. Pp. 386-399 In: W. F. Laurance and R. O. Bierregard, Jr. (eds).
Tropical forest remnants: ecology, management, and conservation of
fragmented communities. University of Chicago Press, Chicago, IL

 

Lamb, D., J. Parotta, R. Keenan, and N. Tucker. 1997. Rejoining habitat
remnants: restoring degraded rainforest lands. Pages 366-385 In: W. F.
Laurance and R. O. Bierregard, Jr. (eds). Tropical forest remnants:
ecology, management, and conservation of fragmented communities.
University of Chicago Press, Chicago, IL

Laurance, W. F. and R. O. Bierregard, Jr. (eds). 1997. Tropical forest remnants:
ecology, management, and conservation of fragmented communities.
University of Chicago Press, Chicago, IL

Laurance, W. F., R. O. Bierregard, Jr., C. Gascon, R. K. Didham, A. P. Smith, A.
J. Lynam, V. M. Viana, T. E. Lovejoy, K. E. Sieving, J. W. Sites, Jr., M.
Andersen, M. D. Tocher, E. A. Kramer, C. Restrepo, and C. Moritz. 1997.
Tropical forest fragmentation: synthesis of a diverse and dynamic
discipline. Pages 502-514 In: W. F. Laurance and R. O. Bierregard,‘Jr.
(eds). Tropical forest remnants: ecology, management, and conservation
of fragmented communities. University of Chicago Press, Chicago, IL

Lenart, L. A., R. Powell, J. S. Parmerlee, Jr., A. Lathrop, and D. D. Smith. 1997.
Anoline diversity in three differentially altered habitats in the Sierra de
Baoruco, Republica Dominica, Hispaniola. Biotropica 29(1): 117-123

Lewis, A. R. 1986. Body size and growth-in two populations of the Puerto Rican
ground lizard (Teiidae). Journal of Herpetology 20(2): 190-195

Lewis, A. R. 1989. Diet selection and depression of prey abundance by and
intensively foraging lizard. .Joumal of Hematology 23(2): 164-170

124

 

Lewis, A. R. and J. E. Saliva. 1987. Effects of sex and size on home range,
dominance, and activity budgets in Ameiva exsul (Lacertilia: Teiidae).
Herpetologica 43(3): 374-383

Lerdau, M., J. Whitbeck and N. M. Holbrook. 1991. Tropical deciduous forest:
death of a biome. Trends in Ecology and Evolution 6(7): 201 -202

Lister, B. C. 1976. The nature of niche expansion in West Indian Anolis lizards. I.
Ecological consequences of reduced competition. Evolution 30: 659-676

Lister, B. C. 1981. Seasonal relationships of rainforest anoles. Ecology 62(6):
1 548-1560

Lord, J. M. and D. A. Norton. 1990. Scale and spatial concept of fragmentation.
Conservation Biology 4: 197-202

Losos, J. B., J. C. Marks, and T. W. Schoener. 1993. Habitat use and ecological
interactions of an introduced and a native species of Anolis lizard on
Grand Cayman, with a review of the outcomes of anole introductions.
Oecologia 95: 525-532

Losos, J. B. and D. J. Irschick. 1996. The effect of perch diameter on escape
behavior of Anolis lizards: laboratory predictions and field tests. Animal
Behaviour 51 :593-602

Losos, J. B. and B. Sinervo. 1989. The effect of morphology and perch diameter
on sprint performance of Anolis lizards. Journal of Experimental Biology
1 45: 23-30

Losos, J. B. and D. A. Spiller. 1999. Differential colonization success and
Assymetrical interactions between two lizard species. Ecology 80(1):
252-258

Lugo, A. E., J. A. Gonzalez-Liboy, B. Cintron, and K. Dugger. 1978. Structure,
productivity, and transpiration of a subtropical dry forest in Puerto Rico.
Biotropica 10: 278-291

Lugo, A. E., 0. Ramos, S. Molina, F. N. Scatena, and L. L. Vélez Rodriguez.
1996. A fifty-three year record of land use change in the Guanica Forest
Biosphere Reserve and its vicinity. USDA Forest Service, lntemational
Institute of Tropical Forestry, Rio Piedras, Puerto Rico

MacArthur, R. H. and E. O. Wilson. 1967. The theory of island biogeography.
Princeton University Press, Princeton, NJ

125

 

 

 

Marcellini, D. L. and T. A. Jenssen. 1983. A character for differentiating the
sympatric lizards. Anolis cooki and Anolis cristatellus. Herpetological
Review 14(4): 113-114

Marcellini, D. L., T. A. Jenssen, and C. A. Pague. 1985. Distribution of the lizard
Anolis cooki, with comments on its possible future extinction.
Herpetological Review 16(4): 99-102

Moerrnond, T. C. 1979. Habitat constraints on the behavior, morphology, and
community structure of Anolis lizards. Ecology 60: 152-164

Morales-Carrion, A. 1983. Puerto Rico, a political and social history. Norton
Press, New York, NY _ '

Morrison, D. F. 1990. Multivariate statistical methods, 3" ed. McGraw-Hill
Publishing Company. New York, NY

Murphy, P. G. and A. E. Lugo. 1986a. Structure and biomass of a subtropical dry
forest in Puerto Rico. Biotropica 18: 89-96

Murphy, P. G. and A. E. Lugo. 1986b. Ecology of tropical dry forest. Annual
Review of Ecology and Systematics 17: 67-88

Murphy, P. G. and A. E. Lugo. 1990. Dry forests of the tropics and subtropics:
Gua’nica Forest in context. Acta Cientifica 4(1-3): 15-24

Murphy, P. G. and A. E. Lugo. 1995. Dry forests of Central America and the
Caribbean. Pp. 9-34 In: S. H. Bullock, H. A. Mooney, and E. Medina
(eds). Seasonally Dry Tropical Forests. Cambridge University Press,
Cambridge, England

Murphy, P.G., A. E. Lugo, A. J. Murphy, and D. C. Nepstad. 1995. The dry forest
of Puerto Rico’s south coast. In: A. E. Lugo and C. Lowe (eds). Tropical
Forests: Management and Ecology. Springer-Verlag, New York, NY

Nagy, K. A. 1987. Field metabolic rates and food requirement scaling in
mammals and birds. Ecological Monographs 57: 111-128

Nason, J. D., P. R. Aldrich and J. L. Hamrick. 1997. Dispersal and the dynamics
of genetic structure in fragmented tropical tree populations. Pp. 304-320
In: W. F. Laurance and R. O. Bierregard, Jr. (eds). Tropical forest
_ remnants: ecology, management, and conservation of fragmented
communities. University of Chicago Press, Chicago, IL

126

 

 

Newmark, W. D. 1987. A land-bridge perspective on mammalian extinctions in
western North American parks. Nature 325: 430-432

Noss, R. F. 1983. A regional landscape approach to maintain diversity
BioScience 33(11): 700-705

Ortiz, P. R. 1990. Status and distribution of Anolis cooki (Reptilia; Sauria;
lguanidae). Acta Cientifica 4(1-3): 157-159

Ortiz, P. R. and T. A. Jenssen. 1982. Interspecific aggression between lizard
competitors, Anolis cooki and Anolis cristatellus. Zeitschrift fuer
Tierpsychologie 60: 227-238

Overton, W. S. 1971. Estimating the number of animals in wildlife populations.
pp. 403-455 in: R. H. Giles (ed.). Wildlife management techniques.
Edwards Brothers, Ann Arbor, MI

Patterson, B. D. and W. Atmar. 1986. Nested subsets and the structure of
insular mammalian faunas and archipelagos. Pp. 65-82 in: L. R. Heany
and B. D. Patterson (eds). Island biogeography of mammals. Biological
Journal of the Linnean Society. 28. Academic Press and the Linnean
Society of London, England

Philibosian, R. 1975. Territorial behavior and population regulation in the lizards,
Anolis acutus and A. cristatellus. Copeia 1975(3): 428-444

Portmann, A. 1967. Animal forms and patterns. Schocken Books, New York, NY

Powell, G. V. N. and R. Bjork. 1995. Implications of intratropical migration on
reserve design: a case study using Pharomachrus mocinno. Conservation
Biology 9: 354-362

Pregill, G. 1981. Late Pleistocene herpetofaunas from Puerto Rico. University of
Kansas Museum of Natural History Misc. Publication. 71

Quinn, J. F. and S. P. Harrison. 1988. Effects of habitat fragmentation and
isolation on species richness: evidence from biogeographic patterns.
Oecologia 75: 132-140

Raffaele, H. A. 1989. A guide to the birds of Puerto Rico and'the Virgin Islands
(revised ed.). Princeton University Press, Princeton, NJ

Ramjohn, l. A., P. G. Murphy, and T. M. Burton. Unpublished. Islands of

biodiversity in altered tropical landscapes. Final project report,
llTF pub. ITF-CA-94-008. Rio Piedras, Puerto Rico

127

 

 

Rand, A. S. 1964. Ecological distribution in anoline lizards of Puerto Rico.
Ecology 45: 745-752

Reagan, D. P. 1986. Foraging behavior of Anolis stratulus in a Puerto Rican rain
forest. Biotropica 18(2): 157-160

Reagan, D. P. 1992. Congeneric species distribution and abundance in a three
dimensional habitat: the rain forest anoles of Puerto Rico. Copeia
1 992(2): 392-403

Reagan, D. P. 1996. Anoline lizards. In: D. P. Reagan and R. B. Waide (eds).
Food web of a tropical rain forest. University of Chicago Press, Chicago,
IL

Redford, K H., A. Taber, and J. A. Simonetti. 1990. There is more to biodiversity —
than the tropical rainforests. Conservation Biology 4(3): 328-330

Rivera, L. W. and T. M. Aide. 1998. Forest recovery in the karst region of Puerto
Rico. Forest Ecology and Management 108: 63-75

Rivero, J. A. 1978. The amphibians and reptiles of Puerto Rico. University of
Puerto Rico, Editorial Universitaria, Barcelona, Spain

Robbins, C. S., D. K. Dawson, and B. A. Dowell. 1989. Habitat area
Requirements of breeding forest birds of the Middle Atlantic States.
Wildlife Monograph 103: 1-34

Roughgarden, J. 1995. Anolis lizards of the Caribbean: ecology, evolution, and
plate tectonics. Oxford University Press, New York, NY

Salzburg, M. A. 1984. Anolis sagrei and Anolis cristatellus in southern Florida: a
case study in interspecific competition. Ecology 65(1): 14-19

Sarre, S. 1995. Size and structure of populations of Oedura retuiculata (Reptilia:
Gekkonidae) in woodland remnants: implications for the future regional
distribution of a currently common species. Australian Journal of Ecology
20: 288-298

Sarre, 8., G. T. Smith, and J. A. Meyers. 1995. Persistence of two species of
gecko (Oedura reticulata and Gehyra van'egata) in remnant habitat.
Biological Conservation 71: 25-33

Saunders, D. A., R. J. Hobbs, and C. R. Margules. 1991. Biological

consequences of ecosystem fragmentation: a review. Conservation
Biology 5(1): 18-32

128

 

Schaller, G. B. and P. G. Crawshaw, Jr. 1980. Movement patterns of jaguar.
Biotropica 12: 161-168

Schoener, T. W. 1968. The Anolis lizards ofBimini: Resource partitioning in a
complex fauna. Ecology 49: 704-726

Schoener, T. W. 1975. Presence and absence of habitat shift in some
widespread lizard species. Ecological Monographs 45: 233-249

Schoener, T. W. 1977. Competition and the niche. Pp. 35-136 in: C. Gans and
D. Tinkle (eds). Biology of the Reptilia, vol. 7. Academic Press, New
York, NY

Schoener, T. W. 1985. Are lizard populations unusually constant through time?
American Naturalist 126: 633-641

Schoener, T. W. 1994. Temporal variability in lizard numbers: what the
appropriate kind of study population? American Naturalist 144(5): 868-
872

Schoener, T. W. and A. Schoener. 1971a. Structural habitats of West Indian
Anolis lizards. l. Jamaican lowlands. Breviora 368: 1-53

Schoener, T. W. and A. Schoener. 1971b. Structural habitats of West Indian
Anolis lizards. ll. Puerto Rican uplands. Breviora 375: 1-39

Schoener, T.W. and A. Schoener. 1978. Inverse correlation of survival in lizards
with island size and avifaunal richness. Nature 274: 685-687

Schoener, T.W. and A. Schoener. 1982. The ecological correlates of survival in
some Bahamian Anolis lizards. Oikos 39: 1-16

Schoener, T.W. and A. Schoener. 1983a. Distribution of vertebrates on some
very small islands. l. Occurrence sequences of individual species. Journal
of Animal Ecology 52: 209-235

Schoener, T.W. and A. Schoener. 1983b. Distribution of vertebrates on some
very small islands. ll. Patterns in species numbers. Journal of Animal .
Ecology 52: 237-262

Schwartz, A. and R. W. Henderson. 1991. Amphibians and reptiles of the West

Indies: descriptions, distributions, and natural history. University of
Florida Press, Gainsville, FL

129

 

 

 

Scott, N. J., Jr., D. E. Wilson, C. Jones, and R. M. Andrews. 1976. The choice of
perch dimensions by lizards of the genus Anolis (Reptilia, Lacertilia,
lguanidae). Journal of Herpetology 10: 75—84

Shochat, D. and H. C. Dessauer. 1981. Comparative immunological study of
Albumins of Anolis lizards of the Caribbean Islands. Comparative
Biochemical Physiology 68A: 67-73

Smith, G. T., G. W. Arnold, S. Sarre, M. Abensperg-Traun, and D. E. Steven.
1996. The effect of habitat fragmentation and grazing on animal
communities in remnants of gimlet Eucalyptus salubn’s woodland in the
Western Australian wheatbelt. II. Lizards. Journal of Applied Ecology 33:
1302-1 310

Soulé, M. E., B. A. Wilcox, and C. Holtby. 1979. Benign neglect: a model of
faunal collapse in the game reserves of East Africa. Biological
Conservation 15: 259-272

Stamps, J. A., 1977. The relationship between resource competition, risk, and
aggression in a tropical territorial lizard. Ecology 58: 349-358

Stohlgren, T. J., M. B. Coughenour, G. W. Chang, D. Binkley, M. A. Kalkhan, L.
D. Schell, D. J. Buckley, and J. K Berry. 1997. Landscape analysis of
plant diversity. Landscape Ecology 12: 155-170

Thomas, R. and A. Schwartz. 1966. Sphaerodactylus (Gekkonidae) in the
Greater Puerto Rico region. Bulletin of the Florida State Museum 10(6):
1 93-260

Thomlinson, J. R., M. l. Serrano, T. M. Lopez, T. M. Aide, and J. K. Zimmerman.
1996. Land-use dynamics in a post-agricultural Puerto Rican landscape
(1936-1988). Biotropica 28(4a): 525-536

Tokarz, R. R. and J. W. Beck, Jr. 1987. Behaviour of the suspected lizard
competitors Anolis sagrei and Anolis carolinensis: an experimental test
For behavioural interference. Animal Behaviour 35: 722-734

Tosl, J. A., Jr., and R. F. Voertman. 1964. Some environmental factors in the
economic development of the tropics. Economic Geography 40: 189-205

Turner, F. B. 1977. The dynamics of populations of squamates, crocodylians,

and rhynchocephalians. II. Lizards. Pp. 158-201 In: C. Gans and D. W.
Tinkle (eds). Biology of the Reptilia, vol. 7. Academic Press, New York,
NY ’

130

Turner, F. B., R. l. Jennrich, and J. O. Weintaub. 1969. Home ranges and body
size of lizards. Ecology 50: 1076-1081

Turner, I. M. 1996. Species loss in fragments of tropical rain forest: a review of
the evidence. Journal of Applied Ecology 33: 200-209

Turner, l. M. and R. T. Corlett. 1996. The conservation value of small, isolated
fragments of lowland tropical rain forest. Trends in Ecology and Evolution
1 1(8): 330-333

Vélez Rodriguez, L. L. 1995a. Land use and land cover 1936 Guanica
Commonwealth Forest. USDA Forest Service, lntemational Institute of
Tropical Forestry, Rio Piedras, Puerto Rico (map)

Vélez Rodriguez, L. L. 1995b. Land use and land cover '1 950-51 Guanica
Commonwealth Forest. USDA Forest Service, lntemational Institute of
Tropical Forestry, Rio Piedras, Puerto Rico (map)

Vélez Rodriguez, L. L. 1995c. Land use and land cover 1963 Guanica
Commonwealth Forest. USDA Forest Service, lntemational Institute of
Tropical Forestry, Rio Piedras, Puerto Rico (map)

Vélez Rodriguez, L. L. 1995d. Land use and land cover 1971 Guanica
Commonwealth Forest. USDA Forest Service, lntemational Institute of
Tropical Forestry, Rio Piedras, Puerto Rico (map)

Vélez Rodriguez, L. L. 1995a. Land use and land cover 1983 Guanica
Commonwealth Forest. USDA Forest Service, International Institute of
Tropical Forestry, Rio Piedras, Puerto Rico (map)

Vélez Rodriguez, L. L. 1995f. Land use and land cover 1989 Guanica
Commonwealth Forest. USDA Forest Service, lntemational Institute of
Tropical Forestry, Rio Piedras, Puerto Rico (map)

Whitmore, T. C. 1997. Tropical forest disturbance, disappearance, and species
loss. Pp. 3-12 In: W. F. Laurance and R. O. Bierregard, Jr. (eds).
Tropical forest remnants: ecology, management, and conservation of
fragmented communities. University of Chicago Press, Chicago, IL

Wilcox, B. A. and D. D. Murphy. 1985. Conservation strategy: the effects of
fragmentation on extinction. American Naturalist 125: 879-887

Williams, E. E. 1969. The ecology of colonization as seen in the zoogeography

of anoline lizards on small islands. Quarterly Review of Biology 44: 345
389

131

Williams, E. E. 1972. The origin of faunas. Evolution of lizard congeners in a
complex island fauna: atrial analysis. Pp. 47-89 In T. Dobzhansky, M.
Hecht, and W. Steere (eds) Evolutionary Biology, 6. Appleton-Century
Crofts New York, NY

Williams, E. E. 1976. West Indian anoles: a taxonomic and evolutionary
summary. I. Introduction and a species list. Breviora 440: 1-21

Williams, E. E. 1983. Eoomorphs, faunas, island size, and diverse end points in
island radiations of Anolis. Pages 326-370 In: R. B. Huey, E. R. Pianka,
and T. W. Schoener (eds). Lizard ecology: studies of a model organism.
Harvard University Press, Cambridge, MA

Zar, J. H. 1984. Biostatistical Analysis, second edition. Prentice Hall,
Englewood Cliffs, NJ

132