t :1 ‘
..¥ 1 Y ‘ .I z

: \I.V> .

It 1111.394115.)

421111.... . .IHNAMVL

W. f.“ .v FYI
: ln

t. ..

‘3. 2
.
.l

.1 ;

Lt“). . ‘
533:: x...
kn». ,

.9. .y. .2.
$3.5 Attyi...‘

W 3::

3.4,! .
. ‘ t. .r 4 $
alivtI:
. V : ’» «I’thixlv
51...}. . ; ... . .
{I}; 5. .{Ix

.lllh' a

 

54:1.
firmnfiguafiflno
. . 3.1.1

91112144

I. (51".11.
titrflio.‘
t. It

 

 

 

iitfiaggéé .. fiafimyggg ., nfiwfifir

 

THESIS

aorx-N

llllUHllHl[IllUlllllllllllllllllllllllllllllll

301841 0245

 

 

 

a ? A

Mitt it ”all State
University

 

 

 

This is to certify that the

thesis entitled

DIVERSITY AND ABUNDANCE OF TERMITES (ISOPTERA)
IN A FRAGMENTED SUBTROPICAL DRY FOREST
LANDSCAPE

presented by

John A. Genet III

has been accepted towards fulfillment
of the requirements for

Masters degree in Zoolggx

Date-W /???

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

we chlHC/DdoDuopGfi-p.“

THE DIVERSITY AND ABUNDANCE OF TERMITES (ISOPTERA) IN A
FRAGMENTED SUBTROPICAL DRY FOREST LANDSCAPE

By

John A. Genet III

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 DIVERSITY AND ABUNDANCE OF TERMITES (ISOPTERA) IN A
FRAGMENTED SUBTROPICAL DRY FOREST LANDSCAPE

By

John A. Genet III

The fragmentation of natural habitats currently represents one of the most serious
threats to biodiversity. However, it is becoming apparent that not all components of an
ecosystem respond similarly to fragmentation. The efl‘ects of forest fragment size on
termite community composition and wood decomposition was determined in the
subtropical dry forests of southwestern Puerto Rico. These components were measured in
small (<1 ha), medium (1-10 ha), and large (>10 ha) forest fiagments and compared to
those in Guanica Commonwealth Forest, a relatively undisturbed, contiguous tract of dry
forest. Since very little autecological information exists for most of the termite species
included in this study, data were collected concerning termite population densities, diet,
and habitat preferences. The composition and structure of termite communities did not
differ significantly between the forest fragments and Guanica Forest, while rates of wood
decomposition decreased significantly within the fragments as compared to the contiguous
forest. These results indicate the problems associated with inferring the response of an
ecosystem process to forest fragmentation based solely on the response exhibited by the

community of organisms which controls the process.

ACKNOWLEDGEMENTS

I am very grateful for the guidance and support throughout the development and
completion of this thesis provided by T. M. Burton, P. G. Murphy, and C. M. Bristow.
K. Genet, 1. Ramjohn, and S. Van Bloem provided assistance in the field and aided with
the identification of plants. D. Uzarski, J. Gathman, and S. Rifi‘ell provided invaluable
statistical advice. M. E. Canals provided advice and logistical support on numerous
occasions. B. Noyle assisted with the development of the GIS database and subsequent
analyses. C. Rodriguez and 0. Ramos of the International Institute of Tropical Forestry
(IITF), GIS/GPS Lab generously provided digital format data on the study area’s
hydrology, hypsography, and roads. The Puerto Rican Departamento de Recursos
Naturales y Ambientales granted permission to collect specimens (DNRA permits 97-10
66 and 98-IC-75). Thanks also to R. Scheflrahn (University of Florida) for commenting
on an earlier draft of chapter 1 and providing taxonomic advice. Funding was provided by
the Department of Zoology and the Ecology, Evolutionary Biology and Behavior Program
of Michigan State University. Special thanks to AE. Lugo, Director of IITF, who
provided fimding for the associated plant community studies and assisted in numerous

ways, including methodological advice and access to the resources of IITF.

iii

TABLE OF CONTENTS

LIST OF TABLES vi
LIST OF FIGURES viii
INTRODUCTION 1
Study Region 5
Guanica Forest 8
Study Sites 10
CHAPTER 1 .
The Termites of Puerto Rico’s Southwestern Dry Forests 12
Materials & Methods 14
The Study Region 14
Nest Density 15
Sample Size 17
Colony Composition 18
Nest Characteristics of Nasutitermes spp. 18
Results & Discussion 19
Sampling Efi‘ectiveness 19
Kaloterrnitidae 22
Rhinotermitidae 27
Tennitidae 28
CHAPTER 2
Environmental Influences on the Distribution and Abundance Termite Species in a
Subtropical Dry Forest 38
Materials & Methods 40
The Study Region 40
Nest Density 41
Habitat Variables 42
Results ’ 48
10 mx 10mPlots 48
Forest Fragments 51
Discussion 54
Termite Diversity S4
Termite Abundance 57
CHAPTER 3
Contrasting Responses of Termite Community Composition and Wood Decomposition to
Habitat Fragmentation in a Subtropical Dry Forest 60

Materials & Methods 63

The Study Region
Fragment and Landscape Characterization
Termite Sampling
Sample Size
Wood Decomposition
Analyses
Results
Sampling Efi‘ectiveness
Comparison of Reference and Fragment Communities
Minimum Fragment Area
Fragment and Landscape Characteristics
Wood Decomposition
Discussion
Termite Community Response to Fragmentation
Effects of Fragmentation on the Decomposition of Wood
Conservation Priorities
Conclusions

APPENDIX A
APPENDIX B
APPENDIX C
APPENDIX D

REFERENCES

63

68
70
7O
74
76
76
78
80
81
83
88
88
94
96
96

98

100

105

110

111

LIST OF TABLES

Page

1. Characteristics of forest fiagments and reference site selected for study. 10
2. Summary of sampling design at forest fragments and reference site

(Guanica Commonwealth Forest). 16
3. Mean termite nest densities within a relatively undisturbed dry forest

(Guanica Forest). 20
4. List of plant species colonized by Procryptotennes comiceps and the

frequency each species was colonized (when identification was possible).

Type of host species also included: t = tree, 8 = shrub, v = vine. 23
5. Colony composition of Procryptotermes corniceps. 25
6. Location of Nasufitermes spp. arboreal nests, including the frequency for

each tree species or general position if not situated in a live tree. 35

7. The termite groupings used in determining associations with habitat variables. 44

8. Mean number of nests/ha for individual temrite species at the 12 forest
fragments and the reference site (Guanica Forest). 46

9. Speannan’s rank correlation coefficients (n) between termite diversity and
abundance estimates and habitat variables within the 10 m x 10 m plots. 49

10. Pearson correlation coefficients (r) between termite diversity and abundance
estimates and characteristics of the plant community within forest fragments. 52

11. Selected fragment and landscape characteristics of the twelve study sites and
the reference site (Guanica Forest). 67

12. The occurrence of termite species within the 12 forest fragments (listed by
site codes) and reference site (Guanica Forest). 77

13. Pearson correlation coefficients (r) among termite diversity and abundance
(no. nests/ha) measures and fragment characteristics. 81

14. Pearson correlation coefficients (r) among termite diversity and abundance
(no. nests/ha) measures and landscape characteristics. 82

15. Pearson correlation coefficients (r) between termite diversity and abundance
(no.nests/ha) measures and the percentage of the study site perimeter occupied
by the four major land-use types. 82

16. List of taxa collected from wood bundles including their frequency (the no.
of bundles in which they were found), location within the bundles, and

indication of whether or notthey consume wood. . 85

17. The mean density (no. nests/ ha) of the three termite species contributing to
the decomposition of wood within the bundles at the four study sites. 95

vi

10.

11.

LIST OF FIGURES

. Map of Puerto Rico illustrating the extent of the subtropical dry forest life

zone on the island. Enlarged area showing the relative size and location
of the twelve forest fi'agments and the reference site (Guanica Forest)
within the southwest region of the island.

Performance curve for the mean nest density (no. nest/ha) of Incisitermes
bequaerti at the reference site (all three sites combined).

Arboreal nest of N. acajutlae in T amarindus indica.

Nest characteristics of Nasutitermes spp., (A) nest volume and (B) height
of the nest from ground.

Mean (: SE) (A) abundance of P. wolcotti, (B) abundance of arboreal termite

species, and (C) termite species richness within the three canopy categories.

Association between termite and lizard species richness among forest
fragments.

Dendrograms constructed using plant community variables and temrite
community variables fi'om eleven of the forest fragment sites.

Land use map of Guanica Commonwealth Forest and its vicinity, including
the location of interior forest study sites (REF C, S, N) and fragments 7,
10, and 4. Inset: Map of Puerto Rico illustrating the extent of the
subtropical dry forest life zone on the island.

Design summary of wood decomposition study. (A) Arrangement of wood
bundles around N. acajutlae arboreal nests. (B) Sample size and design for
each of the study sites. C = control bundle, E = experimental bundle.

Dendrograrn of site similarities based on termite densities from the small
(S), medium (M), and large (L) fragments and the reference sites
(REF C, S, N).

Graph summarizing the fragment area required to support proportional
subsets of the termite community found in the reference site.

viii

21

31

33

52

53'

65

79

80

12. Mean (1 SE) loss of biomass from (A) experimental and (B) control bundles
at the <1ha, 1-10 ha, >10 ha, and reference forest study sites. 84

13. Boxplots of reductions in wood bundle biomass categorized by the
presence/absence and relative amount of the two major wood-feeding
invertebrate groups encountered in this analysis. None = no major
decomposer activity (group 1). Horizontal lines show the 10'”, 25‘“, 75‘“,
and 90'h percentiles increasing up the ordinate. Open circles represent all
data outside the 10"I and 90"“ percentiles. 87

ix

INTRODUCTION

Rates of deforestation in the tropics and subtropics have continued to increase
over the past several decades (F A0 1993, Whitmore 1997). Concomitantly, the majority
of remaining forests throughout these regions are represented by small patches set in a
matrix of non-forested vegetation. Due to their high degree of biodiversity and the rapid
rate at which they are currently being deforested, the conversion of wet tropical forests
has received much attention in the past few decades (e. g., Lovejoy & Bierregaard 1990,
Whitmore & Sayer 1992, see also Turner 1996). However, as a result of their productive
soils and reasonably comfortable climate, it has been the dry forests that have historically
undergone the greatest alterations (Tosy & Voertman 1964, Murphy & Lugo 1986,
Janzen 198 8a).

Dry forests constitute approximately 42 % of the world’s tropical and subtropical
forests (Brown & Lugo 1982), yet they have been disturbed to such a great extent that
they could be considered endangered ecosystems (Janzen 1988a, 1988b, Murphy & Lugo
1990, 1995). Only within the past few years have studies emerged concerning the impacts
of fi'agmentation in dry forest life zones (Aizen & Feinsinger 1994a, 1994b, Kramer
1997). While there are a few exceptions (e.g., Corlett & Turner 1997), the majority of
fragments created by the deforestation of tropical wet forests have only existed for a
relatively short period of time (less than a decade) (Turner 1996, Turner & Corlett 1996).
Therefore, questions concerning the long-term survival of species within fragments may be
better answered through studies of dry forest fi'agments, the majority of which have been

isolated for many decades.

In Central America and the Caribbean there exists only a few intact representatives
of dry forest (Janzen 1988a, 1988b, Murphy &_ Lugo 1995). Clearly, the conservation of
these ecosystems will depend on the protection and management of forest remnants as
well as the restoration of areas following disturbance (Janzen 1988a, 1988b). The
culmination of such processes would be the coalescence of habitat fiagments back into a
contiguous forest with the fragments providing the ‘seeds’ for the re-establishment of
native flora and fauna in the recovering landscape (Janzen 1988b, Turner & Corlett 1996).
While it is unlikely that this will produce tracts of forest of the same extent and ecological
integrity as the original forest, it will nonetheless result in a significant contribution to the
maintenance of ecosystem firnction and preservation of biodiversity. The success of such
initiatives, however, relies on the knowledge of how species have responded to habitat
fragmentation and the factors that control these responses.

Habitat fragmentation is currently one of the most serious threats to biological
diversity (Wilcox & Murphy 1985, Turner 1996, Laurance & Bierregaard 1997). The
primary mechanism of this decline in diversity is the decrease in local species richness
associated with the division of a continuous habitat into isolated remnants of vegetation
set in a matrix of alternate land uses. There are a multitude of ways, however, that
fiagmentation can lead to local extinctions of native populations, including: (1) reduction
in population sizes; (2) reduction in immigration rates; (3) increased human disturbance
during and after fi'agmentation; (4) deleterious edge efi’ects; (5) changes in community
interactions; and (6) immigration of exotic species (Saunder et a1. 1991, Turner 1996). In

addition, biogeographic factors such as remnant size and shape, degree of isolation, and

time since excision from the, continuous habitat can affect the intensity of the above
mentioned circumstances (Saunder et a1. 1991, Turner 1996).

Given the variety of possible outcomes, it is problematic to formulate
generalizations about the responses of species to habitat fragmentation without knowledge
of their basic biology (Margules et a1. 1994). The response of individual species is
dictated by their autecologies (McCoy & Mushinsky 1994) and the efi'ects of
fragmentation on other species in the community with which they interact. Once this
information is obtained it then becomes possible to identify the particular traits that make a
species sensitive to habitat fragmentation, thus allowing one to generalize about the
responses of species that share these traits. For instance, it has been previously
demonstrated that specialists, rare species, patchily distributed species, and wide ranging
species are particularly susceptible to fragmentation (Turner 1996). Unfortunately, the
necessary biological information for many species is sparse or lacking completely.

In an attempt to gain insight into the processes responsible for the observed
impacts of fragmentation, a compilation of the sparse amount of existing information and
the descriptive .data collected during this study was put together for termites, the
organisms at the focus of this investigation. Hence, the first chapter of this thesis deals
with the descriptive information collected for each species, with some species getting
more attention than others due to their abundance in the study region. The second chapter
continues by looking at the relationships between habitat variables and the diversity of the
termite community as well as the abundance of individual species. Once this information
is reviewed, an analysis of the impacts of habitat fragmentation is undertaken in chapter 3.

This last chapter. investigates the effects of fragmentation on the composition of the

termite community and decomposition of wood by comparing these phenomena in habitat
fiagments and relatively undisturbed contiguous dry forest.
The main objectives of this study are as follows:

(1) supplement the sparse amount of existing biological data for termites of the Subtropical
dry forests of Puerto Rico by determining the population densities, diet preferences,
colony size, and nesting locations of selected species.

(2) identify characteristics of the habitat that influence the composition and structure of
termite communities in a subtropical dry forest. It was hypothesized that structural
characteristics which provide cover from direct exposure to sunlight would increase
termite diversity and abundance.

(3) determine the composition and structure of the termite community in a relatively
undisturbed, contiguous tract of subtropical dry forest (Guanica Forest) and compare
this reference community to termite communities found within forest fragments of
various sizes. It was hypothesized that individual forest fragments would only support
a subset of the termite community found within the reference forest, but when forest
fragments were viewed collectively, the diversity and abundance of termites would be
comparable to that of the reference forest.

(4) determine the minimum fi'agment area required to support subsets (50 and 75%) of the
termite community found in the reference site.

(5) identify fragment and landscape characteristics that influence the composition and
structure of termite communities in forest fragments. Utilizing the principles of island
biogeography, it was hypothesized that as fragments became smaller and more

isolated, the diversity and abundance of termites would decrease.

(6) compare rates of wood decomposition in a contiguous subtropical dry forest to those
in forest fragments of various sizes. It was hypothesized that wood decomposition
would proceed at a significantly slower rate in forest fragments as compared to the
reference forest. The basis for this hypothesis was the investigation by Klein (1989)
which demonstrated declines in the rate of dung decomposition within Amazonian
forest remnants.

(7) determine the relative contribution of tennites to wood decomposition in a subtropical
dry forest ecosystem. It was hypothesized that termites are important contributors to
the decomposition of wood within subtropical dry forests. This hypothesis was
suggested based on, the seasonality of precipitation in subtropical dry forest
ecosystems, which would seem to confine the activity of microbial decomposers to the
rainy season.

(8) provide recommendations for the conservation and management of southwestern

Puerto Rico’s fragmented subtropical dry forest landscape.

Study Region
The southwestern portion of Puerto Rico has been categorized as subtropical dry
forest (sensu Holdridge 1967) by Ewel & Whitmore (1973) (Figure 1). This life zone is
characterized by an annual precipitation that ranges fi'om 250 to 2000 mm, an average
annual biotemperature greater than 17 °C, a ratio of potential evapotranspiration to
precipitation that is greater than one and less than two, and absence of frost throughout

the year (Holdridge 1967).

    
      

.283 2: Co sew“: 303538 2: 55:5 085.,— 8:833
86 oocouomoe 2: was ficofimflm $809 m. 2: we connec— wce cum 02:22 2: mes/0% no; women—cm
65E 9: so econ om: 32$ be Quiche; 2: mo E038 or: 95835:. 002 8.5.5 mo amE ._ oSwE

lit
95525:! E. h o

 

K

 

mm?
A umofifi

m.ceN o5
«moan. >5 .moiobnsw

 

Puerto Rico’s dry southwestern region is due to a rain shadow effect resulting
from the northeasterly trade winds passing over the island’s central mountain system. This
region extends east along the south coast approximately 120 km fiom the southwest
corner of the island and reaches between 3 and 20 km inland, depending on local
topography (Ewel & Whitmore 1973). Although this southwest region contains the
majority of Puerto Rico’s dry forests, it still only represents roughly 4 % (5000 ha) of the
island’s original dry forests (Murphy & Lugo 1990, Murphy et al. 1995).

Annual precipitation in this region ranges from 600 mm in the west of the dry
forest life zone to about 1,000 mm in the east. The majority of the precipitation occurs
during two wet seasons, one long period that extends from August to November and a
shorter one during the month of May. Variation of year-to-year rainfall is significant and
is attributed to the unpredictable passage of large storms and hurricanes over or near the
region (Murphy et al. 1995). For instance, thesouthwestem region recently (September
1998) experienced a major storm as Hurricane Georges traversed the entire length of the
island and soaked the dry forest life zone with heavy downpours for several hours (Van
Bloem pers. comm). The impacts of such severe storms have not been previously
determined for subtropical dry forests.

The dry forests of southwestern Puerto Rico represent an ecosystem that has been
heavily impacted through human activities. This general trend extends to the rest of the
island as well. During pre-colonial times the island was 95% forested, however, this
amount was reduced to a mere 5% by 1948 with most of the remaining forests restricted
to isolated mountainous areas (Birdsey & Weaver 1982). During the last 60 years a

. reverse in this trend occurred due to a dramatic shift in the Puerto Rican economy from an

agricultural system to a more urbanized industry with an emphasis on manufactured
products (Morales-Carrion 1983, Dietz 1986). This significantly reduced the pressure on
the remainder of the island’s forests and even led to a recovery of previously deforested
areas, resulting in forest. cover increasing to 34% by 1985 (Birdsey & Weaver 1987). An
example from the dry forest life zone can be found in the vicinity of Guanica Forest (see
below), where closed canopy forest increased from 3181 to 5502 ha between the years
1936 and 1989 (Lugo et al. 1996). This was accompanied by a decrease in agricultural
land from 3517 to 1799 ha during the same time period.

The current state of Puerto Rico’s southwestern dry forest life zone can best be
described as a series of forest fragments sparsely distributed over a highly impacted
landscape. A recent characterization of the landscape (Ramjohn et al. unpublished)
yielded a total of 308 closed and 312 open canopy forest fragments that together only
accounted for 23% (16960 ha) of the total land area of the region. The majority of these
fragments were less than 5 ha in area and there was only 14 fiagments that represented
substantial tracts (>100 ha) of closed canopy forest. While the mean fragment size was

42.6 ha, the median fragment size for closed canopy forests was less than 5 ha.

Gua’nica Forest
Although considered a forest fragment by most measures, Guénica Commonwealth
Forest represents one of the best remaining examples of subtropical dry forest in the world
(Murphy & Lugo 1990). Situated in the center of Puerto Rico’s dry forest life zone
(Figure 1), this relatively large (~4000 ha) reserve has been protected at various

magnitudes since the 1930’s, culminating with its designation as a Biosphere Reserve by

the UNESCO Man and the Biosphere Program in 1982. Before Guanica Forest oficially
received protected status it supported a myriad of services, including tree plantations
(Swietenia mahagoni and Haematoxylum campechiam‘um), fencepost production,
charcoal production, cultivated crops (maize, tobacco, sugar cane), cattle pastures, and
even human settlements (Murphy et al. 1995). Only fencepost production persisted past
the 19403, continuing into the mid-1970s (Canals Mora pers. comm). This complex
disturbance history has resulted in a mosaic of forest stands of widely varying structure
and composition (Murphy & Lugo 1990).

Guanica Forest is divided into two portions separated by Guanica Bay, with the
more extensive and less disturbed tract of the reserve located on the eastern side on the
bay (Figure 1). This area forms a continuous habitat from the shoreline of the Caribbean
Sea up to elevations exceeding 200 m, supporting six floristically and structurally
distinguishable plant associations (Lugo et al. 1978). The three major intergrading plant
associations of Guanica Forest are scrub forest located predominantly on limestone
outcrOppings, deciduous forest located on upland slopes and valleys, and sernievergreen
forest that is confined to narrow ravines. While there are no permanent water courses
within the forest, narrow canyons formed by collapsed cave systems (F amsworth 1993)
run in a north-south orientation throughout the reserve and often flood during large storm
events (Murphy et al. 1995).

Regardless of its past disturbance history and present conditions, Guanica Forest
serves as the best reference system in the region for which comparisons with forest
fragments can be made. Although most of its vegetation has been disturbed in the

relatively recent past, the forest is reaching a state of maturity that is rivaled nowhere else

within Puerto Rico’s dry forest life zone (Murphy & Lugo 1990, 1995, Murphy et al.

1995)

Study Sites
Twelve forest fragments were selected for study fiom a set of 40 fragments that
had previously been described by Ramjohn et al. (unpublished) in terms of their vegetative
composition and structure (Appendix A, Figure 1). These fragments 'vary greatly in size,
shape, topographic location, vegetative composition and structure, amount of past and
present disturbance, as well as the type of substrate (Table 1). For instance, a few of the
fiagments are in close proximity to the coast and are situated on limestone outcroppings

with an almost complete lack of soil. Other fragments are located on abandoned pastures

Table 1. Characteristics of forest fragments and reference site selected for study.

 

 

 

Primary Elevational Plant Sp. Tree Sp.
Site-IDT Area (ha) Perimeter Lm) Aspect Range (m) Richness’ RichnessT
40 0.006 27 S ~30 21 7
36 0.06 92 S 30-35 70 32
30 0.26 226 B 30-40 76 36
27 0.69 408 W 170-200 71 27
28 2.0 835 E 40-100 68 31
13 3.1 919 Flat 15-26 69 31
16 3.4 832 E + S 0-26 69 34
10 6.3 1723 N 50-100 101 42
7 33.6 3517 All 25-108 149 61
5 97.1 5258 All 10-100 152 79
4 136.8 7562 All 40-190 153 73
2 854.1 23277 All 0-130 128 67
Guanica 3724 47885 All 0-228

 

1 from Ramjohn et al. (unpublished)
area and perimeter measurements are for entire tract of continuous forest rn which the
eastern portion of Guanica Forest lies.

10

and have well developed soils. In addition, some sites have been under extreme pressure
from human activities for prolonged periods of time while others have remained virtually
untouched following their isolation from adjacent forests. Most of the fragments are
located in relatively inaccessible areas such as ravines, steep slopes, and hilltops that were
not necessarily suitable for agricultural practices or urban development.

Sites within the reference system, Guanica Forest, were chosen on the basis of
having areas that reflected the variability of characteristics displayed by the fiagments.
For instance, since a few of the sites were in close proximity to the sea and consisted
primarily of vegetation characteristic of the scrub forest, a site was selected within
Guanica Forest that was located near the sea within the scrub forest. Overall, three sites
were selected within Guanica Forest which represented coastal scrub forest, upland

deciduous forest with a predominantly south-facing aspect, and upland deciduous forest

with a predominantly north-facing aspect.

I]

CHAPTER ONE

The Termites of Puerto Rico’s Southwestern Dry Forests

A recent survey of the termites (Isoptera) of Puerto Rico (Jones & Schefi‘rahn
unpublished) increased the number of recorded species on the island from 16 to 22, with
10 of these occurring within the island’s subtropical dry forest life zone (sensu Holdridge
1967). Three termite families (Kalotermitidae, Rhinotermitidae, Termitidae) are
represented throughout Puerto Rico as well as within the subtropical dry forests.

Species belonging to the primitive family Kalotermitidae are commonly known as
drywood termites because their diet and therefore nests are almost exclusively restricted to
dry, dead wood. Colonies are established and remain completely contained within their

food source with the end result of foraging and nest expansion being one and the same.
Nests are constructed primarily within fallen dead branches and standing dead wood,
although, the colony does not require contact with the soil for moisture. Instead, the
termites are able to obtain water as a metabolic by-product of cellular metabolism as well
as from external water and living plant tissue. The typical mature colony consists of a pair
of primary reproductives (king and queen) or replacement reproductives, nymphs, worker-
like pseudergates, soldiers, larvae, and eggs. In addition, winged irnagos (alates) may be
present within the nest at certain times of the year before they disperse to establish new
colonies. Mature colonies generally do not exceed a few thousand individuals (Nutting

1969, Lenz 1994).

12

Most members of the family Rhinotermitidae have subterranean nests constructed
directly in the soil. However, some species build their nests within buried, rotting wood
or in logs above ground. All species consume wood, and many are economically
important pests, including Reticulitermesflavipes (Kollar) in the eastern United States and
Heterotermes spp. in the.West Indies (Gillot 1995).

Three-quarters of all termite species belong to the diverse family Terrnitidae. It is
within this family that the greatest range of social development and specialization exists
(Gillot 1995). Members of this family from the West Indies belong to three subfamilies:
(1) Tennitinae, (2) Apicotermitinae, and (3) Nasutitennitinae. The Nasutitennitinae are
characterized by the evolutionary development in soldiers of a rostrum (nasus) from which
defensive secretions of the frontal gland can be exuded or forcefully ejected onto nest
invaders.

Little information exists concerning the basic biology of the majority of termite
species native to Puerto Rico beyond the few comments and observations within their
original descriptions and a recent paper by Jones et al. (1995). In fact, it still is unclear
exactly how many species there are in Puerto Rico. As recently as 1994 there was a new
species described from the island (Scheffrahn 1994), and the taxonomic status of others
remains unresolved. Therefore, the objective of this chapter is to augment the sparse
information that currently exists for termites native to Puerto Rico’s southwestern dry

forests.

l3

MATERIALS & METHODS
The Stuay Region

The southwestern portion of Puerto Rico includes the majority of the island’s
subtropical dry forest life zone. This region extends east along the south coast
approximately 120 km from the southwest corner of the island (Ewel & Whitmore 1973)
and reaches between 3 and 20 km inland, depending on local topography. Although there
is considerable year-to-year variation, annual precipitation ranges from 600 mm in the
' west of the dry forest life zone to about 1,000 mm in the east. Mean annual temperature
is 25.1 °C. The majority of the region is highly disturbed due to industrial, agricultural,
and urban land uses, resulting in the fragmentation of natural habitats. However, Guanica
Commonwealth Forest, situated within the center of the dry forest life zone (Figure 1),
represents one of the best remaining examples of subtropical dry forest in the world
(Murphy & Lugo 1990). This area of approximately 4000 ha has been protected since the
1930’s and encompasses a variety of edaphically determined associations.

Twelve forest fragments of various sizes were randomly selected for study from a
total of 40 sites in which the vegetative composition and structure had previously been
documented by Ramjohn et al. (unpublished) (Appendix A, Figure 1). In addition, three
sites within Guanica Forest were selected in order to obtain baseline data for termite
communities in a subtropical dry forest. These sites within Guanica Forest were located
within a coastal scmb forest and upland deciduous forests on north- and south-facing
slopes. A more complete description of the study sites, Guanica Forest, as well as the dry

forest life zone of Puerto Rico can be found in the main Introduction of this thesis.

l4

Nest Density

A quantitative survey of the termites of the southwestern dry forests of Puerto
Rico was conducted during the summer months of 1997 (May, June, and July) and 1998
(July and August). Due to the variability in nest sizes among the different termite species,
two sampling regimes were necessary in order to obtain accurate estimates of abundance.
Two species (Nasutitermes costalis (Holrrrgren) and Nasuititermes acajutlae (Holmgren))
build large, conspicuous arboreal nests and were sampled using 100 m x 10 m belt
transects. The location for establishing the transects was determined by randomly
selecting points on a grid overlay of the aerial photos of each site. Large fragments (>10
ha) and Guanica sites were stratified using four aspect categories (N -, S-,E-, W-facing
slopes) and one category for ravine areas. The number of samples selected from each
category was largely determined by the proportion of the total area that each category
represented.

The remainder of the termites have less conspicuous nesting locations such as
within small pieces of wood or underneath rocks and logs. Nest abundance for these
species was determined by exhaustively searching 2 m x 2 m plots randomly selected fiom
within the first 10 m segment of the belt transect. However, transects were not
established at all of the selected sampling points. Instead, a single 10 m x 10 m plot was
demarcated from which the 2 m x 2 m plots were randomly selected (see Table 2 for
summary of sampling design at all sites). All dead wood within the plot, including
standing dead wood (SDW) and dead branches up to a height of 2 m, was searched. The
host plant was identified whenever possible using the taxonomic keys and descriptions in

Little & Wadsworth (1964) and Little et al. (1974). The litter layer was then searched,

15

Table 2. Summary of sampling design at forest fragments and reference site (Guanica

 

 

 

 

Commonwealth Forest).
Percentage of
Total Area of Site Searched

Site Area Belt Transects (m2) No. of Plots Belt 2 x 2 Site
ID # (ha) (No. of transects) 10 x 10' 2 x 2 Transect Plots Walk? ”
40 0.006 0.006 I - 4 100 33 No
36 0.06 0.06 l - 4 100 3.2 No
30 0.26 0.26 * - 5 100 0.8 Yes
27 0.69 2000 (3)" 4 12 29 0.7 No
28 2.0 2000 (3)" 4 12 10 0.2 Yes
13 3.1 2000 (2) 4 12 6.4 0.2 No
16 3.4 2000(2) 4 12 5.9 0.1 Yes
10 6.3 2000 (2) 4 12 3.2 0.1 Yes
7 33.6 6000 (6) 8 24 1.8 0.03 No
5 97.1 8000 (8) 12 36 0.8 0.02 Yes
4 136.8 8000(8) 12 36 0.6 0.01 Yes
2 854.1 8000 (8) 12 36 0.1 0.002 Yes
reference site‘:

C 3724 6000 (6) 8 24 - - -
S 3724 6000 (6) 8 24 - - -
N 3724 6000 (6) 8 24 - - -
Total 3724 18000 (18) 24 72 0.05 0.001 Yes

 

’ Includes first 10 m segments of belt transects and individually established plots.

" Presence/absence data only.
* Entire site was searched.

1’ Sites were not large enough to establish two complete 100 m belt transects.

‘ Reference forest sites: C = Coastal scrub, S = Upland deciduous on predominantly

south-facing slope, N = Upland deciduous on predominantly north-facing slope.

16

and all rocks were cleared in order to locate subterranean nests.

Whenever possible, soldiers and winged reproductives (alates) were collected from
each colony and field preserved in 80% ethanol for later identification. One soldierless
Species (Anoplotermes sp.) was identified to genus using characteristics of the workers
since alates were rarely found within the nest. Identifications were made using the original
descriptions of Snyder (1923, 1924, 1929) and taxonomic key (Snyder 1956). Two
questionable determinations were verified by R. H. Schefiiahn (University of Florida) and

included comparisons with reference specimens.‘

Sample Size

The total number of samples taken fiom each site was largely a function of area;
the number of samples increased as the area of the site increased (Table 2). However, due
to the amount of time required for each sample (between 1 and 2 hours on average to
search a plot) it was impossible to achieve a constant proportion of area sampled across all
sites (Table 2). Ultimately, the determination of the number of samples taken from each
site was based on maximizing the number of sites that could be sampled with the given
time constraints while maintaining some degree of consistency for the proportion of area
sampled. In order to evaluate the effectiveness of this sampling design, performance
curves were constructed for each site using data fiom the 2m x 2m plots. The mean nest
density for one of the most common species in the region, Procryptotermes corniceps

(Snyder), was used to create the performance curves.

l7

Colony Composition

Entire colonies of Procoptotermes comiceps were collected during directed
searches for their nests. Collecting these colonies proved to be too time conng to
conduct during the 2 m x 2 m plot sampling. Dead branches and SDW were searched
until a colony was discovered, whereupon every attempt was made to collect the entire
colony into 80% ethanol. Each colony was later examined and enumerated using a
dissecting microscope. All termites were placed in one of the following categories: 1)
primary reproductives, 2) alates, 3) soldiers, 4) pseudergates and nymphs with short wing
pads (tip of mesothoracic wing pad not extending beyond thoracic segments), 5) nymphs
with long wing pads, and 6) larvae (first three instars). Lightly pigmented replacement
(neotenic) reproductives were unable to be differentiated. The total number of termites
within each category was recorded for each colony. In addition, the number of eggs
collected was noted, but it is unlikely that these figures accurately represent the amount of

eggs within each colony.

Nest Characteristics of Nasutitermes spp.

For each N. acajutlae and N. costalis nest encountered along the belt transects, a
series of measurements was taken in order to reveal any patterns in their choice of nesting
locations and to determine if there are any differences between the two species with
respect to nesting location and nest size. These measurements included: 1) tree species in
which the nest was situated, 2) height of nest (from ground to base of nest), 3) diameter of
all supporting stems (measured at point of entry into nest), 4) general shape of the nest,

and 5) nest diameters. Most nests had an ellipsoid shape and therefore required the

18

measurement of three diameters: length, width, and depth. These diameters were used to
obtain three radii values for each nest. Depending on the general shape of the nest,
volume was calculated using one of three formulas: 1) V=4/31tr’ (sphere), 2) V=4/3rtr.r2r3
(ellipsoid), and 3) V=1/31rr2h (cone). The volume of all satellite nests was also determined
and combined with that of the main nest in order to obtain a value for total nest volume.

Basal area (A=1tr2) of all supporting stems was calculated for each nest.

RESULTS & DISCUSSION
The survey of temrites from the southwestern region of Puerto Rico revealed 10
species fiom 8 genera and 3 families (Table 3). One species, Caribitermes discolor
(Banks), had not been previously documented from this region, while another species
(Incisitermes snyderi (Light)) listed for this region was not present in any of the samples
and‘could not be located during systematic searches of the sites (Jones & Schefi’rahn

unpublished).

Sampling Effectiveness
Examination of the performance curves revealed that even though a large
proportion of the‘small (< 1 ha) sites was sampled (Table 2), the estimates of P. comiceps
abundance may not be as accurate as those obtained from larger sites (Appendix B). For
instance, the mean nest density at site 36 was still fluctuating widely from sample to
sample at the end of the curve. In general, increasing the number of samples as the area of

a site increased resulted in more reliable estimates of P. comiceps abundance. However,

19

Table 3. Mean termite nest densities within a relatively undisturbed dry forest (Guanica
Forest). + = present but not quantified.

 

 

Species Nest Density (/ha)
Kalotermitidae

Incisitermes bequaerti (Snyder) 3 5

Neotermes mona (Banks) --

Procryptotermes corniceps (Snyder) 2222
Rhinotermitidae

Heterotermes spp. 13

H. tenuis (Hagen)
H. convexinotatus (Snyder)

Tennitidae

Apicoterrnitinae
Anoplotermes sp. (meridianus) Emerson +

Nasutitermitinae
Caribitermes discolor (Banks) 69
Nasutitermes acajutlae (Holmgren) 6.5
Nasutitermes costalis (Holmgren) ‘ +
Parvitermes wolcotti (Snyder) 243

 

the tremendous number of samples required to obtain relatively accurate density estimates
becomes evident when examining the performance curve for all reference sites combined
(Appendix B). Approximately 50 samples (200 m2 area sampled) were required before the
cumulative mean density of P. corniceps became insensitive to fluctuations in the data.
The abundance estimates for the remainder of the termite species in this study was
not as reliable as the abundance estimates obtained for P. corniceps. Due to their low
fi'equency of occurrence, performance curves for the other termite species never reached a
point where the cumulative mean was insensitive to fluctuations in the data. For example,
even with all the reference sites combined (96 samples total), the cumulative mean nest
density for Incisitermes bequaerti (Snyder) would double with the addition of 1 nest in a

subsequent sample (Figure 2).

20

 

    

 

 

 

 

120
=1 100 E Efi‘ect ofadding 1 nest
E to density estimate
g 80 -
B—
0
g 60 -
E
2 40 _
.‘a’
E 20
3 samples taken
c.) i

0 2
l l l I l l l

 

0 50 100-: 150 200 250 300 ~350

Cumulative area sampled (m2)

Figure 2. Performance curve for the mean nest density (no. nest/ha) of Incisitermes
bequaerti at the reference site (all three sites combined).

21

Eggleton and Bignell (1995) discuss the numerous problems associated with
sampling termite populations in terms of the number of individuals. They suggest that the
aggregated distribution of temrites within colonies and the variety of microhabitats that
different species occupy make ecological studies of termites difiicult to undertake. It
appears that these. same problems can be extended to studies which utilize the termite
colony rather than individuals as the sampling unit. For example, the majority of the
termite species encountered in this study had a patchy distribution of colonies, and there
was a wide variety of rrricrohabitat types in which termite colonies were located. In
addition, the extremely rare occurrence of some termite species in the southwestern dry
forests ‘of Puerto Rico and their inconspicuous nest locations exacerbated the above

mentioned difficulties.

Kalotermitidae

Procnptotermes corniceps was the most widespread and common termite species
in the dry forests of southwestern Puerto Rico. Likewise, it has also been reported as the
most common kaloterrnitid in natural vegetation on Providenciales and Grand Turk Islands
(Turks & Caicos Islands) (Schefl‘rahn et al. 1990) and as the most common termite
species on Mona Island (Jones et al. 1995). It occurred in 14 of the 15 sites sampled (not
collected from site 40; Figure 1) and had densities that ranged from 500-4375 nests/ha.
The nest density at the three sites within Guanica Forest ranged from 1042-343 8/ha (mean
= 2222) (Table 3). P. corniceps colonies were found in wood spanning a wide range of

densities (Table 4). Moreover, Wolcott (1950) rates B. simaruba and C. rosea as “very

22

Table 4. List of plant species colonized by Procoptotermes corniceps and the frequency
each species was colonized (when identification was possible). Type of host
species also included: t = tree, 5 = shrub, v = vine.

 

 

Species ' Type Frequency Specific Gravity’
Acacia retusa v 2

Amyris elemifera t 1 1.0-1 . 1
Bursera simaruba t 2 0.29
Canella winterana t 2 0.9-1.0
C Iusea rosea t 1 0.67
Coccoloba diversifolia t 1 0. 8
C occoloba krugii t 2

Coccoloba microstachya t 2

C ordia bullata humilis s l

Croton asteroites s 2

Eugenia rhombea t 1

Gymnanthes lucida t 6 1.1
Krugiodendronferreum t 1 1 3-1.4
Lantana involucrata s 2

Leucaena leucocephala t 17 0. 7
Pithecellobium unguis-cati t 4

T abebuia heterophylla t 1 0.58
unidentifiable dead branches - 88

unidentifiable SDW - 33

unidentifiable dead vines v 8

 

’ fi'om Little & Wadsworth (1964)

23

susceptible”, K. ferreum as “very resistant”, and A. elemifera as “repellent” to the closely
related drywood termite, Cryptotermes brevis. These findings corroborate previous
evidence (Jones et al. 1995, Jones & Schefiahn unpublished) that P. corniceps is a
generalist in subtropical dry forests, where it attacks tree species that span a broad range
of wood densities and degrees of relative susceptibility.

The colony size of P. corniceps ranged from 4 to 2,394 individuals (not including
eggs) (Table 5). Eggs were found in colonies of all sizes, including the largest and
smallest colonies collected. Alates were only found in colonies that had 92 or more
individuals, although not consistently. For instance, alates were not found in two of the
larger colonies, one with n = 203 termites and the other with n = 488 temrites. When
present, alate percentages ranged from 0.9% to 4.2% of the colony, while soldier
percentages ranged from 1.5% to 25% of the colony.

Jones et al. (1995) reported similar percentages for P. corniceps alates (1.5 -
9.8%) and soldiers (0.7 - 20.5%) from another location within Puerto Rico’s subtropical
dry forest life zone, Mona Island. It is worth noting that the slightly higher percentages of
alates within the colonies sampled by Jones et a1. (1995) could reflect the seasonal
variability of alate occurrence within the nest. The dispersal of alates in the tropics
generally occurs during a single rainy season (Nutting 1969). Colonies from Mona Island
were collected in March, prior to the short rainy season that often occurs in this region
during the month of May (Ewel & Whitmore 1973). The present study collected colonies
in July, after this short rainy season, but before the extended wet season that lasts from

August to November. Even though both studies collected P. corniceps alates just prior to

24

.2228 go x62 9 one BEEN—co on No: :8 .828 no cozeoczeoe— c.
.20382 2% Com ~ oSwE com .

 

25

 

NNNN exaeNN _N N 3%.: ON 3 8N 6N N255
NN §NNV N _ N $.93 N N 8 N 8:35
NN o v N EEC e o R o 8S5
N o z N $9».sz N o 2 N 8&5
v o _ N §NNV _ o _ o 355
N o N N .§q8 o o o o 855
N o o N 31.5 _ o e o 855
8 o N N €53 _ o t o 8S5
Na. 0 N N Axes N c on. o. v
N_ o 2 N axes o o o. _ v
e o o _ €03: _ o N _ 85:0
8 §N$ a o 2 Asia N 6 NN o 83:0
8 o o _ Gee I. N N_ o 83.6
NON o o N. £64; 2 o E 3 83:0
NNN @532 2 N §QNV : 2 SN 2 8E5
8 o o N axes N o 8 N 8E5
2 o o N. 333 _ o : N 835
N o _ N §NNNV N o N o N

.88. News? mmwm Negozeoaom NEE—om £952 use £9:th outer— .ozm

bus—cm meg? wee: gamete, team
um NoSwuovaomm

 

«3.6.5.50 meExeNoEABEmmo :oEmomEoo E530 .m 033.

a rainy season, the extent to which these alates disperse during each rainy season has not
been described for this region. Thus, an accurate explanation of the discrepancy between
the two studies may not be achieved until the seasonal flight pattern of P. corniceps alates
is described for this region.

The largest termite encountered during the two summers of sampling was
Neotermes mona (Banks). This species was originally described from Mona Island (Banks
1919) and has also been reported from Hispaniola (Dominican Republic) (Schefi’rahn et al.
1994) and the Turks & Caicos (Grand Turk, Parrot Cay, and Providenciales) (Schefi‘rahn
et al. 1990). This species was neither widespread nor common, occurring in only 4 of the
15 sites (sites 4, 7, 27, and 28; Figure 1) with densities that ranged from 104-208 nests/ha.
Despite extensive searches on several occasions, this species was not located within
Guanica Forest. Colonies of N. mona were collected exclusively from L. leucocephala.
However, it has also been collected from B. simaruba and Meliococcus bijugatus (Jones
et al. 1995).

Incisitermes bequaerti was collected on only one occasion (Guanica Forest),
which yielded a density estimate of 35 nests/ha for that site (Table 3). This species has
been reported fiom Bermuda, Cuba, Hispaniola (Dominican Republic), Mona, the Turks
& Caicos (Grand Turk and Providenciales), and the US Virgin Islands (St. Croix and St.
Thomas) (see Schefi‘rahn et al. 1994 for compilation of locality references). Another
member of this genus, I. snyderi, has been reported from the subtropical dry forest life
zone of Puerto Rico (Jones & Schefi’rahn unpublished), but was not located during this
study. Both 1. bequaerti and I. snyderi have been listed as more common in the eastern

dry forests of Puerto Rico than that of southern Puerto Rico (Jones & Scheffrahn

26

unpublished). In addition, recent evidence indicates that I. snyderi may actually be I.

incisus (S cheffi’ahn pers. comm).

Rhinotermitidae

Heterotermes tenuis (Hagen) and Heterotermes convexinotatus (Snyder) have both
been reported to occur within the dry forests of southwestern Puerto Rico (Jones &
Schefl‘rahn unpublished). However, differentiating between these two species often
proves problematic considering the vagueness of Snyder’s (1956) taxonomic key. It is still
unclear whether these species represent a single extremely variable species or a complex of
closely related distinct species (Schefii’ahn et al. 1994). Therefore, species identifications
for Heterotermes could not be accomplished, although they presumably are H. tenuis and
H. convexinotatus (Snyder 195 6).

Heterotermes spp. occurred at 13 of the 15 sites (at all sites except 27 and 36;
Figure 1), being one of the most prevalent temrites of the dry forest. Similarly, Jones and
Schefialrn (unpublished) report Heterotermes spp. as being the most abundant
subterranean termites in the subtropical dry forest life zone of Puerto Rico. Nest densities
ranged from 13-167/ha at all sites, but was consistently 13 nests/ha at the three sites
within Guanica Forest (Table 3). However, these estimates should be interpreted with
caution due to the difficulties in determining the extent of the foraging territories of
individual colonies which likely extend to distances of 10 m or greater (Scheffrahn pers.
comm). For instance, these temrites were often collected in SDW with galleries leading
underground. Ifthis situation occurred multiple times within the same 10m x 10m plot

then these termites were treated as foragers from the same colony, and only one colony

27

was recorded for the entire plot. Thus, the results reported here represent conservative
estimates for Heterotermes spp. nest density.

Heterotermes spp. were collected under rocks and in dead wood, including
Pisonia albida, Leucaena leucocephala, and Bursera simaruba. Foraging groups of these
species could often be located by scraping away the light brown tunnels that they
construct on the surface of and within dead wood. These tunnels closely resembled those

of Parvitermes wolcotti which are constructed out of soil (Snyder 1923 ).

T ermitidae

The only soldierless termite species of the subtropical dry forests of Puerto Rico
belongs to the genus Anoplotermes. However, species identifications could not be
assigned for this genus due to the absence of alates within the colonies encountered during
sampling. Alates were collected once within the colony, but the species identification has
not yet been designated (Schefi’rahn pers. comm). It is currently believed that the only
species from this genus that occurs in Puerto Rico is A. meridianus Emerson (Snyder
1956, Jones & Schefi‘rahn unpublished). This species was present at 7 of the 15 sites
(sites 2, 4, 5, 7, 28, 30 and Guanica N; Figure 1), and had nest densities in the range of
69-1000/ha. These densities may overestimate the abundance of this species because
colonies were on occasion assumed to lie within the 2 m x 2 m plots when only foraging
workers were found in the litter layer. Colonies of Anoplotermes sp. were not
encountered during quantitative samng at Guanica Forest, but were discovered during
qualitative searches of the site. In addition to the litter layer, this species was also

frequently found under rocks. .

28

One of the smallest temrites of Puerto Rico’s dry forests is Parvitermes wolcotti
(Snyder). This species was originally described from specimens collected at Boqueron,
Puerto Rico (Snyder 1924), but has since been reported from the British Virgin Islands
(Great Camanoe, Guana, Eustatia, Virgin Gorda) (Schefiiahn et al. 1994). This
subterranean species occurred at 9 of the 15 sites (sites 4, 5, 7, 10, 27, 28, 30, and
Guanica Forest sites S and N; Figure 1) with densities ranging from 69-500 nests/ha. At
the three sites within Guanica Forest the range was 0-417 nests/ha (mean = 243) (Table
3). Once again, these estimates should be interpreted with caution as nothing is known of
the foraging territory for this species. Colonies were located mostly under rocks, but
foraging individuals were on occasion collected from dead wood, including
Pithecellobium unguis-cati.

P. wolcotti workers and soldiers were also collected from soil tunnels and
sheetings covering dead branches in the litter layer and SDW. These could easily be
distinguished from the carton galleries of Nasutitermes spp. based on size and color. P.
wolcotti tunnels were about halfthe width of the tunnels of Nasutitermes spp. and were
light tan in color, while those of Nasutitennes spp. were dark brown. As mentioned
earlier, it was impossible to distinguish P. wolcotti tunnels from those of Heterotermes
spp. based upon visual inspection alone.

Caribitermes discolor had not been previously documented fi'om the subtropical
dry forest life zone of Puerto Rico. This species was recently reclassified from the genus
Parvitermes to the new monotypic genus Caribitermes (Roisin 1996). C. discolor was
collected on only two occasions, both within an area of Guanica Forest that has had

minimal human disturbance over the past 60 years. These occurrences yielded a nest

29

density of 208/ha for the relatively undisturbed site and an average density of 69 nests/ha
for Guanica Forest (Table 3). This species is listed as common in Puerto Rico’s
subtropical moist forest, subtropical wet forest, subtropical rain forest, and lower montane
wet forest life zones (Jones & Schefi‘rahn unpublished). During surveys within the
subtropical wet forest life zone at El Verde, Puerto Rico, C. discolor foragers were found
infesting one in six small (<5 cm diameter) dead branches on the forest floor (McMahan
1996). The environmental factors limiting the range of this species remain unclear.
However, judging from its abundance in the other life zones, moisture probably plays a
major role by limiting the colonization and initial breakdown of wood by other
decomposers such as fungi, which may be a prerequisite for consumption by C. discolor.
One of the most prominent features of the dry forest are the arboreal nests of
Nasutitermes acajutlae and Nasutitennes costalis. These nests ranged in density from
3 .6-18/ha for N. acajutlae and 1.2-4.8/ha for N. costalis at the sites in which they were
recorded during quantitative sampling. N. acajutlae was also more widespread than N.
costalis, occurring at 13 of the 15 sites (at all sites except 10 and 40, Figure 1) while N.
costalis only occurred at 7 sites (sites 2, 4, 7, 10, 16, 27, and Guanica S; Figure 1).
Within Guanica Forest, N. acajutlae nest density ranged fiom 4.8-9.7/ha (mean = 6.5)
(Table 3). N. costalis was not encountered in Guanica Forest while sampling along the
belt transects, however, it was recorded during a qualitative search. Both Nasutitermes
spp. have been reported from the subtropical dry, moist, and wet forest life zones of
Puerto Rico (Jones & Scheffrahn unpublished). The above results contradict earlier
reports (Martorell 1945, Wolcott 1948) that list N. acajutlae (=N. creolina) as scarce in

Puerto Rico. Perhaps, N. acajutlae was often misidentified as N. costalis during these

30

 

Figure 3. Arboreal nest of N. acajutlae in T amarindus indica.

earlier reports, as the two species are very similar. Their distinguishing characteristics
(pilosity of soldier head capsule and abdominal tergites) can not be easily evaluated in the
field.

In comparison to the dry forest life zone, the opposite trend in terms of
Nasutitermes spp. abundance exists in the wet forest life zone of Puerto Rico. At El
. Verde N. costalis appears to be the common Nasutitermes sp. with a density of 4.5
nests/ha (McMahan 1996). Nasutitermes acajutlae has much lower nest densities at El
Verde, with only 0.47 nests/ha. However, nests of N. acajutlae are generally much larger
than those of N. costalis, a trend that is repeated in the dry forest life zone.

The size of the arboreal nests of N. acajutlae was often quite large, with some
exceeding 1 m in height. In fact, this species builds arboreal nests that are among the 1
largest known for termites (Thome et al. 1994). Most large nests of N. acajutlae had an
ellipsoid shape (Figure 3). The volume of N. acajutlae nests was significantly greater than
those of N. costalis (Mann-Whitney: U=161, n1=5, n2=36, P<0.0025; Figure 4a). Nests of
N. costalis generally did not exceed the size of a basketball in the dry forest.

Nest volume has been implemented as a way to estimate Nasutitermes spp. colony
size by determining a termite density-nest volume relationship (W eigert 1970, Clarke &
Garraway 1994). Applying a model created using N. costalis nests from the wet forest life
zone (W eigert 1970) to dry forest nests yields colony sizes ranging from 9,600-79,000
individuals. Much higher colony sizes were estimated using the nest volumes of N.
costalis from the wet forest, however, there is too little data to support any causal
relationships between climatic variables and colony size. Ifthis same model were applied

to the nest volume estimates of N. acajutlae, colony size would range fi'om 12,800-

32

its

=5

nest volume (cnf)
or
i

N A
O O
8 O
O
O O
I
I

 

 

O
"I

 

N. costalis N. acajutlae

W
00
u

1

hei ht (m
'31

1 “ =5.

 

 

N. costalis . N. acajutlae

Figure 4. Nest characteristics of Nasutitermes spp., (A) nest volume and (B) height of
the nest fi'om ground. Vertical bars represent :1 SE of the mean.

33

800,000 termites/nest. A similar analysis by Clarke and Garraway (1994) of N. nigriceps
nests fiom mangrove forests in Jamaica yielded lower estimates of colony size (see below
for taxonomic discussion of N. nigriceps and N. acajutlae).

Even though these models are being applied to other species and life zones, they
may still be a relatively accurate way to obtain estimates of colony size for species that
build carton nests. Their utility depends on whether colony size is mainly determined by
characteristics of the species, physical properties of the nest, environmental variables, or
some combination of these factors. If colony size is solely determined by the physical
properties of the nest, then these models should be applicable to other regions. as well as
closely related species. One suggestion has been that surface area of the nest afi‘ects the

overall rate at which oxygen and carbon dioxide can diffuse throughout the nest, thereby
limiting the number of tenrrites that can be accommodated within the nest (W eigert 1970).

Martorell (1945, 1941) lists 108 tree species that are afi’ected by N. costalis in
Puerto Rico. The present study did not include quantitative sampling to determine which
tree species were being consumed by Nasutitermes spp., as only nest locations were
recorded. N. acajutlae exhibited a preference for the tree species Bucida buceras, Pisonia
albida, and Pilosocereus royenii for its nesting location, while there was not enough data
to determine any preferences for N. costalis (Table 6).

Selection of the nesting site probably depends more on the size rather than the
species of the supporting plant. The mean basal area of the stems supporting N. acajutlae
nests was 0.039 m2. The diameter of a single supporting stem needed to yield such a basal
area is 22 cm. Considering that 95% of the trees in a reference site within Guanica Forest

have a DBH less than 7.5 cm (Murphy & Lugo 1986), suitable nesting locations are

34

Table 6. Location of Nasutitermes spp. arboreal nests, including the frequency for each
tree species or general position if not situated in a live tree.

 

 

 

F regency
Location N. acajutlae N. costalis
Bucida buceras 18 (25.7%)
Pisonia albida 9 (12.9%)
Pilosocereus royenii 7 (10.0%) 1 (16.7%)
Leucaena leucocephala 5 (7.1%) 1 (16.7%)
Clusia rosea 3 (4.3%)
Sweitenia mahogani 3 (4.3%)
Capparis hastata 2 (2.9%)
combination of treesT 2 (2.9%)
dead tree (horizontal) 2 (2.9%)
Erythroxylum areolatum 2 (2.9%)
Erythroxylum rotundifolium 2 (2.9%)
Leptocereus quadricostatus 2 (2.9%)
Acacia famesiana 1 (1.4%)
Calyptranthes pallens 1 (1.4%)
Capparis cynophallophora l (1.4%) 1 (1 6. 7%)
Coccoloba diversrfolia l (1 .4%)
Colubrina elliptica l (1 .4%)
Delonix regia 1 (1.4%)
Ficus cimfolia l (1.4%)
Haematoxylum campechiaum 1 (1.4%)
Pictetia aculeata l (1.4%) 1 (16.7%)
Prosopis pallida 1 (1.4%)
Randia aculeata 1 (1.4%)
T amarindus indica l (1.4%)
on ground 1 (1.4%)
Gymnanthes lucida 1 ( 16.7%)
T abebuia heterohylla l (16. 7%)
Total 7 0 6

 

1 Bursera simaruba/Zanthoxylum flavum and Bursera simaruba/Capparisflexuosa/Pithe-

cellobium unguis-cati.

35

limited to tree species capable of producing adequately large trunks and branches. The
two most preferred trees, B. buceras (26%) and P. albida (13%), are also two of the most
common large (>200m DBH) tree species of the dry forest life zone. Bursera simaruba is
another common tree of the dry forest life zone capable of exceeding 20 cm DBH, but its
wood is very lightweight (specific gravity 0.29) and may not be able to support the very '
large nests of N. acajutlae. It also becomes evident after examining the tree preference
data that these termites do not attack the tree in which their nest is situated, as the pipe-
organ cactus (Pilosocereus royenii) is one of the most preferred species.

The average height of the nest ofi the ground was another variable that was
significantly different between the two Nasutitermes spp. N. acajutlae nests were located
at greater heights in trees than the nests of N. costalis (Mann-Whitney: U=190.5, n1=5,
n2=41, P<0.0005; Figure 4b). Three out of the five N. costalis nests recorded were
located on the ground. Their. nests have also been reported occurring on the ground in
Puerto Rico’s rain forests (W eigert 1970) and in the Turks & Caicos Islands (Schefiahn
et al. 1990). It is unclear why this species seems to be limited to the ground, although
moisture requirements may be a major factor in subtropical dry forest life zones.

N. acajutlae represents a recent taxonomic revision by Thome et al. (1994) that is
still being questioned by some Caribbean termite taxonomists (Jones & Schefi’rahn
unpublished). Previously this species was designated as N. nigriceps (Haldeman) by
Snyder (1956). The current designation is based on the following morphological
characteristics of the soldier: (1) pilosity of the head capsule, (2) indentation shape on the

anterior margin of the pronotum, and (3) the presence of bristles on regions flanking the

36

postmentum (Thome et al. 1994). Thome et al. (1994) proposed an eastern distribution
throughout the Caribbean for N. acajutlae (Puerto Rico east and south to Trinidad, and
Guyana) and a western distribution for N. nigriceps (J arnaica, Cayman Islands, Mexico
south to Panama). Even though the issue is yet to be resolved, N. acajutlae was the name
chosen to represent this species in this paper as it coincides with the current literature
(Schefiahn et al. 1994).

In conclusion, very little is known about the temrites of Puerto Rico’s subtropical
dry forest life zone. With a few exceptions, previous knowledge of these termites was
limited to the comments and observations listed in their original descriptions. This chapter
surrunarized the data and observations that were collected during a quantitative survey of
these termites throughout the dry forest life zone. A species was recorded that is new to
this life zone, while a previously documented species occurrence was not confirmed. Nest
densities were reported for the first time from a representative subtropical dry forest,
Guanica Forest, providing new insight into termite community structure for this type of
ecosystem. The drywood termite, P. corniceps, exhibited a colony composition and diet
similar to previous findings (Jones & Schefi‘rahn 1995, Jones et al. 1995). A comparison
of Nasutitermes spp. revealed many differences in terms of nest size, location, and
abundance. Nasutitermes acajutlae was more common than its congener N. costalis in the
subtropical dry forests of southwestern Puerto Rico, a result which contradicts earlier

reports (Martorell 1945, Wolcott 1948).

37

CHAPTER TWO

Environmental Influences on the Distribution and Abundance Termite Species
in a Subtropical Dry Forest

There are a multitude of environmental factors that afi‘ect the activity, distribution,
and population density of termites. Of these factors, temperature, moisture, and
availability of a food source have been implicated as major contributors (Lee & Wood
1971, Ueckert et al. 1976, Davis & Karnble 1994, Leponce et al. 1995, Cabrera & Rust
1996, Pearce 1997). In addition, certain structural characteristics of the habitat can
greatly influence these variables, resulting in a diverse array of microhabitat conditions.
For instance, temperature extremes experienced by termites are influenced by their ability
to move underground or build soil runways, the amount of shade provided by various
vegetation types (Sands 1965), and the presence of objects on the soil surface (i.e., rocks
and litter) which provide a thermal shadow in the soil below (Lee & Wood 1971,
Ettershank et al. 1980, Pearce 1997).

The subtropical dry forest life zone of Puerto Rico has a mean annual temperature
of 25. 1°C, a ratio of potential evapotranspiration to precipitation that exceeds unity, and
soils that exhibit moisture deficits 10 months out of the year (Murphy & Lugo 1986).
Suitable habitat for temrites in this hot, dry environment may be limited to areas where
enough cover exists to provide refugia from high temperatures and allow the retention of
moisture in the soil for relatively extended periods.

Termites have differing temperature and moisture tolerances, which can vary

among genera and families as well as at the species level (Pearce 1997). For instance,

38

temrites in the family Kalotermitidae build their nests in sound dry wood and do not
require contact with the soil. Their nesting location combined with their resistance to
desiccation (Pearce 1997) would suggest that Kalotermitids are not going to be afi‘ected
by soil moisture deficits as much as other temrites.

The objective of this chapter is to determine whether structural variables of the
habitat that influence its temperature and moisture levels can also affect termite
distributions in a subtropical dry forest. The knowledge gained fi'om such an investigation
coupled with a landscape-level analysis (Chapter 3) is of great importance to conservation
planning. Knowledge of species’ habitat requirements or preferences can be used to limit
the extent of area in a landscape considered as ‘suitable’ and can therefore focus efi’orts
towards habitats with greater conservation value (Scott et al. 1993). The value of such
habitat studies increases when the variables that affect the distribution of a species can be
detected by remote sensing (i.e., percent canopy cover, vegetation type) or determined
from pre-existing spatially referenced data sources (i.e., soil type, elevation, aspect).

Another area of conservation planning involves the use of indicator groups to
predict the occurrence of a wide range of other unrelated groups in order to pinpoint
diversity ‘hotspots’(Kremen 1992, Pearson & Cassola 1992, Kremen 1994, Beccaloni &
Gaston 1995, Abensperg-Traun et al. 1996a). The determination of a potential indicator
group is accomplished by looking for a group of organisms (e.g., butterflies, tiger beetles)
whose species richness is correlated with that of a number of other groups of organisms.
While it is not the objective of this chapter to make a case for termites as biodiversity
indicators (they do not fit the criteria established by Pearson & Cassola 1992), it would

still be useful for conservation purposes to determine whether the diversity of the termite

39

community is correlated to the diversity of other communities of the dry forest life zone.
Therefore, this chapter also investigates whether any correlations exist between temrite

species richness and the species richness of plant and lizard communities.

MATERIALS & METHODS
The Study Region

The southwestern portion of Puerto Rico includes the majority of the island’s
subtropical dry forest life zone. This region extends east along the south coast
approximately 120 km from the southwest comer of the island (Figure 1) and reaches
between 3 and 20 km inland, depending on local topography (Ewel & Whitmore 1973).
Although there is considerable temporal variability, annual precipitation ranges from 600
mm in the west of the dry forest life zone to about 1,000 mm in the east.

The majority of the region is highly disturbed due to industrial, agricultural, and
urban land uses, resulting in the fragmentation of natural habitats. However, Guanica
Commonwealth Forest, situated in the center of the dry forest life zone, represents one of
the best remaining examples of subtropical dry forest in the world (Murphy & Lugo
1990). This area of approximately 4000 ha has been protected since the 1930’s and
encompasses a variety of edaphically determined associations. Three sites within Guanica
Forest were selected in order to obtain baseline data for termite communities in a relatively
undisturbed, contiguous subtropical dry forest. In addition, twelve forest fragments of
various sizes (Figure l) were randomly selected for study from a total of 40 sites in which
the vegetation composition and structure had previously been documented (Ramjohn et al.

unpublished; Appendix A). . A more complete description of the study sites, Guanica

40

Forest, as well as the dry forest life zone of Puerto Rico can be found in the main

Introduction of this thesis.

Nest Density

A quantitative survey 0f the termites of the southwestern dry forests of Puerto
Rico was conducted during the summer months of 1997 (May, June, and July) and 1998
(July and August). The location of the sampling plots (10 m x 10 m) was determined by
randomly selecting points on a grid overlay of the aerial photos of each site. Large
fragments (>10 ha) and sites within Guanica Forest were stratified using four aspect
categories (N-, S-,E-, W-facing slopes) and one category for ravine areas. The number of
samples selected from each category was largely determined by the proportion of the total
area that each represented.

The total number of samples taken from each site was largely a function of area;
the number of samples increased as the area of the site increased (Table 2). However, due
to the amount of time required for each sample (between 1 and 2 hours on average to
search a plot) it was impossible to achieve a constant proportion of area sampled across all
sites (Table 2). Ultimately, the determination of the number of samples taken from each
site was based-on maximizing the number of sites that could be sampled with the given
time constraints while maintaining some degree of consistency for the proportion of area
sampled.

The abundance of termites was determined by exhaustively searching three 2 m x 2
m subplots which were randomly selected within each 10 m x 10 m plot. The majority of

tennite Species encountered build relatively small nests and have foraging territories that

41

generally did not extend beyond the boundary of the 2 m x 2 m subplot. For these species,
the total number of nests recorded fi'om the three nested 2 m x 2 m subplots was used as a
measure their abundance. For termite species that forage great distances (>10 m), the
three nested 2 m x 2 m subplots may lie entirely within the foraging territory of a single
colony (Schefirahn pers. comm). Multiple occurrences of such species (e.g.,
Heterotermes spp., Nasutitermes spp.) within the same 10 m x 10 m plot were treated as

foragers from the same colony, and only one colony was recorded for the entire plot.

- Habitat Variables

Associations between termites and characteristics of the habitat were examined on
two spatial scales. The present study collected habitat structural data within randomly
selected 10 m x 10 m plots located within the forest fragments. The composition and
structure of the plant communities within the twelve forest fragment study sites was
previously described by Ramjohn et al. (unpublished) (see Appendix A) as part of a
related study and was incorporated into this study to investigate the effects of habitat
structure on termite communities at a spatial scale greater than the 10 m x 10 m plots. In
addition, associations between termite and lizard species richness were also examined
using the forest fi'agment as the sampling unit.
10 m x 10 m Plots

A number of variables describing habitat structure and composition were recorded
for each 10 m x. 10 m plot. A compass was used to determine the primary aspect of the
plot using eight direction categories (N, NE, E, SE, S, SW, W, NW) and one category for

plots that were located on flat terrain. The canopy of each plot was classified by visual

42

inspection as either open (<3 3% cover), partially open (34-66%) or closed (67-100%).
For relatively short canopies (< 4 m), ‘ canOpy height was measured with a 2 1n measuring
stick held at a known height above the ground. Canopy height for taller tracts of forest
had to be estimated by visual inspection. Species richness of the woody vegetation (tree,
shrub, and vine) within a plot was accomplished by thoroughly searching each plot and
recording the number of species present:

Other variables that were estimated for each plot include the percentage of rock
covering the ground and the average depth of the litter layer. The percentage of the
substrate within a plot covered by rock outcroppings was estimated by visual inspection.
Litter depth was measured by vertically inserting a metric ruler into the litter layer until it
reached the soil surface. Measurements were taken at five random locations within the
plot to obtain an average litter depth.

Spearman’s rank correlation was used to examine the associations between the
diversity and abundance of temrites and the following habitat variables: canopy height,
percent rock, litter depth, and the species richness of woody vegetation. The percentage
of rock was arcsine transformed in order to approach an underlying normal distribution
(Zar 1984). Correlations were performed between the habitat variables and termite
species richness as well as with the individual densities of the most abundant termite
species fi'om each of the three families. Limiting the analysis to only the most abundant
species in each family was necessary because the rare occurrence of other species
precluded determination of any patterns in their distributions.

The abundances of termite species which nest in similar locations were pooled (see

Table 7) and included in the correlation analysis in order to determine whether any

43

associations exist between these groups and the habitat variables. Drywood temrites refer
to species belonging to the family Kalotermitidae. These species excavate their nest within
their food source, dead wood that can either be on the ground or still standing. The
subterranean group includes temrites fi'om the families Rhinotermitidae and Termitidae.
These species construct their nests within the soil but may forage above ground. The
arboreal group consists of two species from the family Termitidae which build carton nests

on the exterior of trees.

Table 7. The termite groupings used in determining associations with habitat

 

 

variables.

Group Species Included

Drywood termites Incisitermes bequaerti
Neotermes mona
Procrjptotermes corniceps

Subterranean termites Anoplotermes sp.

~ " ' Caribitermes discolor
Heterotermes spp.
Parvitermes wolcotti

Arboreal temrites Nasutitermes acajutlae
Nasutitermes costalis

 

In order to evaluate whether aspect or canopy cover has any influence on the
species richness of termite communities or the abundance of individual termite species, a

Kruskal-Wallis nonparametric analysis of variance (AN OVA) was performed. The

combined total abundance of the species within each of the groups (Table 7) was also

analyzed using the Kruskal-Wallis ANOVA. In all cases with significant ANOVAs,
multiple comparisons were made with Tukey’s Honestly Significant Difference (HSD). In
addition, the diversity and abundance data from plots on primarily north-facing slopes (N,
NE, NW) were grouped together in order to make the comparison with termite
communities on south-facing slopes (S, SE, SW). In general, north-facing slopes in
Puerto Rico’s dry forest life zone receive more precipitation than south-facing slopes
(Ewel & Whitmore 1973) which may affect the distribution of temrites in this region.
Forest .Frgments

Plant species composition as well as vegetative structural data for the twelve sites
were acquired from Ramjohn et al. (unpublished) (see Appendix A). The species richness
of the lizard community within the study sites was iobtained from K. Genet (pers. comm).
The vegetation data were collected during the previous two years as well as the same year
that the termite and lizard data were collected.

Pearson correlation was used to examine individual associations between termite
community attributes (e. g., species richness and individual abundances) and characteristics
of the plant community. The vegetation variables included in the analysis were: (1)
number of plant species, (2) number of tree species, (3) number of stems/ha greater than
1 cm DBH, (4) number of stems/ha greater than 2.5 cm, (5) number of stems/ha greater
than 5 cnr, and (6) mean basal area (mzlha). In addition, associations between termite
richness and the species richness of plants, trees, and lizards within forest fragments were

examined using Pearson correlation.

45

 

 

46

 

 

when Shem 00.0 00.0 00.0 00$ 00.5 005$ K.va nmdmmm vmhm 8:530
2.3: :.v0_ 00.0 00.0 0nd. 0m.N_ hNéo— $0; 05.500 omsw—N Vme Z
00.0 00.0 00.0 00.0 0m.N_ 0m.N_ ova: 0m.N_m enamm 00>va *4th m
00.0 00.0 00.0 00.0 0m.N_ 0m.N_ 00.0 00.0 2;: 50.300— vmhm O
00.0 00.0 00.0 00.0 3.3 mm.mm 00.0 00.0 3.05 00.30 menu N
00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 3.8V NNNR 0.02 v
00.0 00.0 00.0 00.0 00.0 00.0 vvdo 36.00 0N00~ mm.mmw ~60 m
00.0 00.0 560— :40— 0m.w— 0m8m ova: 002m wed; Cams 0.mm b
00.0 00.0 00.0 00.0 Eda 00.0w 00.0 00.0 00.0 00.0 no 0—
00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 Savfl $002 in 0—
00.0 00.0 00.0 00.0 Swm 00.0w 00.0 00.0 0082 00.0mhm mm m—
00.0 00.0 mm.wo~ mm.w0N 00.0 00.0 00.0 00.0 mumNmo 0003— ON mm
00.0 00.0 mm.w0N 2.00m 00.0 00.0 «6.00m mmwom 09903 mm.mw0N 00.0 .8
00.0 00.0 00.0 00.0 00.0 00.0 00.00m 00.00m 00.00m 00.00m 0N0 0m
00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.313 006me 00.0 cm
00.0 00.0 00.0 00.0 .1. 50.02 00.0 00.0 00.0 00.0 000.0 0».
j:— Ammu =8:— G—mv 53:. 950 53.: am 53:. 7:: so...“ 8:
33:03 .N ESE .2 0% amEamBNmEm .0803: N 3.6858 N
Neda

 

coho 03056 n um
.Cmouom 8:20:00 86 8:80.08 2: can Eco—ems.“ “880 S 2: 3 860% 8:58 33.365 No.0 «£38: 00 58:3: :82 .m 2an

.0000 05:0 05 E 00.0800. 000: 28 Eco 003 0005 00005 .80 00008000 000000 0008 005 .0300 0033 98:0 0.80030 ”08 Z

.000 0.05 05 mo 000000 0 so 0080 2.0 0008000 36:00 0000000 005800000 00 No: 0300 5:0 0.00530 2.

 

 

 

05.00 00.00 00.0 00.0 _0._ 00.0 05.00 00.00 00.0 00.0 0050 000020
00.0 00.0 00.0 00.0 05.0 00.0 00.0 00.0 00.0 00.0 0050 Z
5_ .02 2.02 00.0 00.0 05.0 00.0 00.03 00.000 00.0 00.0 0050 0
2.02 5_ .02 00.0 00.0 05.0 00.0 00.0 00.0 00.0 00.0 0050 U
0000— 05.550 2 ._ 0. ._ 00.0 05.0 00.0 00.0 00.00 00.00 0.000 m
05.0: 00.000 00.0 00.0 00.0 05.0 00.0 00.0 00.0 00.0 0.00_ 0
05.0: 00.000 00.0 00.0 00.0 50 .0 00.0 00.0 _ 00.00 00.00 _ _.50 m
05. _ _0 00.000 00.0 00.0 05.0 00.0 00.0 00.0 0000— 00.000 0.00 5
00.000 00.000 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0.0 0_
00.000 50.00 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0.0 0_
00.000 00.000 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 _.0 00
00.0 00.0 00.0 00.0 05.0 05.0 00.0 00.0 00.0 00.0 0.0 00
00.0 00.0 05.0 05.0 05.0 05.0 00.0 00.0 00.0 00.0 00.0 50
00.000 00.000 .2. 00.0 .2. 00.0 00.0 00.0 50.30 00.002 00.0 00
00.0 00.0 .2. 00.0 .2. 50.: 00.0 00.0 00.0 00.0 00.0 00
00.0 00.0 .2. 00.0 .2. 00.0 00.0 00.0 00.0 00.0 000.0 00
A050 5.0:. 40000 50:. 350 .80.: £00 :02: Q00 :00... A05 «0.... 0000
0.008000% 030505 0.08.80 .2 033.0000 .2 32000.00 U .00 ESSBEENV

0100400

 

00:8 .N 635

47

dis

Single-linkage cluster analysis (SYSTAT version 8.0) measuring euclidean
distance was performed using the composition and structure of the plant community at
each site (see Appendix A). The resulting dendrogram was then compared to another
cluster analysis created using the abundance data for the termites present at each site
(Table 8) in order to determine whether vegetative characteristics of the habitat influence
the structure and composition of termite communities in forest fragments. Ifthe two
dendrograms exhibited similar groupings then it was speculated that the structure of the
two communities being compared was. some how interrelated. Sites were not included in
the cluster analysis if they did not have a complete set of both termite and plant
community data. For example, site 40 could not be included in the analysis because mean

basal area was not reported for this site (Ramjohn et al. unpublished).

RESULTS
10 m x 10 m Plots

Nine (32%) of the 28 correlations between the termite diversity and abundance
measures and the habitat variables were statistically significant (Table 9). The height of
the canopy accounted for the majority of these significant correlations. However, all of
the significant associations between the termite and habitat variables were weak; none of
the habitat variables explained more than 12% of the variation in any of the termite
abundance and diversity measures. There were no significant correlations between the

species richness of woody vegetation and the termite variables.

48

 

The amount of can0py cover had a significant efi‘ect on the distribution of certain
termite species within the study sites (Figure 5). The abundance of P. wolcotti, arboreal
species, and termite species richness were greatest in areas that had a closed canopy. In
addition, a number of other termite abundance measures exhibited marginally significant

(P<0.10) difi'erences between the three canopy levels. These include the abundance of P.

Table 9. Spearman’s rank correlation coefficients (r.) between termite diversity and
abundance estimates and habitat variables within the 10 m x 10 m plots (N = 88
plots). ’ = P < 0.01, “ = P < 0.005, ns = P > 0.05.

 

 

 

Termite Variables Structural Variables
Canopy Litter Woody Vegetation
Height % Rocl_< Depth Species Richness
Procryptotermes corniceps 0.228 ns ns ns
Parvitermes wolcotti 0.342“ ns 0.282. ns
Heterotermes spp. ns ns ns ns
Subterranean termites 0305” ns 0.224 0.217
Arboreal termites ns ns ns ns
Drywood termites 0.222 ns ns ns
Termite species richness 0.250 ns 0.216 ns

 

corniceps (Kruskal—Wallis: H = 5.734, P = 0.057), subterranean termites (H = 5.413, P =
0.067), and drywood termites (H = 5.498, P = 0.064). In all of the above cases, the
greatest abundance was in areas that had a closed canopy.

There were no significant differences in any of the termite diversity and abundance
measures between the eight aspect categories. There were also no statistically significant
differences when the termite data from predominantly north- and south-facing slopes were

compared.

49

Arboreal Termite Abundance p. wolcotti Nest Density

Termite Species Richness

Figure 5.

Mean (3: SE) (A) abundance of P. wolcotti, (B) abundance of
arboreal termite species, and (C) termite species richness within
the three canopy categories. Points within a graph with at least
one of the same superscripts are significantly different (P < 0.05,
Tukey’s HSD). See text for canopy category definitions.

 

 

, H '= 9.241

 

 

 

 

 

 

 

P = 0.010
ab
b
a l
o
I l T
H = 7.979
P = 0.019
ab
0
a b
I I
I l l
H = 10.735
P = 0.005
a
a I
E
l l 1
Open Partial Closed

(N = 88 plots)

Canopy

50

 

 

Forest Fragments
When forest fragments were utilized as the sampling unit, termite species richness and the
abundance of Neotermes mona were the only two termite variables that exhibited
significant correlations with characteristics of the plant community (Table 10). All the
significant correlations were relatively strong, each explaining more than 35% of the
variance. The two most widespread species, Procryptotermes corniceps and Nasutitermes
acajutlae, were not significantly correlated with any of the plant community variables.
Termite species richness, in addition to the significant associations with the number of
plant and tree species, was also significantly correlated with lizard species richness (Figure
6).

The two resulting dendrograms produced by the cluster analysis of plant and
termite community data exhibited distinct groups of fiagments (Figure 7). The cluster
analysis of the plant data revealed two large groups, one group consisting of sites 2, 5, 7,
27, and 28, and a second group consisting of sites 4, 10, 13, 16, 30, and 36. Examining
the dendrogram produced by the termite data reveals a large group consisting of sites 2, 4,
5, 7, 10, 16, 27, 28, and 30, and a much smaller group consisting of sites 13 and 36
(Figure 7). Therefore, it appears that the combination of plant community variables
included in this analysis do not greatly influence the structure of the termite community in

a subtropical dry forest ecosystem.

51

Table 10. Pearson correlation coeflicients (r) between termite diversity and abundance
estimates and characteristics of the plant community within forest fragments (N
= 12 fragments). ' = P < 0.01, " = P < 0005, ns = P > 0.05.

 

 

 

 

 

Termite Variables Plant Sp. Tree Sp. No. Stems/ha Mean Basal
Richness Richness >1 cm >2.5 cm >5 cm Area (m2/h_a)_
Anoplotermes sp. ns ns ns ns ns ns
Heterotermes spp. ns ns ns ns ns ns
Neotermes mona ns ns ns 0.756" 0.694 ns
N. acajutlae ns ns ns ns ns ns
N. costalis ns ns ns ns ns ns
P. wolcotti ns ns ns ns ns ns
P. corniceps ns ns ns ns ns ns
Species Richness 0.741. 0.678 0.608 ns ns ns
1 2

a 10 — o

0

= o

r:

i 8 - o o o

8 I

.8 6 __ .

0

a?

3 4 0 o o o

E

r. 2 — .

[—

r = 0.701
0 — ' P = 0.011
I I l I l l l

 

 

 

01'2345678

Lizard Species Richness

Figure 6. Association between termiteand lizard species richness among forest fragments.

52

.0000 E08000

0008.0 05 00 0053 :80 0032005 00008800 008000 000 003.300.» 00:20:80 000.0 0503 000250000 088080009 .5 0.505

0000905

000 000 000 000 000 000 0
_ _ 0 _ _ _ . _

68 =05
6 0 BE
£3 =05
_ . a: one
. ll 00 82
E 090.
.60 090.

Q 090.
E 090.

a: 006
68 :08...
3%

3% E0809...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

900 3058800 0:50...

comm OOON com? Door 000

—

0000205

“-0

_ b P —

 

88:85
68:05
8:005
6:008
8:008
$0090.
£02.95

600005
$090.

 

 

 

 

 

 

H m

 

 

 

 

 

 

 

 

@090.

 

 

 

 

A5090.
00.080.24.00

35 000808“.

0000 3:25:00 000.1

53

DISCUSSION
Termite Diversity

The diversity of termites in the subtropical dry forests of Puerto Rico appears to be
influenced by certain characteristics of the habitat which may be important for providing
cover in this water-limited environment. Termite species richness increased with litter
depth, canopy height, and the amount of cover provided by the forest canopy. It is
difficult to determine whether these results are due to the increased amount of shade
provided by the canopy (and litter for subterranean species), the relative degree of
disturbance these sites have experienced, or if these two factors are inextricably linked.
For instance, the height of the 'canopy is probably a good reflection of the amount of time
since the forest was last cut or burnt, while the amount of canopy cover may indicate the
current degree of disturbance at a site (e.g., fencepost harvesting). Litter would also be
more developed in older, less disturbed tracts of forest. In fact, litter depth and canopy
height have the strongest association of all the correlations examined (r, = 0.583, P <
0.001). Clearly, a more rigorous approach is needed to determine the causal agent of
these observed patterns.

Examination of the associations with plant variables on the scale of entire forest
fiagments revealed that termite species richness increased with the number of stems/ha
that had a diameter at breast height (DBH) greater than 1 cm. In addition to serving as a
food source, trees and their dead branches serve as a substrate in which certain termite
species construct their nests. For instance, drywood termites build their nests in dead
branches and standing dead wood while the Nasutitermes spp. construct large carton nests

on the exterior of trees. Therefore, as the density of stems greater than 1 cm DBH

54

increases the arnount of suitable habitat for drywood termites increases. However, the
large carton nests of Nasutitermes spp. require stems much greater than 1 cm DBH to
provide adequate support (see Chapter 1). Overall, the increase in termite species richness
is likely due to the combination of the increase in suitable habitat for drywood termites and
the increase in the amount of potential food for other termite species.

Even without knowing the causal factors, these results still have implications for
establishing conservation priorities. Tracts of closed canopy forest with relatively high
stem densities support the richest communities of temrites and should therefore be given
the highest priority for conserving these species. Other factors such as aspect, amount of
rock, and mean basal area are not major associates of termite diversity and therefore do
not warrant as much attention if conservation of termites is the main goal of a
management plan.

The amount of readily available food is another consideration when trying to
elucidate the factors that affect the distribution and abundance of termites. However,
termite species feed on a wide variety of plant material which makes the quantification of
food availability for all the termites present within a habitat a rather dimcult task. For
instance, temrites may eat wood (living or dead), herbaceous plant material, plant litter,
humus, fungi, or dung, depending on the species in question (Lee & Wood 1971). While
the variety of diets exhibited by the termites of southwest Puerto Rico was not as diverse
as the list above, as all of them primarily feed on dead wood, there was variability in the
condition and type of wood that each preferred (Chapter 1). For example, P. corniceps
fed on tree species with a wide variety of densities while Heterotermes spp. seemed to

prefer only lightweight woods. In addition, P. wolcotti and C. discolor primarily fed on

55

‘wood that was in an advanced state of decay (pers. obs. ). Therefore, quantification of the
food source for all of the species that comprise the entire termite community would need
to consider the type of wood as well as its stage of decomposition.

In addition to the dificulties associated with quantifying the availability of all the
various types and conditions of wood, the task of measuring the volume of wood litter has
traditionally been problematic (Warren & Olsen 1964). In attempts to make the process
of estimating the amount of wood litter reasonably accurate and less time consuming, a
number of line-intersect sampling methods have been introduced (Warren & Olsen 1964,
Van Wagner 1968, De Vries 1974). However, these techniques fail to include standing
dead wood and are still time consuming if the habitat contains dense understory vegetation
which is characteristic of the subtropical dry forests of southwestern Puerto Rico. The
above considerations led to the omission of food availability as a measured variable from
the present study.

The data fi'om the 10 me): 10 m plots produced an insignificant correlation between
the species richness of termites and ‘woody vegetation, lending support to the notion that
species-rich areas (‘diversity «hotspots’) frequently do not coincide for different taxa
(Prendergast et al.. 1993, Pirnm & Lawton 1998, van J aarsveld et al. 1998). However,
significant correlations between the species richness of termites, plants, trees, and lizards
were obtained when data fi'om entire forest fi'agments were pooled. These results
coincide with another study that investigated patterns of species richness for a variety of
taxa. Abensperg-Traun et al. (1996a) found significant associations between the species
richness of temrites and other taxa (plants, lizards, scorpions) in Eucalyptus woodlands of

the central wheatbelt region of Western Australia. The contrasting results obtained in this

56

study depending on whether data from the 10 m x 10 m plots or entire forest fragments
were used illustrates the importance of selecting the appropriate spatial scale on which to
conduct investigations of this nature. Obviously, the appropriate scale of the study will
depend on the research question being asked. Clearly, if selecting areas to designate as
biodiversity reserves is the main goal of a project, then examining patterns of species
richness among forest fragments is more appropriate than examining these same patterns

among 10 m x 10 m plots.

Termite Abundance

Only a few species were influenced by the habitat characteristics that were
measured in this study. Canopy height and the amount of canopy cover were the most
important habitat variables affecting termite abundance. The nest density of
Procwtotermes corniceps and Parvitennes wolcotti both increased with canopy height.
Again, canopy height may just be a surrogate measure of disturbance, reflecting the time
since the forest was last cleared. The abundance of P. wolcotti and arboreal termites was
also greatest in areas that had a closed canopy. Tracts of closed canopy forest tended to
be mature stands with large trees, perhaps explaining why arboreal termites were found in
these areas at a greater frequency (see Chapter 1 for a discussion of tree size preferences).
Similarly, Leponce et al. (1995) demonstrated that the proportion of trees occupied by
arboreal termite colonies and average nest volume was greater in dense coconut
plantations than in open or standard plantations. The authors suggested that these results
were mainly the effect of varying sun exposure and density of available resources among

the three plantation types.

57

The subterranean temrite P. wolcotti and the combined abundances of all
subterranean species exhibited a positive relationship with litter depth. These results may
reflect a reliance on cover at the soil surface for providing a thermal shadow, however,
there were no significant correlations with percent rock (a more effective barrier of solar
radiation). Once again, the observed association may reflect the influence of disturbance
on these temrite species, assuming that areas with more developed litter layers are less
disturbed. In fact, the relationship between litter depth and the abundance of subterranean
species may have no biological significance at all, and could merely be a spurious
correlation.

The only species that exhibited significant associations with the plant variables
measured for entire forest fragments was the drywood termite Neotermes mona. This was
the largest species encountered during the sampling period of this study and was only
collected from standing dead wood that had a DBH of at least 3 cm. Due to its large size,
N. mona may not be able to excavate nests out of wood that has a DBH less than 2.5 cm.
Therefore, as the density of larger stems (> 2.5 cm) increases, there is a concomitant
increase in the abundance of N. mona. However, this rationale is not supported within the
other drywood termite species. The most abundant drywood species in southwest Puerto
Rico, P. corniceps, did not exhibit any significant relationships with the stem density data.

The composition and structure of the plant community within subtropical dry
forest fragments does not appear to influence the overall structure of the termite
community. The comparison of the two dendrograms did not reveal any patterns which
would indicate that a certain type of plant community supports a termite community

distinct fi'om those within other plant communities. This type of analysis was not the idea]

58

way to compare these two sets of variables, however, the extremely low sample size (12
fragments) precluded the use of more sophisticated multivariate analyses.

In summary, litter depth, height of the canopy, tree stem density, and the amount
of cover within a habitat are important characteristics that influence the diversity of
termite communities and abundance of certain termite species. While it cannot be
determined from this study whether the observed patterns are due to the effects of
shading, disturbance, availability of nesting substrate, or some other unmeasured factor,
the results still have implications for determining conservation priorities. The availability
of food is another major factor to consider when trying to determine the environmental
variables that affect the distribution and abundance of termites. However, this variable is
difficult to quantify in the subtropical dry forests of Puerto Rico and thus was not included
in this analysis. The notion of ‘diversity hotspots’ is supported by the significant
correlations between termite, plant, tree, and lizard species richness when forest fragments
were used as the sampling unit. Due to their scarcity, it was impossible to determine any
relationships between the abundance of rare species and habitat variables. The information '
gained from this habitat analysis will nonetheless provide guidance for refining and

focusing a landscape-level applications.

59

CHAPTER THREE

Contrasting Responses of Termite Community Composition and Wood
Decomposition to Habitat Fragmentation in a Subtropical Dry Forest

The fi'agmentation of terrestrial ecosystems currently represents one of the leading
threats to biological diversity (Wilcox & Murphy 1985, Saunders et a1. 1991,Tumer 1996,
Laurance & Bierregaard 1997). While the majority of studies thus far have supported this
assessment by demonstrating declines in variables such as species richness and densities of
individual species within fiagments (Turner 1996), it is becoming evident that not all
native species are adversely afi‘ected by the fiagmentation of their habitat (Robinson et al.
1992, Margules et al. 1994). Once such species have been identified, the traits that confer
resilience to fragmentation can be ascertained, thus contributing to the development of a
framework concerning species. responses to fragmentation. Though often neglected, this
approach should prove just as beneficial as determining the traits which render species
susceptible to local population decline and extinction (Didham et al. 1998).

The preponderance of fragmentation studies focusing on vertebrates (especially
birds and mammals) has led to the development of generalizations concerning the
relationship between specific traits and responses to fi'agmentation. For example, species
that are large-bodied, poor dispersers, rare, or intolerant of disturbed conditions are
particularly predisposed to local extinction as a result of habitat fragmentation (Turner
1996, Laurance et al. 1997). However, due to the limited number of studies investigating
invertebrate responses to fragmentation, it has not yet been determined (whether such

broad generalities can be extended to this group of organisms.

60

Considering the importance of arthropods in ecosystem functions such as nutrient
cycling, pollination, and seed parasitism (J anzen 1987), it is surprising that there has not
been more of an emphasis on determining their response to fragmentation until recently.
Over the past decade there has been a dramatic increase in the number of studies
investigating the efl‘ects of fragmentation on arthropods (e.g., Klein 1989, Aizen &
Feinsinger 1994a, F onseca de Souza & Brown 1994, Margules et al. 1994, Daily &
Ehrlich 1995, Brown & Hutchings 1997, Didharn 1997, Weisharnpel et al. 1997, Didham
et al. 1998). However, only a few of these studies have actually measured the efl‘ccts of
habitat fiagmentation on the ecosystem processes mediated by the organisms under
investigation (e.g., Klein 1989, Aizen & Feinsinger 1994b, Didharn 1998). In order to
gain a better understanding of the relationship between biodiversity and ecosystem
processes in fragmented habitats, research needs to address the alteration of ecosystem
functions in addition to the changes in species abundance and diversity resulting from
fragmentation (Robinson et al. 1992, Didharn et al. 1996).

Throughout the tropics, termites (Isoptera) are important mediators of ecosystem
processes such as soil turnover (Lee & Wood 1971, Lobry de Bruyn & Conacher 1990)
and nutrient cycling (Peakin & Josens 1978, Wood & Sands 1978, Wood & Johnson
1986). The importance of temrites as decomposers of organic material in savanna
ecosystems has been confirmed (Wood & Sands 1978, Collins 1981, Holt 1990, Jones
1990); although high termite biomass in tropical forests implies a critical role in
decomposition, their significance in these regions has yet to be quantified (Wood &

Johnson 1986). In addition to their contribution to ecosystem processes, termites also

61

have negative impacts as highly destructive pests of residential, agricultural, and
silvicultural interests (Wood 1978, Wood & Pearce 1991, Pearce 1997).

Given the significant role of termites in essential ecosystem processes, it is
imperative for the development of conservation and management guidelines for
fragmented landscapes to determine the impacts of habitat fiagmentation on these
communities. Previous studies of the efl‘ects of disturbance and fiagmentation on termite
communities have yielded similar results: soil-feeding termites appear to be more sensitive
to disturbance than wood-feeding species (Collins 1980, Wood et al. 1982, Eggleton et al.
1995) and more sensitive to fragmentation than litter-feeders and species intermediate
between soil- and wood-feeding (F onseca de Souza & Brown 1994). However, these
investigations looked at the response of termite communities to disturbance events that
had happened less than 15 years prior to the time of the study. In order to allow suficient
time for faunal relaxation, the process whereby the number of species inhabiting an area
approaches some dynamic equilibrium between colonization and extinction rates
(Diamond 1972, 1973, Terbourgh 1974), studies need to focus on areas that have been
disturbed or fi'agmented for decades not years. Information concerning the long term
viability of fiagments and the new equilibrium point that they reach will contribute the
most to developing conservation strategies (Turner 1996).

Due to their favorable climate and productive soils, the dry forests of Puerto Rico
represent an ecosystem that has been historically impacted through human activities
(Murphy & Lugo 1986). As a- result, the landscape currently resembles a mosaic
dominated by agriculture and urban developments with interspersed patches of natural

vegetation. These patches of habitat have been isolated for many decades (at least 60

62

years), therefore making them suitable for investigations into the long term survival of
species in forest fi'agments.

The objectives of this study were to determine: (1) whether termite communities in
subtropical dry forest fragments of varying sizes are distinct from those in a relatively
undisturbed tract of contiguous dry forest, (2) the minimum fragment area required to
support subsets (50 and 75%) of the termite community found in the reference site, (3)
how the composition and structure of termite communities vary relative to characteristics
of fiagments and their surrounding landscape, (3) the impact of fi'agmentation on an
important ecosystem fimction: wood decomposition, and (4) the relative importance of
termites in the decomposition of wood within a subtropical dry forest ecosystem. In
addition, this study examines what is currently known about the biology of the termites
native to the subtropical dry forests of Puerto Rico in an attempt to determine the traits

which may account for the observed responses to fi'agmentation.

MATERIALS & METHODS
The Study Region

The southwestern portion of Puerto Rico comprises the majority of the island’s
subtropical dry forest life zone (sensu Holdridge 1967) (Figure 1). This region extends
east along the south coast approximately 120 km from the southwest corner of the island
and reaches between 3 and 20 km inland, depending on local topography (Ewel &
Whitmore 1973). Although there is considerable year-to-year variation, annual
precipitation ranges fi'om 600 mm in the west of the dry forest life zone to about 1,000

mm in the east. Mean annual temperature is 251° C. The majority of the region is highly

disturbed due to industrial, agricultural, and urban land uses, resulting in the fragmentation
of natural habitats. However, Guanica Commonwealth Forest, situated within the center
of the dry forest life zone, represents one of the best remaining examples of subtropical
dry forest in the world (Murphy & Lugo 1990). This area of approximately 4000 ha has
been protected since the 1930’s and encompasses a variety of edaphically determined
associations.

This study was conducted in the southwestern coastal region of Puerto Rico
(approximately 18 °N and 66 ° 35’W to 66 °12’W). Twelve forest fragments (Figure 1)
were randomly selected for study fi'om a total of 40 sites in which the vegetative
composition and structure had previously been documented (Ramjohn et al. unpublished;
Appendix A). Site selection was stratified such that there were four fragments in each of
the following size categories: “small” (<1 ha), “medium” (1-10 ha), or “large” (>10 ha)
(Table 11). All sites were located within a few kilometers of the coast and to the west of
the city of Ponce. In addition, three sites within Guénica Forest were selected in order to
obtain baseline data for termite communities in a subtropical dry forest (Figure 8). These
sites within Guanica Forest include a coastal scrub forest and upland deciduous forests on

north- and south-facing slopes.

Fragment and Landscape Characterization
Photointerpretation of color aerial photographs (1220,000 scale) taken in February
1998 was used to create land-use maps for the areas surrounding the 12 forest fragments
and the reference site (Figure 8). Polygons were delineated fiom the aerial photographs

by tracing their outlines onto mylar overlays. Each polygon was assigned to one of nine

 

 

acm— cmtmn _m_o_.=t<
0:0. 5:00 6.202 I
0:000; 00000000002 I
0:000; 0000000”. I
000:0.— >00:mo :000 I
000:2 30:00 0000.0 I
.225 D

0:0. .0::0_:0_:0< .

25 93:3 5 :25 I
003 0:3

 

  

 

 

0:0N 00:
Ali 000:0“. >5 $06255

gt)

0:02 05 :0 0:00 0.0— 00080 0001030003 05 .00 0:800 05 050.503:
0010 200:0 .00 002 ”0005 .0 0.8 .0_ .5 3:00.080 0:0 AZ .0 .U .0va 0000 .0030 0020.0 8:005 .00
:00002 05 050205 .0050; 0: 0:0 00.0.8.0 5303:0500 0005.25 .00 BE 00: 0:3 .0 00:05

 

ESoEBE 0.. N o .0.

          

00000“. 02:90

/{

 

65

land-use classes using the criteria specified in Lugo et al. (1996). Photointerpreted
polygons were then digitized into the ARC/INFO version 7.1 geographic information
system (GIS.) package. Once topology was constructed, the digitized data were then
geroreferenced and transformed into Universal Tranverse Mercator (UTM) Grid Zone 19
coordinates to allow accurate measurements and overlay with other geographic data sets.
Adjacent coverages were subsequently edgematched and permanently linked via the
MAPJOIN command. Attribute data, including land-use classes, termite species richness,
and the nest density of each species, were then assigned to each polygon in order to create
a final useable coverage of the 12 forest fragments, the reference site, and their
surroundings.

A number of fragment and landscape characteristics were determined from the
completed coverage in order to investigate spatial patterns of termite species richness and
abundance. Fragment characteristics included in this study were area, perimeter, and
compactness (Table 11). Compactness (K.) is a measure of the shape of a polygon based
on perimeter and area estimates (Bosch 1978, Davis 1986; K, = ZVnarea/perimeter). As
compactness decreases from a theoretical maximum of one (a perfectly circular polygon),
it indicates an increasing perimeter to area ratio and thus a greater exposure of the
fragment core to its edge.

In order to ascertain the degree of isolation and disturbance experienced by the 12
study sites, a number of landscape variables were examined. Using the BUFFER
command of ARC/INFO, the surrounding area within 1000 m of the study sites was
delineated (see Appendix D). This new polygon coverage was then overlaid with land-use

in order to create a new coverage from which queries could be performed concerning the

66

Table 11. Selected fragment and landscape characteristics of the twelve study sites and
the reference site (Guanica Forest).

 

Patch Distance to Nearest Primary

Site-ID“ Ara (ha) Perimeter (m) Compactness Forest Fragment ( m) As_p:ect
Small:

 

40 0.006 27 0.9989 392 S
36 0.06 92 0.9010 110 S
30 0.26 . 226 » 0.7911 , 364 . E
27 0.69 408 0.7191 ‘ 10 W
Medium: .

28 2.0 835 0.5978 7.5 E
13 3.1 - 919 . 0.6821 . 20 Flat
16 3.4 832 0.7868 278 E + S
10 6.3 - 1723 0.5182 15 N
Large:

7 33.6 3517 0.5846 10 All
5 97.1 5258 0.6641 30 All
4 136.8 7562 0.5482 10 All
2 854.1 23277 0.4451 20 All
Guénica' 3724 47885 0.4518 A11

 

* fiom Ramjohn et al. (unpublished); see Appendix A.

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

landscape within 1 km of the site. Within this specified area, the following landscape

variables were measured: (1) number of closed canopy forest fragments, (2) percentage of

closed canopy forest, (3) percentage of open canopy forest, (4) percentage of total forest,

(5) percentage of agriculture, and (6) percentage of urban land-use. The percentage of

urban land-use in close proximity to the study sites was investigated in order to determine

if areas of concentrated light sources (e. g., residential areas, factories, etc.) were acting as

barriers to termite dispersal by attracting reproductives exhibiting positive phototactic

behavior during nocturnal flights (Nutting 1969). The distance to the nearest closed

67

canopy forest fragment was also measured in order to characterize the isolation of the
study sites.

The influence of land-use directly adjacent to the fragments was assessed by
calculating the percentage of the study site perimeter comprised by each land—use. Only
the major types of land-use (closed forest, open forest, urban, and agriculture) were
investigated, as the other land-use types rarely occurred adjacent to the study sites.
Water, artificial barren land, natural barren land, and forested wetland were only adjacent
to two sites each, while non-forested wetland was not adjacent to any of the study sites.
Closed canopy forest separated only by a road from the study sites was considered

adjacent.

Termite Sampling

A quantitative survey of the termites of the southwestern dry forests of Puerto
Rico was conducted during the summer months of 1997 (May, June, and July) and 1998
(July and August). Due to the variability in nest sizes among the different termite species,
two sampling regimes were necessary to obtain accurate estimates of abundance. Two
species (Nasutitermes costalis and Nasuititennes acajutlae) build large, conspicuous
arboreal nests and were sampled using 100 m x 10 m belt transects. The location for
establishing the transects was determined by randomly selecting points on a grid overlay of
the aerial photos of each site. Large fragments (>10 ha) and Guanica Forest sites were
stratified using four aspect categories (N—, S-,E-, W-facing slopes) and one category for
ravine areas. The number of samples selected from each category was largely determined

by the proportion of the total area that each category represented.

68

The remainder of the termites have less conspicuous nesting locations such as
within small pieces of wood or underneath rocks and logs. Nest abundance for these
species was determined by exhaustively searching 2 m x 2 m plots randomly selected from
within the first 10 m segment of the belt transect. All dead wood within the plot, including
standing dead wood (SDW) and attached dead branches up to a height of 2 m, was
sampled for termites. The litter layer was also searched, and all rocks were cleared in
order to detect foraging termites and subterranean nests.

The vast area of some of the sites precluded thorough sampling with the plot and
transect methods. Therefore, qualitative searches of the sites (‘site walks’) were necessary
in order to document species that were not detected within the samples. These searches
were used only to supplement the species richness data for the sites (Table 2).

Whenever possible, soldiers and winged reproductives (alates) were collected from
each colony and field preserved in 80% ethanol for subsequent identification. One
soldierless species (Anoplotermes sp.) was identified to genus using characteristics of the
workers since alates were rarely found within the nest. Identifications were made using
the original descriptions of Snyder (1923, 1924, 1929) and taxonomic key (Snyder 1956).

Abundance estimates for each species represent nest densities (# nests/ha), not the
number of individuals per hectare. Certain assumptions were necessary in order to
estimate the nest densities of subterranean species. For instance, termites belonging to the
genus Heterotermes have foraging territories that likely extend 10 m or more from the
nest (R. Schefirahn pers. comm.) Ifthis species was detected multiple times within the
same 10 m x 10 m plot then these termites were considered foragers fi'om the same

colony, and only one colony was recorded for the entire plot. Thus, the results reported

69

here represent conservative estimates for Heterotermes spp. nest density. The above
assumption was not applied to the other subterranean species (Anoplotermes sp.,
Parvitermes wolcotti, Caribitermes' discolor) as nothing is known of the foraging territory
for these species. The abundance estimates for these species may therefore be positively

biased.

Sample Size

The total number of samples taken from each site was largely a function of area;
the number of samples increased as the area of the site increased (Table 2). However, due
to the amount of time required for each sample (between 1 and 2 hours on average to
search a plot) it was impossible to achieve a constant proportion of area sampled across all
sites (Table 2). Ultimately, the determination of the number of samples taken fi'om each
site was based on maximizing the number of sites that could be sampled with the given
time constraints while maintaining some degree of consistency for the proportion of area
sampled. In order to evaluate the efl‘ectiveness of this sampling design, species
accumulation and performance curves were constructed for each site using data fi'om the
2m x 2m plots. The mean nest density for one of the most common species in the region,

Procnptotermes corniceps, was used to create the performance curves.

Wood Decomposition
At the end of the dry season (August 1997) wood was harvested from two sites
within the subtropical dry forest life zone. Only fresh-cut wood from the common

secondary growth tree species, Leucaena leucocephala (Leguminosae: Mimosoideae),

7O

was collected. This species is very abundant throughout the dry forests of Puerto Rico
and is probably the only tree species common to all of the study sites. The relatively hard,
heavy wood (specific gravity 0.7) has traditionally only had uses as fuelwood in Puerto
Rico (Little & Wadsworth 1964), although this species has tremendous value in
reforestation projects in other regions.

The collected wood was cut into 50 cm lengths and wrapped in wire mesh.
Subsarnples were periodically collected, immediately weighed, and later oven-dried in the
laboratory at 68 °C until a constant weight was attained in order to estimate the dry
weight of the wood. Each bundle consisted of one “log” (diameter between 4-6 cm), two
“branches” (2-4 cm), and two “twigs” (1-2 cm). This criterion for bundle assembly was an
attempt to minimize the initial variability among the bundles. The initial wet weight of the
wood within the bundles ranged from 936 to 2285 g (mean 1; SE, 1545 i- 42 g), while the
estimates of initial dry weight ranged from 639 to 1637 g (mean 1 SE, 1108 i 32 g).

Two types of bundles were assembled, an experimental and control bundle. The
wood in the control bundle was wrapped in a fine aluminum mesh (1 mm) in order to
exclude termites and other macroinvertebrate decomposers. Experimental bundles were
constructed with a wider mesh (5 m) that would not prevent invertebrates from
attacking the wood inside. The two mesh treatments were used in order to ascertain the
relative role of termites and other smaller invertebrates in the decomposition of wood.
Bundles were placed in pairs (control and experimental) around the large nests of the
common arboreal termite, Nasutitermes acajutlae, at one site in each of the three size

categories (<1 ha, 1-10 ha, >10 ha) and at two locations within the reference site (Guanica

71

 

 

 

 

 

 

: /control
\exp
20m
B
Size Categories
<1 ha; l-lO ha >10 ha reference site
Away Near Away Near Away Near Away Near
CECE CECE CECE NestAzCECE
CECE CECE CECE CECE
CECE CECE CECE CECE
NestB: C E C E
C E C E
C E

 

 

 

Figure 9. Design summary of wood decomposition study. (A) Arrangement of wood

bundles around N. acajutlae arboreal nests. (B) Sample size and design for each
of the study sites. C = control bundle, E = experimental bundle.

72

Forest). Three replicates (3 pairs) were placed within 2 m of the base of the tree which
supported the nest and were equidistant fi'om each other (Figure 9a). In addition, three
replicates were placed approximately 20 m fiom the nest, making sure that they were not
near other Nasutitermes spp. colonies (Figure 9a). This stratification of the sampling units
was an attempt to account for the contribution made by the patchily distributed and
numerically dominant N. acajutlae (see Chapter 1) as well as the contribution made by

. other invertebrates to the decomposition of wood. A total of 58 bundles were assembled
and placed in the field (see Figure 9b for experimental design).

Bundles were retrieved approximately one year after they had been placed at the
study sites. The wood was carefully removed from the mesh and termite mud (or carton)
was removed prior to weighing. Once again, subsamples were taken for dry weight
estimation. In addition, the condition of the wood from each size category (log, branch,
twig) within the bundle was noted.

Upon collection of the bundles, the discovery was made that the mesh size of the
control bundles failed to exclude invertebrates, including termites. Therefore, a list of
invertebrates was compiled for each bundle in order to determine if bundles attacked by
temrites lost more weight than those that were attacked by other invertebrate
decomposers (e. g., beetles) or those that weren’t attacked by any invertebrates (natural
controls). Immature insects were identified to the lowest taxonomic unit using the keys
and descriptions of Stehr (1987, 1991), while adults were identified using the taxonomic
keys of Bland and Jacques (1978); In addition, evidence of decomposer activity was
classified as either termite or beetle based on the size and shape of the remnant tunnels

within the wood.

73

Analyses

In order to determine whether termite communities in forest fiagments were
distinct from those in the reference site, a one-way analysis of variance (AN OVA) was
performed. Since it could not be determined if the samples came from a population with a
normal distribution both parametric and nonparametric (Kruskal-Wallis) AN OVA was
used to analyze the data. These AN OVAs were applied to test for differences in species
richness and individual species’ abundances among the three fragment size classes and
contiguous forest sites. In all cases with significant ANOVAs, multiple comparisons were
performed with Dunnett’s test. To determine if community structure varied relative to
area of the fragment, a single-linkage cluster analysis (SYSTAT, version 8.0) measuring
euclidean distance was used to compare individual species densities among the sites (Table
8). The resulting dendrogram was then examined in order to determine if termite
communities clustered together according to size category.

Inspection of the species richness data allowed the-determination of the minimum
fragment area required to support subsets of the termite community found in the reference
site. The proportion of the reference site species composition represented by the species
richness of each fiagment was calculated. The subsets were arbitrarily set at 50 and 75 %
of the species richness within the reference site.

Pearson correlation analysis was used to examine the associations between the
termite diversity and abundance measures and the characteristics of fragments and their
surrounding landscape. Fragment area was log transformed in order to convert its
positively skewed distribution into a symmetrical one and percentage data were arcsine

transformed in order to approach a normal distribution (Zar 1984). Multiple regression

74

was also used to generate predictive models of termite species richness and individual
termite densities for the four most abundant species.

The efi’ect of fi'agrnentation on wood decomposition was analyzed using a one-way
AN OVA to test for difierences in biomass loss fi'om the bundles among the sites (<1 ha,
1-10 ha, <10 ha, reference). Once again, in the case of a significant ANOVA, Dunnett’s
test was used for multiple comparisons. In order to determine whether control and
experimental bundles could be grouped in the analysis, a paired-sample t test was
performed. Similarly, a two-sample t test was used to test for differences in biomass loss
between bundles that were near and bundles that were away from N. acajutlae nests.

Examination of the invertebrate lists from the bundles revealed two major
decomposer groups: termites and beetles. The relative contribution of these groups was
assessed by categorizing the bundles based on the presence and amount or evidence of
each group. The categories used to classify the bundles were as follows: (1) no major
decomposer activity (less than 10 wood-consuming individuals present), (2) termite
dominated activity (temrites present with less than 10 wood-consuming beetles), (3) beetle
dominated activity (wood-consuming beetles present with no termites), and (4)
combination of termite and beetle activity (termites present with more than 10 wood-
consuming beetles). One-way ANOVA (both parametric and nonparametric) was used to
test for differences in biomass loss among the above decomposer categories. Multiple

comparisons were made using Tukey’s Honestly Significant Difl‘erence (HSD).

75

RESULTS
Sampling Eflectiveness

The species accumulation curves indicated that some sites were more adequately
sampled than others (Appendix C). However, the majority of sites were adequately
sampled in terms of termite species richness by thoroughly searching 2 m x 2 m plots. The
only sites which did not demonstrate a clear plateau in their species accumulation curves
were sites 7, 27, and 28.

Exaniination of the performance curves revealed that even though a large
proportion of the small (< 1 ha) sites was sampled (Table 2), the estimates of P. corniceps
abundance may not be as accurate as those obtained fi'om larger sites (Appendix B). For
instance, the mean nest density at site 36 was still fluctuating widely from sample to
sample at the end of the curve. In general, increasing the number of samples as the area of
a site increased resulted in more reliable estimates of P. corniceps abundance. However,
the tremendous number of samples required to obtain relatively accurate density estimates
becomes evident when examining the performance curve for all reference sites combined
(Appendix B). Approximately 50 samples (200 m2 area sampled) were required before the
cumulative mean density of P. corniceps became insensitive to fluctuations in the data.

The abundance estimates for the remainder of the termite species in this study was
not as reliable as the abundance estimates obtained for P. corniceps. Due to their low
frequency of occurrence, performance curves for the other termite species never reached a
point where the cumulative mean was insensitive to fluctuations in the data. For example,

even with all the reference sites combined (96 samples total), the cumulative mean nest

76

Table 12. The occurrence of temlite species within the 12 forest fragments (listed by site
codes) and reference site (Guanica Forest). X = present, detected fi'om plot
and transect sampling methods. P = present, detected during qualitative site

 

 

 

 

search.
<1 ha 1-10 h_a >10 ha reference gig"
SpecieL 4O 36 3O 27 28 13 16 10 7 5 4 2 C S N T
Family Kalotermitidae
Incisitermes bequaerti X X
Neotermes mona X X X P
Procryptotermes corniceps ' X X X X X X P X X X X X X X X
Family Rhinotermitidae
Hererorennes spp." x P P x P x x P P x x x x X
Family Termitidae
Anoplotermes sp. X P X X P X P P
Caribitermes discolor X X
Nasutitermes costalis X X X X X P P
N asutitermes acajutlae X X X X X X X X X X X X X X
Parvitermes wolcotti X X P P X X P X X X
Site species richness l 2 5 5 6 3 4 4 7 5 7 5 3 6 6 8

 

* C = coastal scrub site; S = upland deciduous forest on predominantly south-facing slope;
N = upland deciduous forest on predominantly north-facing slope; T = total of all three
reference sites.

" H. tenuis and H. convexinotatus have both been documented within the dry forests of
southwestern Puerto Rico (Jones & Schefl’rahn unpublished), but could not be
distinguished in this study. Thus, individuals exhibiting traits of this genus were treated
as one species for the purposes of this analysis.

77

density for Incisitermes bequaerti would double with the addition of l nest in a

subsequent sample (Figure 2).

Comparison of Reference and Fragment Communities

A total of 9 species from 8 genera and 3 families were encountered during this
study (Table 12). Only one species (Neotermes mona) was not detected within the
reference site and all but two species (Caribitermes discolor and Incisitermes bequaerti)
were represented in the fragments. Of the seven species present in the fragments, all were
represented in each of the size categories (Table 12). Species richness was not
significantly different between the <1 ha, 1-10 ha, >10 ha, and reference forest sites (P =
0.152, ANOVA; P = 0.173, Kruskal-Wallis). In addition, none of the species had
densities that were significantly difi‘erent among the four size categories (all P > 0.05,
ANOVA and Kruskal-Wallis).

Cluster analysis did not group termite communities according to the size categories
(Figure 10). In fact, none of the small, medium, or contiguous forest sites clustered with
other sites within the same category. Large fragments exhibited a partial clustering of
three out of the four sites with a contiguous forest site included in the group. The
minimum distance occurred between a large fragment and a contiguous forest site (coastal
scrub). The grouping pattern of the dendrograrn did not change when the densities of
specimens that were unable to be identified were included in the analysis, only the

euclidean distances varied between the two dendrograms.

78

Size
Category (site code):

 

8(30)
3(40) —.
M(10) ——'
L(7)
L(2)
L(5)
L(4)
REF(C)
M(28)
M(16)

8(27) —-r
REF(N) -———-J

REF(S)
M(13) I |

3(36)l '

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l H I l 1
0 100 200 300 400 500
Distances

Figure 10. Dendrogram of site similarities based on termite densities fiom the
small (S), medium (M), and large (L) fragments and the reference sites (REF C, S,

N).

79

Minimum Fragment Area
The minimum fragment area required to support 50 and 75% of the species
composition in the reference site was 0.256 ha (site 30) and 1.985 ha (site 28),
respectively (Figure 11). A fragment area of 3.4 ha or more consistently supported at
least 50 % of the reference termite community. There was no area threshold at which the
fragments surveyed in this study consistently supported 75% of the termite community

found in the reference site.

 

 

 

 

1.0
0 08 C.
5,; ' 1 ——————— o ————————— 75%

:
fig 06. co 0 o
‘25 L ——————— 4* ——————— 50%
@5204: o
§§ o
3:!- o.2 —
C
0-0 l T l I l l
-3 -2 -l 0 l 2 3 4
Log (Fragment Area)

Figure l 1. Graph summarizing the fragment area required to support
proportional subsets of the termite community found in the reference
site.

80

Fragment and Landscape Characteristics
The majority of correlations between termite and fragment variables were not
statistically significant (Table 13). Species richness was the only termite variable to be
significantly correlated with the fragment characteristics (area and compactness).
However, fiagment area and compactness are strongly correlated (r = -0.886, P < 0.001),
and therefore should be treated as one fragment characteristic rather than two separate

ODCS.

Table 13. Pearson correlation coeficients (r) among termite diversity and abundance (no.
nests/ha) measures and fi'agment characteristics. ' P < 0.05

 

 

Termite

Variables Area Perimeter Compactness__
Anoplotermes sp. -0. 141 -O. 100 0.127
Heterotermes spp. -0.422 -0.06O 0.365
Nasutitermes acajutlae -0.179 -0.029 0.261
Nasutitermes costalis - 0.086 -0.046 -0.001
Neotermes mona -0.070 -O.219 -0. 151
Parvitermes wolcotti -0. l 13 -0.201 0.088
Procoptotermes corniceps -0.23 8 -0.262 0.291
species richness 0.710' 0.305 -0.728'

 

Only three (5%) of the 56 correlations between termite diversity and abundance
variables and landscape characteristics were statistically significant (Table 14). The
percentage of open forest and total forest in the surrounding landscape was positively
correlated with N. acajutlae and P. corniceps abundance, respectively. None of the
landscape variables were significantly correlated with termite species richness. In addition,

one (3%) of the 32 correlations between the measured termite variables and percentage of

81

Table 14. Pearson correlation coeflicients (r) among termite diversity and abundance (no.
nests/ha) measures and landscape characteristics. ' P < 0.05

 

 

 

 

 

Termite Landscape Attributes’

Variables 1 2 3 4 5 6 7
Anoplotermes sp. 0.484 -0.454 -0.213 -0.252 -0.372 -O.546 0.566
Heterotermes spp. 0.410 0.101 0.147 -0.232 -0.036 -O. 106 0.301
Nasutitermes acajutlae -0.090 -0.014 -0.412 0.585' 0.027 0.225 -0.110
Nasutitermes costalis -0.026 0.053 -0.369 0.028 -0.273 -0.267 -0. 100
Neotermes mona -0.372 0.492 -0.035 0.510 0.265 -0. 148 0.080
Parvitermes wolcotti 0.275 —0.348 -0.337 -0.057 -0.358 -0.573 0.646'
Procoptotermes corniceps -0.213 -0. 127 0.259 0.399 0.603' -0.033 -0.457
species richness -0.516 0.244 -0.045 -0.009 -0.112 0.118 0.093

 

* 1 = distance to nearest closed canopy forest fiagment, 2 = number of closed canopy
fiagments, 3 = percentage of closed canopy forest, 4 = percentage of open canopy, 5 =
percentage of total forest, 6 = percentage of urban land use, 7 = percentage of
agriculture (2, 3, 4, 5, 6, and 7 represent values within 1 km of fragment).

Table 15. Pearson correlation coeflicients (r) between termite diversity and abundance
(no. nests/ha) measures and the percentage of the study site perimeter occupied

by the four major land-use types. 'P < 0.05

 

Adjacent Land-Use

 

Termite Closed Open

Lariable Fore_st Forest Umap Agg' culture
Anoplotermes sp. -0.308 -0.226 -O.239 0.484
Heterotermes spp. -O.139 -0.236 -0. 155 0.413
Nasutitermes acajutlae -0.302 -0.184 -0.220 0.331
Nasutitermes costalis 0.135 0.232 0.029 -0.334
Neotermes mona 0.145 0.597' -0.360 -0.3 57
Parvitermes wolcotti -0. 137 0.084 -0.272 0.345
Procryptotermes corniceps 0. 177 0. 188 -0.422 0. 13 1
species richness 0.387 0.390 -0.499

0.092

 

82

adjacent land-use types was statistically significant (Table 15). Multiple regression
analysis yielded no models that included more than one significant fragment or landscape
variable for predicting termite species richness, P. corniceps abundance, P. wolcotti

abundance, Heterotermes spp. abundance, or N. acajutlae abundance.

Wood Decomposition

Even though invertebrates were not excluded fi'om the control bundles, the smaller
mesh size significantly reduced the amount of biomass loss compared to the experimental
bundles (T = 2.771, df = 28, P = 0.010). There was no significant difl‘erence in biomass
loss between the bundles that were near and the bundles that were away from N. acajutlae
nests (T = -0.391, df = 56, P > 0.05). Therefore, one-way ANOVAs to test for
difi’erences in biomass loss among the sites were performed separately for the control and
experimental bundles, but data fi'om near and away bundles were pooled within each site.
Biomass loss in both the control and experimental bundles varied significantly among sites
(P < 0.05, ANOVA). Experimental wood bundles lost significantly more biomass in the
reference site than they did in any of the fiagment size categories (all P < 0.05, Dunnett’s
test; Figure 12a). In addition, control bundles in the 1-10 ha site exhibited significantly
reduced levels of wood decomposition compared to the reference site (P = 0.007,
Dunnett’s test; Figure 12b).

Identification of the invertebrates collected fi'om the wood bundles

revealed an invertebrate community composed of 5 classes and at least 13 distinct orders

(Table 16). Coleoptera was the most diverse group with 10 identified families and 12

83

 

300
250 — abc

200 - E

150 — b c

T
100 - a

Decrease in Biomass after 1 year (g)

 

 

 

 

 

50 0
0 I l I I
<1 1-10 >10 reference
300
1B
250 -
200 — a

150 - E

100 — i a i

Decrease in Biomass after 1 year (g)

 

 

 

"100 l l l T

<1 1-10 >10 reference

Figure 12. Mean (: SE) loss of biomass from (A) experimental and (B) control bundles at
the <1 ha, 1-10 ha, >10 ha, and reference forest study sites. Points within a graph with at
least one of the same superscripts are significantly different (P < 0.05, Dunnett’s test).

84

Table 16. List of taxa collected from wood bundles including their frequency (the no. of
bundles in which they were found), location within the bundles, and indication
of whether or not they consume wood.

 

 

 

Wood
Taxa No. of bundles Q/o) Locaflrn’ Consumer
Class Malacostraca
Isopoda 2 (3.4%) M, B, L —
Class Arachnida
Acari 3 (5.2%) B, L —
Aranaea 15(25.9%) M, L —
Scorpionidas species A 26(44.8%) M, T, B, L —
species B l (1.7%) L —
Whip Scorpion 2 (3.4%) M —
Class Diplopoda 4 (6.9%) M —
Class Chilopoda 15(25.9%) M, T, B, L —
Class Insecta
Coleoptera:
Anthribidae 1 (1 .7%) B +
Buprestidae l (1.7%) M +
Bostlichidae 2 (3.4%) M, T +
- Cerambycidae:
Cerarnbycinae (adult) 3 (5.2%) M, T, B —
“ (pupa) 5 (8.6%) T, B —
‘ “ (larvae) 18(31%) T, B, L +
Cryptophagidae 1 (1.7%) M —
Cucujidae 30(51.7%) T, B, L —
Curculionidae 1 (1.7%) M —
Erotylidae 2 (3.4%) L —
Scolytidae 17(29.3%) T, B, L +
Tenebrionidae: adult A 9 (15.5%) M, L ?
adult B 3 (5.2%) M ?
unk. larvae 4 (6.9%) L ?
unknown adult #1 3 (5.2%) M, B ?

 

85

Table 16. cont’d

 

 

 

Wood
'l;a_x_a No. of bun_dles (%) Logltion'r Consumer
unknown immature #1 l (1 .7%) B, L ?
unknown immature #2 36(62.1%) T, B, L ?
unknown immature #3 1 (1 .7%) T ?
unknown immature #4 1 (1.7%) L ?
unknown immature #5 1 (1.7%) T ?
unknown immature #6 l (1.7%) B ?
unknown immature #7 1 (1.7%) T ?
unknown immature #8 1 (1.7%) B ?
unknown immature #9 1 (1.7%) B ?
unknown immature # 10 1 (1.7%) T ?
unknown immature # 11 1 ( 1.7%) T ?
Embioptera 7 (12.1%) T, B, L +
Hemiptera: Heteroptera 1 (1 .7%) M —
Hyrnenoptera: Formicidae: species A 9 (15.5%) T, B, L —
species B 1 (1.7%) M, T —
Isoptera: Rhinotermitidae
Heterotermes spp. 23(39.6%) M, T, B, L +
Termitidae
Nasutitermes acajutlae 17(29.3%) M, T, B, L +
Parvitermes wolcotti 3 (5.2%) T, B, L +
Thysanura 6 (10.3%) M, T, B —
unknown insect # 1 31(53.5%) T, B, L ?
unknown insect # 2 : . 1 (1.7%) M ‘7
Fungi _ . 31(53.5%) T, B, L +

 

* M= within mesh of bundles, T = twigs (1-2 cm. diameter), B = branches (2-4 cm
diameter), L = Log (4-6 cm diameter)

86

 

500.

v

300 — b $-

200 _l

100 — a J—
O

0 -l !
'100 I I I I
None Beetles Termites Both

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Decrease in Biomass after 1 year (g)
0

 

 

 

Invertebrate Category

Figure 13. Boxplots of reductions in wood bundle biomass categorized by the
presence/absence and relative amount of the two major wood-feeding invertebrate
groups encountered in this analysis. None = no major decomposer activity (group
1). Horizontal lines show the 10'“, 25'“, 7 5'”, and 90th percentiles increasing up the
ordinate. Open circles represent all data outside the 10“‘ and 90"I percentiles.
Boxes within a graph with at least one of the same superscripts are significantly
difl'erent (P < 0.05,Tukey’s HSD).

87

species that could not be assigned to families. Of the list of invertebrates, only beetles and
termites emerged as potentially significant consumers of wood due to there fiequency of
occurrence and abundance in the wood bundles (Table 16).

The categorization of the wood bundles based on invertebrate decomposer activity
resulted in 4 bundles being classified as having no major activity, 15 as termite dominated
activity, 16 as beetle dominated activity, and 23 as a combination of both termite and
beetle activity. The amount of biomass loss varied significantly among the four
decomposer categories (P < 0.001, ANOVA; P < 0.001, Kruskal-Wallis). Bundles
dominated by termite activity lost significantly more biomass than those classified as
having no major decomposer activity, beetle dominated activity, or termite and beetle
activity (all P <.0.05, Tukey’s HSD; Figure 13). In addition, bundles which were
classified as having a combination of beetle and termite activity lost more biomass than

those classified as having beetle dominated activity (P < 0.05, Tukey’s HSD).

DISCUSSION
T ermite Community Response to Fragmentation
The composition and structure of termite communities did not differ distinctly
between dry forest fi'agments and contiguous forest. Forest fiagments and contiguous
forest were not significantly difl’erent in terms of species richness or abundance of
individual species compared in the context of the entire termite community. Cluster
analysis revealed no separation of the reference forest sites fi'om fiagments, and grouping

patterns among the fiagments were only apparent with the large sites.

88

There are a multitude of possible mechanisms explaining why termites are able to
persist within forest fragments. First, termites may be able to establish colonies within
non-forested land-use types, eliminating the need for long dispersal flights between
forested areas. Four of the seven species found within the forest fiagments have been
reported to occur in structural lumber (Schefl‘rahn et al. 1990), and a fifth species, P.
corniceps, was commonly observed in natural wood occurring within residential areas
(pers. obs.) Ifterrnites are able to become established within the surrounding matrix, then
populations within forest fiagments may be maintained by relatively high colonization
rates.

Second, even if termites are not able to maintain populations within non-forested
areas, perhaps these alternate land-uses do not serve as efi‘ective barriers of dispersal.
While the majority of termite dispersal studies have focused on the environmental variables
that initiate alate swarming, very little information exists on the factors that afi‘ect the
duration of flights or the maximum intrinsic distances that termite species disperse in
natural habitats (Nutting 1969). The few estimates that exist for the dispersal distance of
termites range fiom a few dozen meters (Grassé 1942, Wilkinson 1962) up to a few
kilometers (Grassé 1949). The greatest distance separating any of the twelve forest
fragments of this study fiom other patches of closed canopy forest was approximately 400
meters. If conditions within the non-forested habitats surrounding forest fragments do not
impede termite dispersal, then it is plausible to suggest that alates could traverse the
distances separating the forest fragments surveyed in this study. Kalshoven (1960) and
Nutting (1969) provide some evidence in support of this idea, both observing the flight of

alates within residential areas. However, much more research on the factors affecting

89

termite dispersal in both natural and human-impacted landscapes needs to be conducted
before solid conclusions regarding the resilience of alates to conditions within non-
forested areas can be drawn.

The two previously discussed possible explanations are further supported by the
lack of observed relationships between termite community attributes and landscape
characteristics. For instance, the distance to the nearest fragment was not significantly
associated with any of the termite community measures, while the percentage of the
various surrounding land-uses (directly adjacent or within 1 km) was only significantly
associated with a few of these attributes. Furthermore, none of the species exhibited a
positive relationship with the number of closed canopy forest fragments or the amount of
closed canopy forest within 1 km. Considering that closed canopy forest supported the
most diverse termite assemblages (see Chapter 2: Figure 5), one would expect that if
termites are sensitive to conditions within the matrix then fi'agments surrounded by the
most preferred habitat type. would harbor the most species, or at least higher abundances
of some species.

Third, termites may not be afl‘ected by the altered rnicroclimatic conditions at the
forest edge. In fact, damage to vegetation at the forest edge, either as a direct result of
increased wind exposure (Caborn 1957, Moen 1974, Grace 1977), or an indirect
consequence of elevated evapotranspiration coupled with reduced humidity and increased
desiccation (Tranquillini 197 9, Lovejoy et al. 1986) may increase available food resources
for wood-feeding termites. Since the majority of the temrites encountered in this study fed
predominantly on wood (Chapter 1), these species might be expected to increase in

abundance at the edge or .in very small fi'agments. Although there were no significant

90

correlations, it is still interesting to note that the abundance of six of the seven species
occurring within the fragments exhibited a negative relationship with fragment area (Table
13). Eggleton et al. (1995) also found that wood-feeding termites in a lowland moist
tropical forest were relatively unafi‘ected by deforestation.

Finally, termite communities may appear to be persisting within fragments only
because there has been insuficient time since the fiagments were isolated to reach a new
equilibrium between rates of colonization and extinction. As Parnilo and Crozier (1997)
acknowledge, the population dynamics of social insects operate over a much longer time
scale in comparison with other insect groups. In extreme cases where species have
colonies which may persist for many years with a single queen or by producing
supplementary reproductives (e.g., Nasutitermes spp.), isolated colonies within fi'agments
may resemble the living dead (J anzen 1986). Without immigration into this isolated
population, the species will inevitably become locally extinct. However, even considering
the slow pace of social insect population dynamics, there has likely been sufiicient time to
observe the effects of faunal relaxation within the forest fragments of this study, since the
majority of these fiagments have existed 20 or more years, with some created more than
60 years ago (Lugo et al. 1996).

It is not clear which of these explanations is responsible for the maintenance of
termite communities within the subtropical dry forest fragments of southwestern Puerto
Rico. In fact, the appropriate explanation may change depending on the species under
consideration. For instance, not all of the species in this study feed on wood (e. g.,
Anoplotermes sp.) and may therefore be less resilient to the altered microclimate at the

forest edge. Moreover, the persistence of termite populations within forest fragments

91

probably results fiom a combination of the above explanations rather than one acting in
isolation. For example, even if extinction rates for populations of wood-feeding termites
were not significantly increased by edge efi‘ects, immigration might still be required to
prevent stochastic extinctions of populations within the fiagments.

Termites and social insects in general may be'more vulnerable to stochastic
extinctions than they appear based on numbers of individuals. The reproductive division
of labor within colonies means that the efl‘ective population size may be small even though
the number of individuals is great (Pamilo & Crozier 1997). Considering the extensive
foraging territories maintained by some species (e.g., Nasutitermes spp., Heterotermes
spp.), the effective population size may be very low within fiagments due to the limited
number of colonies they can support. This suggests that immigration probably has a
significant role in maintaining population viability for such species (Wright & Hubbell
1983).

One of the most frequently implicated traits believed to confer resilience to the
various impacts of fragmentation is generality (Willis 1979, Aizen & Feinsinger 1994a,
Margules et al. 1994, Turner 1996). Polyphagy seems to be the rule rather than the
exception among the termites (Lee & Wood 1971, Wood & Johnson 1986). The most
prevalent species among the fi'agments in this study, P. corniceps, consumed wood fi'om a
number of trees, shrubs, and vines which spanned a broad range of densities (see Chapter
1: Table 4). While there were insuficient data to determine whether the rest of the
species in this study were generalists, there was no evidence proving otherwise.

The two species not represented in the fiagments were rare even in the contiguous

reference forest. C. discolor and I. bequarti were detected in less than 2% of the samples.

92

Failure to detect these species within the fragments may be the result of inadequate
sampling. However, another rare species, N. mona, was found at three of the fiagments,
two of which had only 12 samples taken at each. A possible reason for C. discolor
absence fi'om the fi'agments is that it may be sensitive to the altered conditions within
forest fragments. This species is common in the moist and wet life zones of Puerto Rico
and was previously not reported to even occur within the dry forest life zone (Jones &
Scheffrahn unpublished). Therefore, perhaps only relatively undisturbed tracts of dry
forest with suitable canopy and litter conditions allow this species to exist in the dry forest
life zone. In fact, C. discolor was only sampled within an area of Guanica Forest that was
selected as a study site because it had received a minimal amount of disturbance during the
past 60 years (Vélez Rodriguez 1995a-f, Lugo et al. 1996).

Previous studies of the efi‘ects of forestfragmentation on termite communities have
yielded differing results than those of the present study. Fonseca de Souza and Brown
(1994) discovered that central Amazonia forest fragments supported considerably fewer
termite species than contiguous forest. However, the loss of species within the fiagments
was largely attributable to the decrease of soil-feeding termites, while other feeding guilds
were less afieaed. The present study encountered only one soil-feeding species
(An0plotermes sp.), while the rest primarily consumed wood. Therefore, the contrasting
results of these two studies could possibly be ascribed to the distinct composition of the
termite communities each examined. Another study conducted in the wheatbelt region of
western Australia also demonstrated significant declines in termite species richness as a

result of habitat disturbance and fiagmentation (Abensperg-Traun et al. 1996b).

93

However, habitat disturbance in the form of grazing and invasion by weedy species was

determined to be the major factor contributing to the observed decline in species richness.

Effects of Fragmentation on the Decomposition of Wood

Rates of wood decomposition appear to be markedly afi’ected by the consequences
of forest fiagmentation. It is difficult to determine whether these results can be attributed
to fragment area or the influence of edge efl‘ects. Due to the small size of the <1 ha and l-
10 ha sites, the array of bundles at these sites was within 20 m of the forest edge. In the
>10 ha and reference forest sites, bundles were more than 100 m from the edge of the site,
but were not greater than 50 m from internal fragmenting factors (e. g., roads, right-of-
ways, etc). Since no studies have investigated the extent of edge penetration into
subtropical dry forest fragments, it cannot be assumed that bundles at the >10 ha and the
reference forest sites were beyond the area of altered microclimatic conditions associated
with the forest edge.

Examining the invertebrate decomposer data reveals that bundles classified as
predominantly attacked by termites were mostly from the reference site. Conversely,
bundles classified as predominantly attacked by beetles were mostly fi'om the fi'agments.
Considering the relative contributions made by beetles and temrites (Figure 13), the
differences in decomposition rates probably reflect the difl‘erences in the decomposer
community at the sites. These results are diflicult to reconcile with those of the
community comparisons among the sites and the associations between species abundances
and fragment area. However, a closer examination of the four sites included in the

decomposition investigation (sites 2, 16, 30, and ref S) reveals highly disparate termite

94

communities among these sites (Figure 10). In addition, only three termite species
contributed to the breakdown of wood within the bundles. All three of these species had
relatively high abundances in the reference site, whereas the fiagments only supported one

or two of these species (Table 17). In fact, the species that was found in the most

Table 17. The mean density (no. nests/ ha) of the three termite species contributing to the
decomposition of wood within the bundles at the four study sites.

 

 

 

Parvitermes Heterotermes spp. Nasutitermes
site ar (Lia) wolcotti acaz‘utlae
30 0.26 500 0 3.9
16 3.41 0 0 4.6
2 854.1 1 0 33.3 4.8
refS 3724.75 312.5 12.5 9.7
No. of bundles
attacked 3 23 17

 

bundles, Heterotermes spp., was not even present in the two sites which exhibited the
lowest decomposition rates. An experimental design which encompasses more study sites
is needed to resolve the conflicting responses to fiagmentation exhibited by the termite
community and wood decomposition.

Regardless of the mechanisms responsible, alterations of wood decomposition
rates within small (<10 ha) fragments have important implications for maintaining
ecosystem functions. Declines in the breakdown rate of wood can lead to accumulation of
woody debris on the forest floor, increased retention time of nutrients in the litter,
decreased nutrient availability to plants, and consequent reductions in primary

productivity. Ultimately, the synergistic relationship of these processes may set small

95

 

fragments on a trajectory of gradual deterioration. However, long-term investigations into

this matter will be required before the fate of small fiagments can accurately be predicted.

Conservation Priorities

In general, termite species richness increased relative to fiagment area, suggesting
that large fiagments should receive the highest priority when considering which areas to
protect and manage. The protection of large fragments will also be required in order to
preserve species which may be sensitive to forest fiagmentation (e.g., C. discolor).
However, fiagments as small as 0.3 ha are able to support relatively diverse assemblages
of termites and should also be incorporated into landscape level management strategies.
The percentage of statistically significant relationships between termite community
attributes and landscape characteristics was low (less than 5% without being corrected for
multiple comparisons). Therefore, no recommendations can be made concerning the
surrounding landscape of forest fragments. These results should be interpreted with
caution due to the limited number of sites investigated and the low variability among the
sites with respect to their surrounding landscapes (i.e., three of the twelve sites were
completely surrounded by agriculture), which may preclude determination of existing

relationships.

Conclusions
This study demonstrated that related components of an ecosystem can respond in
difi‘erent ways to habitat fi'agmentation. Although the composition and structure of

termite communities were unafl’ected by fragmentation, a process largely controlled by

96

termites appeared to be altered. These results indicate the problems with inferring
mechanistic responses to fragmentation based on changes in community composition. In
addition, the difl’ering contributions of species to important ecosystem filnctions
necessitates the inclusion of population estimates into fiagmentation studies if the factors
responsible for altering these processes are to be elucidated.

This study emphasizes the importance of forest fiagments for maintaining native
populations within human-altered landscapes. The protection and management of forest
fi'agments will supplement the genetic diversity maintained by termite populations in
Guanica Forest, the only contiguous tract of . subtropical dry forest left in southwestern
Puerto Rico. Additional studies of this nature will greatly contribute to the management
and conservation of biodiversity in a landscape that is inevitably dominated by

anthropogenic influences.

97

APPENDIX A

Compositional and structural data for the vegetation at 40 subtropical dry forest fragments
located throughout southwestern Puerto Rico and three forest types within Guanica
Commonwealth Forest (from Ramj ohn et al. unpublished). Sites included in the present

 

 

 

 

study are in bold print.
Site Area Plant Species Tree Species No. of Stems/ha Mean Basal
Richness Rich_n_e_ss >1 cm >2.5cm >5cm Area (mtha)
1 750 173 87 24825 8600 1550 17.5
2 854 128 67 18800 7600 . 2160 18.4
3 98 28.8
4 137 153 73 1 1861 5687 1357 16.5
5 97 152 79 21680 7120 2840 30.7
6 68. 1 96 40 4600 2733 1200 17.9
7 33.6 149 . 61 25400 8600 2760 26.1
8 9.4 102 42 8333 4600 2900 23 .0
9 5.9 51 26 35840 4960 160 9.5
10 6.3 101 42 8067 3800 1667 17.0
11 3.1 108 52 12250 6300 1450 23.3
12 3 82 33 11752 5214 1239 13.1
13 3.1 69 31 l 1371 5029 1029 9.5
14 2.7 84 46 12800 6050 1700 15.4
15 2.1 71 25 13244 4790 2647 23.0
16 3.4 69 34 10100 2900 967 11.7
17 1.7 59 38 16867 2467 267 5.1
18 1.5 84 37 24686 9543 1600 17.8
19 1.5 62 27 6400 4700 2250 23 .7
20 1.5 47 16 5300 3750 1900 14.8
21 1.2 104 40 6067 3933 1133 10.9
22 1.2 70 44 25543 4571 800 12.5
23 1 65 23 5143 3000 2000
24 1 30 13 6800 3600 700 10.2
25 0.98 98 46 10200 5300 2800 20.8
26 0.9 74 27 7479 4790 2647 14.1
27 0.7 71 27 17067 10267 6333 33.3
28 2 68 31 15840 8240 2880 16.5
29 0.2 44 22 9467 6400 3600
30 0.26 76 36 5134 2723 804 15.7
31 0.1 1 63 34
32 O. 1 41 15

 

98

 

APPENDIX A (cont’d)

 

 

 

 

 

Site Area Plant Species Tree Species No. of Stems/ha Mean Basal

Richness Richness >1cm >2.5cm >5cm Area (mzlha)
33 0.09 27 7 4900 2100 1000 5.6
34 0.07 58 22 2257 1285 771 25.4
35 0.07 46 18 12661 8257 4771 23.0
36 0.06 70 32 7717 3696 1087 19.4 E“
37 0.02 29 12 2750 1800 1500 f;
38 0.01 17 6 20583 9750 2833 '
39 0.01 28 10 10900 5300 3200
40 0.006 21 7 7000 5000 3833
Guanica Forest“: ._
Semi-evergreen forest 1 1275 5125 1800 20.5 E"
Deciduous forest 24857 9857 1905 16.9
Scrub forest 18133 8733 4467 25.8

 

T Richness data not available.

99

APPENDIX B

Performance curves for Procoptotermes corniceps nest density (# nests/ha) at the 12
forest fiagments and reference site (Guanica Forest). Sites where P. corniceps was not
detected by the plot sampling procedure have not been included.

Small (< 1 ha) Fragments:

Cumulative mean number of nests/ha

 

site 36-(0.06 ha)

 

 

 

800-

 

 

Curnulative mean number of nests/ha

 

I I I I

0 4 8 12 16

Cumulative area sampled (m2)
12000

 

site 30 (0.26 ha)

 

 

I I I I

12 16 20 24

Cumulative area sampled (m2)

 

10000 -

  
  
  
  
  

8000 0
6000 -
4000 -
2000 -

Cumulative mean number of nest/ha

 

0

site 27 (0.69 ha)

 

 

 

I

0102030405060

T I I I

Cumulative area sampled (m2)

100

 

Medium (1-10 ha) Fragments:

Cumulative mean number of nest/ha

 

w

1000-

 

   

0

site 28 (2.0 ha)

 

 
 

Olmulative mean number of nest/m

 

   

I I I

 

10 20 30 40 50 60
Qumrlativeareasanpleumz)

2500

 

Site 13 (3.1 ha)

“i???“

 

 

 

 

O I I I I I

0 10 20 30 40 50 60
Cmmlativeareasanpledmz)

 

i

Cumulative mean lumber of nests/ha
l

 

0

site 16 (3.4 ha)

 

 

 

10 20 30 40 50 60
Omrlativeareasanpledmz)

101

 

Lar_'ge (> 10 ha) Fragments:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

31130 1G!) I
site7(33.6ha) site5(97.lha)
g 25n— % g m_
._ 2000— I ,_
0 I 0 60% '
I
E 15004 E N
t W! 2 “°— .1
.9 ,9 '
'3 6
3 w_ E 2m—
0 I I I I I 0 I I I I I T
0 20 40 60 80 1C!) 120 0 20 40 60 80 1m120140160
Omllative area sampled (m2) Oarmlative area sampled (m?)
25C!) 12D a,
site4(l36.8ln) . siteZ(854.lln)
5 20104 1W7
E i aco-
° 15004 °
g g em-
a 1CDO— g I
9 400—
,3 J g
g 500 3 200~
0 I I I I I I 0 I I I I I I
0 20 40 60 801m1201401m 0 20 40 60 801(0120140160

Olrmlatiw area mpled (m2) Oerlative arm mnpled (m2)

102

Reference Site (Guanica Forest):

Cumulative mean number of nests/ha

4000

3500 ~
3000 —
2500 4
2000 ~
1500 a
1000 7

500 4

 

 

0

 

CoamlSerub

 

Cumulative mean number of nests/ha

 

     

I I I I I

20 40 60 80 100120

Cumulative area sampled (012)

4000

3500 ~
3000 —
2500 -
2000 J

1500

1000—
500—

0

 

 

0

Upland Deciduous (south)

    
 
 
 
   
 

 

 

I I I T I

20 40 6O 80 100120

Cumulative area sampled (m2)

 

3000
2500 -
2000 -
1500 -
1000 - /

5000

Cumulative mean number of nets/ha

 

Upland Deciduous (north)

,J’

 

“1..

I

I

 

I

0 20 40 6O 80 100120

Cumulative area sampled (m2)

103

 

 

Reference Sites Combined

350

 

 

 

 

4000

.2. ___|I..0
000 COMO
was was

2:38: .20 30:5: :02: 0330an

2000 0

Cumulative area sampled (m2)

104

APPENDIX C

Species accumulation curves for the 2 m x 2 m plots at the 12 forest fragments and
reference site (Guanica Forest).

Small < 1 ha Fra ents:

 

 

 

 

 

 

 

 

 

 

 

 

3 3
g site 40 (0.006 ha) .g site 36 (0.06 ha)
0 2%
“5 2 “5 2 2
a 7 is
.9 .0
S S-
t: t:
.3 ,2
a 1 - 4 o 5 1 2 1 a 4
.2 .2
:3 :3
E 5
U o
0 '7 I T T 0 I I. I I
0 4 8 1 2 16 0 4 8 12 16
Cumulative area sampled (m2) Cumulative area sampled (m2)
5 6
site 30 (0.26 ha) site 27 (0.69 ha)
4 - o 0 o 0 5 ‘

 

Cumulative number of species
Cumulative number of species

 

 

 

 

      

 

 

0 I T I I I g I 0 I I I I I
0 4 8 12 16 20 24 0 10 20 30 40 50 60
Cumulative area sampled (m2) Cumulative area sampled (m2)

105

Medium (1-10 ha) Fragments:

 

 

site 28 (2.0 ha) site 13 (3.1 ha)

Cumulative number of species
Cumulative number of species

 

 

   

 

 

 

 

0 T I I I , I 0 I I I I I
0 10 20 30 40 50 60 0 10 20 30 4O 50 60
Cumulative area sampled (m2) Cumulative area sampled (m2)
3 4

 

 

site 10 (6.3 ha)

site 16 (3.4 ha)

1.1

Cumulative number of species
Cumulative number of species

 

 

         

 

 

 

 

 

I I I I 0 I I I I I
0 10 2030 40 5060 0 10 20 30 40 50 60

Cumulative area sampled (m2) Cumulative area sampled (m2)

106

L_r_l_rge1> 10 ha) Fragments:

 

 

site 7 (33.6 ha) site 5 (97.1 ha)

Cumulative number of species
Cumulative number of species

 

 

 

 

 

 

 

 

 

0 I I I I T 0 I I I I I I F
0 20 40 60 80 100 120 0 20 40 60 80 100 120 140 160
Cumulative area sampled (m2) Cumulative area sampled (m2)
5 6
site 4 (136.8 ha) site 2 (854.1 ha)

Cumulative number of species
Cumulative number of species

 

 

 

 

 

 

 

I I T I I I I 0 I I I I I I f
0 20 40 60 80 100120140160 0 20 40 60 80 100120140160
Cumulative area sampled (m2) Cumulative area sampled (m2)

107

Reference Site (Guanica Forest):

Cumulative number of species

 

 

 

Coastal Scrub

 

Cumulative number of species

 

I. *0

 

I I I
20 40 60

Cumulative area sampled (m2)

Cumulative number of species

6

5

80 100 120

 

Upland Deciduous (south)

 

 

 

I I 7 I I

0 20 40 60 80 100120

Cumulative area sampled (m2)

 

Upland Deciduous (north)

 

 

0 “f I I I
0 20 40 60 80 100 120

Cumulative area sampled (m2)

108

 

I I

Cumulative number of species

O-KNQAUICDNO

 

 

 

Reference Sites Combined

 

 

 

I I
50 100

Cumulative area sampled (m2)

150

109

200

I

250

300

350

APPENDIX D

   
     
 

I
0.5 0 0.5 1 Kilometers

 

110

Land-use map illustrating an exam le of the 1 km landscape surrounding a fragment.

I

Land-use

- Urban or built-up land
Agricultural land
1:] Water

- Closed canopy forest
- Open canopy forest
- Forested wetland
- Non-forested wetland
- Natural barren land
Artificial barren land

 

REFERENCES

Abensperg-Traun, M., G. W. Arnold, D. E. Steven, G. T Smith, L. Atkins, J. J. Viveen,
and M. Gutter. 1996a. Biodiversity indicators in semi-arid, agricultural Western
Australia Pacific Conservation Biology, 2: 375- 389.

Abensperg-Traun, M. G., G. T. Smith, G. W. Arnold, and D. E. Steven. 1996b. The
efl’ects of habitat fragmentation and livestock-grazing on animal communities in
remnants of gimlet Eucalyptus salublis woodland in the Western Australian
wheatbelt. I. Arthropods. Journal of Applied Ecology, 33: 1281-1301.

Aizen, M. A. and P. Feinsinger. 1994a. Habitat fiagmentation, native insect pollinators,
and feral honey bees in Argentine “Chaco Serrano”. Ecological Applications, 4:
378-3 92.

Aizen, M. A. and P. Feinsinger. 1994b. Forest fragmentation, pollination, and plant
' reproduction in a ChaCo dry forest, Argentina. Ecology, 75: 330-351.

Banks, N. 1919. Antillean Isoptera. Bull. Mus. Comp. Zool. 62: 475-489.

Beccaloni, G. W. and K. J. Gaston. '1995. Predicting the species richness of neotropical
forest butterflies: Ithomiinae (Lepidoptera: Nymphalidae) as indicators. Biological
Conservation, 71:77-86. '

Birdsey, R. A. and P. L. Weaver. 1982. The forest resources of Puerto Rico. Resource
Bulletin SO-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 SO-331. New Orleans, LA.

Bland, 'R. G. and H. E. Jacques. 1978. How to know the insects, third edition. Wm. C.
Brown Company Publishers, Dubuque, IA.

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

Brown, K. 8., Jr. and R. W. Hutchings. 1997. Disturbance, fiagmentation, and the
dynamics of diversity in Amazonian forest butterflies, pp. 91-110 in W. F.
Laurance and R. O. Bierregaard, Jr. [eds] Tropical forest remnants: ecology,
management, and conservation of fragmented communities. The University of
Chicago Press, Chicago, IL.

111

Brown, S. and A E. Lugo. 1982. The storage and production of organic matter in
tropical forests and their role in the global carbon cycle. Biotropica, 14: 161-187.

Caborn, J. M. 1957. Shelterbelts and microclimate. Forestry Commission Bulletin 29.
Her Majesty’s Stationary Ofice, Edinburgh, Scotland.

Cabrera, B. J. and M. K. Rust. 1996. Behavioral responses to light and thermal gradients
by the western drywood termite (Isoptera: Kalotermitidae). Environmental
Entomology, 25: 436-445.

Clarke, P. A. and E. Garraway. 1994. Development of nests and composition of colonies
of Nasutitermes nigriceps (Isoptera: Termitidae) in the mangroves of Jamaica.
Florida Entomologist, 77: 272-280.

Collins, N. M. 1980. The efl‘ect of logging on termite diversity and decomposition
processes in lowland dipterocarp forests. Tropical Ecology and Development,
198: 113-121.

Collins, N. M. 1981. The role of temrites in the decomposition of wood and leaf litter in
the Southern Guinea Savanna of Nigeria. Oecologia, 51: 389-399.

Corlett, R. T. and I. M. Turner. 1997. Long-term survival in tropical forest remnants in
Singapore and Hong Kong, pp. 333-346 in W. F. Laurance and R O. Bierregaard,
Jr. [eds] Tropical forest remnants: ecology, management, and conservation of
fi'agmented communities. The University of Chicago Press, Chicago, IL.

Daily, G. C. and P. R. Ehrlich. 1995. Preservation of biodiversity in small rainforest
patches: rapid evaluations using butterfly trapping. Biodiversity curd
Conservation, 3. 35- 55.

Davis, J. C. 1986. Statistics and data analysis in geology, second edition. John Wiley,
New York, NY.

Davis, R. W. and S. T. Karnble. 1994. Low temperature effects on survival of the eastern
subterranean termite (Isoptera: Rhinotermitidae). Environmental Entomology, 23:
1211-1214. ..

De Vries, P. G. 1974. Multi-stage line intersect sampling. Forest Science, 20: 129-133.

Diamond, J. 1972. Biogeographic kinetics: Estimation of relaxation times for avifaunas
of southwest Pacific islands. Proc. Nat. Acad Sci. USA, 69: 3199-3202.

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

112

 

Didharn, R. K. 1997. The influence of edge efl‘ects and forest fragmentation on leaf litter
invertebrates in central Amazonia, pp. 55-70 in W. F. Laurance and R. O.
Bierregaard, Jr. [eds] Tropical forest remnants: ecology, management, and
conservation of fiagmented communities. The University of Chicago Press,
Chicago, IL.

Didharn, R. K. 1998. Altered leaf-litter decomposition rates in tropical forest fragments.
Oecologia, 116: 397-406.

Didharn, R K., J. Ghazoul, N. E. Stork, and A J. Davis. 1996. Insects in fiagmented
forests: a functional approach. Trends in Ecology and Evolution, 11: 25 5-260.

Didham, R K., P. M. Hammond, J. H. Lawton, P. Eggleton, and N. E. Stork. 1998.
Beetle species responses to tropical forest fiagmentation. Ecological
Monographs, 68: 295-323.

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

Eggleton, P. and D. E. Bignell. 1995. Monitoring the response of tropical insects to
change in the environment: troubles with termites, pp. 474-497 in R. Harrington
and N. E. Stork [eds] Insects in a changing environment. Academic Press, .Inc.,
London.

Eggleton, P., D. E. Bignell, W. A. Sands, B. Waite, T. G. Wood, and J. H. Lawton.
1995. The species richness of temrites (Isoptera) under differing levels of forest
disturbance in the Mbalmayo Forest Reserve, southern Cameroon. Journal of
Tropical Ecology, 1 1: 85-98.

Ettershank, G., J. A. Ettershank, and W. G. Whitford. 1980. Location of food resources
by subterranean temrites. Environmental Entomology, 92645-648.

Ewel, J. J. and 'J. L. Whitmore. 1973. The ecological life zones of Puerto Rico and the
US. Virgin Islands. USDA Forest Service, Institute of Tropical Forestry, Rio
Piedras, PR, ITF-18. 72 pp.

FAO. 1993.. Forest resource assessment 1990 - tropical countries. Forestry Paper No.
112. Food and Agriculture Organization of the United Nations, Rome.

Farnsworth, E. J. 1993. Ecology of semi-evergreen plant assemblages in the Guanica dry
forest, Puerto Rico. Caribbean Journal of Science, 29: 106-123.

Fonseca de Souza, O. F. and V; K. Brown. 1994. Efiects of habitat fragmentation on
Amazonian termite communities. Journal of Tropical Ecology, 10: 197-206.

113

Gillot, C. 1995. Entomology, second edition, pp. 163-173. Plenum Press, New York,
NY.

Grace, J. 1977. Plant response to wind. Academic Press, London.

Grassé, P.-P. 1942. L’essaimage des termites. Essai d’analyse causale d’un complexe
instinctif. Bull. Biol. France Belg, 76: 347-3 82.

Grassé, P.-P. 1949. Ordre des Isoptéres ou temrites, pp.-408-544 in P.-P. Grassé [ed]
Traité de Zoologie, volume IX. Masson, Paris.

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

Holt, J. A. 1990. Carbon mineralization in serni-aridtropical Australia: the role of mound-
building termites. Australian Journal of Ecology, 15: 133-134.

Janzen, D. H. 1986. The future of tropical ecology. Ann. Rev. Ecol. Syst., 17: 305-324.

J anzen, D. H. 1987 . Insect diversity of a Costa Rican dry forest: why keep it, and how?
Biological Journal of the Linnean Society, 30: 343 -356.

J anzen, 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.

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

Jones, J. A. 1990. Termites, soil fertility and carbon cycling in dry tropical Afiica: a
i hypothesis. Journal of Tropical Ecology, 6: 291-305.

Jones, S. C., C. A. Nalepa, E. A. McMahan, and J. A. Torres. 1995. Survey and
ecological studies of the termites (Isoptera: Kalotermitidae) of Mona Island.
Florida Entomologist, 78: 305-313.

Jones, S. C. and R. H. Schefl‘rahn. unpublished manuscript.- Survey of the termites of
Puerto Rico (Isoptera: Kalotermitidae, Rhinotermitidae, Termitidae).

Kalshoven, L. G. E. 1960. Biological notes on the Cryptotermes species of Indonesia.
Acta Twp, 17: 263-272.

Klein, B. C. 1989. Effects of forest fragmentation on dung and carrion beetle
communities in central Amazonia. Ecology, 70: 1715-1725.

114

Kramer, E. A. 1997. Measuring landscape changes in remnant tropical dry forest, pp.
386-3 99 in W. F. Laurance and R. O. Bierregaard, Jr. [eds] Tropical forest
remnants: ecology, management, and conservation of fragmented communities.
The University of Chicago Press, Chicago, IL.

Kremen, C. 1992. Assessing the indicator properties of species assemblages for natural
areas monitoring. Ecological Applications, 22203-217.

Kremen, C. 1994. Biological inventory using target taxa: a case study of the butterflies of
Madagascar. Ecological Applications, 42407-422.

Laurance, W. F. and R. O. Bierregaard, Jr. [eds] 1997. Tropical forest remnants:
ecology, management, and conservation of fragmented communities. The
University of Chicago Press, Chicago, IL.

Laurance, W. F ., R. O. Bierregard, Jr., C. Gascon, R.'K. Didham, A. P. Smith, A. J.
Lynarn, 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
fiagmentation: synthesis of a diverse and dynamic discipline, pp. 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, 1]..

Lee, K. E. and T. G. Wood. 1971. Termites and Soils. Academic Press Inc., London,
England.

Lenz, M. 1994. Food resources, colony growth and caste development in wood-feeding
termites, pp. 159-209 in J. H. Hunt & C. A. Nalepa [eds], Nourishment and
evolution in insect societies. Westview Press, Boulder, CO.

Leponce, M., Y. Roisin, and J. M. Pasteels. 1995. Environmental influences on the
arboreal nesting termite community in New Guinean coconut plantations.
Environmental Entomology, 24: 1442- 1452.

Little, E. L., Jr. and F. H. Wadsworth. 1964. Common trees of Puerto Rico and the
Virgin Islands. USDA Forest Service, Agric. Handbook 249, 556 pp.

Little, E. L., Jr., R. O. Woodbury, and F. H. Wadsworth. 1974. Trees of Puerto Rico and
the Virgin Islands, Second Volume. USDA Forest Service, Agric. Handbook 449,
1024 pp.

Lobry de Bruyn, L. A and A. J. Conacher. 1990. The role of termites and ants in soil
modification: a review. Australian Journal of Soil Research, 28: 55-93.

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

115

Harper, K. S. Brown, Jr., A. H. Powell, G. V. N. Powell, H. O. R. Shubart, and
M. B. Hays. 1986. Edge and other efl‘ects of isolation on Amazon forest
fragments, pp. 257-285 in M. E. Soule’ [ed.] Conservation biology: The science of
scarcity and diversity. Sinauer Associates, Sunderland, MA.

Lovejoy, T. E. and R. O. Bierregaard, Jr. 1990. Central Amazonian forests and the
Minimum Critical Size of Ecosystems Project, pp. 60-71 in A. H. Gentry [ed]
F our neotropical rainforests. Yale University Press, New Haven, CT.

Lugo, A. E., J. A. Gonzalez-Liboy, B. Cintron, and K. Dugger. 1978. Structure,
productivity, and transpiration of a subtropical dry forest. Biotropica, 10: 27 8-
291.

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

Margules, C. R., G. A. Milkovits, and G. T. Smith. 1994. Contrasting efl‘ects of habitat
fiagmentation on the scorpion Cercophonius squama and an amphipod. Ecolog,
75: 2033-2042.

Martorell, L. F. 1945. A survey of the forest insects of Puerto Rico. Journal of
Agriculture of the University of Puerto Rico, 29: 358-3 63.

McCoy, E. D. and H. R. Mushinsky. 1994. Effects of fi‘agmentation on the richness of
vertebrates in the Florida scrub habitat. Ecology, 75: 446-457.

McMahan, E. A. 1996. Termites, pp. 109-135 in D. P. Reagan and R. B. Waide [eds]
The food web of a tropical rain forest, University of Chicago Press, Chicago, Il.

Moen, A. N. 1974. Turbulence and the visualization of wind flow. Ecology, 55: 1420-
1424.

Morales-Carrion, A 1983. Puerto Rico, a political and social history. Norton, New
Yorlg NY.

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

Murphy, P. G. and A. E. Lugo. 1990. Dry forests of the tropics and subtropics: Guanica
Forest in context. Acta Cientzfica, 4:15-24.

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

116

tropical forests. Cambridge University Press, New York, NY.

Murphy, P. G., A. E. Lugo, A. J. Murphy, and D. C. Nepstad. 1995. The dry forest of
Puerto Rico’s south coast, pp. 178-209 in A. E. Lugo and C. Lowe [eds] Tropical
forests: management and ecology. Springer-Verlag, New York, NY.

Nutting, W. L. 1969. Flight and colony foundation, pp. 233-282 in K. Krishna and F. M.
Weesner [eds], Biology of termites, Volume 1. Academic Press, New York, NY.

Parnilo, P. and R. H. Crozier. 1997. Population biology of social insect conservation.
Memoirs of the Museum of Victoria, 56: 411-419.

Peakin, G. J. and G. Josens. 1978. Respiration and energy flow, pp. 111-163 in Brian,
M. V. [ed.] The production ecology of ants and termites. Cambridge University
Press, Cambridge.

Pearce, M. J. 1997. Termites: biology and pest management. CAB International, New
York, NY.

Pearson, D. L. and F. Cassola. 1992. World-wide species richness patterns of tiger
beetles (Coleoptera: Cicindelidae): indicator taxon for biodiversity and
conservation studies. Conservation Biology, 6:3 76-391.

Pirnrn, S. L. and J. H. Lawton. 1998. Planning for biodiversity. Science, 279:2068-2069.

Prendergast, J. R., R. M. Quinn, J. H. Lawton, B. C. Eversharn, and D. W. Gibbons.

1993. Rare species, the coincidence of diversity hotspots and conservation
strategies. Nature, 365:335-337.

Ramjohn, I. A, P. G. Murphy, and T. M. Burton. unpublished. Islands of biodiversity on
altered tropical landscapes. International Institute of Tropical Forestry Report,
Rio Piedras, PR. '

Robinson, G. R., R. D. Holt, M. S. Gaines, M. L. Johnson, H. S. Fitch, E. A. Martinko,
and S. P. Hamburg. 1992. Diverse and contrasting efl‘ects of habitat
fiagmentation. Science, 257: 524-527.

Roisin, Y., R. H. Schem'ahn, and J. Krecek. 1996. Generic revision of the smaller nasute
termites of the Greater Antilles (Isoptera, Termitidae, Nasutitennitinae). Annals of
the Entomological Society of America, 89: 775-787.

Sands, W. A 1965. Termite distribution in man-modified habitats in West Afiica, with

special reference to species segregation in the genus T rinervitermes (Isoptera,
Termitidae, Nasutitennitinae). Journal of Animal Ecology, 34: 557-571.

117

Saunders, D. A, R. J. Hobbs, and C. R. Margules. 1991. Biological consequences of
ecosystem fragmentation: a review. Conservation Biology, 5: 18-32.

Schefl‘rahn, R. H. 1994. Incisitermesfllrvus, a new drywood termite (Isoptera:
Kalotermitidae) fi'om Puerto Rico. Florida Entomologist, 77: 365-3 72.

Schefirahn, R. H., J. P. E. C. Darlington, M. S. Collins, J. Krecek, and N-Y. Su. 1994.
Termites (Isoptera: Kalotermitidae, Rhinotermitidae, Termitidae) of the West
Indies. Sociobiology, 24: 213-238.

Schefiahn, R. H., N-Y. Su, and B. Diehl. 1990. Native, introduced, and structure-
infesting termites of the Turks and Caicos Islands, B.W.I. (Isoptera:
Kalotermitidae, Rhinotermitidae, Termitidae). Florida Entomologist, 73: 622-627.

Scott, J. M., F. Davis, B. Csuti, R. Noss, B. Butterfield, C. Groves, H. Anderson, S.
' Caicco, F. D’Erchia, T. C. Edwards, Jr., J. Ulliman, R. G. Wright. 1993. Gap
Analysis: a geographic approach to protection of biological diversity. Wildlife
Monographs, 123: 1-41.

Snyder, T. E. 1923. A new Glyptotermes from Porto Rico. Proc. Ent. Soc. Washington,
25: 91-94.

Snyder, T. E. 1924. Description of a new termite from Porto Rico. Proc. Ent. Soc.
Washington, 26: 131-132.

Snyder, T. E. 1929. New temrites from the Antilles and Middle America. Proc. Ent.
Soc. Washington, 31: 79-87.

Snyder, T. E. 1956. Termites of the West Indies, the Bahamas and Bermuda (Isoptera).
Journal of Agriculture of the University of Puerto Rico, 40: 189-202.

Stehr, F. W. [ed] 1987. Immature insects. Kendall/Hunt Publishing Company,
Dubuque, IA.

Stehr, F. W. [ed] 1991. Immature insects, volume 2. Kendall/Hunt Publishing
Company, Dubuque, IA.

Terbourgh, J. 1974. Preservation of natural diversity: the problem of extinction prone
species. Bioscience, 24: 715-722.

Thome, B. L., M. I. Haverty, and M. S. Collins. 1994. Taxonomy and biogeography of
Nasutitermes acajutlae and N. nigriceps (Isoptera: Termitidae) in the Caribbean
and Central America. Ann. Ent. Soc. Am., 87: 762-770.

Tosy, J. and R. F. Voertman. 1964. Some environmental factors in the economic

118

development of the tropics. Economic Geography, 40: 189-205.

Tranquillini, W. 1979. Physiological ecology of the alpine timberline; tree existence at
high latitudes with special reference to the European Alps. Springer, New York,
NY.

Turner, 1. M. 1996. Species loss in fiagments of tropical rain forest: a review of the
evidence. Journal of Applied Ecology, 33: ZOO-209.

Turner, 1. M. and R. T. Corlett. 1996. The conservation value of small, isolated
fi'agments of lowland tropical rain forest. Trends in Ecology and Evolution, 11:
330- 333.

Ueckert, D. N., M. C. Bodine, and B. M. Spears. 1976. Population density and biomass
of the desert termite Gnathamitermes tubiformans (Isoptera: Termitidae) in a
shortgrass prairie: relationship to temperature and moisture. Ecology, 57: 1273-
1280.

van Jaarsveld, A. S., S. Freitag,.S. L. Chown, C. Muller, S. Koch, H. Hull, C. Bellamy, M
Kruger, S. Endrody-Younga, M. W. Mansell, and C. H. Scholtz. 1998.
Biodiversity assessment and conservation strategies. Science, 279: 2 1 06-2108.

Van Wagner, C. E. 1968. The line intersect method in forest firel sampling Forest
Science, 14:20-26.

Vélez Rodriguez, L. L. 1995a. Land use and land cover 1936 Guiinica Commonwealth
Forest. USDA Forest Service, Institute of Tropical Forestry, Rio Piedras, PR

(map)

Vélez Rodriguez, L. L. 1995b. Land use and land cover 1950-1951 Guanica
Commonwealth Forest. USDA Forest Service, Institute of Tropical Forestry, Rio
Piedras, PR (map).

Vélez Rodriguez, L. L. 1995c. Land use and land cover 1963 Guanica Commonwealth
Forest. USDA Forest Service, Institute of Tropical Forestry, Rio Piedras, PR

(map).

Ve'lez Rodriguez, L. L. l995d. Land use and land cover 1971 Guénica Commonwealth
Forest. USDA Forest Service, Institute of Tropical Forestry, Rio Piedras, PR

(map)-

Vélez Rodriguez, L. L. 19950. Land use and land cover 1983 Guanica Commonwealth
Forest. USDA Forest Service, Institute of Tropical Forestry, Rio Piedras, PR

(map)-

119

Vélez Rodriguez, L. L. l995f. Land use and land cover 1989 Guénica Commonwealth
Forest. USDA Forest Service, Institute of Tropical Forestry, Rio Piedras, PR

(map)-

Warren, W. G. and P. F. Olson. 1964. A line intercept technique for assessing logging
waste. Forest Science, 10: 267-276.

Weigert, R. G. 1970. Energetics of the nest building Nasutitermes costalis (Holmgren) in
a Puerto Rican rain forest, pp. 57-64 in H. T. Odurn and R. Pigeon [eds] A
tropical rainforest. Division of Technical Information. US. Atomic Energy
Cormnission.

Weisharnpel, J. F., H. H. Shugart, and W. E. Westman. 1997. Phenetic variation in
insular populations of a rainforest centipede, pp. 111-123 in W. F. Laurance and
R. O. Bierregaard, Jr. [eds] Tropical forest remnants: ecology, management, and
conservation of fragmented communities. The University of Chicago Press,
Chicago, IL.

Whitmore, T. C. 1997. Tropical forest disturbance, disappearance, and species loss, pp.
3-12 in W. F. Laurance and R. O. Bierregaard, Jr. [eds] Tropical forest
remnants: ecology, management, and conservation of fiagmented communities.
The University of Chicago Press, Chicago, IL.

Whitmore, T. C. and J. A. Sayer. 1992. Deforestation and species extinction in tropical
moist forests, pp. 1-14 in T. C. Whitmore and J. A. Sayer [eds] Tropical
deforestation and species extinction. Chapman and Hall, London.

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

Wilkinson, W. 1962. Dispersal of alates and establishment of new colonies in
Coptotermes havilandi ( S jdstedt) (Isoptera, Kalotermitidae). Bull. Entomol.
Res, 53: 265-286.

Willis, E. O. 1979. The composition of avian communities in remanescent woodlots in
Southern Brazil. Papéis Avulsos de Zoologia, 33: 1-25.

Wolcott, G. N. 1948. The insects of Puerto Rico. Journal of Agriculture of the
University of Puerto Rico, 32: 1-224.

Wolcott, G. N. 1950. An index to the termite resistance of woods. Agricultural
Experimental Station Bulletin No. 85, Rio Piedras, Puerto Rico.

Wood, T. G. 1978. Food and feeding habits of termites, pp. 55-80 in Brian, M. V. [ed]
The production ecology of ants and termites. Cambridge University Press,

120

Cambridge.

Wood, T. G., R. A. Johnson, 8. Bacchus, M. O. Shittu, and J. M. Anderson. 1982.
Abundance and distribution of termites (Isoptera) in a riparian forest in the
Southern Guinea savanna vegetation zone of Nigeria. Biotropica, 14: 25-39.

Wood, T. G. and R. A. Johnson. 1986. The biology, physiology, and ecology of temrites,
pp. 1-68 in Vinson, S. B. [ed.] The economic impact and control of social insects.
Praeger Publications, New York, NY.

Wood, T. G. and M. J. Pearce. 1991. Termites in Afiica: the environmental impact of
control measures and damage to crops, trees, rangeland, and rural buildings.
Sociobiology, 19: 221-234.

Wood, T. G. and W. A. Sands. 1978. The role of termites in ecosystems, pp. 245-292 in

Brian, M. V. [ed] The production ecology of ants and temrites. Cambridge
University Press, Cambridge.

Wright, S. J. and S. P. Hubbell. 1983. Stochastic extinction and reserve size: a focal
species approach. Oikos, 41: 466—476.

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

121

"71111111111111111113