Professional Documents
Culture Documents
Gray, 2002 PDF
Gray, 2002 PDF
by
A DISSERTATION
IN
WILDLIFE SCIENCE
DOCTOR OF PHILOSOPHY
Approved
May, 2002
©2002, Matthew J. Gray
ACKNOWLEDGEMENTS
Foremost I thank my major advisor, Dr. Loren M. Smith, for providing the
University (TTU). Dr. Smith not only provided me with this great academic opportunity:
his guidance honed me as a scientist. I now am capable of perceiving the edge of science
(see page v). Without question, my future research endeavors will be a product of his
mentorship. I hope my science will make him proud. Thank you so much, Loren!
Mathematics and Statistics at TTU. Dr. Allen sculptured my quantitative abilities and
changed my perception of science. Thank you. Dr. Allen, for making me quantitatively
literate. I also want to thank my other instructors in the math department: Drs. Kamal
Chanda, Wijesuriya Dayawansa, William Gustafson, Ira Wayne Lewis, and James Surles.
In addition to Drs. Smith and Allen, I thank my committee members: Drs. Lou
Densmore (Department of Biological Sciences, TTU), David Haukos (U.S. Fish and
Wildlife Service), and David Wester (Department of Range, Wildlife and Fisheries,
TTU). These individuals provided incredible personal and professional guidance during
Thank you for making my doctoral pursuit a stimulating and intellectual experience!
n
I also want to thank Dr. James Surles of the Department of Mathematics and
Statistics at TTU for quantitative guidance on se\'eral anal> ses in my dissertation and
Drs. Kevin Pope, Mark Wallace, and Gene Wilde of the Department of Range. Wildlife,
and Fisheries at TTU and Dr. Ray Semlitsch of the University of Missouri (Di\ision of
following TTU graduate students for reviewing chapters: J. Avey, J. Brunjes. K. Brunjes,
and S. Sebring. I thank Stephanie Dupree of the Geospatial Technology Lab at TTU for
helping construct Figure 2.1. I am grateful to Dr. Ernest Fish, Cherise Ginsburg. and
gratefulness allowed me to take my research thoughts to the field. I also thank M. Cerda.
T. Grand, L. Lloyd, and A. Wallace of the U.S. Department of Agriculture (USDA) Farm
Service Agency (FSA) for assistance in locating aerial slides and FSA Farm Folders for
my study playas.
This study would not have been possible without field and lab assistance from
ill
Smith. I especially am indebted to Roberto Brenes (now at the University of Texas-
Tyler)—my right-hand man and savior. Roberto assisted me in the field for two years.
Honestly, I do not understand how he tolerated me—and, I am still quite amazed that we
survived. Roberto was more than a technician to me—he was a colleague and great
friend. I am confident my research would have suffered without him. I thank him for all
his dedication, patience, and herpetological insight. I doubt if I will ever be able to repay
This study was financially supported primarily by the Caesar Kleberg Foundation
for Wildlife Conservation and the Department of Range, Wildlife, and Fisheries
Management at TTU. Student research grants also were provided by the Society of
Wetland Scientists and Ducks Unlimited, Inc., and Home Depot donated buckets for
pitfall traps. I thank Dr. Gene Wilde of TTU for use of various sampling equipment and
Program CANOCO for multivariate ordination analyses. I also am very grateful to Dr.
Rob Mitchell of TTU for allowing me use his ATV on days when the west Texas roads
were impassable. Lastly, I thank Dr. Nick Parker of TTU and Dick Lubke of the Texas
Parks and Wildlife Department for use of aquariums and a PIT tag scanner, respectively.
(father), Scott H. Gray (brother), Sharon L. Malloy (mother), and Carly L. Pratt (sister).
No one gets anywhere by himself or herself My family has been an amazing nucleus of
love and support. Without their encouragement, my 12-year academic journey would
have been impossible. I also thank my very dear friend, Jessica L. Emrich. Her love and
IV
I dedicate my dissertation to my four academic mentors. In order of academic
Wildlife Research Unit), and Loren M. Smith (Texas Tech University). Dr. Prince
trained me in the scientific methods. Dr. Kaminski fueled m\ passion for wetlands and
research and de\ eloped my wxiting skills. Dr. \ ilella sparked my interest in conservation
biology, and Dr. Smith directed me in ecological theor>' and landscape ecology. These
individuals made me the person and scientist that I am toda}'. Indeed, I am indebted.
\'
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ii
ABSTRACT ix
LIST OF TABLES xi
CHAPTER
I. INTRODUCTION 1
Abstract 5
Introduction 6
Study Area and Amphibians 9
Methods 13
Experimental Design and Units 13
Sampling Techniques 14
Terrestrial capture 14
Biological processing 15
Statistical Analyses 15
Results 16
Landuse Type and Year Interaction Effects 16
Adults 16
Subadults 17
Metamorphs 17
Landuse Type Main Effect 19
Adults 19
Subadults 19
Metamorphs 19
Year Main Effect 21
Adults 21
Subadults 21
Metamorphs 23
Discussion 23
Landuse Effect 23
Year Effect 27
Competing Body Size Hypotheses 29
VI
Conclusions ^^
Abstract 32
Introduction 33
Methods 36
Results 41
Abundance and Diversity 41
Source-Sink Dynamics 43
Discussion 45
Landuse Effect 47
Year Effect 52
Source-Sink Dynamics 53
Conservation Implications 54
Abstract 56
Introduction 57
Methods 59
Results 63
Discussion 76
Conservation Implications 78
Abstract 80
Introduction 81
Methods 83
Results 85
Discussion 91
Abstract 106
Introduction 107
Methods 110
Vll
Results 117
Discussion 128
Ecological and Mathematical Implications 133
APPENDIX
Vlll
ABSTRACT
landscape use (cultivation vs. grassland) and structure on postmetamorphic body size (a
Southern High Plains (SHP) of Texas during 1999 and 2000. Amphibian populations
were monitored using drift fence and pitfall traps, landscape structure was quantified
using spatial analysis software, and dynamics were assessed using difference equations.
Postmetamorphic body size of all amphibian species and age classes generally
was greater at grassland than cropland playas, and in 1999 (i.e., a wetter year) than 2000.
Abimdance of New Mexico and plains spadefoots (Spea multiplicata and S. bombifrons)
generally was greater at cropland than grassland playas, and greater for barred tiger
landscapes became more complex (e.g., numbers of edges increased) and inter-playa
demographic anah ses indicated that temporal niche partitioning existed in SHP
ix
grassland playas. A stochastic density-dependent Ricker function predicted chaotic
cultivation decreased body size and altered amphibian demographics and dynamics from
2.1 Adult body size of New Mexico spadefoot {Spea multiplicata, NSF), plains
spadefoot {S. bombifrons, PSF), Great Plains toad {Bufo cognatus, GPT),
and barred tiger salamander {Ambystoma tigrinum mavortium, BTS) between
landuse types and years (fixed factorial effects) at 16 playa wetlands (random
nested effect) on the Southern High Plains of Texas, April-September 1999
and 2000 18
2.2 Subadult body size of New Mexico spadefoot {Spea multiplicata, NSF), plains
spadefoot {S. bombifrons, PSF), and Great Plains toad {Bufo cognatus, GPT)
between landuse types and years (fixed factorial effects) at 16 playa wetlands
(random nested effect) on the Southern High Plains of Texas, April-
September 1999 and 2000 20
2.3 Metamorph body size of New Mexico spadefoot {Spea multiplicata, NSF), plains
spadefoot {S. bombifrons, PSF), Great Plains toad {Bufo cognatus, GPT), and
barred tiger salamander {Ambystoma tigrinum mavortium, BTS) between
landuse types and years (fixed factorial effects) at 16 playa wetlands (random
nested effect) on the Southern High Plains of Texas, April-September 1999
and 2000 22
3.1 Relative daily abundance and diversity of amphibians between landuse types
and years and among species at 16 playa wetlands on the Southern High
Plains, Texas, May-October 1999 and April-August 2000 42
3.2 Relative daily abundance of age classes between landuse types and years for
amphibian species that differed significantly in daily abundance (see Table
C.2) at 16 playa wetlands on the Southern High Plains, Texas, May-October
1999 and April-August 2000 44
4.2 Intra-set correlations of landscape* metrics associated with the first two axes
generated by a canonical correspondence analysis of mean daily abundance of
New Mexico spadefoot {Spea multiplicata), plains spadefoot {S. bombifrons),
barred tiger salamander {Ambystoma tigrinum mavortium), and Great Plains toad
{Bufo cognatus) at 16 playa wetlands on the Southern High Plains, Texas, 1999
and 2000 66
XI
4.3 Univariate Pearson coefficients of correlations between mean daily abundance
of amphibians and landscape* metrics at 16 playa wetlands on the Southern
High Plains, Texas, 1999 and 2000 70
4.4 Multiple linear regression between mean daily abundance of amphibians and
landscape* metrics at 16 playa wetlands on the Southern High Plains, Texas,
1999 and 2000 74
C.l Statistics for ANOVAs for species (by age class) that differed significantly in
relative daily abundance between landuse types (cropland vs. grassland) or years
(see Table B.l) at 16 playa wetlands on the Southern High Plains, Texas,
May-October 1999 and April-August 2000 173
xii
LIST OF FIGURES
2.1 Location of study playa wetlands on the Southern High Plains of Texas,
U.S.A 12
3.1 Effect of landuse (top graph) and year (bottom graph) on frequency of directional
movement of amphibian species at 16 playa wetlands on the Southern High
Plains, Texas, May-October 1999 and April-August 2000 46
4.2 Simple linear regression and 95% confidence bands of mean daily abundance
(natural-log transformed) of New Mexico spadefoot {Spea multiplicata) and
percent aerial coverage of playa wetlands (PP), interspersion/juxtaposition index
(IJI, McGarigal and Marks 1995:103), edge density (ED, McGarigal and Marks
1995:106), playa edge density (PED [i.e., mean number of edges to cross from
study playa to surrounding playas), landscape shape index (LSI, McGarigal and
Marks 1995:109), and landuse richness (LR, McGarigal and Marks 1995:119) in
landscapes (i.e., 2,830-ha circular plot) associated with 16 study playas on the
Southern High Plains, Texas, 1999 and 2000 71
4.3 Simple linear regression and 95% confidence bands of mean daily abundance
(natural-log transformed) of plains spadefoot {Spea bombifrons) and percent
aerial coverage of playa wetlands (PP), interspersion/juxtaposition index (IJI,
McGarigal and Marks 1995:103), edge density (ED, McGarigal and Marks
1995:106), and landscape shape index (LSI, McGarigal and Marks 1995:109) in
landscapes (i.e., 2,830-ha circular plot) associated with 16 study playas on the
Southern High Plains, Texas, 1999 and 2000 73
5.1 Time series plots for pairs of amphibian species residing at 16 playa wetlands
in 2 landuses and years on the Southern High Plains of Texas 87
6.1 Time series plots of amphibian assemblage dynamics at 8 playa wetlands during
16 May-17 October 1999 on the Southern High Plains, Texas 118
Xlll
6.2 Time series plots of amphibian assemblage dynamics at 8 playa wetlands during
19 April-18 August 2000 on the Southern High Plains, Texas 120
D.l Anthropogenic and natural cover types in landscapes (i.e., 2,830-ha [3-km radius]
circular plots) surrounding 16 playa wetlands on the Southern High Plains, 1999
and 2000 176
xiv
CHAPTER I
INTRODUCTION
surrounding wetlands has been implicated as a primary cause (Blaustein et al. 1994.
wetlands have been considered ultimate reasons for decline (Alford and Richards 1999).
Interestingly, few data exist in the United States that suggest amphibian populations are
negatively affected by cultivation. Two of the three studies (Knutson et al. 1999,
Kolozsvary and Swihart 1999) indicate that amphibian populations can be elevated in
differences between disturbed and undisturbed wetlands. In Canada and Europe, most
and species richness, and negatively affects fitness correlates (Laan and Verboom 1990,
Oldham and Swan 1991, Hecnar and M'Closkey 1996, Bonin et al. I997a,b). However,
most of these studies were conducted using call surveys during the breeding season,
which have less demographic resolution than other methods (e.g., pitfall sampling, Heyer
et al. 1994). Drift fence and pitfall traps facilitate the most comprehensive examination
disturbance, particularly considering that few sampling efforts were performed outside
populations by altering abiotic and biotic conditions in and around wetlands (Semlitsch
2000). Anthropogenic disturbance can negatively affect water qualit>, which might
affect trophic structuring in the aquatic environment, and ultimately growth and survival
of larvae (Boone and Semlitsch 2001). Agricultural chemicals also can reduce
immunocompetence, mobility, and survival of larvae (Carey et al. 1999, Bridges and
Semlitsch 2000). Sedimentation in wetlands can decrease hydroperiods (Luo et al. 1997).
thus presumably breeding opportunities for adults and de\ elopmental time of larvae.
Finally, landscape disturbance may cause direct mortality, reduce terrestrial habitat and
(Wilbur 1984). Larval amphibians can significantly affect trophic structuring and
nutrient cycling in wetlands via consumption and excrement of fine and course
particulate organic matter and micro- and macro-invertebrates (Scale 1980, Hoff et al.
1999, Petranka and Kennedy 1999). Also, amphibian larvae are prey for various
numerous invertebrates, some of which are agricultural pests, and they are food items for
various terrestrial predators (Duellman and Trueb 1994, Anderson et al. 19996). Larval
(Boutilier et al. 1992, Shoemaker et al. 1992), which can affect their physiology and
(Beebee 1996).
High Plains (Bolen et al. 1989). Cultivation and grassland surround most playas; thus,
they are ideal systems to comparatively and simultaneously investigate the effect of
proposed the following research. My goal was to determine the effect of anthropogenic
over 2 years (1999 and 2000, 4 playas per landuse per year). My specific research
objectives were to determine the effect of (1) landuse and yearly rainfall on
postmetamorphic body size (a fitness correlate) of amphibians (Chapter II), (2) landuse
community composition and relative abundance of amphibians (Chapter IV), (4) landuse
(Chapters V and VI), and to (5) identify potential regulating mechanisms in an amphibian
assemblage (Chapter VI). I used pitfall sampling, the geographic information system and
presented herein provide insight into the effect of anthropogenic disturbance and
biology, landscape ecology, applied and theoretical mathematics, and basic ecological
theo^>^
CHAPTER II
Abstract
fitness. Landuse surrounding wetlands and rainfall may affect body size by influencing
proximate factors (e.g., hydroperiod, food resources) that affect larval and
surrounding wetlands would affect body size of amphibian, but the magnitude of its
size between two landuse types (cultivation versus grassland) surrounding playa wetlands
and two years (1999 versus 2000) with different rainfall (1999 > 2000) for 4 species
and 3 age classes (metamorph, subadult, and adult) of amphibians on the Southern High
Plains of Texas. Sixteen playas (4 per landuse type per year) were partially enclosed with
drift fence and pitfall traps, and mass and snout-vent length (SVL) measured from a
subsample of captured individuals {n = 5,188). Mass and SVL generally were greater at
grassland than cropland playas for all species and age classes. Mass and SVL also were
greater in 1999 (i.e., the wetter year) than in 2000 for most species and age classes.
Moreover, the effect of landuse type and year on postmetamorphic body size generally
was non-additive; cultivation had a greater negative effect on body size in 2000 (i.e., the
drier year) than in 1999 for some of the age classes and species. Differences in body size
between landuse types and > ears may have been related to densit>' independent and
dependent factors, such as reduced wetland hydroperiod, food availabilit\\ and predators
Introduction
affecting survi\'al and reproduction (Wilbur 1984, Werner 1986, Semlitsch et al. 1988).
postmetamorphic body size (Smith 1987, Berven 1990. Goater 1994, Morey and Reznick
2001). Large individuals may have greater survival than small indi\ iduals, because they
can use a greater moisture range of microhabitats (Newman and Dunham 1994). can
consume a greater range of prey sizes (Flowers and Graves 1995. Newman 1999). are
more efficient foragers (Newman 1999), and have lower specific metabolic rates
(Hutchison et al. 1968, Ultsch 1973, 1974). Large individuals also take more time to
reach critical levels of dehydration than small individuals (Newman and Dunham 1994).
In addition, large individuals have greater sprint speeds, jumping ability, and endurance
than small individuals (Taigen and Pough 1981, John-Alder and Morin 1990, Goater et
al. 1993, Beck and Congdon 2000). Mating success of male and female amphibians also
may be positively related to body size (Berven 1981, Verrell 1982, 1985. Howard and
Young 1998, Marco et al. 1998). For territorial amphibians, territory size and quality
(i.e., density of food resources) can be positively correlated with body size (Howard
1978, Mathis 1990, Gabor 1995). Large females also may exhibit greater fecundity than
small females and spawn multiple times (Clarke 1974, Howard 1978, Berven 1982,
Semlitsch 1985a, Krupa 1986, 1994, Fontenot 1999). Finally, large individuals may
reproduce at an earlier age than small individuals (Berven and Gill 1983, Smith 1987.
development and pre- and post-metamorphic conditions that differentially affect growth
and mortality rates of individuals in aquatic and terrestrial habitats (Wilbur and Collins
1973, Werner 1986). Generally, postmetamorphic body size is correlated with body size
at metamorphosis (Smith-Gill and Berven 1979. Semlitsch et al. 1988, Scott 1994).
Wetland hydroperiod, water quality, and density of food resources, conspecifics, and
1984, Wilbur 1984, Semlitsch 1987fl, Alford and Harris 1988, Pfennig et al. 1991,
Bradford et al. 1992, Werner and Anholt 1993). Postmetamorphic body size also is
affected by density of food resources and conspecifics in the terrestrial landscape (Goater
1994, Morey and Reznick 2001). Indeed, small body size at metamorphosis may be
compensated for by high density of terrestrial food resources (as theoretically suggested
by Werner 1986), allowing small individuals to attain a body size at first reproduction
similar to individuals that were larger at metamorphosis (e.g.. Goater 1994, Morey and
Reznick 2001).
amphibians (Hecnar and M'Closkey 1996, Bonin et al. 1997a,b, Dodd 1997, Alford and
Richards 1999, Semlitsch 2000), presumably by influencing some of the aforementioned
mechanisms that affect growth and mortality rates of individuals in aquatic and terrestrial
environments (Werner 1986). Landscape cultivation (i.e., arable cropland) may localize
densities (Knutson et al. 1999, Kolozsvary and Swihart 1999, Chapter III).
(Martin and Hartman 1987, Corn and Bury 1989, Luo et al. 1997), and may reduce
duration of larval development (Brady and Griffiths 2000). Insecticides and herbicides
associated with agricultural practices may affect postmetamorphic body size by reducing
organophosphates) can bioaccumulate and reduce foraging activity and growth of larval
amphibians (Hall and Kolbe 1980, Baker and Waights 1993, 1994, Hecnar 1995,
(Tinsley and Tocque 1995, Camp et al. 2000) and potentially postmetamorphic body size
and condition (Bruce and Hairston 1990, Reading and Clarke 1995). Rainfall can
increase pond hydroperiod hence growth rates and duration of larval development
(Semlitsch et al. 1996, Reading and Clark 1999, Camp et al. 2000). Rainfall also can
increase prey availability in the terrestrial landscape (Dimmitt and Ruibal 1980a, Jaeger
8
1980). Moreover, rainfall reduces probability of desiccation and increases foraging
size of amphibians, but the magnitude of its effect would depend on rainfall. Thus, my
objective was to measure the effect of two landuse types (i.e., cultivation and grassland
[control]) over two years (i.e., 1999 and 2000) on postmetamorphic body size of
amphibians. I quantified the effect of landuse and year for 3 age classes (i.e., metamorph,
subadult, and adult) and 4 species (i.e., Spea multiplicata, S. bombifrons, Bufo cognatus,
Southern High Plains. This information is particularly important for species currently in
decline, such as amphibians (Houlahan et al. 2000), because body size can potentially
My study was conducted on the Southern High Plains (SHP) of Texas. The SHP
(i.e., the Llano Estacado) is the largest continuous plateau in the United States (Sabin and
Holliday 1995), encompassing ca. 8.9 million ha, of which 5.9 million ha are in
northwestern Texas (Haukos and Smith 1994). Primary land uses on the SHP are
agricultural cultivation (2.7 million ha, 46%)) and grassland (2.9 million ha, 49%) [native
and replanted lands combined], Haukos and Smith 1994). The most significant
topographic formations on the SHP are playa wetlands (Bolen et al. 1989).
Playa wetlands (or simply playas) occur at the highest density (ca. 19,340) in
Texas, comprising ca. 2% of the Texas SHP landscape (Haukos and Smith 1994). Playas
are circular wetlands (xsize=6.3 ha) that form dynamically and continually via an
interaction of eolian and alluvial processes (Osterkamp and Wood 1987, Gustavson et al.
1994). Playas are the lowest points on the relatively level SHP landscape, and their
watersheds are not interconnected (Guthery and Bryant 1982). Playas also are not
directly connected to the underlying Ogallala Aquifer (Bolen et al. 1989). Consequently,
playa hydrology is dependent on precipitation and watershed runoff (Haukos and Smith
1994). Because the climate on the SHP is semi-arid (i.e., Xrainfaii=33-45 cm/year,
><evaporation^200-250 cm/ycar) and playa basins are shallow, playa hydroperiods are
variable (Bolen et al. 1989). Playas usually have intermittent to seasonal or annual
Amphibians can be the most abundant vertebrate in playas during summer (Bolen
et al. 1989). Thirteen species exist on the SHP; I selected the 4 most abundant for this
study: New Mexico spadefoot {Spea multiplicata Cope), plains spadefoot {S. bombifrons
Cope), Great Plains toad {Bufo cognatus Say), and barred tiger salamander {Ambystoma
tigrinum mavortium Green) (Rose and Armentrout 1974, Anderson 1997:74, Chapter III).
The above anurans are explosive breeders (i.e., complete breeding in 1-3 d, Sullivan and
Fernandez 1999), although Great Plains toads may breed for several weeks and spawn
multiple times if ambient conditions are favorable (Krupa 1986, 1994). Barred tiger
salamanders can be more prolonged breeders, taking ca. 2-4 weeks to breed (Petranka
10
1998:111). Breeding for these species is concentrated during and after seasonal rains
March-August; however, barred tiger salamander may breed during any month on the
SHP if sufficient water is present (Webb and Roueche 1971. Rose and Armentrout 1976).
Duration of larval development for these amphibians is rapid (i.e.. 2-6 weeks) and likely
quality (Rose and Armentrout 1976, Newman 1992, 1994a. Werner and McPeek 1994,
Sixteen playas (4 per landuse type per year) were selected via aerial and ground
grassland), presence of water (i.e., all playas were inundated initially), landowner
cooperation, and study playa juxtaposition (i.e., clusters of playas within the same
landuse type were sought to increase sampling efficiency). Playas were considered to be
in grassland if >75%) of the surrounding landscape (i.e., <3 km from the playa center) was
undisturbed and vegetated, and cultivation if >75% of the surrounding landscape was
arable cropland (Anderson et al. 1999(^:760, Chapter III). Grassland could have included
native, replanted (via the Conservation Reserve Program, CRP), or grazed lands with
intact grass. I chose the 3-km threshold for landuse designation, because it probably was
the near maximum dispersal distance for my study species (cf Gehlbach 1967, Gehlbach
Grassland playas were located in Castro County northeast of Dimmitt, Texas USA
(Figure 2.1). Playas surrounded by cropland were located in Hale and Floyd counties
11
34M5'0CrN
"'"100"52'3Cr W
Si- / •
.%^'
amoiitt*©
A ^
- • .•• ; ••• •/
• <...
Ca stre <\ ••••' lllv»rWfiV_,
A
1
0 0 0 J
BHn»M ^J-.
-A
"V.
o %
Hale F10 y d ^'••^.
- A
^
Lltti«l«id
A. V
- - I
J
0^ -
8 0 y
<.'^-
/ • '
L»v«llind
Lubboet ._ Crasb^s
• /
":.\.
33*22-3 (TN 1 •;:•
102*37'30W
20 10 0 20 40 60 80 100
Kilometers
Figure 2.1. Location of study playa wetlands on the Southern High Plains of Texas,
U.S.A.
12
near Plainview, Texas in 1999 and north of Ralls, Texas in Crosby County in 2000.
Although grassland playas were located northwest (x=97 km, SD=11) of cropland playas,
soils and historical vegetation were similar between locations (Allen et al. 1972, D. A.
Haukos and L. M. Smith, U. S. Fish and Wildlife Service and Texas Tech University,
unpublished data). Moreover, rainfall (and I assume other ambient variables because
elevation and latitudinal change is minimal) was similar between sites (x=5.5 cm, SE=1.8
[cultivation]; x=6.9 cm, SE=2.7 [grassland]) during field sampling. Thus, I assumed a
location. Average rainfall at the study playas during sampling was greater in 1999
Methods
(Montgomery 2001:569-573). Landuse types and years were crossed factors and
considered fixed effects. Playas were nested within landuses and years and treated as a
landuse and year. Individuals might be considered pseudoreplicates of the crossed effects
in these 2 landuse types and years on Southern High Plains. I also used an unrestricted
mixed model, which allows covariance (i.e., dependency) between observations within
levels of the nested random effect (Hocking 1973). Consequently, results of ratio
significance tests using this model remained unbiased and accurate, despite potential
13
interdependencies between observations (Hocking 1973, Montgomer> 2001:526-527).
the response.
Recent research (e.g., Morey and Reznick 2001) also suggests that terrestrial food
resources might differentially affect postmetamorphic body size of genders. Thus. 1 also
tested the effect of gender and landuse type on body size of adults in preliminary
multivariate analyses. These analyses were not conducted on the other age classes (i.e.,
metamorph and subadult), because captured individuals were not dissected for gender
determination. Body size of females was greater than males for adult New Mexico
spadefoot, plains spadefoot and Great Plains toads. However, because the gender effect
did not interact with landuse type and mean sex ratio across species was similar between
Sampling Techniques
with 60-cm high drift fence (xiength= 330 m per playa) and 19-L pitfall traps (Dodd and
Scott 1994). Pitfall traps were placed on alternate sides of the fence at 10-m intervals
with openings flush to the ground (Dodd and Scott 1994). Pitfall traps in each landuse
type were checked alternate days for captures from 16 May-17 October 1999 and 19
April-18 August 2000. Pitfall traps within one landuse type were opened near dusk (e.g.,
1800-2100 hrs) before the night of intended capture and processed during the next
14
afternoon (e.g., 1400-1800 hrs). Traps were subsequently closed and reopened in 24 hrs.
Simultaneous with closing of pitfall traps in one landuse type, pitfall traps in the other
(metamorph [juveniles <1 yr], subadult [>1 yr but not displaying secondary sexual
characters], adult [>1 yr and possessing reproductive morpholog) such as vocal sacs,
nuptial excrescences, or eggs for anurans, and enlarged cloacae for the salamander];
Duellman and Trueb 1994:33-38, 52-60). I did not designate a subadult category for
barred tiger salamander because distinction is based on color (Webb and Roueche 1971).
length (i.e., tip of snout to posterior of vent; SVL) and mass were measured using calipers
and a spring scale for the first 5 individuals captured per playa per species per age class
per day (« = 2816 and 2372 in cultivation and grassland [years combined]). Individuals
gluconate); recaptured individuals were not measured. Sampling techniques in this and
subsequent chapters followed approved Texas Tech University Animal Use and Care
Statistical Analyses
= 0.05) in body size within species and age classes between landuse types and years and
among levels of the random nested effect (i.e., playas). I assumed the matrix composed
of SVL and body mass response vectors represented a multivariate description of body
15
size. I used an unrestricted mixed MANOVA model with an unstructured covariance
covariance structures (Littell et al. 1996:99). Because sample size generally was large
(i.e., «>30 individuals per level per effect), I assumed average body size per effect
Johnson and Wichem 1998:187). Despite these precautions, I also used Pillai-Bartlett
trace test statistic, because it is more powerful and robust than other statistics (e.g.,
Wilk's likelihood ratio) when multivariate assumptions are violated (Olson 1974, 1976,
1979). I tested additivity of fixed main effects by including an interaction term in the
mixed model (Montgomery 2001:526); analyses were sorted by simple main factors (i.e.,
landuse type by year and year by landuse) with playas nested when this assumption was
(hereafter STEPBOOT) was used to test for simple effect differences in body size
between levels of fixed factors when MANOVA was significant (Westfall et al.
1999:227-239). I did not present results of the random effect (i.e., differences in body
Results
Adults. Landuse and year effects were not additive for adult Great Plains toad
(MANOVA F=12.6; 2, 593 df; P<0.001), New Mexico spadefoot (MANOVA F=49.8; 2.
16
1541 df; P<0.001), and barred tiger salamander (MANOVA F=7.5; 2, 361 df; i'<0.001);
therefore, I separated analyses by main factors (see next 2 sections). Landuse and year
effects were additive for aduft plains spadefoots (MANOVA F=2.5; 2, 240 df; ^^=0.09);
thus, analyses were not separated by main factors for this species. Body size of adult
plains spadefoots was greater at grassland than at cropland playas (MANOVA F=15.0; 2,
240 df; /'<0.001) and greater in 1999 than in 2000 (MANOVA F=4.1; 2, 240 df;
P=0.02): both SVL and mass were significant (STEPBOOT P<0.001, Table 2.1).
Subadults. Landuse and year effects were not additive for subadult Great Plains
toad (MANOVA F=\9A; 2, 325 df; P<0.001) and New Mexico spadefoot (MANOVA
F=12.3; 2, 383 df; P<0.001); therefore, I separated analyses by main factors. Additivity
of main effects was not tested for subadult plains spadefoot because no individuals were
captured in cropland in 1999; thus, these analyses were separated immediately by main
factors. Analyses were not performed for barred tiger salamander, because there was no
Metamorphs. Landuse and year effects were not additive for metamorph New
Mexico spadefoot (MANOVA F=12.2; 2, 577 df; P<0.00\) and barred tiger salamander
(MANOVA F=7.8; 2, 462 df; P<0.001); therefore, I separated analyses by main factors.
Additivity of main effects was not tested for metamorph Great Plains toad and plains
17
Table 2.1. Aduk body size of New Mexico spadefoot {Spea multiplicata. NSF). plains
spadefoot {S. bombifrons. PSF), Great Plains toad {Bufo cognatus. GPT). and barred tiger
salamander {Ambystoma tigrinum mavortium, BTS) between landuse types and \ears
(fixed factorial effects) at 16 playa wetlands (random nested effect) on the Southern High
Plains of Texas, April-September 1999 and 2000.
Size Landuse
Variable Species Year' Cultivation Grassland
n X SE P<0.05' n X SE P<0.05
Snout- NSF 1999 536 42.1 0.1 Aa 409 44.4 0.2 Ba
vent 2000 401 38.7 0.2 Ab 206 43.7 0.2 Bb
length PSF 1999 45 46.1 0.6 Aa 57 49.2 0.7 Ba
(mm) 2000 77 43.9 0.4 Ab 72 46.9 0.4 Bb
GPT 1999 256 72.4 0.5 Aa 164 80.2 0.6 Ba
2000 76 64.9 0.9 Ab 108 75.3 0.8 Bb
BTS 1999 63 92.6 1.8 Aa 107 101.4 0.9 Ba
2000 66 81.3 0.9 Ab 100 86.5 1.2 Bb
Mass NSF 1999 536 8.5 0.1 Aa 409 9.8 0.1 Ba
(g) 2000 401 7.2 0.1 Ab 206 9.4 0.2 Bb
1999 45 11.4 0.5 Aa 57 12.2 0.4 Ba
PSF 2000 77 9.7 0.3 Ab 72 11.5 0.3 Bb
1999 256 51.5 1.1 Aa 164 70.1 1.7 Ba
GPT 2000 76 40.5 1.7 Ab 108 53.8 1.8 Bb
1999 63 39.7 2.8 Aa 107 48.1 1.6 Ba
BTS 2000 66 22.3 0.9 Ab 100 27.1 1.4 Bb
^Analyzed by year because landuse and year main effects were not additi\e (P<0.001) for
all species except PSF (P=0.09); PSF statistics were presented by year for tabular
parsimony.
Within-species means in the same row with unlike uppercase letters are different (i.e..
simple effect test between landuses within years); means in the same column within size
variables and species with unlike lowercase letters are different (i.e., simple effect test
between years within landuses); all analyses performed using Westfall-Young
(1993:113-121) multivariate multiple comparison test after the simple main effect
MANOVA for factorial-nested designs (Montgomery 2001:569-573) was significant
(P<0.05).
18
Landuse Type Main Effect
Adults. Body size was greater at grassland than at cropland playas for adult Great
Plains toad (MANOVA F=47.2; 2, 411 df; P<0.00\), New Mexico spadefoot (MANOVA
F=9A; 2, 936 df; P<0.001), and barred tiger salamander (MANOVA F=12.5; 2, 161 df;
P<0.001) in 1999; SVL and mass were significant for all species (STEPBOOT P<0.005,
Table 2.1). Body size also was greater at grassland than at cropland playas for adult
Great Plains toad (MANOVA F=18.3; 2, 175 df; P<0.001), New Mexico spadefoot
(MANOVA F=90.6; 2, 598 df; P<Om\), and barred tiger salamander (MANOVA
F=5.4; 2, 157 df; /*<0.005) in 2000; SVL and mass were significant for all species
Subadults. Body size was greater at grassland than at cropland playas for
subaduh Great Plains toad (MANOVA F=\3.7; 2, 256 df; P<0.001), New Mexico
spadefoot (MANOVA F=\ 12.2; 2, 298 df; P<0.001), and plains spadefoot (MANOVA
F=24.8; 2, 64 df; P<0.00\) in 2000; SVL and mass were significant for all species
(STEPBOOT P<0.001, Table 2.2). Body size did not differ between landuses for
subadult Great Plains toad (MANOVA F=0.7; 2, 62 df; P=0.49) and New Mexico
Metamorphs. Body size was greater at grassland than at cropland playas for
metamorph Great Plains toad (MANOVA F=l 14.1; 2, 335 df; P<0.001), New Mexico
spadefoot (MANOVA F=19.2; 2, 494 df; P<0.00\), and plains spadefoot (MANOVA
F=21.9; 2, 156 df; P<0.001) in 1999; SVL and mass were significant for all species
19
Table 2.2. Subadult body size of New Mexico spadefoot {Spea multiplicata, NSF), plains
spadefoot {S. bombifrons, PSF), and Great Plains toad {Bufo cognatus, GPT) between
landuse types and years (fixed factorial effects) at 16 playa wetlands (random nested
effect) on the Southern High Plains of Texas, April-September 1999 and 2000.
Size Landuse
Variable Species Year' CLiltivation Grassland
n X SE P<0.05'' n X SE P<0.05
Snout- NSF 1999 26 34.1 0.9 Aa 61 36.2 0.5 Aa
vent 2000 221 32.5 0.2 Ab 86 39.1 0.4 Bb
length PSF 1999 0 0 0 NT 15 38.1 0.8 a
(mm) 2000 55 34.2 0.5 A 18 43.2 0.7 Bb
GPT 1999 41 58.4 0.9 Aa 30 56.5 1.3 Aa
2000 64 47.1 0.8 Ab 201 59.0 0.6 Ba
Mass NSF 1999 26 4.8 0.3 Aa 61 5.3 0.3 Aa
(g) 2000 221 4.6 0.2 Aa 86 6.6 0.2 Bb
PSF 1999 0 0 0 NT 15 6.5 0.4 a
2000 55 4.7 0.2 A 18 7.8 0.4 Bb
GPT 1999 41 23.3 0.9 Aa 30 24.4 1.8 Aa
2000 64 14.7 0.7 Ab 201 26.9 0.7 Ba
Analyzed by year because landuse and year main effects were not additive (P<0.001) for
NSF and GPT; PSF analyzed for 2000 only.
Within-species means in the same row with unlike uppercase letters are different (i.e.,
simple effect test between landuses within years); means in the same column within size
variables and species with unlike lowercase letters are different (i.e., simple effect test
between years within landuses); all analyses performed using Westfall-Young
(1993:113-121) multivariate multiple comparison test after the simple main effect
MANOVA for factorial-nested designs (Montgomery 2001:569-573) was significant
(/*<0.05); NT indicates no test performed because SE=0 for >1 level of the effect.
20
(STEPBOOT P<0.001, Table 2.3). Body size did not differ between landuses for
metamorph barred tiger salamander (MANOVA F-0.5; 2, 267 df; P=0.62) in 1999.
Body size was greater at grassland than at cropland playas for metamorph barred tiger
salamander (MANOVA F=18.4; 2, 188 df; P<0.001) in 2000; SVL and mass were
significant (STEPBOOT P<0.005, Table 2.3). Body size was greater at cropland than at
grassland playas for metamorph New Mexico spadefoot (MANOVA F=7.9; 2, 78 df;
P<0.001) in 2000; SVL and mass were significant (STEPBOOT P<0.006, Table 2.3).
Adults. Body size was greater in 1999 than in 2000 for aduk Great Plains toad
(MANOVA F-15.6; 2, 323 df; P<0.001), New Mexico spadefoot (MANOVA F=l 10.6;
2, 928 df; P<0.001), and barred tiger salamander (MANOVA F=22.5; 2, 120 df;
P<0.001) at cropland playas; SVL and mass were significant for all species (STEPBOOT
P<Om\, Table 2.1). Body size also was greater in 1999 than in 2000 for aduU Great
Plains toad (MANOVA F=63; 2, 263 df; 7^=0.002), New Mexico spadefoot (MANOVA
F=3.1; 2, 606 df; P=0.05), and barred tiger salamander (MANOVA F-65.1; 2, 198 df;
P<Q.OO\) at grassland playas; SVL and mass were significant for all species
Subadults. Body size was greater in 1999 than in 2000 for subaduh Great Plains
toad (MANOVA F=12.4; 2, 96 df; /'<0.001) and New Mexico spadefoot (MANOVA
F=3.7; 2, 239 df; P=0.03) at cropland playas; SVL and mass were significant for Great
Plains toad (STEPBOOT P<0.001), and SVL was significant for New Mexico spadefoot
(STEPBOOT P=0.04, Table 2.2). Body size was greater in 2000 than in 1999 for
21
Table 2.3. Metamorph body size of New Mexico spadefoot {Spea multiplicata. NSF).
plains spadefoot {S. bombifrons, PSF), Great Plains toad {Bufo cognatus, GPT). and
barred tiger salamander {Ambystoma tigrinum mavortium, BTS) between landuse types
and years (fixed factorial effects) at 16 playa wetlands (random nested effect) on the
Southern High Plains of Texas, April-September 1999 and 2000.
Size Landuse
Variable Species Year" Cultivation Grassland
n X SE P<0.05'' n X SE P<0.05
Snout- NSF 1999 384 28.1 0.2 Aa 119 31.9 0.4 Ba
vent 2000 51 24.5 0.4 Ab 34 22.9 0.4 Bb
length PSF 1999 76 31.8 0.4 Aa 89 36.5 0.4 B
(mm) 2000 27 25.9 0.4 b 0 0 0 NT
GPT 1999 142 25.1 0.7 Aa 202 38.2 0.7 B
2000 39 25.5 1.1 a 0 0 0 NT
BTS 1999 103 77.4 1.0 Aa 173 78.5 0.6 Aa
2000 76 72.4 0.7 Ab 121 79.4 0.7 Ba
Mass NSF 1999 384 2.8 0.1 Aa 119 3.7 0.1 Ba
(g) 2000 51 2.3 0.1 Ab 34 1.2 0.1 Bb
1999 76 3.6 0.2 Aa 89 5.6 0.2 B
PSF 2000 27 2.7 0.1 b 0 0 0 NT
1999 142 2.5 0.2 Aa 202 8.6 0.5 B
GPT 2000 39 2.8 0.3 a 0 0 0 NT
1999 103 21.4 0.9 Aa 173 22.0 0.6 Aa
BTS 2000 76 15.4 0.9 Ab 121 18.8 0.8 Bb
^Analyzed by year because landuse and year main effects were not additive (P^O.OOl) for
NSF and BTS; PSF and GPT analyzed for 1999 only.
Within-species means in the same row with unlike uppercase letters are different (i.e.,
simple effect test between landuses within years); means in the same column within size
variables and species with unlike lowercase letters are different (i.e., simple effect test
between years within landuses); all analyses performed using Westfall-Young
(1993:113-121) multivariate multiple comparison test after the simple main effect
MANOVA for factorial-nested designs (Montgomery 2001:569-573) was significant
(/*<0.001); NT indicates no test performed because SE=0 for >1 level of the effect.
22
subaduh New Mexico spadefoot (MANOVA F=3.2; 2. 139 df: P-0.04) and plains
spadefoot (MANOVA F=3.5; 2, 24 df; P=0.05) at grassland playas; SVL and mass were
significant (STEPBOOT P<0.01, Table 2.2). Body size did not differ between years for
subadult Great Plains toad (MANOVA F=0.2; 2, 222 df; P=0.78) at grassland playas.
Metamorphs. Body size was greater in 1999 than in 2000 for metamorph New
Mexico spadefoot (MANOVA F=17.9; 2, 427 df; P<0.00\), plains spadefoot (MANO\'A
F=37.5; 2. 96 df; P<0.001), and barred tiger salamander (MANOVA F=10.8: 2. 170 df
P<0.00\) at cropland playas: SVL and mass were significant for all species (STEPBOOT
P<0.006, Table 2.3). Body size did not differ between years for metamorph Great Plains
toad at cropland playas (MANOVA F=0.3; 2, 174 df; ^=0.73). Body size was greater in
1999 than in 2000 for metamorph New Mexico spadefoot (MANOVA F=16.4; 2. 145 df:
P<0.001) and barred tiger salamander (MANOVA F=20.7; 2, 285 df; P<0.001) at
grassland playas; SVL and mass were significant for New Mexico spadefoot
(STEPBOOT P<0.00\), and mass was significant for barred tiger salamander
Discussion
Landuse Effect
Body size of amphibians generally was greater at grassland playas than at playas
surrounded by cultivation for all age classes (metamorph, subadult, and adult) on the
Southern High Plains. To my knowledge, this is the first documentation of the potential
(1983) and Oldham (1985) documented in the United Kingdom that body size of Bufo
23
bufo and two species of newts {Triturus vulgaris, T. helveticus) at wetlands in cultivation
was greater than at wetlands with an uncultivated watershed (i.e.. opposite of my results).
Oldham (1985) related his resuhs to postmetamorphic density, where spring influx of
adults was 14X greater at the uncultivated wetland than at the wetland surrounded by
individuals in pitfall traps (all species combined) at cropland playas (x=648. SE=226
[1999]: x=157. SE=29 [2000]) was greater than (P<0.03, Wilcoxon 2-sample
nonparametric test, Conover 1980:216) at grassland playas (x=135, SE=20 [1999]; x=90.
SE=23 [2000]) both years (also see Chapter III). Density of conspecifics, and possibly
congeners, can affect postmetamorphic growth rate and body size of amphibians (Goater
1994, Pechmann 1994. Morey and Reznick 2001) by increasing competition for food
resources.
Beebee (1983) and Oldham (1985) also speculated that body size might be a
abundance, capture of beetles, springtails, and spiders appeared greater in pitfall traps at
grassland playas than at cropland playas (M. J. Gray and R. Brenes. Texas Tech
stomachs {n=70) of metamorph Great Plains toads was greater (P=0.005, Wilcoxon test)
vegetation and soil may negatively affect terrestrial food resources available for
24
amphibians at wetlands surrounded by cultivation (Freemark and Boutin 1995); however,
Postmetamorphic body size also may have been affected by conditions in the
larval em ironment (Wilbur and Collins 1973, Werner 1986, Brady and Griffiths 2000,
More> and Reznick 2001). Hydroperiod can affect postmetamorphic body size b}
reducing duration of larval development (e.g., Newman 1988, 1989. Denver et al. 1998).
Mean h\ droperiod (i.e., measured from the date amphibian sampling started until playas
dried) of grassland playas (x=139 d, SE=14) was greater than (P=0.05, Wilcoxon test)
cropland playas (x=89 d. SE=22) in 1999. Although 1 did not detect differences (F=0.18.
Wilcoxon test) between landuse types, mean hydroperiod in grassland playas (x=107 d.
SE=13) was greater than in cropland playas (x=69 d, SE=18) during 2000. Because
rainfall at playas was similar {P>0.32, Wilcoxon test) between landuses (x=6.4 cm,
SE=1.3 [cultivation, 1999]; x=9.2 cm, SE=1.5 [grassland. 1999]; x=4.4 cm, SE=2.2
[cultivation, 2000]; x=4.5 cm, SE=3.8 [grassland, 2000]) both years, hydroperiod
probably was shorter in cropland playas because of differential sedimentation (Luo et al.
1997).
Although I did not quantify potential herbaceous (e.g., nektonic algae. Savage
1952) and animal (e.g., anostracan shrimp. Pfennig 1992) foods between landuse types,
amphibian densities may have been lower in cropland playas because of increased
mortality of these organisms associated with agricultural runoff or drift (Freemark and
Boutin 1995). Reduced aquatic food resources can negatively influence larval growth
(e.g.. Wilbur 1977, Steinwascher 1979, Newman 19946, Walls 1998), and consequently
25
postmetamorphic body size. Sediments and agricultural chemicals also ma> have altered
the percent composition of aquatic invertebrate taxa in cropland playas (Hall 1997:68-69.
85-87), which could have influenced potential predators (Skelly 1997). If predator)
insect density was reduced in cropland playas, competition for food resources may have
been greater than in grassland playas (e.g., Werner and McPeek 1994). perhaps
directh reduce foraging acti\'ity and growth of larval amphibians, which can decrease
body size at metamorphosis (Semlitsch et al. 1995, Fioramonti et al. 1997. Boone and
Semlitsch 2001).
Amphibian predator density and size also may have been related positively to
hydroperiods (Skelly 1996). Mean daily capture of barred tiger salamander larvae and
neonates (a potential predator and cannibal. Rose and Armentrout 1976, Collins and
Holomuzki 1984) in seine net samples (0.002 ha sample area) was greater (P<0.01.
Wilcoxon test) in grassland playas (x=108, SE=16 [1999]; x=l 1. SE=4 [2000]) than in
cropland playas (x=26, SE=6 [1999]; x=0.2, SE=0.1 [2000]). Predators can positiveh
larvae, thus competition for food resources (Wilbur 1984). Alternatively, predators may
activity and size at metamorphosis (Skelly and Werner 1990), yet this effect may be
presence is primarily reducing competition (e.g., Morin 1983), thus resulting in large
26
postmetamorphic body size associated with predator-rich wetlands (i.e., grassland
playas).
One different resuU existed in my landuse results; body size of metamorph New
Mexico spadefoot tadpoles metamorphosed ca. one month earlier from playas in
cultivation (06/28/00) than from grassland playas (07/25/00), which was ca. 25 and 52 d
following breeding, respectively. Thus, metamorph New Mexico spadefoots had more
documented for spadefoot tadpoles (see references in Newman 1992 and Denver 1997).
Presumably, New Mexico spadefoot tadpoles assayed the ratio between their size-specific
mortality and growth rates (Wilbur 1984, Werner 1986) and delayed metamorphosis in
grassland playas, perhaps because water levels were more stable than in cropland playas
(as suggested by my hydroperiod data). Indeed, body size (e.g., SVL) of newly
metamorphosed individuals (i.e., captured the first day) at grassland playas (x=21.9 mm,
SE=0.4) was slightly greater than at cropland playas (x=20.7 mm, SE=0.4), but it
apparently did not compensate for the longer duration of postmetamorphic growth of
individuals in cultivation.
Year Effect
Postmetamorphic body size of amphibians was greater in 1999 (i.e., the wetter
year) than in 2000 generally for all species and age classes. Moreover, the effect of
27
landuse type and year on body size generally was non-additive. Differences in body size
between landuse treatments increased 33-545% between 1999 and 2000 for adult and
subadult New Mexico spadefoot and Great Plains toads, and metamorph barred tiger
salamanders. Rainfall can affect postmetamorphic growth rates (Tinsley and Tocque
1995) and presumably body size (Bruce and Hairston 1990, Reading 1990) by reducing
probability of desiccation (Newman and Dunham 1994) and increasing foraging (Jaeger
1980) and movement activity (e.g., Hurlbert 1969, Semlitsch \9S5b, Sinsch 1988). Mean
number of individuals immigrating and emigrating daily (all species and movement
directions combined) at my playas was greater {P<0.00\, Wilcoxon test) in 1999 (x=391,
SE=115) than in 2000 (x=129, SE=20) for both landuses. Increased opportunity for
movement may increase foraging success of amphibians (Jaeger 1980). Prey abundance
also may increase with rainfall (Dimmitt and Ruibal \9S0a). Diet of metamorphs and
adults of my study species was primarily composed of beetles on the Southern High
1981:380-381,391-396).
larval development (Reading and Clarke 1999, Camp et al. 2000). Mean larval
emergence at my playas was 43.5 and 38.5 d after breeding commenced (determined as
per ancillary call surveys) in 1999 and 2000. Interestingly, this difference in larval
(x=36 d, SE=4 [1999], 25 d, SE=2 [2000]) not in grassland playas (x=51 d, SE=2 [1999],
28
52 d, SE=2 [2000]). Thus, postmetamorphic body size may have been different between
years, and landuse and year main effects non-additive, because of greater available food
resources and duration of larval development associated with 1999 and grassland playas.
Body size of subadult New Mexico and plains spadefoots was greater in 2000
than in 1999 at grassland playas, which was contradictory to the above results. Because
body size of subadult anurans reflects growth during the previous year (see Turner 1960),
body size of subadults may have been elevated in 2000 because of additional growth
experienced during more favorable ambient conditions in 1999. However, if this was
true, then subadults captured at cropland playas should have been larger in 2000 than in
1999 (which was not observed). Possibly, the negative effect of cultivation on
postmetamorphic body size of subadults during both years counteracted the positive
Wilbur and Collins (1973) suggested that body size at metamorphosis would be
empirical studies (e.g., Semlitsch et al. 1988, Berven 1990, Scott 1994) have supported
their prediction. Werner (1986) suggested that body size at metamorphosis (and
ultimately adult body size and fitness) would be a consequence of size-specific mortality
(u) and growth rates {g) in aquatic and terrestrial habitats. Thus, an individual that
metamorphosed at a small body size due to a high larval \i/g ratio might exhibit ''catch-
up" growth because of a low postmetamorphic \i/g ratio. Morey and Reznick (2001) and
Goater (1994) supported this prediction (at least in part). Although I did not measure the
29
same indi\'iduals through time, because individuals were random samples from the
population residing within main effect treatments and body size of amphibians can var>'
references therein). I can assume that postmetamorphic bod\ size for each age class
represented mean population response of age-specific growth rate to main effects. Thus,
my results seemingly provide support for U'ilbur and Collins' (1973) h\ pothesis. because
differences in body size between levels of main effects generally were maintained
throughout all age classes (i.e.. metamorphs, subadults, and adults were consistenth
Conclusions
Landuse type surrounding wetlands and rainfall can affect postmetamorphic body
grassland playas generally were larger than those caught at playas surrounded by
agricultural cultivation, and individuals in 1999 (i.e., a wetter year) usually were larger
than those captured in 2000. Moreover, I found the effect of landuse type and year
usually was non-additi\'e; cultivation negati\ eh affected body size greater during the
amphibians, especially during drier years. Density independent and dependent factors
both may be related to landuse and year effects, and proximally responsible for
postmetamorphic bod> size can be correlated with individual survival and fitness, thus
30
population persistence of amphibians. Indeed, retention of grasslands surrounding
31
CHAPTER III
Abstract
factor in local and global amphibian declines. Few data exist on the effects of landscape
objective was to test the effect of landuse (cultivation vs. grassland [control]) on
(1999 vs. 2000) on the Southern High Plains of Texas. I partially enclosed 16 playa
wetlands (4 per landuse per year) with drift fence and pitfall traps, and monitored relative
abundance and diversity daily from 16 May-17 October 1999 and 19 April-18 August
2000. I also monitored immigration and emigration rates to investigate the effect of
landuse and year on possible source-sink dynamics. In general, abundance (i.e., daily
capture rate) of New Mexico and plains spadefoots {Spea multiplicata, S. bombifrons)
was greater at cropland than grassland playas; abundance of other species and diversity of
the amphibian assemblage was not affected by landuse. Also, abundance generally was
greater in 1999 than 2000 for metamorph spadefoots and barred tiger salamanders
(i.e., source dynamics) was greater at grassland than cropland playas and in 1999 than
2000. Differences in spadefoot abundance between landuse types may have been related
32
landscapes and absence of a keystone intraguild predator (i.e.. neotenic or large larval
frequency of direcfional movements likely was related to armual rainfall (i.e., 1999 >
Introduction
organisms (Findlay and Houlahan 1997). Most research suggests species richness and
et al. 1993, Hecnar and M'Closkey 1998, Scribner et al. 2001). However, some recent
disturbed landscapes (Anderson et al. 1999a, Knutson et al. 1999, Kolozsvary and
physically altering the aquatic (i.e., wetland) and terrestrial (i.e., upland) environments
(Wilbur 1980, Semlitsch 2000). For example, agricultural cultivation can affect water
quality and increase sedimentation in wetlands (Hanson et al. 1994, Luo et al. 1997).
Pesticides in drift and runoff can directly increase mortality of amphibian larvae or
33
indirectly affect survivorship by reducing swimming mobility, thus probability of
predator avoidance and food item acquisition (Bridges 1997, Bridges and Semlitsch
2000, Relyea and Mills 2001). Pesticides also may reduce invertebrate densities in
wetlands (Boone and Semlitsch 2001), which can function as predators, competitors, and
food for amphibian larvae (Morin et al. 1988, Skelly 1997). Herbicides in runoff can
reduce densities of phytoplankton and periphyton (Freemark and Boutin 1995), which
could reduce foraging carrying capacity of amphibian larvae (Wilbur 1997). In contrast,
nitrogen fertilizers can increase net primary productivity in wetlands (Hanson et al.
1994), which might increase foraging opportunities for larval amphibians (Leibold and
Wilbur 1992). Finally, sedimentation decreases wetland volume and hydroperiod (Luo et
al. 1997), which may decrease time available for larvae to develop (Chapter II).
cultivation can cause direct mortality of fossorial amphibians or it can unbury and expose
disturbance also reduces live and detrital vegetation, which can be important in retaining
soil moisture and ftjnction as foraging, retreat, and borrowing sites for amphibians (Dodd
1996, deMaynadier and Hunter 1998, Herbeck and Larsen 1999, Naughton et al. 2000).
(Welsh 1990), thus decrease burrowing efficiency of amphibians (Jansen et al. 2001),
which can be particularly important for fossorial species inhabiting xeric environments
34
Landscape disturbance also can affect connectivity of spatially structured
Gibbs 1998a). For example, agricultural landscapes may be more complex (e.g., contain
greater number of edges to cross. Chapter IV), thus be less permeable to interdemic
movement (Stamps et al. 1987). Agricultural fields also may contain vegetation with less
vertical structure than grasslands or forests thus be less viscous and more permeable to
movement (Wiens et al. 1993, Vos and Chardon 1997). Additionally, pesticides applied
individuals can have a greater probability of extinction, because smaller individuals are
less evolutionarily fit than larger individuals (Semlitsch et al. 1988, Berven 1990).
Several studies have been conducted in North America examining the effects of
deMaynadier and Hunter 1995, Gibbs \99Sb, Herbeck and Larsen 1999, Naughton et al.
2000). Interestingly, few data exist (cf Knutson et al. 1999, Kolozsvary and Swihart
However, recent evidence suggests that some amphibian populations may become nested
in wetlands within disturbed landscapes (Hecnar and M'Closkey 1997, Wright et al.
35
(Knutson et al. 1999. Kolozsvary and Swihart 1999). Addifionally. some species (e.g.,
American toad, Bufo americanus Holbrook) may actually benefit from disturbance
Two primary landuse types exist on the Southern High Plains (SHP) of Texas:
cultivation and intact grassland (Haukos and Smith 1994). Amphibians on the SHP
inhabit spatially structured playa wetlands (Bolen et al. 1989). Because playas are
numerous, similar structurally, and exist in their own watershed (Smith and Haukos
2002), they are ideal systems to test the effects of anthropogenic disturbance on
playa wetlands on amphibian demographics during 1999 and 2000. In agreement with
the status quo (e.g., Berger 1989, Laan and Verboom 1990, Oldham and Swan 1991:147.
Jung 1993, Bonin et al. 1997a, Hecnar and M'Closkey 1998), I hypothesized that playa
dynamics (i.e., immigration > emigration) more frequently than source dynamics (i.e..
emigration > immigration, PuUiam 1988). Thus, I examined whether relative daily
amphibians residing in playa wetlands on the SHP varied by landuse (cultivation vs.
grassland) and between years (1999 vs. 2000) that had different rainfall (Chapter II).
Methods
This study was conducted on the Southern High Plains of Texas (Sabin and
Holliday 1995). I used playa wetlands as experimental units of landuse and year effects.
36
On average, playas located in grassland and cropland were surrounded by >75% natural
playas (4 per landuse per year) via aerial and ground reconnaissance in March 1999 and
2000. Because wildlife population dynamics can be highly variable among playas (Smith
and Haukos 2002), I restricted experimental inferences to these landuses and years and
my study playas. A more detailed description of my study area, playas, and years was
Playas were partially enclosed (i.e., 25% of circumference) with 60-cm high drift
fence and 19-L pitfall traps (Dodd and Scott 1994); a cardinal quadrant was randomly
selected for placement. Fence and pitfalls were placed ca. 10 m upslope from and
parallel to the playa edge (i.e., clay-silt loam line, Luo et al. 1997). One 10-m
perpendicular lead centered at the main fence existed at each end. Vegetation underneath
the main fence and leads was removed, and the bases covered with soil to reduce
probability of trespass (Dodd and Scott 1994). Pitfalls were placed on alternate sides of
the fence at 10-m intervals with openings flush to the ground (Dodd and Scott 1994).
Water and sponges were placed in pitfalls to reduce desiccation of captured individuals
(Daoust 1991). Pitfall traps were checked alternate days in each landuse for captures
from 16 May-17 October 1999 and 19 April-18 August 2000. Pitfalls were open for
subadult, and adult). Morphological characters used to distinguish among age classes
have been described (Chapter II). Also, I did not designate a subadult category for
37
captured barred tiger salamanders (Chapter II). Individuals were toe clipped uniquely
with respect to study playa and direction of movement (i.e., immigration, emigration)
were assumed to be immigrating and emigrating if captured on the upslope and wetland
side of the fence, respectively (Dodd and Scott 1994). Individuals were released on the
opposite side of the fence after marking (Dodd and Scott 1994).
I used mean daily capture per playa as an estimate of relative daily abundance,
and captures on the upslope and wetland side of the fence as an estimate of relative daily
immigration and emigration of amphibians (Dodd and Scott 1994 and references therein.
Heyer et al. 1994:77). I assumed if any bias existed by using mean daily capture to
landuses and years. It follows that results of ratio tests of significance between levels of
effects would not be affected as per the rules for expectation (Milton and Arnold
1995:53). Mean daily diversity was calculated using the Shannon-Weaver algorithm
(Hair 1980:273). Because individuals were marked uniquely with respect to direction of
movement at capture, trespass rate per direction could be estimated from subsequent
captures. For example, if an individual emigrated during its first and second captures, it
percent trespass per direction (M. J. Gray, unpublished data) and correct immigration and
emigration equaled immigration was recorded per playa to examine source-sink dynamics
38
of the amphibian assemblage (Pulliam 1988, 1996). I assumed any bias associated with
combining species to examine source-sink dynamics was constant between landuses and
years, thus did not affect resuhs of ratio tests of significance as discussed previously.
2001:194-196), where landuse, year, and species captured were crossed-fixed effects. 1
analyzed population demographics for 7 species {Spea multiplicata Cope [New Mexico
spadefoot, NSF], S. bombifrons Cope [plains spadefoot, PSF], Bufo cognatus Say [Great
Plains toad, GPT], B. woodhousii Girard [Woodhouse's toad, WHT], Pseudacris clarkii
[spotted chorus frog, SCF], Rana blairi Mecham et al. 1973 [plains leopard frog, PLF],
Ambystoma tigrinum mavortium Green [barred tiger salamander, BTS]). Two additional
species were captured {B. debilis Girard [green toad], Gastrophryne olivacea Hallowell
[Great Plains narrowmouth toad]); however, there was insufficient capture data for
to test for differences (a = 0.05) in relative abundance of individuals (i.e., daily capture)
between landuse types and years and among species (Montgomery 2001:194-196). I
used a 2-factor ANOVA model (landuse and year main effects) with days as subsamples
index thus it was not available as an effect. Nonadditivity was tested by including all 2-
way and the 3-way interaction in the models (Montgomery 2001:194). When
nonadditivity was violated (i.e., >1 interaction significant), analyses were separated by
species to test for differences between landuses and years (i.e., 2-way ANOVAs) and by
39
landuse and year (i.e., 1-way ANOVAs) to test the species effect. For species that
way ANOVAs using landuse and year as crossed factors, but by age class, to discern
which demographic category was responsible for the observed abundance differences for
that species. Age was not included as a factor for the initial analyses, because there was
no subadult category for BTS (i.e., design was not balanced) and I assume levels of age
were not independent. Also, this analytical approach conserved experimentwise error
rate at the lowest probability for the main effect ANOVAs, because only significant
species were analyzed in the post-hoc tests for each age class. Tukey's HSD multiple
comparison test was used to test for differences among species when the main effect
ANOVA was significant (Milliken and Johnson 1992:36-38, Westfall et al. 1999:179). 1
natural-log transformed all data to meet linear model assumptions of normality and
homoscedasticity {P > 0.08 as per Shapiro-Wilk and Levene's tests, Milliken and Johnson
I used logistic regression with a generalized logit model to test for differences (a
(IM), IM exceeded EM, and EM equaled IM between landuses and years; nonaddivity
was tested by including an interaction term in the model (Agresti 1990:307, Stokes et al.
distribution per effect, and frequency distributions were independent (Agresti 1990:306).
0.015, were used to test pairwise differences in frequency distributions when the main
40
effect logit model was significant (Zar 1984:395-396, Milton and Arnold
1995:667-670).
Results
Mean daily abundance of amphibians was different between landuses and years
and among species (Appendix B). Cropland playas (x=71.5, SE=1.4) had greater
abundance of amphibians than grassland playas (x=22.4, SE=1.4), and abundance was
greater in 1999 (x=67.8, SE=1.4) than 2000 (x-23.5, SE=1.4). However, the species
effect interacted with landuse and marginally interacted with year (Appendix B);
Mean daily abundance of NSF and PSF was greater at cropland than grassland
playas, and abundance was greater in 1999 than 2000 for BTS (Appendix B, Table 3.1).
Mean daily abundance also differed among species across and within levels of landuse
and year main effects. New Mexico spadefoot were more abundant than all species at
cropland playas. Mean daily abundance of BTS, GPT, and PSF was greater than SCF,
PLF, and WHT at cropland playas. New Mexico spadefoot, BTS, GPT, and PSF were
more abundant than SCF, PLF, and WHT at grassland playas. Mean daily abundance of
NSF was greater than all other species except BTS in 1999 and 2000. Barred tiger
salamanders were more abtmdant than all other species except NSF and GPT in 1999,
and they were more abundant than other species except NSF, GPT, and PSF in 2000.
Mean daily abundance of GPT was greater than SCF, PLF, and WHT in 1999 and 2000.
41
en
^ p c« W -o -a
o < en < < < < < <
V II
'o
o »y^mmr^ooO'— 'C'r-
a, 04
00
on
C •n- 0 0 — — — —
<N
o ^»y-ir^r<^ — — — —
a
G
t-c
o
o
o
cj _o
< ^ < ^ < <
o -a -O -O "O
<
s^l
o V
o
T3 (N
C *^
c3 oi
D c/5
ON<Nm(N — — — —
C/)
O —
^ CI. • ^ '^^ f S VO - ^ (N (N
IX ^ C^ ^ ^ _ _ _
C <
iH T3
c
cd
rt ca rt C3 X5 X3 X5
2 ] < < CQ < < <
^ ON
on <U
J2
a o 00 T t m - ^ f S — 0 0 —
:a o
C cd
cd >^ — n T;!- p rsj — —, —
IX
O vd S "^ ("^ — —' ^ —
.ti X c
>
p CO x> x> X) u u u
o < < < < < < < £-c
T3 C V
cd ex. 3 II
-a T3
C
c
cd 00 a. u^ r i m m — — o
(U
o Jd i i
o
c U
cd
C
;3
^ — — — mcNcsm
cd IX ro r^ w^ ^- ^ _• ^ ^
>% o
cd
T3
t i - C/D H tin tl.1 UH C
X J«
> O 00 H Cu c/3 U J ^
o a. ; z CQ O cu C/D Cu ^
C/O
cd
o
c
Cd
« cd
.2 T3 (/I
C
3
cd ^ > X>
< •
Plains spadefoot were more abundant than PLF and WHT in 2000. No differences were
detected in amphibian diversity between landuses and years (Appendix B, Table 3.1).
Mean daily abundance of adult NSF was greater at cropland than grassland
playas; landuse and year effects interacted for metamorph and subadult NSF thus
analyses were separated by main effects (Appendix C, Table 3.2). Metamorph NSF were
more abundant at cropland than grassland playas in 1999, and they were more abundant
in 1999 than 2000 at cropland and grassland playas. Mean daily abundance of subadult
NSF was greater at cropland than grassland playas in 2000, and abundance was greater in
2000 than 1999 at cropland playas. Abundance of metamorph PSF was greater in 1999
than 2000, and adult PSF were more abundant in 2000 than 1999; landuse and year
effects interacted for subadult PSF thus analyses were separated by main effects. Mean
daily abundance of subadult PSF was greater at cropland than grassland playas in 2000,
and abundance was greater in 2000 than 1999 at cropland playas. Metamorph BTS were
more abundant in 1999 than 2000. No additional differences were detected between
landuses and years in mean daily abundance for age classes of significant species
Source-Sink Dynamics
(EM) exceeded immigration (IM), IM exceeded EM, and EM equaled IM were different
between landuses (xV26.1, P<0.001) and years (xV27.1, P<0.001); main effects did
not interact (% 2^2.8, P=0.25). Percent days EM exceeded IM (i.e., source dynamics)
43
Table 3.2. Relafive daily abundance of age classes between landuse types and years for
amphibian species that differed significantly in daily abundance (see Appendix B) at 16
playa wetlands on the Southern High Plains, Texas, May-October 1999 and
April-August 2000.
Landuse
Species^ Age Year^ Cultivation Grassland
X SE' P<0.05^ X SE P<0.05
NSF Metamorph 1999 100.5 1.7 Aa 3.1 1.8 Ba
2000 1.6 1.5 Ab 2.2 2.1 Aa
Subadult 1999 1.2 1.0 Aa 1.3 1.2 Aa
2000 11.8 1.4 Ab 2.1 1.1 Bb
Aduh 1999 8.9 1.2 Aa 3.5 1.5 Ba
2000 12.2 1.2 Aa 3.2 1.5 Ba
PSF Metamorph 1999 2.6 1.2 Aa 1.9 1.3 Aa
2000 1.3 1.2 Ab 1.0 1.0 Ab
Subadult 1999 1.0 1.0 Aa 1.1 1.0 Aa
2000 3.3 1.4 Ab 1.2 1.0 Ba
Adult 1999 1.3 1.2 Aa 1.3 1.1 Aa
2000 2.9 1.3 Ab 1.6 1.1 Ab
BTS Metamorph 1999 8.9 1.7 Aa 14.5 1.1 Aa
2000 2.6 1.2 Ab 4.7 1.5 Ab
Adult 1999 2.2 1.2 Aa 2.4 1.3 Aa
2000 2.7 1.3 Aa 3.2 1.4 Aa
^ S F = New Mexico spadefoot {Spea multiplicata), PSF = plains spadefoot {S.
bombifrons), and BTS = barred tiger salamander {Ambystoma tigrinum mavortium).
''Analyzed by year for metamorph NSF and subadult NSF and PSF because landuse and
year main effects interacted (Appendix C); statistics for remaining ages and species
presented by year for tabular parsimony.
'^n=A playa wetlands per landuse per year; xs and SEs were back-transformed from
natural-logs for presentation; transformation was necessary to meet linear model
assumptions of ANOVA.
''within-species means in the same row (i.e., within ages and years) with unlike
uppercase letters are different (i.e., landuse effect test); means in the same column within
species and ages with unlike lowercase letters are different (i.e., year effect test).
44
was greater at grassland than cropland playas (Zi=2.7, /'=0.007, Figure 3.1), and greater
in 1999 than 2000 (Zi=2.7, P<0.001). Percent days IM exceeded EM (i.e., sink
dynamics) was greater at cropland than grassland playas (Zi=4.9, P<0.00\), and greater
in 2000 than 1999 (Zi=2.4, P=0.015). Percent days EM equaled IM (i.e., neutral
dynamics) was greater at grassland than cropland playas (Zi=3.1, P=0.002), and greater
in 2000 than 1999 (Zi=2.9, P=0.004). At cropland playas, percent days IM exceeded EM
P<0.001). Also, percent days EM exceeded IM was greater than percent days EM
percent days EM exceeded IM and IM exceeded EM were greater than percent days EM
equaled IM (Zi=5.4-10.6, P<0.001), but the former did not differ (Zi=0.9-1.5,
P=0.147-0.322) between each other. In 2000, percent days IM exceeded EM was greater
than percent days EM exceeded IM and EM equaled IM (Zi=8.3-9.3, P<0.00\), but latter
did not differ (Zi=l.l, P=0.292) between each other (Figure 3.1).
Discussion
by cultivation than at grassland playas, and abundance was greater in 1999 than 2000.
However, when I combined data for all species, frequency of days emigration exceeded
immigration for the amphibian assemblage was greater at grassland than cropland playas;
emigration was greater at cropland than at grassland playas and greater in 2000 than
1999. In general, New Mexico spadefoots were more abundant than all other species,
45
0.6
0.5 :
!
ays 0.4 Ba
Q
Aa Cropland
0.3
Percei
Grassland
#»f#
0.2
0.1
4 •, * ]
0 -—^ * *
EM IM EQ
Directional Movement
11999
12000
EM IM EQ
Directional Movement
Figure 3.1. Effect of landuse (top graph) and year (bottom graph) on frequency of
directional movement of amphibians at 16 playa wetlands on the Southern High Plains,
Texas, May-October 1999 and April-August 2000; EM = percent days emigration
exceeded immigration (i.e., source dynamics), IM = percent days immigration exceeded
emigration (i.e., sink dynamics), EQ = percent days emigration equaled immigration (i.e.,
neutral dynamics); unlike uppercase letters within directional movements and unlike
lowercase letters within landuses are different (P<0.007) by a main effect logistic
regression (generalized logit model) and pairwise chi-square tests of homogeneity
Bonferroni corrected at a = 0.015 (Milton and Arnold 1995:667-670, Stokes et al.
2000:257- 259).
46
followed numerically by barred tiger salamander and Great Plains toad. No difference in
mean daily diversity of amphibians was detected between landuses and years.
Landuse Effect
landscape disturbance may have confined individuals to the remnant available habitat
(i.e., the playa, Kolozsvary and Swihart 1999). Spadefoots may have become more
nested in cropland wetlands than other species, because of their relative vagility and
perception to patch viscosity and edge permeability in adjacent agricultural fields (Wiens
1997). Indeed, geometric complexity (i.e., mean fractal dimension, FD) and edge density
(i.e., m edge/ha, ED) were greater {P<0.0\, Wilcoxon 2-sample nonparametric test,
also was greater in cropland than grassland landscapes (Appendix A), which can decrease
permeability (Yanes et al. 1995). Thus, dispersing spadefoots may have been unable to
Spadefoot vagility may have been less than other species on the Southern High
Plains, because of their relatively small postmetamorphic body size (With and Crist
positively correlated with body size for various organisms (e.g., Peters 1983:89-91, Crist
et al. 1992, With 1994). Excepting the spotted chorus frog, spadefoots were smaller
physically than other species that I monitored (Degenhardt et al. 1996:61, 81, Chapter II).
47
I would expect the spotted chorus frog to be affected similarly by landscape disturbance
connectivity between them. Differences may not have been detected in demographics
between landuses for spotted chorus frog, because they have relatively low abundance on
the Southern High Plains (Anderson et al. 1999fl, this study). Alternatively, other
responses to landuse.
consequence of changes in the trophic structure of the aquatic environment. Barred tiger
salamander may fiinction as a keystone intraguild predator {sensu Paine 1969, Polls and
Holt 1992) in playas similar to different species in other wetland systems (e.g.,
Ambystoma opacum. Walls and Williams 2001), because they can establish larval and
neotenic populations prior to anurans if water is present during winter or early spring
(Rose and Armentrout 1974, 1976). Indeed, density of larval and neotenic barred tiger
presumably due to longer hydroperiods associated with grassland playas (Chapter II).
Grassland playas generally have greater volume and longer hydroperiods than cropland
playas because of differential sedimentation (Luo et al. 1997). Consequently, barred tiger
salamander populations became established in grassland playas before anurans bred both
years (M. J. Gray, unpublished data), which may have resulted in top-down control on
spadefoot populations via predation on their eggs and larvae (Henrickson 1990, Walls
and Williams 2001). Insect predators (e.g., Anax, Dytiscus) also may have been abundant
48
in grassland playas due to their longer hydroperiods (Schneider and Frost 1996). In
contrast, cropland playas had shorter hydroperiods and were dry until intense rains early
salamanders, anurans, and insects. Thus, salamander and insect larvae likely interacted
with spadefoot larvae as competitors not predators in cropland playas (Morin 1983.
Morin et al. 1988). Other predators of amphibian larvae (e.g., wading birds, fish) occur
infrequently in playa wetlands during summer (Bolen et al. 1989); thus, cropland and
grassland playas may have represented relatively predator-free and -rich environments,
respectively (Heyer et al. 1975). The aforementioned may have facilitated greater
grassland.
Spadefoot larvae may have been more susceptible to predation in grassland pla\ as
than other species on the Southern High Plains, because of their relative palatabilit\ and
environments worldwide and exploit ephemeral water sources with few vertebrate or
evolved physiological mechanisms for rapid growth (Richmand 1947, Newman 1992) but
lack the ability (e.g., toxicity, cryptic coloration, altered microhabitat use) to efficienth
escape predators (e.g., Spea bombifrons, Kruse and Francis 1977). For example,
spadefoot tadpoles are more active than tadpoles of certain Bufo, Rana, and Pseudacris
species; thus, they are more conspicuous and vulnerable to predators (Morin 1983.
Woodward 1983, Dayton and Fitzgerald 2001). Spadefoot tadpoles also lack dermal
49
toxins and cryptic coloration unlike many species of Bufo, Rana, and Pseudacris (e.g.,
Walters 1975, Formanowicz and Brodie 1982, Peterson and Blaustein 1991, Smith and
VanBuskirk 1995).
High Plains (e.g., Morin 1983, Wilbur 1987, Dayton and Fitzgerald 2001). Thus, in the
absence of predation, spadefoots might outcompete other species for food and other
resources resulting in increased recruitment (Woodward 1982, Wilbur 1987, Dayton and
Fitzgerald 2001). Indeed, several elegant lab experiments have documented that
predators are absent (e.g., Morin 1981, 1983, Wilbur 1987). Unfortunately, no studies
ability among all my species simultaneously; thus, I only can speculate that
have positively influenced spadefoot populations at cropland playas. Smith and Haukos
(2002) noted that plant diversity and structure could be greater in cropland playas, which
could increase food resources and escape cover for larvae. Nitrogen influx from
fertilizers may have increased food resources for larval amphibians in cropland playas
(Leibold and Wilbur 1992). Additionally, pesticide drift and runoff into cropland playas
may have reduced aquatic insect densities (Boone and Semlitsch 2001), which can
compete with amphibian larvae for food resources (Morin et al. 1988). Similarly,
50
cultivafion and chemical application may have reduced invertebrate densities and cover
in the surrounding landscape near cropland playas (Freemark and Boutin 1995), inducing
suggestions were strictly true, I should have observed similar demographic responses by
species other than spadefoots. Thus, these hypotheses probably are not ultimate factors
Lastly, subadults were the only age class of plains spadefoots that was greater in
cropland than grassland playas. I am uncertain why the other age classes did not respond
individuals >1 year that did not exhibit sexual characteristics thus probably did not breed
(Chapter II). Amphibians may forego breeding if resource conditions are not adequate
(Semlitsch et al. 1996). Perhaps, cropland landscapes contained less food and other
resources than grassland landscapes (Freemark and Boutin 1995), resulting in greater
number of individuals at cropland playas that did not breed. However, if this was true, I
should have observed similar elevated subadult levels with other species. Alternatively,
increased New Mexico spadefoot abundance at cropland playas may have induced
delayed breeding of plains spadefoots, because these species are known to compete for
resources (Pfermig and Murphy 2000). Thus, abundance of subadult plains spadefoots
may have been a consequence of interspecific competition with New Mexico spadefoot
51
Year Effect
metamorph New Mexico and plains spadefoots, and barred tiger salamanders in 1999
than 2000. Demographic differences between years likely were related to rainfall
(Berven 1995, Semlitsch et al. 1996). Average monthly rainfall at my playas was greater
in 1999 than 2000 (Chapter II). Rainfall stimulates amphibian emergence (Dimmitt and
Ruibal \9S0b) and positively affects intra- and inter-demic movement (Hurlbert 1969,
Sinsch 1988). Adult survival also can be positively related to rainfall (e.g., Berven
1990), presumably because density of food resources increases (Dimmitt and Ruibal
1980a) and probability of desiccation decreases (Jaeger 1980, Newman and Dunham
1994). Probability of breeding, larval survival, and juvenile recruitment also increases
with rainfall, because wetland hydroperiods are positively correlated with rainfall
(Pechmann et al. 1989, Sinsch and Seidel 1995, Semlitsch et al. 1996). Indeed, mean
duration of hydroperiods at my study playas was greater in 1999 than 2000 (Chapter II).
cropland playas, suggesting that disturbance may affect recruitment greater during drier
Mexico and plains spadefoots was greater in 2000 than 1999 at cropland playas, and adult
plains spadefoots were more abundant in 2000 than 1999. Subadult spadefoot abundance
cropland playas (Semlitsch et al. 1996, Meyer et al. 1998). Also, food densities may have
52
been lower during the drier year (i.e., 2000) at cropland playas (Dimmitt and Ruibal
subadults. I am uncertain why adult plains spadefoots were more abundant during 2000
in both landuses, but perhaps drier conditions during this year facilitated competitive
Source-Sink Dynamics
the assemblage was greater at grassland than cropland playas and in 1999 (i.e., the wetter
year) than 2000. Grassland playas may have served as sources more often than cropland
playas, because of their longer hydroperiods (Chapter II). Consequently, larvae may
have metamorphosed more continuously from grassland than cropland playas, facilitating
the former to function as a more constant source of emigrating propagules. Also, abiotic
conditions (e.g., relative humidity) may have been more ideal in landscapes surrounding
grassland than cropland playas for interdemic movement, particularly the exodus of
Source dynamics likely were greater in 1999 than 2000, because favorable
Lannoo (1998) suggested that source dynamics would occur more frequently during
dry before larvae can attain a critical body-size threshold for metamorphosis (Semlitsch
53
Conservation Implications
and chorus frogs, may be affected more by disturbance, resulting in greater nestedness of
movement or reduced home ranges. Moreover, more individuals does not necessarily
imply better conditions for a population (Yeargers et al. 1996). Populations with elevated
abundance may be less stable, particularly if they exceed their carrying capacity or have a
high intrinsic rate of increase (Edelstein-Keshet 1988). Pathogenic incidence also can be
disturbance may negatively affect genetic structure of populations (Reh and Seitz 1990)
and fitness correlates, such as postmetamorphic body size of amphibians (Chapter II).
My source-sink results for combined species also suggest that playas in cultivated
landscapes may serve as sinks more often than sources, perhaps because of wetland
chemical and mechanical disturbance in and around wetlands on the trophic structure of
54
invertebrate and amphibian communities needs to be investigated. Pathogenic incidence
in amphibians (i.e., viruses, fungi, bacteria, and parasites) also should be quanfified in
disturbed and undisturbed wetland systems. Estimates of landuse type and boundary
permeability to amphibian movement also are needed. This latter suggestion would
enable landscape ecologists to assign permeability and hardness indices to cover types
and boundaries in spatial analysis programs (e.g., FRAGSTATS, McGarigal and Marks
1995), and subsequently estimate relative connectivity of habitat patches and extinction
probabilities for local populations. Finally, my year results suggest that annual rainfall
55
CHAPTER IV
Abstract
populations, particularly with less \agile organisms such as amphibians. I examined the
amphibians inhabiting the Southern High Plains of Texas. Amphibian populations were
monitored using pitfall traps and drift fence located at 16 playa wetlands (8 playas/year)
in 1999 and 2000. I quantified landscape structure surrounding each playa via estimating
13 spatial metrics that represented playa isolation and landscape complexity using remote
sensing from aerial photographs and geocorrection and spatial analysis software.
complexity. Great Plains toads {Bufo cognatus) and barred tiger salamanders
{Ambystoma tigrinum mavortium) usually were negatively associated with spadefoots but
viscosity and edge permeability. Spatial separation of these species in the ordination also
may have been a consequence of differential competitive ability for food or other
56
resources in aquatic and terrestrial environments. Results in Chapter III suggested
amphibian populations. I also present multiple regression models for predicting relative
information can be used by landscape ecologists to locate potential sites for amphibian
conservation initiatives.
Introduction
and the geometric matrix in which they are embedded (Fahrig and Merriam 1994).
are not optimally juxtaposed (Lefkovitch and Fahrig 1985, Burel 1989). Moreover, as
decrease because boundary density increases (Stamps et al. 1987, Wiens 1997).
probability of extinction than those more optimally positioned within simpler landscapes
Anthropogenic modifications of habitat patches and the landscape matrix have been
57
implicated in many declines (Blaustein et al. 1994). Various studies have demonstrated
Marsh and Pearman 1997. Pope et al. 2000, Scribner et al. 2001). Anthropogenic
anthropogenic cover types or permeability of edges (Durelli et al. 1990. With 1994,
edge density and less terrestrial habitat, resulting in increased nestedness and abundance
On the Southern High Plains (SHP), amphibians exist primarily in playa wetlands
(Bolen et al. 1989). General landscape use (i.e., cultivation vs. grassland) can affect
amphibians can interact among playas via dispersal (M. J. Gray, unpublished data),
estimating various spatial metrics and relating them to relative daily abtmdance of 4 SHP
species {Spea multiplicata Cope [New Mexico spadefoot], S. bombifrons Cope [plains
58
spadefoot], Bufo cognatus Say [Great Plains toad], Ambystoma tigrinum mavortium
Green [barred figer salamander) at 16 playas (8/year) during 1999 and 2000. I
playa isolation (Loman 1988, Laan and Verboom 1990, Vos and Stumpel 1995, Lehfinen
et al. 1999). I also hypothesized that landscape complexity would be positively related to
abundance (Knutson et al. 1999, Kolozsvary and Swihart 1999). Understanding the
Methods
This study was conducted at 16 playa wetlands on the Southern High Plains of
Texas. Playas were partially enclosed (i.e., 25% of circumference) with 60-cm high drift
fence and 19-L pitfall traps (Dodd and Scott 1994). Pitfall traps were checked alternate
days for captures from 16 May-17 October 1999 and 19 April-18 August 2000.
more detailed account of my study area, amphibians, and sampling protocol was
sensing from aerial photos and geographical information system and spatial analysis
software. Landscapes («=16) were 2,830-ha circular plots (i.e., 3-km radius) with their
origins positioned at the center of each study playa (Appendix D). I chose a 3-km radius
to delineate the landscape plot, because this distance probably was the near maximum
59
dispersal distance for my species (cf Gehlbach 1967, Gehlbach et al. 1969, Sinsch 1990,
1997, Miaud et al. 2000). Aerial images of my landscapes in summers 1999 and 2000
were obtained from the Farm Service Agency (FSA) of the U.S. Department of
Agriculture in Crosby, Floyd, Hale, and Castro counties, Texas. Ground control points
were identified on images and U.S. Geological Survey 7.5-minute quadrangle maps then
ERDAS® and cover types (i.e., crop type, replanted Conservation Reserve Program
grass, and natural grass) were identified in each landscape plot using 1999 and 2000 FSA
farm folders. Boundaries of cover types were digitized in ERDAS® then exported as an
ESRI® ARC/INFO coverage. Coverages were cleaned, built, and polygons classified as
per cover types in ESRI® ARC/INFO. Landscape structure was quantified per plot using
following 13 spatial metrics. Shape index of the study playa (PSI), study playa size (PS),
mean nearest-neighbor distance from the study playa to surrounding playas (PNN), mean
nearest-neighbor distance from all playas to each other (MNN), number of playas (NP),
(IJI) were used to quantify relative spatial positioning and isolation of playa wetlands
(McGarigal and Marks 1995). Mean number of edges to cross from the study playa to
surrounding playas (PED), edge density (m edge/ha, ED), landscape shape index (i.e.,
measure of geometric complexity, LSI), landuse (i.e., cover type) richness (LR), Shannon
evenness index of landuses (SEI), and Shannon diversity index of landuses (SDI) were
used as measures relative landscape complexity (McGarigal and Marks 1995). I direct
readers to McGarigal and Marks (1995) for spatial metric formulae and algorithms.
60
Unity was assigned to cover type and edge permeability in FRAGSTATS* ARC®
(McGarigal and Marks 1995), because relafive viscosity for my species is unknown.
I used mean daily capture per species per playa as an index of relative daily
abundance (Dodd and Scott 1994). Canonical correspondence analysis (CCA) was used
the amphibian assemblage (ter Braak 1986, 1994). Correlated metrics (P<0.05) were
removed prior to CCA to reduce probability of an arch effect (ter Braak 1995:139).
Because CCA is sensitive to outliers and bimodally distributed data, mean daily
abundance per species was natural-log transformed prior to analysis (ter Braak 1995); all
resulting distributions were unimodal with <2 outliers. A global Monte Carlo
permutation test was performed to test for existence of a relationship between landscape
V
metrics and species composition (ter Braak and Smilauer 1998:47-49, 124-125). A
graphically examine the pattern of variation in relative species abundance with landscape
metrics (ter Braak 1995, ter Braak and Smilauer 1998). Biplot axes were interpreted by
examining the sign and magnitude of their respective intra-set correlations and using
(i.e., the arrows) in the biplot was interpreted as the strength of correlation between the
metric and species abundance (ter Braak 1995:141-142). Therefore, long eigenvectors
were most important in affecting the species assemblage (ter Braak 1995:141). Relative
correlative ranking of species with respect to the metrics was graphically represented by
extending each eigenvector through the origin of the biplot and intersecting it with
orthogonal lines drawn from the species (ter Braak 1995:143). Species positioned near
61
the arrow- and blunt-end of the eigenvector were most positively and negafively
Braak and Smilauer 1998). Thus, I also calculated Pearson coefficients of correlafion and
tested the relationship of each metric with species-specific abundance (Milton and Arnold
also were regressed linearly using least-squares estimation, and univariate prediction
models developed (Milton and Arnold 1995:386-391). Finally, I used multiple linear
regression with stepwise selection (entry and stay significance level at a=0.15) to develop
models for predicting relative daily abundance per species using all landscape metrics
and 2 indicator variables (i.e., general landuse [O=cropland, l=grassland] and year [0=wet
year, l=dry year]) as potential explanatory variables (Milton and Arnold 1995:497-501,
because results in Chapter III suggested that general landuse and annual rainfall could
[intercept adjusted]) among predictor variables in the final models was examined via
estimating variance inflation factors (VIF) and condition numbers (CN); VIF > 10 and
CN > 30 were suggestive of coUinearity between >1 variable (Freund and Littell
linear interpretation and predicting relative daily abundance of amphibians using field
Coefficients of determination adjusted for number of variables in the model (i.e., R adj)
62
were presented as a measure of model performance (Freund and Littell 2000:23). Mean
daily abundance per species were natural-log transformed for all correlation and
Results
Six of the 13 landscape metrics (PNN, NP, PED, ED, LSI, and SDI) were not
used in the canonical correspondence analysis, because they were correlated with other
metrics (Table 4.1). The global Monte Carlo permutation test based on 199 permutations
revealed that species composition was associated (F=2.201, P=0.05) with landscape
metrics. The first two canonical axes were most important in explaining variation in the
these SHP amphibian species. Intra-set correlations suggested that axes 1 and 2
landscape complexity, respectively, because of the large absolute magnitude of PP and IJI
for axis 1 and SEI for axis 2 (Table 4.2). The dimensionless species-environmental biplot
of axes 1 (ordinate) and 2 (abscissa) suggested that SEI was most positively correlated (as
per eigenvector length) with the pattern of species abundance (Figure 4.1a). Playa size,
PP, IJI, and LR were moderately correlated and PSI and MNN least correlated with
eigenvectors suggested that both spadefoot species were most positively associated with
PS, PP, IJI, and LR; barred tiger salamander and Great Plains toad generally were
63
O
to O
OS d
V
oo Os C7\ m
m oo oo o
T^ o so o
t) H d d d d
o m CN -^ •<:l-
o (N r<-) r^ o
o
d d ON CN r<^
V o o o
O C3 ON »o -^ ON
OS O Q -^ »r^ OS vo
OS . O oo csl (N
o ^ O "^ d d d d
"^o*^c:50<50\J2r--S5<N
(-5 »'i v^w -^ 0\ 2 ON
• ^ O ^. "^ "^ ON
^ ' ^ ^ ^ d d o o o o
r - m r - — fNtNONONSIJ-^moN
O N O C N l O C N O V O i / O " — ' o - ^ v o
voot~-or~-om-^^r^ooo
o o o o o o o o o o o
oo -^ t ^ ON o o r -
r-. vo r- — oo m oo
O <N CN ON ^ ON
CO ^ CU 'sf o lO o u-^ voocncs<^m*^r-
o o o o o o o o o o ' ^ o ' ^ o
voONmf<^ONT}-ONONOO^^u-ioo^m^^(N
G- o - ^ v o f N ' ^ o o m O N c n o o ON W-i CO —
c n t N i r j O r n — c o - ^ m c N ^ - vo ON O 0 \
d d d d d d d d d d d d oI d d d
^»r^zroo;3^r~-r"r-^ ^ ^ o
- vo -^ oo
^. O
_ ' p. ON ^ . O
o ^~. u-) <N r«-, CN (^ —;. v_o —- VO O oo
< = ' d < = = ' d < = > d ' = ' d o d < = > d- • ^' " ' S- • -<z>
• d <=i ^
-^•^r-^co-^ONcnoooovovovouooor^iriOomr^tN^^
r^ONO-^ — »/^r-)Oor~-mcnoosvocncnr-i^j^^'-'ON
o r ~ - ( N ' ^ o a \ c n c N O O N O o o — ^O'— i r > c s c n < ^ ' ^ ' ^ . vo
d o d d o d d o o o o o o o o o d d p d
*/^ r~- vo ^,
en f!:^voOOcMOsco-<^cnONOO«y^Oscnr^^ — r J t N
t-- !_ o ^ •£'^^cnvOc<^'^NOcnOcnONCNcnS^^O
-: u-^ ^ o
^ i r j P o o - - » n c s c n - ^ ^ o o o o o s g ' ^ <^. e n < ^ <N
'^<d>'^<6ci<6c><zi<d><zi<d><6d><:6 ,- d ^O <zi '^ <z:>
<=> d 9 d
Q
CO 00
a. OH
OH
Q C/3
00
Q
C/3
64
T3
• * - '
O 2 eu
<V
o
<u
•4—'
c
o
<U
O
C/3 a-^
O
Q3 .^
OX)
o
..—t
T3
>-i
c ^<
en N 00
on
OH
00
>^-s
3
o
C/3 CO C/5
»-l II H <L>
*o
I
<U
o
CO
oo
(N
<L)
c«
-a VO
c
o
w
u. (U
ft (/3 -TH
^-H
cd
'St o
<U T3
c«
3
Cd
§ a ^ g | >>&
h-1
H ca Quo c/j Cd DH O
65
Table 4.2. Intra-set correlations of landscape^ metrics associated with the first two axes
generated by a canonical correspondence analysis of mean daily abundance of New
Mexico spadefoot {Spea multiplicata), plains spadefoot {S. bombifrons). barred tiger
salamander {Ambystoma tigrinum mavortium). and Great Plains toad {Bufo cognatus) at
16 playa wetlands on the Southern High Plains, Texas, 1999 and 2000.
Metric^ Correlations^
Axis 1 Axis 2
PSI 0.0202 -0.1854
PP -0.6970 0.1859
IJI -0.7625 -0.0788
MNN 0.1489 -0.0134
LR -0.5559 -0.2485
SEI 0.0092 -0.7729
PS -0.3993 0.4379
Landscapes {n=\6) were 2830-ha circular plots (i.e., 3-km radius) with their origins
positioned at the center of each study pla\a.
PSI = shape index of study playa. PP = percent aerial coverage of playas. IJI =
interspersion/juxtaposition index of playas, MNN = mean nearest-neighbor distance from
all playas to each other. LR = landuse richness, SEI = Shannon evenness index of
landuses, and PS = study playa size (McGarigal and Marks 1995).
'^Intra-set correlations are standardized and dimensionless thus the\ can be interpreted as
the strength and direction of metric-specific correlation and species abundance in the
presence of all other metrics (ter Braak 1995:140): I infer that axes 1 and 2 explained
variation in species abundance with respect to wetland positioning and landscape
complexity because of the large absolute magnitude of PP and IJI for axis 1 and SEI for
axis 2.
66
67
Figure 4.1. Canonical correspondence analysis of mean daily capture (natural-log
transformed) of amphibians and uncorrelated metrics in landscapes (i.e., 2,830-ha circular
plot) associated with 16 study playas on the Southern High Plains, Texas, 1999 and 2000.
a) Species-environmental (i.e., -landscape metric) biplot (ter Braak 1995:142); axes 1
(ordinate) and 2 (abscissa) are dimensionless and represent wetland positioning and
landscape complexity, respectively (see corresponding intra-set correlations in Table 2);
length of eigenvectors (i.e., their respective eigenvalue) indicates the strength of
correlation between the variable and pattem of variation in species composition (ter
Braak 1995:141); species more close to ends of eigenvectors are more positively
correlated with it; PS = study playa size, SEI = Shannon evermess index of landuses, PP
= percent aerial coverage of playas, PSI = shape index of study playa, MNN = mean
nearest-neighbor distance from all playas to each other, IJI = interspersion and
juxtaposition index of playas, and LR = landuse richness (McGarigal and Marks 1995).
b) Inferred ranking of species along variables based on biplot interpretation of Part a of
figure (ter Braak 1995:143); ranking constructed by extending eigenvectors through the
origin and intersecting with orthogonal lines from the species to the vector; the vertical
segment represents the origin (i.e., centroid) of the biplot and is the grand mean of each
variable; species more close to the arrow or blunt end are positively or negatively
correlated with the variable, respectively. NSF = New Mexico spadefoot {Spea
multiplicata), PSF = plains spadefoot {S. bombifrons), GPT = Great Plains toad {Bufo
cognatus), and BTS = barred tiger salamander {Ambystoma tigrinum mavortium).
a)
PS
PP
' ..^^^pspy i GPT
•
•
IJI ^,..-'^^^?.Y
I
PSI
LR
-
' jSEI
I r — - - - " - T — - t f T • T ---—,-- f 1 f T T
-1.0 + 1.0
p N G B P G B N
s S P T S P T S
PS F T S F T S F SEI
b)
P N G B p G B N
S S P T s P T S
PP F F T S F T S F PSI
^ - •
N P G B P N G B
S S P T S S P T
IJI F F T S F F T S MNN
^4- —•
N P B G
S S T P
LR F F S T
68
negatively associated with these spatial metrics (Figure 4.\b). In contrast, barred tiger
salamander and Great Plains toad were most positively and the spadefoots most
negatively related with MNN. Plains spadefoot also appeared to be negatively associated
Mean daily abundance was correlated linearly with PP, IJI, ED, LSI, PED. and
LR for New Mexico spadefoot and the first 4 aforementioned metrics for plains
spadefoot; significant correlations were not detected for Great Plains toad or barred tiger
salamander (Table 4.3). Approximately 26%, 27%, 45%, 47%, 48%, and 53% of the
respectively (Figure 4.2). Approximately 30%, 33%, 35%, and 35% of the variation in
daily abundance for these amphibian species (Table 4.4). Approximately 78% of the
variation in mean daily abundance of plains spadefoot was explained by PP, SEI, PSI, IJI,
and PNN (F5,io=l 1 -62, F<0.001). Approximately 73% of the variation in mean daily
abundance of New Mexico spadefoot was explained by LU, SDI, MNN, and PSI
(F4,i 1=11.02, F<0.001). Forty-one percent of the variation in mean daily abundance of
Great Plains toad was explained by MNN, PSI, NP, and PNN (F4,ii=3.59, F=0.042).
69
Table 4.3. Univariate Pearson coefficients of correlation between mean daily abundance
of amphibians and landscape^ metrics at 16 playa wetlands on the Southern High Plains.
Texas, 1999 and 2000.
Species'^
Variable'' NSF PSF BTS GPT
r P r P r P r P
PSI -0.138 0.612 -0.175 0.515 -0.102 0.708 -0.284 0.286
PS -0.011 0.970 0.274 0.304 -0.477 0.062 -0.138 0.611
PNN -0.199 0.459 -0.284 0.287 0.102 0.708 -0.166 0.538
MNN -0.405 0.119 -0.299 0.259 -0.322 0.224 -0.414 0.111
NP 0.432 0.095 0.388 0.138 -0.023 0.933 0.082 0.763
PP 0.506 0.046 0.592 0.016 -0.352 0.182 -0.073 0.787
IJI 0.726 0.002 0.589 0.016 -0.036 0.896 0.063 0.816
PED 0.518 0.039 0.457 0.075 -0.353 0.180 -0.186 0.492
ED 0.694 0.003 0.548 0.028 -0.234 0.383 0.048 0.858
LSI 0.687 0.003 0.571 0.021 -0.256 0.338 0.046 0.863
LR 0.670 0.005 0.249 0.351 -0.085 0.754 0.247 0.356
SEI 0.219 0.414 -0.473 0.065 0.104 0.701 -0.077 0.777
SDI 0.421 0.105 -0.297 0.263 0.055 0.838 0.032 0.905
landscapes (« = 16) were 2830-ha circular plots (i.e., 3-km radius) with their origins
positioned at the center of each study playa.
PSI = shape index of study playa, PS = study playa size, PNN = mean nearest-neighbor
distance from study playa to surrounding playas, MNN = mean nearest-neighbor distance
from all playas to each other, NP = number of playas, PP = percent aerial coverage of
playas, IJI = interspersion/juxtaposition index of playas, PED = mean number of edges to
cross from study playa to surrounding playas, ED = edge density (m edge/ha), LSI =
landscape shape index (i.e., measure of geometric complexity), LR = landuse richness,
SEI = Shannon evenness index of landuses, and SDI = Shannon diversity index of
landuses (McGarigal and Marks 1995).
"^NSF = New Mexico spadefoot {Spea multiplicata), PSF = plains spadefoot {S.
bombifrons), BTS = barred tiger salamander {Ambystoma tigrinum mavortium), and GPT
= Great Plains toad {Bufo cognatus).
70
71
Figure 4.2. Simple linear regression and 95% confidence bands of mean daily abundance
(natural-log transformed) of New Mexico spadefoot {Spea multiplicata) and percent
aerial coverage of playa wetlands (PP), interspersion/juxtaposition index (IJI, McGarigal
and Marks 1995:103), edge density (ED, McGarigal and Marks 1995:106), playa edge
density (PED [i.e., mean number of edges to cross from study playa to surrounding
playas), landscape shape index (LSI, McGarigal and Marks 1995:109), and landuse
richness (LR, McGarigal and Marks 1995:119) in landscapes (i.e., 2,830-ha circular plot)
associated with 16 study playas on the Southern High Plains, Texas, 1999 and 2000.
6 —r
0}
o a>
c o
ca c
^—^ OJ
• • . ^ — - ' • ' ' ' ^
•D
c ^^^-^"''''^ ' . C
13
jQ
ailyA
< 3 —
ro
2 —
c Q
to
c
nj
Y = 0 6905 + 0.7074X
R-square = 0.256 Y = 0 4447 + 0.0534X
R-square = 0.527
0 —
~~\ I
10 20 30 40 50 60
PP
IJI
6 —r
(V 5 —
o <u
c o
ro c
c T3
Q
< c
3 —
TO
<
2 —
TO 2 —
C
CD Q
i? 1
Y = -0.92 + 0 0726X c 1 — Y = 0.2842 + 0 3078X
TO R-square = 0.269
R-square = 0 482
"T" —r 0 —
T 1 r I
30 40 so 60 70 6 7 8 13
9 10 11 12 14
ED PED
6 —r
5 —
0) 5 —
O
c
CO
4 — •D
c c
< 3 —
<
CO
:o Q
c
(0 1 —
1 —
Y =-1.3868+ 0 3254X
Y = -0.84 + 0.4482X 0 — R-square = 0.449
R-square = 0.472
I I I r
9 10 11 13 1S
LSI LR
72
25 — /! S —
O •
C 2.0 —
O 20 — CO
C TD
(0 C
•D 3
C 15 —
13 1.5 —
n <
^^,-^^^
. . * • • • '
<
'co 1.0 —
= 10 Q
(0 •• .
c
Q CO
c CU 05 — • ,.-"
CO 05
0) •
Y = -0.033 + 0 3498X Y = 0.2139+ 0.0184X
00 — R-Square = 0 348
R-Square = 0 35
0.0 — 1 1 1 1
10 20 30 40 50 60 70
PP IJI
0) 2 — (U 2 —
O
o c
c
m CO
T3
"D C
c 3
3 XI
.Q <
< CO
Q Q
c c
CO CO
0)
Y =-0 22 + 0.0243X o
Y = -0.26 + 0 1577X
0 — R-Square = 0.301
R-Square = 0.326
"T" "T
30 40 50 60 70 10 11
ED LSI
Figure 4.3. Simple linear regression and 95% confidence bands of mean daily abundance
(natural-log transformed) of plains spadefoot {Spea bombifrons) and percent aerial
coverage of playa wetlands (PP), interspersion/juxtaposition index (IJI, McGarigal and
Marks 1995:103), edge density (ED, McGarigal and Marks 1995:106), and landscape
shape index (LSI, McGarigal and Marks 1995:109) in landscapes (i.e., 2,830-ha circular
plot) associated with 16 study playas on the Southern High Plains, Texas, 1999 and 2000.
73
Table 4.4. Multiple linear regression between mean daily abundance of amphibians and
landscape^ metrics at 16 playa wetlands on the Southern High Plains, Texas, 1999 and
2000.
Estimates*^ Collinlearity'
Species^ Variable"'^ Parameter Standardized / P VIF CN
NSF Intercept 9.869 0 2.35 0.039 0 0
LU -2.077 -0.703 -4.91 <0.001 1.13 1.0
SDI 1.203 0.319 2.24 0.046 1.12 1.14
MNN -0.002 -0.278 -1.98 0.073 1.08 1.39
PSI -5.634 -0.244 -1.67 0.124 1.19 1.58
PSF Intercept 9.199 0 4.53 0.001 0 0
PP 0.190 0.322 2.23 0.049 1.418 1.0
SEI -2.299 -0.458 -3.56 0.005 1.125 1.453
PSI -6.469 -0.665 -4.23 0.002 1.642 1.546
IJI 0.012 0.379 2.48 0.033 1.598 2.175
PNN -0.0005 -0.372 2.26 0.047 1.843 2.587
BTS Intercept 2.619 0 11.51 <0.001 0 0
YR -0.939 -0.615 -2.92 0.011 1.0 1.0
GPT Intercept 17.36 0 3.72 0.003 0 0
MNN -0.003 -0.759 -2.74 0.019 1.957 1.0
PSI 9.220 -0.756 -2.87 0.015 1.754 1.231
NP -0.157 -0.599 -2.14 0.056 1.990 2.285
PNN -0.001 -0.516 -1.92 0.082 1.838 2.731
^Landscapes {n= 16) were 2830-ha circular plots (i.e., 3-km radius) with their origins
positioned at the center of each study playa.
hSlSF = New Mexico spadefoot {Spea multiplicata), PSF = plains spadefoot {S.
bombifrons), BTS = barred tiger salamander {Ambystoma tigrinum mavortium), and GPT
= Great Plains toad {Bufo cognatus).
"^Landscape metrics retained as per stepwise selection using default entry and stay
significance level at a=0.15; all overall F-tests on final models were significant
{P<Q.042) and coefficients of determination adjusted for number of variables in the
model (i.e., /?%) = 0.728, 0.779, 0.378,and 0.409 for NSF, PSF, BTS, and GPT,
respectively.
PSI = shape index of study playa, PNN = mean nearest-neighbor distance from study
playa to surrounding playas, MNN = mean nearest-neighbor distance from all playas to
each other, NP = number of playas, PP = percent aerial coverage of playas, IJI =
interspersion and juxtaposition index of playas, SEI = Shannon evenness index of
landuses, and SDI = Sharmon diversity index of landuses (McGarigal and Marks 1995);
LU (landuse) and YR (year) were indicator variables using 0 = cropland watershed and 1
= grassland watershed, and 0 = 1999 (wet year) and 1 = 2000 (dry year), respectively.
74
Table 4.4. Continued.
^Relative natural-log (LN) abundance per species can be estimated via multiplying field
estimates of landscape metrics or indicator variables by parameter estimates, and
summing them and the intercept within species; predictions can be back-transformed to
original units by exponentiation; magnitude and sign of species-specific standardized
estimates can be interpreted as the strength and relation of the variable with LN
abundance in the presence of the other retained explanatory variables (Myers
1990:384-385).
CoUinearity diagnostics (intercept adjusted) were variance inflation factors (VIF) and
condition numbers (CN); VIF > 10 and CN > 30 are suggestive of a linear dependency
between >1 variable (Freund and Littell 2000:98-101).
75
Finally, 38% of the variation in mean daily abundance of barred tiger salamander was
(Table 4.4).
Discussion
Subsequent univariate analyses indicated New Mexico and plains spadefoots generally
were positively associated with spatial metrics representing optimal spatial positioning of
landscape complexity (i.e., landscape shape index, edge density). Great Plains toad and
barred tiger salamander usually were negatively associated with spadefoots (as per CCA),
but corresponding univariate analyses indicated they were not associated with spatial
metrics. Thus, these results suggest Great Plains toad and barred tiger salamander may
be less affected by spatial positioning of playas and complexity of the landscape matrix
but perhaps are poorer competitors than spadefoots because of their negative association
Spadefoots probably are associated with landscape structure more than the other
species because of their relatively small body size (Chapter II). Body size has been
shown to connote vagility (Peters 1983:89-91, With and Crist 1995), seemingly due to
differential perception to patch viscosity and edge permeability (Crist et al. 1992, With
1994, Wiens et al. 1997, Mclntyre 2000). Although relative vagility has not been
compared among these amphibians, dispersing spadefoots may have been unable to
76
penetrate geometrically complex landscapes, resulting in increased abundance in natal
1999, Kolozsvary and Swihart 1999, Chapter III). Similarly, abundance of spadefoots
may have been positively related with less isolated and more optimally juxtaposed playas
can increase with decreasing inter-patch distance and increasing patch juxtaposition
(Bascompte and Sole 1996, Hess 1996). Consequently, abundance and persistence of
less vagile species can be positively related to habitat patch positioning thus the exchange
of individuals among spatially structured local populations (Ritchie 1997, Stacey et al.
1997). Therefore, these results support my hypotheses and previous applied and
Great Plains toad and barred tiger salamander may have been negatively
(Wilbur 1984). Spadefoot larvae have been shown to be competitively dominant over
various genera (e.g., Bufo, Gastrophryne, Hyla, Rana, Morin 1983, Wilbur 1987, Dayton
may compete for food resources as suggested by diet composition studies (Anderson et
al. 1999^, L. M. Smith and M. J. Gray, unpublished data). Thus, Great Plains toad and
barred tiger salamander may have been displaced or experienced reduced survival in
77
terrestrial landscapes with greater spadefoot abundance, because of potential diet overlap
microhabitat use might explain species separation. Spadefoots seem to prefer playas with
less vegetation for breeding unlike Great Plains toad and barred tiger salamander
Conservation Implications
Landscape structure may favor certain species as per species-specific vagility and their
relative perception to patch viscosity and edge permeability (Wiens 1997). Competitive
ability also may interact individually or synergistically with landscape structure and
FRAGSTATS* ARC®, RAMAS® GIS) have options for inclusion of relative patch
viscosity, boundary permeability, and species vagility data; however, generally unity or
educated estimates are used because the aforementioned data do not exist. Future
anthropogenic and natural cover types and the relative permeability of their boundaries
for realistic estimation of extrinsic connectivity among habitat patches (e.g., With and
78
Crist 1995, Wiens et al. 1997). Indeed, various theoretical studies (e.g., O'Neill et al.
1988, With and Crist 1995) that used percolation models have shown this information is
of amphibians on the SHP using field estimates of landscape metrics (e.g., Knutson et al.
1999, Scribner et al. 2001). Landscape ecologists may use this and ground-truthing
79
CHAPTER V
Abstract
of resource use has been recorded among amphibian species, it has not been examined for
may affect temporal resource use of amphibians. My objective was to test for temporal
(cultivation vs. grassland) affected temporal resource use among 4 common species {Spea
during 2 years (1999 and 2000) was compared among 3 within-year time periods and
between 2 landuses. Inspection of time series plots and hypothesis testing indicated that
temporal resource use is unlikely for these amphibian species. Thus, amphibians that use
cropland wetlands later in a year may have greater probability of extinction in grassland
80
landscapes. It may be beneficial to preserve grassland playas and restore hydroperiods in
Introduction
and allows biological diversity (Schoener 1974). The way in which species partition
resources defines their ecological niche (Pianka 1976). Hutchinson (1957) suggested that
each axis represented abiotic and biotic resources necessary for survival and
reproduction. Niche dimensions also may include axes for time and space (Cody 1968.
The competitive exclusion principle {sensu Gause 1934, Hardin 1960) states that species
273).
Niche hypervolumes must be dissimilar along at least one axis to facilitate species
coexistence (Schoener 1974). The most basic hyper\olume includes axes for a resource,
space, and time (Pianka 1974). Temporal segregation is necessary for coexistence if
species use a common limiting resource, such as food, in the same location (Schoener
1974). Use of shared resources at different times is called temporal niche partitioning
(Huey and Pianka 1983), and a fundamental construct to niche complementarit}' and
biodiversity (Loreau 2000. Loreau and Hector 2001). Presumably, temporal niche
competition (MacArthur and Levins 1967, Case and Gilpin 1974). Indeed, it has been
81
suggested that temporal niche partitiomng in^lies competition along space or resource
patterns among amphibian species (Blair 1%1, Murphy 1963, Creusure and UTiitford
1976, Eraser 1976flA Bowker and Bov^^er 1979, Toft and Duelhnan 1979. W iest 1982).
Following Pianka (1976) and others (e.g., Schoener 1974), it has been hypothesized that
breeding sites or food resources (Wilbur 1980, Toft 1985). For amphibians that
most for food resources, because few species establish breeding territories (Duellman and
Trueb 1994). Indeed, dietary overly has been documented in adults and juveniles of
several coexisting explosive breeding species (Whitaker et al. 1977, Dimmitt and Ruibal
wetlands and tbey are explosive breeders (Bolen et al. 1989, CHiapters II and HI).
Because diets of Southern High Plains amphibians overlap (Anderson et al. 1999ft, L. M.
Smith and M. J. Oray, itnpublished data), species may exhibit temporal niche partitioning
to reduce exploitative competition. Several Southem High Plains species also use similar
habitat (e.g., vegetation structure, percent water coverage) for breeding (Anderson et al.
1999a); thus, species may temporally segregate to reduce interference competition (albeit
82
Southem High Plains amphibian populations (Chapters III and VI). Therefore, my
objective was to test for the existence of temporal niche partitioning in Southem High
Plains amphibian populations, and examine the effect of landuse (cultivation vs.
High Plains amphibians would temporally partition resources, and landuse would affect
Methods
This study was conducted at 16 playa wetlands (8/year) on the Southern High
Plains of Texas during 1999 and 2000 (Bolen et al. 1989). Playas were considered to be
in grassland if on average >75% of the surrounding landscape (2830-ha circular plot) was
undisturbed and vegetated, and cropland if >75% of the landscape was cultivated
(Chapter III). Playas were partially enclosed (i.e., 25% of the circumference) with 60-cm
high drift fence and 19-L pitfall traps (Dodd and Scott 1994), and pitfalls checked
altemate days from 16 May-17 October 1999 and 19 April-18 August 2000. Captured
amphibians were enumerated by species, marked uniquely, and released. After releasing
niche partitioning was examined for the 4 most common species of amphibians on the
Southem High Plains (New Mexico spadefoot [NSF, Spea multiplicata], plains spadefoot
[PSF, S. bombifrons]. Great Plains toad [GPT, Bufo cognatus], and barred tiger
83
account of the study area, amphibians, and sampling protocol was described in Chapters
II and III.
I used total daily capture per species (i.e., summed across playas within landuses
and years) as an estimate of daily species-specific abundance (Dodd and Scott 1994).
Summing across playas (i.e., local populations) was a reasonable estimate of relative
population size within landuses each year, assuming playas were mutually independent
capture (i.e., A^i, sum of captures over t days) within landuses each year (Hurlbert 1978).
the /• species present on the t day. Standardization was necessary to remove abundance
effects and compare niche overlap of species equally among time periods and between
landuses (Hurlbert 1978, Paton and Crouch 2002). Time periods (i.e., 1, 2, and 3)
represented 3 equal sequential time series in each year (i.e., 22 and 17 days per period in
1999 and 2000). I chose to divide each yearly time series into 3 periods, because
partition resources into ca. 2-3 within-year time periods (e.g.. Murphy 1963, Wemer and
McCune 1979, Wilbur 1980). A different temporal division (e.g., 2 or 4) of the within-
Temporal niche partitioning was determined by inspecting time series plots and
comparing frequency distributions of percent abundance for pairs of species among the 3
time periods within years. Frequency distributions also were compared between cropland
84
and grassland landscapes. Frequency distributions were constmcted by tallying number
of days within time periods, landuses and years that percent abundance of pairs of species
exceeded each other. Thus, each frequency distribution represented the pairwise species
abundance of species pairs was compared to the null distribution of equal relative
frequency among time periods and between years (see below). This approach to
and between landuses by each year using logistic regression with a generalized logit
model (Agresti 1990:307, Stokes et al. 2000:257- 259). Nonaddivity of main effects
(i.e., time and landuse) was tested by including an interaction term in the model (Stokes
corrected were used to test pairwise differences in frequency distributions when the main
effect logit model was significant (Zar 1984:395-396, Milton and Amold
1995:667-670). If main effects were not additive, analyses were separated by periods
and landuses, and pairwise Z-tests performed. I asstimed the categorical response of t
days followed a multinomial distribution per effect and frequency distributions were
Results
amphibian species temporally segregated resources on the Southem High Plains (Figure
85
5.1). The main effect logistic regression indicated that frequency distributions were
different among time periods both years between GPT and NSF (x 2=14.9, P<0.001
[1999]; 5cV9.6, P=0.008 [2000]), PSF and NSF (xV29.9, P<0.001 [1999]; xVl8.4,
BTS and PSF (xV31.2, P<0.001 [1999]; xV8.6, P=0.014 [2000]), and GPT and BTS
(xV33.5, P<0.001 [1999]; x V 9 . 1 , P=0.0\ 1 [2000]), and GPT and PSF in 1999
not differ between landuses for all species pairs both years (x^i=0.001-1.75,
P=0.19-0.98).
Nonadditivity was violated for GPT and NSF in 1999 (xVl4.1, P<0.001), BTS
and NSF in 2000 ( x V l 1-9, P<0.001), GPT and PSF in 1999 (xV9.4, P=0.009), BTS
and PSF in 2000 (xVl4.4, P<0.001), and GPT and BTS in 2000 ( x V l 1.2, P=0.004);
different among time periods between GPT and NSF in grassland in 1999 (x^2=15.9,
P<0.001), between GPT and PSF in cropland in 1999 (xVl6.8, P<0.001), and between
BTS and NSF (x^2=15.4, P<0.00\), BTS and PSF (xVl6.8, P<0.001), and GPT and BTS
distributions among time periods for GPT and NSF in cropland in 1999 (x 2=4.3,
P=0.114), GPT and PSF in grassland in 1999 (xV4.6, P=0.099), and for BTS and NSF
(X^2=1.5, P=0.482), BTS and PSF (xVo.65, P=0.723), and GPT and BTS (xVo.76,
86
87
Figure 5.1. Time series plots for pairs of amphibian species residing at 16 playa wetlands
in 2 landuses and years on the Southem High Plains of Texas. Linear trajectories
represent total daily capture per species summed across playas (i.e., 4 per landuse per
year). Daily capture per species was standardized using species-specific capture for the
entire time series (Hurlbert 1978), thus each node on the trajectory represents the
proportion of the population present during that day (i.e., percent abundance). The solid
line is the first species listed for the pair in the upper right comer of the graph. NSF =
New Mexico spadefoot {Spea multiplicata), PSF = plains spadefoot {S. bombifrons), BTS
= barred tiger salamander {Ambystoma tigrinum mavortium), and GPT = Great Plains
toad {Bufo cognatus).
Cropland, 1999 GPT vs. NSF Grassland, 199£
GPT vs. NSF
c
T3
T3
c J3
€ 2 - <c 2 H
o
(1) CL
a.
10 20 30 40 50 60 30 40
Days Days
1 \U \ V \W
10 20 30 Days
Days
0)
o
c
ra
T3 3 -
c
3
<c 2 -
a.
0 -
30n
Days 40
Days
Cropland, 2000 PSF vs. NSF Grassland, 2000 PSF vs. NSF
4 -
8 8
c c
n
•o (D
c 3 - •D
C
3
< <
8
« 0)
Q. a.
••»<o«»
Days Days
88
Cropland, 1999 Grassland, 1999p^ B T S v s . NSF
B T S v s NSF
8 '-^ 3 -
c c
•a
i 3 ^/l ra
T3
c
€ 3
§ 2 <
0)
CL 1 ^
10 20 30 40 50 60 30 r, 40
Days Days
/ - I
Cropland, 2000 Grassland, 2000
BTS vs. NSF BTS vs. NSF
(U
o
c
ra
•D
c
3
<
Q. 2 -
1
\ \! 1/\A ^
i i y 0
0 -
" 1 ~ ~r~ I
10 40 50 Days
20 Days 3 °
c ra
3 T3
3 -
8 c
2 -
0. 8
<u
Q.
1 -
0 -
30 40 30_ 40
Days Days
j, Grassland, 2000
G P T vs. PSF
ra
•o ro
c
3 = 4 -
<
QL
0 -
20 Days 30 20 Days ^0
89
5 - 5 -'
Cropland, 1999 i
BTSvs PSF
o c
c ra
ra T3
•o c
c
3
3 - ! \ 3
.a 3 -
<
i> •
o
o 2 -
i_
<D
a. I ' r, • . . .
Q.
1 - I
:;' 4 ' J : ' •
0 - n ;i 0 -I
10 20 30 40 50 60 30^ 40
Days Days
Grassland, 2000
BTSvs PSF
u
c
ra ra
•D
•o c II f
c 3
3 1.1 1
<
AV
0)
Q.
a.
V m4:
•J I' \ii J ? vyv
I i 1 i u
10 20 30 40 50 10 20 Days 30 40 50
Days
8
c
3 -
ra
•D
C
3
A ra
•o
c
3
.a
< < 2 -
01
^U/1 0)
0.
1 -
a. 1 -
0 _ ( u '— 0 -
10 20 30 40 50 60 10 20 30 40 50 60
Days Days
8c
c ro
ro T3
•D 3 - c
c
3 3
n <
<
c 2 H
Q.
a. 1 -
Days Days
90
In general, pairwise Z-tests indicated that a larger proportion of the GPT
population was present during time 2 than the NSF populafion (Figure 5.2). Also, a
larger proportion of the PSF population was present during time 2 than NSF and GPT
populations. Barred tiger salamander was most active during time 3 compared to NSF,
PSF, and GPT in 1999. However, a larger proportion of the BTS population was present
in time 1 than NSF, PSF, and GPT populations in cropland in 2000 (Figure 5.2).
Discussion
Great Plains amphibian communities. Even though amphibians on the Southem High
Plains generally emerge and breed simultaneously and clutch times are similar (see
Chapter II life history review. Rose and Armentrout 1974, 1976, Degenhardt 1996:38, 40.
51, Sullivan and Femandez 1999), plains spadefoot was more active during the middle of
both years compared to New Mexico spadefoot and Great Plains toad. Creusere and
Whitford (1976) also observed this temporal relationship with plains spadefoot and Great
Plains toad and a different spadefoot {Scaphiopus couchii) in New Mexico. Plains
spadefoot may be more active after New Mexico spadefoot and Great Plains toad because
of exploitative competition for food resources (Schoener 1974, 1983). Anderson et al.
(1999^) documented some diet overlap among adults of these 3 species. Moreover,
plains and New Mexico spadefoots use similar habitats for breeding and can breed
simultaneously (Bragg 1965, Anderson et al. 1999a, M. J. Gray, unpublished call survey
interference competition for breeding sites (Wilbur 1980, Toft 1985). Similarly, Great
91
92
-O 1=J
u
"T^ •*—'
C3'
J3
; ^ 0^ Q PH
W3
(/3
O)
C/3
Cd
(/3
<N
<U
UH TS PC -.cj
CO CO
© m o © ©
19%
LL E o u. E
DGP T DNSF
(0 ©^
o -;- E ©^
DNS
•7 00 VO
V
•1 - vo
o>
o
o
o
I—
Q. Q. CM
O
%3
•a T3
81%
D
c
j2 c o c
J2 D JS
Q. (A Q.
ro
OS
^r\
-^
O
O (A
(Tl
1
n O
(N O o w
II II ^ ^
N Q.,
^ ^ o
JC
i ci i i i c C3 A i i i i c
CO CD
CM CM
© Tt o CM
©
IL
V©
E E o ©
24%
DGP T DNSF
©*>
VO
o E
D f^ at at V
at at o
' i - j ^ at , • . >
o
1— at
Q. o
•a CM
o •o cta
76%
©^ c •a
D c
oa
(/)
vo (0 J2
oo vo ra Q.
oo O o o
— o
II II
N C
i A i i A c
C3
CI3 oo t-O -"^ C--.J
©
"'^4M&'^ ©
E E
©
E
u. >^^ffiss|^
'^'.^^^^ H H
at o
999,
at o
"•'^W' at o
CM
-o
0. '" <u
•a - ::s
O •D IL
c
Ti
a
D ©^ C C
in re w SS re •*->
z IT} (/) a c
iJrop
-^ r^ .—i
n Tj-
o
(A
re
o o
H o O CJ
o o CL vl
O O
II V CN
N (^ U N cx. ir^
(D
S—S—S—3—S" S—S S—S 5" ;-i
C3 oo to •^r Osi ;3
ou
eoudJjnooo l u a o j a d aouajjnooo fuaojad eouajjnooo fuaojad UH
93
CO CO CO
© © ©
82%
E u. E
DPSF DNSF
E
NSF
©^ '©^
DNS
>*^
D *n o VO O)
o at CT>
o at IL at
<N
Q. CO
O •a 0. T3
18%
c C C
D ©^
SS
D j2
SS in
in Q.
Tf in o o fO in
r- -^ re vo rt ro
m r-~ o rt — k-
o o o — O O
II II V II II
N Q. CI, N a.
JC
i c> o A i c t A X A i c;
DPSF DNSF
CM
/'<0.001
CM
Z=5.47
Z=8.94
CM
© o ©
18%
E o ©
DGPT DNSF
DPSF DNSF
10%
©^
OS
o v.. E
V
<SI o 0. at
o at at
o at at
CM at
TJ
82%
71%
T3
•a
90%
C C
J5 SS c
(A a. SS
oo (A o in
in
so o ro ro
o k.
^ o O
A ci <±. A A c c-i C3 O O O C3
CD OO CO ^d- c%j
A
C3
A
oo
A
CO
A A c^
©
^SjB ©
© E
b
DNSF
E i
*
1- t-
at
'.\o -
99,
o at
o at
o IL * " M *^'fcf
o> X!
CM CO »- . <D
•a Q. •o •D 3
Gra sslan
ntin
c
c
SS
D ©^
VO
re
in fo Q.
in oo VD o o
ra
^
oo o O U
-^ o
II II
N Q. oi
u^
A AOO A A A c3 <D
C3 (-0 • « - c-%1 Ui
:3
tJj
eoudJjnooo ^uaojed eouajjnooo tuaojad eouajjnooo )uaojad |J^
94
LL
CO
CO r^ .
CO
©
^ TT O CO
© D ^' o ©
E E E
DNSF
• ^ « r - < •:•--•
©^ o
BTS
1^
Jl ^
V
t^ o N c ITi
o o D a>
o o at
o LL CM
at
CM CO
Q. •D
TJ
D ^ C ^ •a
C SS
SS ON in IT/ c
a. SS
o ^ o n in
re
ON
a.
o
^ o ^
O
JL JL JL JL JL „ JC JL JL 1 1 1
3C CO DC ;0
LL
LL
CO , , CO
z CM oc CM
z
P<0. 001
m o
n •r © CM
Z=2.6
©
o E D
29%
E OJ
DPSF DNSF
LL ©^
, Tim
6=Z
' • . ' ;
BTS
CO
Q.
<s II
o
D t o
o
o
CM
o
o
CM
D
ON
at
at
at
•D '"
71%
•D*
nd,
^ C C
00 SS SS ^
in
00 a. in re
o re ON
>op
k.
^
o
j r — -T. X •X •Tr—
= oc CO c:; oc CO A ^
c; OC
r^
00 o ©
o ©
©
T E ^PP^^^^P-
\XX\i
LL II
E ©^
DPSF BNSF
DNS
t
53%
Tj" CO
C D c c C
SS
SS z ra • ^^
in yr,
a.
o
in
re
• r~ — a
o
r
o
k.
CO rr o
ok- U
O O (D m — o
il i' D il y (N
N 0.- in
c; CD c; c:: c; A S" -TT •nr o
A A A A A c: OC cC •«:- c-^ L<
c: oc 3
CO
eouajjnooo luaojad eouajjnooo ^uaojad eouajjnooo ^uaojad U,
95
"
u. LL
CO r- ^— CO
z ^ o CO CO z CO
D • ^ o P © © D ©
CO
«—1
dV E ©^ E CO ©^
E
1- ^ ex., ©^ h- l>
CQ V) 1 ON CQ o
i o V) o
U at o
at o u o
CM
CM
•o LL
•6
c c ^ c
©^ CO ©^ JS j5
J5
Vi
OS (A zD ci ^—
a (A
(A
(A ON o o
o o o ro
ra CO
H
d II II k.
k. ^—^ oV
m II N ex. O
O D N ex.
A A A A c c5 S" C3 "S—3" A A A o c^ c=
CO
o oo
CM IL
TS DNS
CM
© CM
©
E E ©
E
©^
at o Vi
at o CO f*^ o
at o o
CM D o
CM
•o T3"
c
_re C
•6
(A J5 ^ c
in a. V5
re o OS r- VO iS
(A
k.
O
r- p (A
ro
II II
N 0.,
A A A A A c-
/'<0.001
Z= 10.93
©
© ©
LUJi
©^
Vi
E ^ 1 E
OS vo f o o
at o
o o TJ
at
CM
o 3
CM
•D
•6 c
94%
TJ
u.
CO ©^
c
j2
c
j2 _rre . ^^
z V) (A a in
o
D r- ^- (A o re
(A
O
• ^
o re O
CO o
H ^t
CQ <—> <N
II V U^
U N a. <u
LH
•s—s—s—5" A
C3
A
oo
A
CO
A
-^
A
(Tsl
c= ? A—S—S—3—S—5 3
OX)
PH
eouajjnooQ luaojad eouajjnooo luaojad eouajjnooo ^uaojad
96
DGPT DPSF
DGPT DPSF
CO
CO
©
36%l
CO
©
29%
E
DGPT DPSF
E ©
33%
E
at
at
at at o
at o
o
•a
71%
•a c
©^ c •6
67%
SS
SS in c
so -^ vo a. in SS
o •^ o re r3- to Q.
OS o
O
II
o <^ d o
II iT II II
N C
-c i 1 • JC ! 1 1 1
CM LL
CM
PT DPS
© © CM
©^ E LL ©^
E ©
"^ Q. Tf E
1-H at
at
at D . .vo at
at 1-
at O r- o
o
Q.
D o
CM
u. •a •a
CO ©^ c D c •6
Q. — SS SS ^ c
c vo
00 a. oo so in ON SS
D O o o oo o r^i in oo
re fS a.
t^ d o
h-
Q. V
o T II o
tii. N Ci, k.
o
D ^J o
JU JL JL JC,
oc CO oc iO -^ j _ J- JL X
cr; DC ;-C •^^ rx
u.
CO r~ o
Q. • ^
o ..-..
ime '
D u-i
o ©
II
-^? ©
V
^
E CO ©^ • - E
N H
Q.
00 1- 0. ^ •.
nd. 000,
O ^ 4
•
at
at D '^
D
^ •
at at
' >.- at t-
at
^ •a
Q.
O
CM o
nd.
^
c
SS D ^
_re
c
<N ra in vo
00 Q. in
re — oo
V) Q.c
O o
o O d d o
Jl
N
II
ci.
o u
Csj
A" - x — X - TTT iri
S S S S S ~ C3 oc
=5
97
il. LL
CO (O (^ ^^
001
CO Q. ro Q. • ^
o
D O ^ o CO
u.
©
E o
CO
© • ^— dV ©
CO CO II ^ E CO II ^k
E
Q. 1- N V H N ex. is
D o m ex. -^ m Vi
o D ^* at D at
1- o at at
Q. CM at at
o •6
D c •o X3
SS ^ c i< c
in SS SS
• ^ i/~, in vo a. V, in
OS ra 00 in
o o ON ro
— o ^
u
N c..
jrr
JC I- JL- JC 1 1 1 1
=— oc ^
re ^^
aouajjnooo iuaojad
aouajjnooo luaojad aouajjnooo tuaojad
CM
CM
© CM
LL ©
E LL ^ E ©
CO ©^ ^ LL E
Q. O O CO ©^
Q.
D Vi o Vi Q. 00
o D at
at D SO at
o
GPT
at
CM CO CO at
H H
•o"
D c
m •D
m T3
%z
C D
SS
in
D o SS C
Vi a. SS
in V/ in
c5 O ra o o o OS — f^- in
1.0
1.0
c ^ o o lO o ra
c o
1 II II II II II
^s! C , N ex, N ex,
-c— 1 1 JC X 1- -C X "— X — -J- JL.
JC nr x X X 1-
re -o
c_; oc CO
© "•••-NP'-;' © ©
E . © ^ . - ••
E ©^
E
H
o os: at OS o>
o at at
o at at -TD
T-
CM <U
•a •o T3 P
DBTS DPSF
C
6%
C
c c u.
CO ^ ro • ^ M
SS ra Q. «A
a c
Z= 10.93
in
(A
in o D ra o
S2 ok. ^ § O
r
CO
o 1-
ffl csi
D N ^
»n
A A A A A
jr- -X- -X- -X- <u
LH
oc
D
00
98
miiimimKiiiiit>mtRi.iifJiim!x:f, CO
fO
© ©
DPSF
©
E E E
i- ©^ ©^
o"
Vi
o '^ o
o 0\ at
o CO o at
CM 1- • • " •
at
CM
ffl
DBTS DPSF
•D T3
C D CO T3
©^ ©^ C
SS SS »- C
OS Q. 00 in m r- •—' SS
V)
£ i *—I
o
o
ro
00
—
"^
Vi (A
ro
D
H
• ^
^
^—
.
o
o
dV
a.
o
o
"^ d dII IId
Q.
O N
II
Q,
N ex. D
JL JL X JL
-J— -X-
A A A A A c= A- • 3 — ^
aouajjnooo juaojad
aouajjnooo )uaojad aouajjnooo )uaojad
CM CM
f ,^ - "i^ y A + A ^ '^•*t ''*• '* ' © CM
LL
©
CO E E ©
©^ V
Q. E
r«< o
D 00 * o
o o at
CO ., o o at
1- CM CM at
m
D 13 •o"
C c T3
00 SS J2 C
a. (A SS
, ^H «A
o a.
ro o
°) o %..
CD
o
A A A A A cr
CO
IL i— t^ —
0.00
CO m
14.4
Q. (N -
ON « D
D ©
©
LUJi
CO
^ E 1-
i' y
II E 0. ^
1- t^ij
m O N ex. V)
o D at
D o o at
o
o
o at -a
CM
T- <L»
CM .
TJ" •o n
•D" ^ c
c
c c ra
SS SS Vi ••—>
a c
a. in
in
OS
o o
o ra
k.
k.
O L)
o O
(N
ir^
s—s—s—S—S" <u
;-!
CTJ oo CO •^r c\ 3
DD
aouajjnooo )uaojad aouajjnooo )uaojad tin
aouajjnooo )uaojad
99
oo O
CO so CO
© o CO
ro d ©
E ©
DBTS
©^ V ©^ E E
VO ex. ^-
00 <n
at rs| o o
at o o
n o o
CM
CM
O
(0 c Tj"
^ D •o
C
CQ TJ SS ^ c SS
in OS SS
001
in
D in
ra 1^ a. in
so k.
o re
k. ^
Q. sd o O o
O til^ V
D
^
c;
CD
oc
A A A c: i> 1
C5 CD A ci c.
CM T CM CM
© SI
R o ©
24%
r -> d ©
E E
DBTS
E
, V
at o
at o o
at o o
y—
a. '^M CM
o
CM
•o
c o ©^ c •a
c
2. SS SS
in
in
D vo
1^ a. in
o in
ra ra
o k.
O
A A A A A c
C^ oo CD
(0
H r- ,—,
CD • ^
o
Ii rf o ©
©
©
H •—1
d E E
Q. Jl V ©^ E H
O N ti. V)
o o
D o
en o
o o T 5.
CM
CM <U
•o
•o"
TJ P
c C
SS
CO
H
c c
ro ..—<
.t->
in SS in
tn CQ VO a. in a
2 [1 OS o o re
k.
o
O H d <=? k. U
O
Q.
— o o cs
o i y ir*
D (U
A^ A S S S 5
cr3 "s—s—s—5"
oo CO
c5"
C3 oo CO s—^ CT) oo CO -"3- C^
>-l
3
QU
aouajjnooo )U80jad aouajjnooo )U80Jad aouajjnooo )uaojad
100
Plains toad was more acti\e during the second time period than New Mexico spadefoot.
which may ha\e been a consequence of diet o\erlap, because these species use different
Barred tiger salamander was most acti\e later in 1999 than all other species. I
desiccation not competition (Toft 1985). Although barred tiger salamander can develop
as rapidly as anurans on the Southem High Plains. the> may prolong development or
become neotenic during >ears of favorable rainfall (Petranka 1998). Rainfall and
hydroperiod in playas was greater in 1999 than 2000 (Chapter II): thus, barred tiger
salamander may have been more active later in 1999 than other species because
favorable conditions in the aquatic en\ ironment prolong development hence metamorph
emergence. Rainfall also was greater during the last time period (17.5 cm) than the 2
previous periods (14.5 cm and 10.5 cm for 1 and 2, respectively) in 1999. possibly
facilitating dispersal of all age classes (Petranka 1998). If temporal pattems were a
barred tiger salamander should have been more active later in both years; this was not
observed.
Barred tiger salamanders also were most active earlier than all other species in
hydroperiods between landuses (Chapter II). Two of 4 cropland playas were dry prior to
time period 2, whereas all grassland playas retained water through time period 3 in 2000.
101
Perhaps, early playa drying in cropland landscapes caused barred tiger salamanders to
(Semlitsch 19876). I speculate that anurans remained near playa bottoms and aestivated
least one other niche dimension (Schoener 1974. Huey and Pianka 1983, Pianka 1994).
thus seemingly a competitive dominant exists among Southem High Plains anuran
species. Data from Chapter III suggested that New Mexico spadefoots can become
competitively dominant in the absence of larval and neotenic barred tiger salamanders.
Laboratory studies (e.g., Morin 1983, Wilbur 1987, Dayton and Fitzgerald 2001) indicate
that certain spadefoot tadpoles are superior competitors for food resources compared to
other anurans (e.g., Bufo, Rana). Therefore, New Mexico spadefoots may have displaced
the other 2 anurans from earlier activity over evolutionary time, resulting in plains
spadefoot and Great Plains toad being more active during the middle portion of the
breeding season. The time series plots (Figure 5.1) support this inference, because
species-specific percent abundance was > 0 for most days. However, no experiments
exist comparing the relative competitive ability among these species for food or other
resotirce (e.g., food, cover) would need to be identified among these species (Pianka
compete for the limiting resource in allopatric and sympatric scenarios (Pianka 1976).
102
Numerical or fitness response could be compared subsequently among species (Pianka
1976).
is low or nonexistent compared to more stable habitats (Wiens 1977, Huston 1979,
Strong 1983). This idea was stimulated from the observation that biodiversity is greater
in stable climates (e.g., tropics) than in variable ones (e.g., temperate regions, Pianka
1966). Theoretical studies since then have shown that competition and competitive
Holt 1985, Chesson and Huntly 1997); however, few empirical examples exist. Because
climate on the Southem High Plains is semi-arid and hydroperiod in playa wetlands is
variable (Haukos and Smith 1994), these results seem to support the prediction that
Moreover, it has been suggested that temporal niche partitioning exists most often with
species, such as insects and larval amphibians, that exploit transient resources (Schoener
1974, Toft 1985). Considering the environmental variability in playas and the explosive
breeding strategy of Southem High Plains amphibians, these results also seem to support
this hypothesis.
Southem High Plains amphibians (Chapters III and VI), it did not affect temporal niche
103
developmental time in amphibians. For example, spadefoots may develop carnivorous
invertebrate food resources (Pfennig 1992). Also, many amphibian species may
accelerate growth and undergo metamorphosis at a smaller body size in wetlands with
reasonable to hypothesize that activity pattems would follow water availability, hence
temporal resource use be shortened in wetlands with rapid hydroperiods (i.e., cropland
playas, Luo et al. 1997). Thus, temporal niche partitioning should have been affected by
landuse, but this was not observed. Accordingly, I hypothesize that change in temporal
resources use may be beyond the genotypic norm of reaction for Southem High Plains
amphibian species (Via and Lande 1985, Gomulkiewicz and Kirkpatrick 1992).
partitioning are important. Paton and Crouch (2002) and Snodgrass et al. (2000) suggest
that duration and timing of wetland hydroperiods are critical to survival and reproduction
results support their suggestions. Species, such as plains spadefoot and Great Plains toad,
that use resources later than other species (e.g., New Mexico spadefoot) may have
may decrease probability of survival and reproduction. Consequently, these species may
104
have a greater probability of local extinction. I recommend retention and restoration of
grasslands around playa wetlands, because hydroperiods generally are shorter in cropland
105
CHAPTER VI
Abstract
various discrete biological models; however, it rarely has been documented in natural
spatially structured individuals, and its presence can reduce probability of metapopulation
extinction. It also has been hypothesized that anthropogenic disturbance may influence
wetlands during 1999 and 2000 (4 playas per landuse per year). Also, deterministic and
iteration and least squares techniques. The amphibian assemblage time series was
constructed from total daily captures in pitfall traps summed across species at each playa.
Existence of chaos was determined via estimating Liapunov exponents per playa and
106
using published stability estimates. Occurrence of chaos was greater at grassland (7/8)
than at cropland (1/8) playas, indicating that biological chaos was a fundamental
characteristic of the SHP amphibian assemblage and its occurrence was altered by
predicted that intrinsic rate of increase at grassland playas was greater than at cropland
playas. Intrinsic rates of increase also were greater using an iterative method rather than
the least squares method for estimation in all difference equations. Anthropogenic
stmctured amphibian assemblage on the Southem High Plains, and perhaps, increased
chaos. Parameter estimates for the stochastic Ricker function are presented for future
systems.
Introduction
1988) and initiating conservation endeavors. Most biological theories assume abundance
(DeAngelis and Waterhouse 1987). However, complex dynamics, such as chaos, have
107
been demonstrated in simple discrete deterministic models (May 1974. 1975fl. 1976a).
Interestingly though, chaos has been detected in few natural assemblages (Turchin and
Taylor 1992). Assemblages that exhibit chaos may have a greater probability of
extinction than those near positive stable equilibria (Hastings et al. 1993, Shulenburger et
increasing local abundance noise (Allen et al. 1993). Anthropogenic modifications of the
landscape also may induce transition from stable equilibria to chaos or vice versa
(Berryman and Millstein 1989, Berryman 1991). This may be especially critical for
2000).
assemblages (May 1976b). Competition and predation can be modeled over discrete time
growth and Ricker functions are commonly used to model competition (Hastings 1996).
108
EUner 2000). Parameters of these models can be estimated using observed data and least-
High Plains (SHP, Bolen et al. 1989). Anthropogenic landuse (i.e., grassland and
spatially structured on the SHP, the existence of chaotic dynamics may be critical.
dynamics. Thus, my first objective was to determine the presence of chaos in a SHP
amphibian assemblage at 16 playa wetlands from an existing time series and test for
chaos detection (Turchin 1991, 1993, Turchin and Taylor 1992). My second objective
estimating parameters for deterministic and stochastic versions of the discrete logistic and
109
regulating mechanisms (e.g., density dependence) in populations also operated in
assemblages. Turchin (1991) also made this assumption in his chaotic analysis of multi-
species vole data. Lastly. I quantified differences in parameters between iteration and
Methods
SHP of Texas during 1999 and 2000. Playas were considered to be in grassland if >75%
of the surrounding landscape (i.e., <3 km from the playa center) was undisturbed and
vegetated, and cropland if >75% of the surrounding landscape was cultivated (Chapter
III). Playas were partially enclosed (i.e., 25% of circumference) with 60-cm high drift
fence and 19-L pitfall traps (Dodd and Scott 1994). Pitfall traps were checked altemate
days for captures from 16 May-17 October 1999 and 19 April-18 August 2000. Thus,
the amphibian time series represented altemate equally spaced days nested within 2
years. Captured amphibians were enumerated, marked uniquely, and released. A more
detailed account of the study area, amphibians, and sampling protocol has been described
relative assemblage size (i.e., Nt, Dodd and Scott 1994). As in Chapter III, I assumed if
any bias existed by using mean daily capture to estimate relative daily abimdance of the
110
assemblage, it was constant between landuses and years. Daily capture was summed
across species for Nt to incorporate positive and negative feedbacks into dynamics
(Schaffer and Kot 1986). As mentioned, Turchin (1991, 1993) summed across vole
species using data of Henttonen et al. (1984) for chaos analyses. Turchin (1991) also
suggested and later demonstrated (Turchin and Taylor 1992) that modeling a nonlinear
time series at lower dimensionality (e.g., number of species, time lags) than the tme
overall dimension of the system can decrease detection of real chaos. I did not sum
captures across playas (i.e., local assemblages), because simulations have demonstrated
that it can reduce probability of detecting actual chaos (Rohde and Rohde 2001). I also
used da>s as time units instead of years or generations. Thus, I assumed regulating
each playa. Then, I assumed the amphibian assemblage was govemed by an underlying
difference equation of the form Nt, Nt+i=F(N(). The presence of chaos was determined
where J(t) was the first derivative of F(N) along the assemblage trajectory and m was the
playa-specific length of the time series (Ellner 2000). Liapunov exponents measure the
average deviation in slope of F(N) over time (Abarbanel 1996). They also measure the
conditions, which is tantamount to chaos (Ellner 2000). Positive ^^{t) indicates the
111
presence of chaos; negative exponents suggest convergence to some stable or periodic
equilibrium (Ellner 2000). Inasmuch as equation (6.1) is the natural-log geometric mean
of DF(N), the presence of chaos merely implied the average slope of the trajectory of
F(N) > 1. For detecting chaos in the SHP amphibian assemblage, I initially assumed no
specific form of F(N) (i.e., Nt+i=F(Nt)), because regulating mechanisms (i.e., the
underlying ftanction) were unknown. Thus, Liapunov exponents were estimated directly
by computing the slope of the linear trajectory between time periods (i.e., equally spaced
I assumed the slope of each linear trajectory was a reasonable estimate of the derivative
of A^,. Differences in occurrence of chaos (i.e., positive 'km{t)s) between landuses was
mechanism in the SHP amphibian assemblage and followed discrete logistic growth.
where, r = intrinsic rate of increase and A'= carrying capacity (May 19756). Equation
(6.3) simplifies to
(6.4)
Nt K
simplifies to
^ ^ = a-bNr, (6.5)
Nt
112
which can be solved using 2 successive iterates (Edelstein-Keshet 1988). For example, if
A^i, N2, and N3 = 4.9, 3.6, and 3.7, it follows that NiINx ~ 0.735 and N^^INi = 1.028,
resulting in 2 equations of the form (6.5) with 2 unknowns (i.e., 0.735 = a-4.9b , 1.028 =
a-3.6b). These equations can be solved for a (0.622) and b (-0.113) and subsequently r
(-0.378) and K (3.36). Using daily capture per playa and equation (6.5), I estimated r
and K for each successive pair of iterates in the amphibian assemblage time series, and
computed the arithmetic mean for each parameter per playa. It should be noted that
because my time series was within years and I used difference instead of differential
equations, r and K may not represent the true intrinsic rate of increase and carrying
capacity of the assemblage; they were merely control parameters of these functions.
Because Mean become negative using equation (6.3) if M > K{\+r)lr, which is
logistic growth,
Nt+\ = Nt^^V
'I'-i (6.6)
which is positive for all positive initial conditions (May 19756). Equation (6.6)
simplifies to
In Nt+\ = '-^Nt • (6.7)
\ Nt J
Finally, making the following change of variables {a = r,b = r/K), equation (6.7)
simplifies to
In ^Nt.i^ = a-bNt , (6.8)
V Nt )
113
Previous research (e.g., Morin 1981, Chapters II and III) also suggests that
growth then added a negative quadratic term for predation to equation (6.6),
Finally, making the following change of variables {a = r, b = r/K, and c=c), equation
(6.10) simplifies to
Parameters in equations (6.5), (6.8) and (6.11) also can be approximated using
least-squares estimation (e.g., Turchin 1990). Letting, yt = M+i/M for equation (6.5), yt =
ln(M+i/M) for equation (6.8), and x, = M for both, equations (6.5) and (6.8) reduce to
y- = a + bxi . (6.12)
which is a linear function. Parameters a and b can be solved by fitting a line to the entire
data set and using the least-squares algorithm (Milton and Amold 1995:386). Similarly,
y. = a + bxi-cxj , (6.13)
114
which is a quadratic function. Parameters a, b, and c can be solved similarly by fitting a
quadratic curve to the data set and using least-squares estimation. Thus, I estimated
parameters in equations (6.3), (6.6), and (6.9) per playa using an iterative method and
computing the difference {D) between estimates per playa then performing a one-sample
Mest on D (HQ: D=0). The aforementioned was necessary because estimation samples
and Turchin 1995). This might be especially tme for amphibian assemblages inhabiting
semi-arid environments, such as on the SHP, because their aboveground dynamics can be
realism and account for variability in dynamics associated with the environment. Thus, I
added St to equations (6.6) and (6.9) to create stochastic versions of the Ricker logistic
and.
Parameters in equations (6.14) and (6.15) were approximated by fitting the functions to
the observed data and using least-squares estimation (Turchin and Ellner 2000), which
115
was done using PROC NLIN in SAS® (Freund and Littell 2000). A reasonable estimate
of ft for the function is mean-squared error of the nonlinear regression (Turchin and
Ellner 2000). Residuals can be used as time-specific CtS, which were needed to classify
(e.g., equations 6.3, 6.6, 6.9, 6.14, and 6.15) should be constmcted with lags to increase
which is the embedding dimension of the time series. Because no information exists on
the presence of delayed density dependence in the amphibian assemblage on the SHP, I
reasonable for many vertebrate assemblages because density dependence and predation
usually most strongly affects individuals in the adjacent time period (Turchin and Ellner
2000). Moreover, graphs of autocorrelation functions from my time series (M. J. Gray.
unpublished data) suggested lack of correlation after 1 time lag (Turchin 1990, 1996).
chaotic using published stability estimates from May (19756) for equations (6.3) and
(6.6), and estimating Liapunov exponents for equations (6.9), (6.14), and (6.15). If r >
2.59 and 2.69, dynamics were classified as chaotic for equations (6.3) and (6.6),
respectively (May 19756). Estimation of Liapunov exponents for the other difference
equations was straightforward using equation (6.1), because F{N^ (i.e., the right hand
side of the difference equations) was a constant (c) after solving using parameter
116
estimates. £t and M (Ellner 2000). It follows. DF(M) was mereh' e' for equations (6.9).
(6.14), and (6.15), because the time-specific derivative {J(t)) of M+i= e'N^ was e' (Ellner
2000). Posifive Liapunov exponents indicated chaotic dynamics (Olsen and Degn 1985).
Predicted occurrence of chaos for each difference equation was tested between
landuses and actual dynamics (i.e., chaos occurrence estimated assuming no specific
F(N)) using a main effect logistic regression with a generalized logit model (Stokes et al.
2000:257). Parameter estimates per playa were presented for functions that predicted
chaotic frequency similar to the actual dynamics. Lastly, estimates of parameters for all
difference equations were tested between landuses using a 2-sample /-test (Milton and
Amold 1995:347).
Results
Time series plots of relative abundance at each playa clearly suggested that
within-year dynamics of the SHP amphibian assemblage were not linear and may not
approach a single positive stable equilibrium (Figures 6.1-6.2). Measures of stability for
actual and predicted dynamics from difference equations indicated that chaos was a
occurrence of chaos was greater at grassland (7/8) than cropland (1/8) playas (Table 6.2).
The stochastic Ricker function predicted occurrence of chaos similar to actual dynamics
(i.e., significant and non-significant landuse and model effect, respectively, Table 6.2);
least-squares estimates of its parameters per playa are presented (Table 6.3). The
deterministic and stochastic Ricker fiinctions with a predator term also predicted
dynamics similar to actual dynamics (i.e., no model effect); however, no landuse effect
117
118
Figure 6.1. Time series plots of amphibian assemblage dynamics at 8 playa wetlands
during 16 May-17 October 1999 on the Southem High Plains, Texas. Relative
abundance was total daily capture (natural-log transformed) of all species in pitfall traps;
days were altemate equally spaced days within the time series. Time series for playas 1
and 4 were delayed because rain caused flooding over pitfalls. Playas 1-4 and 5-8 were
located in cropland and grassland, respectively.
6.5
5.5
8c
I 4.5 nj
•o
c c
< 35 XI
> <
TO >
"S 2.5
cr
1.5
0.5
7 -f
0) c
o
c 5 - as
nj T3
-a c
c
XI
< <(1)
<U >
>
"35
<u
ca
1 -
6 -
(U
5 - 0)
o
o c
T3
nj
C
c XI
D
3 -
< 3 -
< >
I
a:
1 -
0 -
10 20 30 40 50 60
Days
b —(
5 -
c c
to n
•o
J3 c
C
X3 X3
< <
>
s
a:
0 -
10 20 30 40 50 60
Days
119
120
Figure 6.2. Time series plots of amphibian assemblage dynamics at 8 playa wetlands
during 19 April-18 August 2000 on the Southem High Plains, Texas. Relative
abundance was total daily capture (natural-log transformed) of all species in pitfall traps;
days were altemate equally spaced days within the time series. Playas 9-12 and 13-16
were located in cropland and grassland, respectively.
6 -f
6 -
5 -
5 -
0)
o o
c c
01 ro
"O •o
c c:
X3 XI
< 3 - 3 -
> <
ro >
"83 j5 2
OJ
1 -
20 ^ 30
Days
6 -f
1 1J Playa 11
5 - (U
o
c
ro ro
•o •o
c c
3 3
X3 XI
< < 3 -
41
> >
— 9
1 - \ 1 -
1 1 1 1 1
10 20 r^ 30 40 50
Days
lU 5 -
8 o
c
c ro
ro 2 -
•D TJ
c c
3
X3 3
< XI
<
> >
-5
0)
1I ro
2 -
•33
Q:
0 -
0)
o
c c
ro ro
TJ TJ
c 4 - c
3 3
XI X3
< 3 - <
!« >
2 -
(U
cr
20 30
Days
121
Table 6.1. Stability of amphibian assemblage trajectories at 16 playa wetlands during 16
May-17 October 1999 and 19 April-18 August 2000 on the Southem High Plains, Texas.
Stability
F(Ny Year Landuse Playa Parameter Classification'^
NSF 1999 Cropland 1 -0.004 Not Chaotic
2 -0.015 Not Chaotic
3 -0.009 Not Chaotic
4 0.001 Chaotic
Grassland 5 0.003 Chaotic
6 0.001 Chaotic
7 0.001 Chaotic
8 0.008 Chaotic
2000 Cropland 9 -0.007 Not Chaotic
10 -0.014 Not Chaotic
11 -0.019 Not Chaotic
12 -0.029 Not Chaotic
Grassland 13 0.023 Chaotic
14 0.015 Chaotic
15 0.015 Chaotic
16 -0.011 Not Chaotic
DDL 1999 Cropland 1 4.465 Chaotic
2 3.691 Chaotic
3 2.693 Chaotic
4 3.931 Chaotic
Grassland 5 2.808 Chaotic
6 3.242 Chaotic
7 2.791 Chaotic
8 2.841 Chaotic
2000 Cropland 9 2.516 Not Chaotic
10 2.842 Chaotic
11 3.087 Chaotic
12 2.271 Not Chaotic
Grassland 13 3.031 Chaotic
14 2.101 Not Chaotic
15 2.219 Not Chaotic
16 2.861 Chaotic
122
Table 6.1. Continued.
Stability
F(Nf Year Landuse Playa Parameter'' Classification'^
DDR 1999 Cropland 1 3.881 Chaotic
2 3.105 Chaotic
3 3.029 Chaotic
4 3.706 Chaotic
Grassland 5 2.615 Not Chaotic
6 2.922 Chaotic
7 2.665 Not Chaotic
8 3.021 Chaotic
2000 Cropland 9 2.305 Not Chaotic
10 2.706 Chaotic
11 2.817 Chaotic
12 1.848 Not Chaotic
Grassland 13 3.141 Chaotic
14 1.995 Not Chaotic
15 2.064 Not Chaotic
16 2.021 Not Chaotic
DSR 1999 Cropland 1 0.006 Chaotic
2 0.004 Chaotic
3 -0.001 Not Chaotic
4 -0.003 Not Chaotic
Grassland 5 0.009 Chaotic
6 0.009 Chaotic
7 -0.002 Not Chaotic
8 0.008 Chaotic
2000 Cropland 9 -0.001 Not Chaotic
10 -0.007 Not Chaotic
11 -0.012 Not Chaotic
12 -0.027 Not Chaotic
Grassland 13 0.001 Chaotic
14 0.024 Chaotic
15 0.015 Chaotic
16 0.011 Chaotic
123
Table 6.1. Continued.
Stability
F(nf Year Landuse Playa Parameter' Classification'
DRP 1999 Cropland 1 -0.036 Not Chaotic
2 0.023 Chaotic
3 -0.021 Not Chaotic
4 -0.419 Not Chaotic
Grassland 5 -0.093 Not Chaotic
6 -0.037 Not Chaotic
7 0.002 Chaotic
8 0.087 Chaotic
2000 Cropland 9 -0.014 Not Chaotic
10 -0.004 Not Chaotic
11 -0.006 Not Chaotic
12 -0.108 Not Chaotic
Grassland 13 -0.099 Not Chaotic
14 -0.033 Not Chaotic
15 0.101 Chaotic
16 0.252 Chaotic
SRP 1999 Cropland 1 -0.029 Not Chaotic
2 0.021 Chaotic
3 -0.019 Not Chaotic
4 -0.423 Not Chaofic
Grassland 5 -0.082 Not Chaotic
6 -0.029 Not Chaotic
7 0.006 Chaotic
8 0.098 Chaotic
2000 Cropland 9 -0.009 Not Chaotic
10 -0.015 Not Chaofic
11 -0.012 Not Chaotic
12 -0.139 Not Chaotic
Grassland 13 -0.108 Not Chaotic
14 -0.014 Not Chaotic
15 0.099 Chaotic
16 0.248 Chaotic
124
Table 6.1. Continued.
125
Table 6.2. Occurrence of chaos between landuses and models in amphibian assemblages
at 16 playa wefiands during 16 May-17 October 1999 and 19 April-18 August 2000 on
the Southem High Plains, Texas.
Percent Chaos Landuse Effect Model Effect"
F(N/ Cropland Grassland X'^ P X^ P
NSF 12.5 87.5 9.0 0.003 NT NT
DDL 75.0 75.0 0.0 1.0 4.26 0.039
DDR 75.0 37.5 2.29 0.131 6.35 0.012
DSR 25.0 87.5 6.35 0.012 0.18 0.675
DRP 12.5 50.0 2.62 0.106 0.96 0.326
DSP 12.5 50.0 2.62 0.106 0.96 0.326
^ S F = no specific form assumed (i.e., Nt+\=F(Nt). true population dynamics), DDL
discrete deterministic logistic (M+i=M + rNt[\- Nt/K]), DDR = discrete deterministic
Ricker {Nt+i=NtQxp[r{l- Nt/K)]), DSR = discrete stochastic Ricker {Nt+i=Ntexp[r{\- Nt/K)
+ Et]), DRP = discrete deterministic Ricker with predator term {Nt+\=Ntexp[r{\- Nt/K) -
cNt ]), and SRP = discrete stochastic Ricker with predator term {Nt+\^Nttxp\r{\-Nt/K) -
cM^ + et]) (May 19756).
''Model effect was pairwise Fisher's exact tests in percent chaos between NSF (i.e., tme
dynamics) and other functions; no test (NT) was performed on NSF and itself
126
Table 6.3. Least-squares esfimates of parameters for the discrete stochastic Ricker
equation^ from amphibian assemblages at 16 playa wetlands during 16 May-17 October
1999 and 19 April-18 August 2000 on the Southem High Plains, Texas.
Parameter Estimate
Year Landuse Playa r K Et MSreg" F" P
1999 Cropland 1 0.273 4.493 0.057 0.168 5.65 0.022
2 0.292 3.551 0.174 0.798 9.14 0.004
3 0.308 3.795 0.113 0.392 6.60 0.013
4 0.118 3.407 0.116 0.189 2.46 0.123
Grassland 5 0.593 4.064 0.247 1.578 12.71 <0.001
6 0.673 3.187 0.211 2.561 24.27 <0.001
7 0.813 3.924 0.175 2.394 27.35 <0.001
8 0.486 3.251 0.209 1.568 15.02 <0.001
2000 Cropland 9 0.491 3.041 0.228 1.277 11.03 0.002
10 0.346 3.807 0.131 0.421 6.07 0.018
11 0.457 3.996 0.111 0.643 11.04 0.002
12 0.677 2.992 0.317 2.955 18.55 <0.001
Grassland 13 0.525 2.113 0.189 1.186 12.54 0.001
14 0.467 2.731 0.174 1.059 12.19 0.001
15 0.654 3.679 0.189 1.571 16.55 <0.001
16 0.511 3.509 0.159 1.025 12.83 <0.001
^ h e discrete stochastic Ricker equation is A^t+i=Mexp(r[l-A^t/-^ + ^t); parameter
estimates can be used for predicting playa- and time-specific assemblage dynamics.
''Mean-square regression (MSreg) is variation in assemblage dynamics explained by the
discrete stochastic Ricker equation (Freund and Littell 2000:185-189).
'^Numerator, denominator degrees of freedom of F were 2,65 and 2,50 for 1999 and 2000,
respectively.
127
was detected, which was inconsistent with observed dynamics (Table 6.2). The
stochastic Ricker function predicted mean intrinsic rate of increase was greater at
grassland than cropland playas; carrying capacities were similar (Table 6.4). No
differences in parameters between landuses were detected for other functions (Table 6.4).
Lastly, estimates of parameters for the deterministic logistic, Ricker, and Ricker with
predator term functions were different the between iterative method and least-squares
estimation (Table 6.5). Estimates of intrinsic rate of increase were greater using the
iterative method than by least squares (Table 6.5). Parametric assumptions of residual
normality (P>0.79) and homoscedasticity {P>0.23) were satisfied for all ^tests.
Discussion
than cropland playas. These results represent the first empirical evidence of chaos in
spatially stmctured amphibian assemblages, and more importantly, that its frequency can
among spatially stmctured populations. Agricultural landscape use and stmcture can
affect local population demographics and body size of amphibians (Chapters II-IV). In
particular, mean daily abundance of some species was greater at cropland than grassland
128
Table 6.4. Iterafive and least-squares estimates of parameters for deterministic difference
equafions used to model amphibian assemblages at 16 playa wetlands during 16 May-17
October 1999 and 19 April-18 August 2000 on the Southem High Plains, Texas.
Estimation Technique Paired
F(N/ Parameter Iteration Least-Squares Difference'^
X SE X SE X SE /15 P
DDL r 3.17 0.15 0.78 0.09 2.39 0.21 11.42 <0.001
K 3.94 0.16 4.12 0.16 0.17 0.07 2.49 0.025
DDR r 2.74 0.15 0.48 0.05 2.26 0.18 12.57 <0.001
K 3.89 0.15 3.47 0.15 0.43 0.08 5.27 <0.001
DRP r -1.52 4.10 0.89 0.13 11.33 3.02 3.74 0.002
K 2.40 0.26 2.89 0.60 1.48 0.61 2.42 0.029
c 0.10 0.22 -0.03 0.008 0.61 0.17 3.66 0.002
"^DL = discrete determinisfic logisfic (M+i=M + ^ ^ [ 1 - Nt/K]), DDR = discrete
deterministic Ricker {Nt+i=NtQxp[r{\- Nt/K)]), and DRP = discrete deterministic Ricker
with predator term (M+i=Mexp[r(l- Nt/K) -cM^]) (May 19756).
Iterative estimates were computed by making a change of variable {a=\+r and b=r/K for
DDL, a=r and b=r/K for DDR, and a=r, b=r/k, and c=c for DRP), solving 2-3 iterates
using daily capture per playa for M+i and Nt, and computing the arithmetic mean within
then across playas; least-squares estimates were ordinary least-squares computed by
fitting the deterministic function to the entire data set and estimating parameters that
allow the best available fit (Milton and Amold 1995:386).
'^Mean paired difference {D) was the absolute difference between iterative and least-
square estimates averaged across playas; a one-sample /-test (HQ: D = 0 ) was performed on
arithmetic mean of D to test for difference between parameter estimates because samples
were not independent (Milton and Amold 1995:353).
129
Table 6.5. Esfimates of parameters between landuses for difference equations used to
model amphibian assemblages at 16 playa wefiands during 16 May-17 October 1999 and
19 April-18 August 2000 on the Southem High Plains, Texas.
Cropland Grassland
F(nr Parameter X SE X SE tu P
DDL r 3.19 0.27 2.74 0.14 1.48 0.161
k 4.19 0.18 3.75 0.27 1.36 0.196
DDR r 2.92 0.24 2.56 0.17 1.27 0.225
k 4.09 0.19 3.69 0.22 1.35 0.199
DSR r 0.37 0.06 0.59 0.04 3.02 0.009
k 3.64 0.18 2.77 0.23 1.14 0.275
DRP r 4.07 6.78 -7.02 4.21 1.39 0.186
k 2.52 0.46 2.28 0.26 0.46 0.651
c -0.17 0.39 0.37 0.21 1.23 0.239
DSP r 0.62 0.18 1.02 0.16 1.64 0.123
k 1.85 0.32 2.35 0.23 1.25 0.231
c -0.02 0.01 -0.05 0.02 2.06 0.361
"^DL = discrete deterministic logisfic (M+i=M + rNt[\- NJK]), DDR = discrete
deterministic Ricker {Nt+\=NtQxp[r{\- Nt/K)\), DSR = discrete stochastic Ricker
(M+i'^Mexp[r(l- Nt/K) + et]), DRP = discrete deterministic Ricker with predator term
{Nt+\=Nttxp[r{\- Nt/K) -cN^]), and SRP = discrete stochastic Ricker with predator term
(M+i=Mexp[r(l- Nt/K) -cN^ + Et]) (May 19756).
130
individuals among local assemblages. Existence of predafion and interdemic movement
owing to the increased occurrence of chaos in undisturbed grassland playas. Allen et al.
have been related to density dependence and environmental stochasticity. The stochastic
assemblage. Density dependence has been hypothesized and shown to regulate some
mechanism in the SHP amphibian assemblage, because most species are explosive
breeders that emerge simultaneously and produce numerous offspring (see species review
in Chapter II). Consequently, relative abundance of individuals can increase rapidly. For
example, total daily capture increased from 30 to 14,623 individuals in 1 week at 1 playa
in 1999. I suspect competition for food, cover, and mating resources may become
limiting under these and less extreme dynamics, inducing density dependent regulation.
assemblage, because the deterministic Ricker model did not predict occurrence of chaos
131
similar to actual dynamics. Thus, these results support Ellner and Turchin (1995)
The stochastic Ricker function also predicted that intrinsic rate of increase was
greater in grassland than cropland playas. Elementary stability analyses demonstrate that
progression to a chaotic state also increases (e.g.. May 1975). The aforementioned may
density dependence between landuses, and the mathematical reason for differences in
predicted mean carrying capacity was greater at cropland (x=38.1 captures/day, SE=1.19)
and III, carrying capacity may be lower in grassland playas because larval and neotenic
barred tiger salamanders may be exhibiting top-down control. Indeed, deterministic and
stochastic Ricker functions with predator terms predicted occurrence identical and similar
A
to actual dynamics in cropland and grassland, respectively. Nutrient loading also may be
greater in cropland than grassland playas due to agriculture (Freemark and Boutin 1995),
abundance (Chapter III). Although abundance and carrying capacity may be higher in
cropland playas, the chaos results herein indicate that dynamics of an amphibian
assemblage in disturbed wetlands are altered from their natural state in grasslands.
132
Estimates of intrinsic rate of increase were greater using the iterati\ e method than
least squares for all difference equations. This probably was a consequence of
difference equations were solved and parameters estimated for 2-3 sequential time
iterates for the iterative technique. Thus, parameters per playa were approximated from
m/2 or m/3 samples (where m is the duration of the time series) for non-predator and
predator difference functions, respectively. The arithmetic average then was calculated
per playa for each parameter estimate. In contrast, the least-squares technique fit the
difference function to the entire data such that residual sum-of-squares v\ as minimized,
and obtained one sample estimate per parameter per playa. Consequently, iteration may
capture the true dynamics of a chaotic time series better than least-squares estimation
because it is based on more samples. However, limited simulations (J. Surles, Texas
Tech University, unpublished data) suggest that least-squares estimates are less \ ariable
assemblage on the SHP, and anthropogenic disturbance around playa wetlands can
dismpt its occurrence. Chaos may exist naturally in this assemblage because of densit>
predation and interdemic movement (Chapters III and IV). Additional research is needed
altered from their natural state in grasslands. These results and those of Allen et al.
133
(1993) on simulated metapopulations suggest that spatially stmctured amphibians
(e.g., wetland buffer restoration, corridor construction) that restore dynamics and
disturbed wetlands can be predicted initially by using the stochastic Ricker model and
parameter estimates presented herein (Table 6.3). The true underlying function
amphibian assemblage likely is more complex than the stochastic Ricker function. Thus,
N,+]^aNt/[l+{bNty], Maynard Smith 1974) and predation. Future models also should
independent state variables (e.g., rainfall, water chemistry) could be added to refine
models. Cross validation techniques and information statistics (e.g., Akaike information
criterion) can be used to compare predictability among difference equations (Turchin and
the population-level may give insight into the embedding dimension of the amphibian
assemblage, and provide support for combined assemblage analyses as presented here.
134
CHAPTER VII
CONCLUSION
wetlands affects body size, demographics, and chaotic dynamics of amphibians on the
(Chapter II). The presence of biological chaos also was altered by anthropogenic
New Mexico and plains spadefoots {Spea multiplicata, S. bombifrons) was positively
assemblage was not affected (Chapter III). Grassland playas exhibited source dynamics
most often, while playas in cultivated landscapes generally were sinks (Chapter III).
Landuse did not affect temporal niche partitioning among amphibian species (Chapter V).
were positively related with amphibian abundance and affected community composition
(Chapter IV). A stochastic Ricker equation predicted that density dependence and
size between cultivated and grassland landscapes may have been related to wetland
135
hydroperiod, agricultural chemicals, availability of food resources, or density of
grassland landscapes, because intrinsic rate of increase was greater at grassland than
(Chapter III). Rigidity in the genie norm of reaction for temporal resource use may have
increased abundance near natal wetlands (Chapter IV). Most of these inferences are
predictions by difference equations; thus, they are speculative. More research is needed
(discussed later). Regardless, results herein strongly suggest that agricultural cultivation
around wetlands affects amphibian populations, thus the relative importance of various
136
Cropland playas exhibited sink dynamics whereas playas in grassland landscapes were
most often sources of individuals (Chapter III). Inasmuch as body size is positively
cropland landscapes (Carey et al. 1999). Lastly, I suspect that interdemic movement is
reduced in cropland landscapes, which can have negative demographic and genetic
Harrison 1994).
Southem High Plains (SHP) amphibians. I recommend the following studies: (1)
determine the effect of landscape use and structure on interdemic genetic variability of
SHP amphibians, (2) estimate species-specific vagility of SHP amphibians in natural and
anthropogenic cover types, (3) estimate home ranges, dispersal probabilities, and micro-
habitat use of SHP amphibians, (4) simultaneously examine the effect of anthropogenic
landuse on sedimentation rate, water chemistry, aquatic and terrestrial food resources,
and vegetation characteristics in playa wetlands, (5) investigate the effect of landuse on
body size, demographics, and dynamics of SHP amphibian larvae populations, (6)
137
examine pathogenic incidence between landuses in SHP amphibians, (7) determine the
effect of UV-B radiation on hatching and survival of SHP amphibian larvae, and (8)
determine the effect of landuse on the relative importance of regulafing mechanisms (e.g..
studies could be accomplished through multiple concurrent lab and field experiments. A
and after grassland restoration, or alternatively in a grassland landscape before and after
disturbance, would be novel and provide strong inferential insight. Results from the
using the geographic information system to predict the effect of restoration techniques
established on the SHP, and demographic information of playa amphibians tied to global
declines.
wildlife populations with substantial statistical replication. Playas also are spatially
stmctured; thus, they are ideal systems to test various hypotheses of metapopulation
138
opportunity. I encourage continued amphibian research on the Southem High Plains.
Indeed, Ann Anderson's (1997) research and this study only have scratched the surface of
our understanding of basic life history for SHP amphibians and the effect of human
Tech University will guarantee the progression of science and our knowledge. Buena
139
LITERATURE CITED
Agresti, A. 1990. Categorical data analysis. John Wiley and Sons, New York, New
York, USA.
Alford, R. A. 1999. Ecology: resource use, competition, and predation. Pages 240-278
in R. W. McDiarmid and R. Altig, editors. Tadpoles: the biology of anuran
larvae. University of Chicago, Chicago, Illinois, USA.
Alford, R. A., and R. N. Harris. 1988. Effects of larval growth history on anuran
metamorphosis. American Naturalist 131:91-106.
Alford, R. A., and S. J. Richards. 1999. Global amphibian declines: a problem in applied
ecology. Annual Review of Ecology and Systematics 30:133-165.
Allen, B. L., B. L. Harris, K. R. Davis, and G. B. Miller. 1972. The mineralogy and
chemistry of High Plains playa lake soils and sediments. Water Resources Center
Publication WRC-72-4, Texas Tech University, Lubbock, Texas, USA.
Anderson, A. M. 1997. Habitat use and diet of amphibians breeding in playa wetlands
on the Southem High Plains of Texas. Thesis, Texas Tech University, Lubbock,
USA.
Andrewartha, H. G., and L. C. Birch. 1954. The distribution and abundance of animals.
University of Chicago, Chicago, Illinois, USA.
Baker, J. M. R., and V. Waights. 1993. The effect of sodium nitrate on the growth and
survival of toad larvae {Bufo bufo) in the laboratory. HerpetologicalJournal
3:147-148.
140
Baker, J. M. R., and V. Waights. 1994. The effects of nitrate on tadpoles of the tree frog
{Litoria caerulea). Herpetological Journal 4:106-108.
Bardsley, L. S. 1998. Use and selecfion of terrestrial resources by adult common toads
{Bufo bufo) in agricultural landscapes. Dissertation. De Montfort Univerisit>.
Leicester, United Kingdom.
Bascompte, J., and R. V. Sole. 1996. Habitat fragmentation and extinction thresholds in
spatially explicit models. Journal of Animal Ecology 65:465-473.
Beck, C. W., and J. D. Congdon. 2000. Effects of age and size at metamorphosis on
performance and metabolic rates of Southern Toad, Bufo terrestris. metamorphs.
Functional Ecology 14:32-38.
Berryman, A. A. 1991. Chaos in ecology and resource management: what causes it and
how to avoid it. Pages 23-38 in J. A. Logan and F. P. Hain, editors. Chaos and
insect ecology. Virginia Polytechnic Institute and State University, Virginia
Experiment Station Information Series 91-3, Blacksburg, Virginia, USA.
Berryman, A. A., and J. A. Millstein. 1989. Are ecological systems chaotic-And if not,
why not? Trends in Ecology and Evolution 4:26-28.
Berven, K. A. 1981. Mate choice in the wood frog, Rana sylvatica. Evolution
35:707-722.
Berven, K. A. 1982. The genetic basis of altitudinal variation in the wood frog Rana
sylvatica. I. An experimental analysis of life history traits. Evolution
36:962-983.
Berven, K. A. 1990. Factors affecting population fluctuations in larval and adult stages
of the wood frog {Rana sylvatica). Ecology 71:1599-1608.
Berven, K. A. 1995. Populafion regulation in the wood frog, Rana sylvatica, from three
diverse geographic localities. Australian Journal of Ecology 20:385-392.
141
Berven, K. A., and D. E. Gill. 1983. Interpreting geographic variation in life-history
traits. American Zoologist 23:85-97.
Bolen, E. G., L. M. Smith, and H. L. Schramm, Jr. 1989. Playa lakes: prairie wetlands of
the Southern High Plains. BioScience 39:615-623.
Bonin, J., J.-L. DesGranges, J. Rodrigue, and M. Ouellet. 1997a. Anuran species
richness in agricultural landscapes of Quebec: foreseeing long-term results of road
call surveys. Pages 141-149 m D. M. Green, editor. Amphibians in decline:
Canadian studies of a global problem. Society for the Study of Amphibians and
Reptiles, St. Louis, Missouri, USA.
Bonin, J., M. Ouellet, J. Rodrigue, and J.-L. DesGranges. 19976. Measuring the health
of frogs in agricultural habitats subjected to pesticides. Pages 246-257 in D. M.
Green, editor. Amphibians in decline: Canadian studies of a global problem.
Society for the Study of Amphibians and Reptiles, St. Louis, Missouri, USA.
Bradford, D. F., C. Swanson, and M. S. Gordon. 1992. Effects of low pH and aluminum
on two declining species of amphibians in the Sierra Nevada, California. Journal
ofHerpetology 26:369-377.
142
Bragg, A. N. 1965. Gnomes of the night: the spadefoot toads. University of
Pennsylvania, Philadelphia, Pennsylvania, USA.
Burel, F. 1989. Landscape structure effects on carabid beefies spatial pattems in westem
France. Landscape Ecology 2:215-226.
Camp, C. D., J. L. Marshall, and R. M. Ausfin, Jr. 2000. The evolution of adult body-
size in black-bellied salamanders {Desmognathus quadramaculatus). Canadian
Journal of Zoology 78:1712-1722.
Case, T. J., and M. E. Gilpin. 1974. Interference competition and niche theory.
Proceedings of the National Academy of Sciences, U.S.A. 71:3073-3077.
Chesson, P., and N. Huntly. 1997. The roles of harsh and fluctuating conditions in the
dynamics of ecological communities. American Naturalist 150:519-553.
Collins, J. P., and J. R. Holomuzki. 1984. Intraspecific variation in diet within and
between trophic morphs in larval tiger salamanders {Ambystoma tigrinum
nebulosum). Canadian Journal of Zoology 62:168-174.
143
Connell, J. H. 1980. Diversity and coevolution of competitors, or the ghost of
competition past. Oikos 35:131-138.
Com, P. S., and R. B. Bury. 1989. Logging in westem Oregon: responses of headwater
habitats and stream amphibians. Forest Ecology and Management 29:39-57.
Dayton. G. H., and L. A. Fitzgerald. 2001. Competition, predation, and the distributions
of four desert anurans. Oecologia 129:430-435.
deMaynadier, P. G., and M. L. Hunter, Jr. 1995. The relationship between forest
management and amphibian ecology: a review of the North American literature.
Environmental Reviews 3:230-261.
deMaynadier, P. G., and M. L. Hunter, Jr. 1998. Effects of silvicultural edges on the
distribution and abundance of amphibians in Maine. Conservation Biology
12:340-352.
144
Dimmitt, M. A. 1975. Terrestrial ecology of spadefoot toads {Scaphiopus): emergence
cues, nutrition, and burrowing habits. Dissertation, University of Califomia,
Riverside, Califomia, USA.
Dodd, C. K., Jr. 1996. Use of terrestrial habitats by amphibians in the sandhill uplands
of north-central Florida. Alytes 14:42-52.
Dodd, C. K., Jr. 1997. Imperiled amphibians: a historical perspective. Pages 163-200 in
G. W. Benz and D. E. Collins, editors. Aquatic fauna in peril: the southeastern
perspective. Special Publication 1, Southeast Aquatic Research Institute, Lenz
Design and Communications, Decatur, Georgia, USA.
Dodd, C. K., Jr., and D. E. Scott. 1994. Drift fences encircling breeding sites. Pages
125-130 in W. R. Heyer. M. A. Donnelly. R. W. McDiarmid. L.-A. C. Hayek.
and M. S. Foster, editors. Measuring and monitoring biological diversity:
standard methods for amphibians. Smithsonian Institution. Washington, D. C.
USA.
145
Ellner, S. P. 2000. Defining chaos for real, noisy data: local Lyapunov exponents and
sensitive response to perturbations. Pages 1-32 in J. N. Perry. R. H. Smith, I. P.
Woiwod, and D. R. Morse, editors. Chaos in real data: the analysis of non-linear
dynamics from short ecological time series. Kluwer Academic Publishers.
Dordrecht, Netherlands.
Ellner, S. P., and P. Turchin. 1995. Chaos in a noisy world: new methods and evidence
from time-series analysis. American Naturalist 145:343-374.
Fahrig. L., and G. Merriam. 1985. Habitat patch connectivity and population survival.
Ecolog}-66:\762-\76^.
Fahrig. L., and J. Paloheimo. 1988. Determinants of local population size in patchy
habitats. Theoretical Population Biology 34:194-212.
Fioramonti, E., R. D. Semlitsch, H.-U. Reyer, and K. Fent. 1997. Effects of triphenyltin
and pH on the growth and development of Rana lessonae and Rana esculenta
tadpoles. Environmental Toxicology and Chemistry 16:1940-1947.
Flowers, M. A., and B. M. Graves. 1995. Prey selecti^ity and size-specific diet changes
in Bufo cognatus and B. woodhousii during earh- postmetamorphic ontogen\.
Journal ofHerpetology 29:608-612.
Fontenot, C. L., Jr. 1999. Reproductive biology of the aquatic salamander yimp/imma
tridactylum in Louisiana. Journal ofHerpetology 33:100-105.
Fomianowicz, D. R.. Jr., and E. D. Brodie, Jr. 1982. Relati\e palatability of members of
the larval amphibian community. Copeia 1982:91-97.
146
Freemark, K., and C. Boutin. 1995. Impacts of agricultural herbicide use on terrestrial
wildlife in temperate landscapes: a re\ iew with special reference to North
America. Agriculture, Ecosystems and Environment 52:67-91.
Freund, R. J, and R. C. Littell. 2000. SAS® system for regression. Third edition. SAS
Institute, Gary. North Carolina, USA.
Gause, G. F. 1934. The struggle for existence. Williams and Wilkins, Baltimore,
Maryland, USA.
Gehlbach, F. R., J. R. Kimmel, and W. A. Weems. 1969. Aggregations and body water
relations in tiger salamanders {Ambystoma tigrinum) from the Grand Canyon
rims, Arizona. Physiological Zoology 42:173-182.
Goater, C. P., R. D. Semlitsch, and M. V. Bemasconi. 1993. Effects of body size and
parasite infection on the locomotion performance of juvenile toads, Bufo bufo.
Oikos 66:129-136.
Gomulkiewicz, R., and M. Kirkpatrick. 1992. Quantitative genetics and the evolution of
reaction norms. Evolution 46:390-411.
147
Gustavson, T. C , V. T. Holliday, and S. D. Hovorka. 1994. Development of playa
basins, Southem High Plains, Texas and New Mexico. Pages 5-14 in L. V.
Urban and A. W. Wyatt, editors. Proceedings of the playa basin symposium.
Texas Tech University, Lubbock, Texas, USA.
Guthery, F. S., and F. C. Bryant. 1982. Status of playas in the Southem Great Plains.
Wildlife Society Bulletin 10:309-317.
Hastings, A., and S. Harrison. 1994. Metapopulation dynamics and genetics. Annual
Review of Ecology and Systematics 25:167-188.
Haukos, D. A., and L. M. Smith. 1994. The importance of playa wetlands to biodiversity
of the Southem High Plains. Landscape and Urban Planning 28:83-98.
148
Hecnar, S. J. 1995. Acute and chronic toxicity of ammonium nitrate fertilizer to
amphibians from southem Ontario. Environmental Toxicology and Chemistry
14:2131-2137.
Hecnar, S. J., and R. T. M'Closkey. 1996. Regional dynamics and the status of
amphibians. Ecology 77:2091-2097.
Hecnar, S. J., and R. T. M'Closkey. 1997. Pattems of nestedness and species association
in a pond-dwelling amphibian fauna. Oikos 80:371-381.
Henttonen, H., J. Tast, J. Viitala, and A. Kaikusalo. 1984. Ecology of cyclic rodents in
northem Finland. Memoranda Societatis pro Fauna et Flora Fennica 60:84-92.
149
Houlahan, J. E., C. S. Findlay, B. R. Schmidt, A. H. Meyers, and S. L. Kuzmin. 2000.
Quantitative evidence for global amphibian population declines. Nature
404:752-755.
Howard, R. D., and J. R. Young. 1998. Individual variation in male vocal traits and
female mating preference in Bufo americanus. Animal Behaviour 55:1165-1179.
Huey, R. B., and E. R. Pianka. 1983. Temporal separafion of acfivity and interspecific
dietary overlap. Pages 281-290 in R. B. Huey, E. R. Pianka, and T. W. Schoener.
editors. Lizard ecology: studies of a model organism. Harvard University,
Cambridge, Massachusetts, USA.
Hurlbert, S. H. 1978. The measurement of niche overlap and some relatives. Ecology
59:67-77.
Hutchison, V. H., W. G. Whitford, and M. Kohl. 1968. Relation of body size and
surface area to gas exchange in anurans. Physiological Zoology 41:65-85.
Jaeger, R. G. 1980. Fluctuations in prey availability and food limitation for a terrestrial
salamander. Oecologia 44:335-34\.
John-Alder, H. B., and P. J. Morin. 1990. Effects of larval density on jumping ability
and stamina in newly metamorphosed Bufo woodhousii fowleri. Copeia
1990:856-860.
150
Johnson, R. A., and D. W. Wichern. 1998. Applied multivariate statistical analysis.
Fourth edition. Prentice Hall, Upper Saddle River. New Jersey, USA.
Kolozsvary, M. B., and R. K. Swihart. 1999. Habitat fragmentation and the distribution
of amphibians: patch and landscape correlates in farmland. Canadian Journal of
Zoology 77:\2SS-1299.
Krupa, J. J. 1986. Multiple egg clutch production in the Great Plains toad. Prairie
Naturalist \S:\5\-\52.
Krupa, J. J. 1994. Breeding biology of the Great Plains toad in Oklahoma. Journal of
Herpetology 28:217-224.
Laan, R., and B. Verboom. 1990. Effects of pool size and isolation on amphibian
communities. Biological Conservation 54:251-262.
Lefkovitch, L. P., and L. Fahrig. 1985. Spafial characteristics of habitat patches and
population survival. Ecological Modelling 30:297-308.
Littell, R. C , R. J. Freund, and R. C. Spector. 1991. SAS® system for linear models.
Third edifion. SAS Institute, Gary, North Carolina, USA.
151
Littell, R. C , G. A. Milliken, W. W. Stroup, and R. D. Wolfinger. 1996. SAS® system
for mixed models. SAS Institute, Gary, North Carolina, USA.
Loman, J. 1988. Breeding by Rana temporaria; the importance of pond size and
isolation. Memoranda Societatis pro Fauna et Flora Fennica 64:113-115.
MacArthur, R. H., and R. Levins. 1967. The limiting similarity, convergence, and
divergence of coexisting species. American Naturalist 101:337-385.
152
May, R. M. 19756. Biological populafions obeying difference equations: stable points,
stable cycles, and chaos. Journal of Theoretical Biology 51:511-524.
McClanahan, L. L., R. Ruibal, and V. H. Shoemaker. 1994. Frogs and toads in deserts.
Scientific American 270:64-70.
McGarigal, K., and B. J. Marks. 1995. FRAGSTATS: spatial pattem analysis program
for quantifying landscape structure. U.S. Forest Service General Technical
Report PNW-GTR-351.
Miaud, C , D. Sanuy, and J.-N. Avrillier. 2000. Terrestrial movements of the natterjack
toad Bufo calamita (Amphibia, Anura) in a semi-arid, agricultural landscape.
Amphibia-Reptilia 21:357-369.
Milliken, G. A., and D. E. Johnson. 1992. Analysis of messy data. Chapman and Hall,
New York, New York, USA.
Milton, J. S., and J. C. Amold. 1995. Introduction to probability and statistics. Third
edition. McGraw-Hill, New York, New York, USA.
153
Montgomery, D. C. 2001. Design and analysis of experiments. Fifth edifion. John
Wiley and Sons, New York, New York, USA.
Morin, P. J. 1983. Predation, competition, and the composition of larval anuran guilds.
Ecological Monographs 53:119-138.
Myers, R. H. 1990. Classical and modern regression with applications. Second edition.
PWS-KENT, Boston, Massachusetts, USA.
Newman, R. A. 1994a. Genefic variafion for phenotypic plasficity in the larval life
history of spadefoot toads {Scaphiopus couchii). Evolution 48:1773-1785.
154
Newman, R. A. 19946. Effects of changing density and food level on metamorphosis of
a desert amphibian, Scaphiopus couchii. Ecology 75:1085-1096.
Newman, R. A. 1999. Body size and diet of recently metamorphosed spadefoot toads
{Scaphiopus couchii). Herpetologica 55:507-515.
Newman, R. A., and A. E. Dunham. 1994. Size at metamorphosis and water loss in a
desert anuran {Scaphiopus couchii). Copeia 1994:372-381.
Olsen, L. F., and H. Degn. 1985. Chaos in biological systems. Quarterly Review of
Biophysics 18:165-225.
Osterkamp, W. R., and W. W. Wood. 1987. Playa-lake basins on the Southem High
Plains of Texas and New Mexico. Parti. Hydrologic, geomorphic, and geologic
evidence for their development. Geological Society of America Bulletin
99:215-223.
Paton, P. W., and W. B. Crouch III. 2002. Using the phenology of pond-breeding
amphibians to develop conservation strategies. Conservation Biology
16:194-204.
155
Pechmann, J. H. K. 1994. Populafion regulation in complex life cycles: aquatic and
terrestrial density-dependence in pond-breeding amphibians. Dissertation, Duke
University, Durham, North Carolina, USA.
Petranka. J. W., and C. A. Kennedy. 1999. Pond tadpoles with generalized morphology:
is it time to reconsider their functional roles in aquatic communities? Oecologia
120:621-631.
Pianka, E. R. 1974. Niche overlap and diffuse competition. Proceedings of the National
Academy of Sciences, U.S.A. 71:2141-2145.
Pianka, E. R. 1976. Competition and niche theory. Pages 114-141 in R. M. May, editor.
Theoretical ecology: principles and applications. W. B. Saunders, Philadelphia,
Pennsylvania, USA.
156
Pianka, E.R. 1994. Evolutionary ecology. Fifth edition. HarperCollins, New York,
New York, USA.
Polls, G. A., and R. D. Holt. 1992. Intraguild predafion: the dynamics of complex
trophic interactions. Trends in Ecology and Evolution 7:151-154.
Reading, C. J. 1990. A comparison of size and body weights of common toads {Bufo
bufo) from two sites in Southem England. Amphibia-Reptilia 11:155-163.
Reading, C. J., and R. T. Clarke. 1995. The effects of density, rainfall and
environmental temperature on body condition and fecundity in the common toad,
Bufo bufo. Oecologia 102:453-459.
Reading, C. J., and R. T. Clarke. 1999. Impacts of climate and density on the duration of
the tadpole stage of the common toad Bufo bufo. Oecologia 121:310-315.
Reh, W., and A. Seitz. 1990. The influence of land use on the genetic stmcture of
populations of the common frog Rana temporaria. Biological Conservation
54:239-249.
Relyea, R. A., and N. Mills. 2001. Predator-induced stress makes the pesticide carbaryl
more deadly to gray treefrog tadpoles {Hyla versicolor). Proceedings of the
National Academy of Sciences, U.S.A. 98:2491-2496.
157
Rohde, K., and P. P. Rohde. 2001. Fuzzy chaos: reduced chaos in combined dynamics
of several independently chaotic populations. American Naturalist 158:553-556.
Roughgarden, J. 1998. Primer of ecological theory. Prentice Hall, Upper Saddle River,
New Jersey, USA.
Sabin, T. J., and V. T. Holliday. 1995. Playas and lunettes on the Southem High Plains:
morphometric and spatial relationships. Annals of the Association of American
Geographers 85:286-305.
Schaffer, W. M., and M. Kot. 1986. Chaos in ecological systems: the coals that
Newcastle forgot. Trends in Ecology and Evolution 1:5 8-63.
Schneider, D. W., and T. M. Frost. 1996. Habitat duration and community stmcture in
temporary ponds. Journal of the North American Benthological Society
15:64-86.
158
Scale. D. B. 1980. Influence of amphibian larvae on primar\ production, nutrient flux,
and competition in a pond ecosystem. £co/6>g>' 61:1531-1550.
Shulenburger, L., Y-C. Lai, T. Yalcinkaya, and R. D. Holt. 1999. Controlling transient
chaos to prevent species extinction. Physics Letters A 260:156-161.
Sinsch, U. 1988. Seasonal changes in the migratory behaviour of the toad Bufo bufo:
direction and magnitude of movements. Oecologia 76:390-398.
159
Sinsch, U. 1990. Migrafion and orientafion in anuran amphibians. Ethology Ecology
and Evolution 2:65-79.
Sinsch, U., and D. Seidel. 1995. Dynamics of local and temporal breeding assemblages
in a Bufo calamita metapopulation. Australian Journal of Ecology 20:351-361.
Sjogren, P. 1991. Extinction and isolation gradients in metapopulations: the case of the
pool frog {Rana lessonae). Biological Journal of the Linnean Society
42:135-147-
Skelly, D. K. 1996. Pond drying, predators, and the distribution of Pseudacris tadpoles.
Copeia 1996:599-605.
Skelly, D. K., and E. E. Wemer. 1990. Behavioral and life-historical responses of larval
American toads to an odonate predator. Ecology 71:2313-2322.
Smith, D. C. 1987. Adult recmitment in choms frogs: effects of size and date at
metamorphosis. Ecology 6^:344-350.
Smith, D. C , and J. Van Burskirk. 1995. Phenotypic design, plasticity, and ecological
performance in two tadpole species. American Naturalist 145:211-233.
Smith, L. M., and D. A. Haukos. 2002. Floral diversity in relation to playa wetland area
and watershed disturbance. Conservation Biology 16:in press.
160
Stacey, P. B., V. A. Johnson, and M. L. Taper. 1997. Migration within metapopulations:
the impact upon local populafion dynamics. Pages 267-291 in I. A. Hanski and
M. E. Gilpin, editors. Metapopulation biology: ecology, genetics, and evolution.
Academic Press, San Diego, Califomia, USA.
Stamps, J. A., M. Buechner, and V. V. Krishnan. 1987. The effects of edge permeability
and habitat geometry on emigration from patches of habitat. American Naturalist
129:533-552.
Stokes, M. E., C. S. Davis, and G. G. Koch. 2000. Second edifion. Categorical data
analysis using the SAS® system. SAS Institute, Gary, North Carolina, USA.
Strong, D. R., Jr. 1983. Natural variability and the manifold mechanisms of
ecological communities. American Naturalist 122:636-660.
Szacki, J. 1999. Spatially stmctured populations: how much do they match the classic
metapopulation concept? Landscape Ecology 14:369-379.
Taigen, T. L., and F. H. Pough. 1981. Activity metabolism of the toad {Bufo
americanus): ecological consequences of ontogenetic change. Journal of
Comparative Physiology B 144:247-252.
ter Braak, C. J. F., and P. Smilauer. 1998. CANOCO reference manual and user's guide
to Canocofor windows: software for canonical community ordination (version 4).
Microcomputer Power, Ithaca, New York, USA.
161
Tinsley, R. C , and K. Tocque. 1995. The populafion dynamics of a desert anuran.
Scaphiopus couchii. Australian Journal of Ecology 20:376-384.
Toft, C. A., and W. E. Duellman. 1979. Anurans of the lower Rio LluUapichis,
Amazonian Peru: a preliminary analysis of community stmcture. Herpetologica
35:71-77.
Turchin, P. 1991. Nonlinear modeling of time-series data: limit cycles and chaos in
forest insects, voles, and epidemics. Pages 39-61 in J. A. Logan and F. P. Hain.
editors. Chaos and insect ecology. Virginia Polytechnic Institute and State
University, Virginia Experiment Station Information Series 91-3, Blacksburg,
Virginia, USA.
Turchin, P. 1993. Chaos and stability in rodent population dynamics: evidence from
non-linear time series analysis. Oikos 68:167-172.
Turchin, P., and S. P. Ellner. 2000. Modelling time-series data. Pages 33-48 in J. N.
Perry, R. H. Smith, I. P. Woiwod, and D. R. Morse, editors. Chaos in real data:
the analysis of non-linear dynamics from short ecological time series. Kluwer
Academic Publishers, Dordrecht, Netherlands.
Turchin, P., and A. D. Taylor. 1992. Complex dynamics in ecological time series.
Ecology 73:2^9-305.
162
Ultsch, G. R. 1973. A theoretical and experimental investigation of the relationship
between metabolic rate, body size, and oxygen exchange capacity. Respiration
Physiology n:l43-l60.
Verrell, P. A. 1982. Male newts prefer large females as mates. Animal Behaviour
30:1254-1255.
Verrell, P. A. 1985. Male mate choice for large, fecund females in the red-spotted newt,
Notophthalmus viridescens: how is the size assessed? Herpetologica 41:3S2-3S6.
Via, S., and R. Lande. 1985. Genotype-environment interaction and the evolution of
phenotypic plasticity. Evolution 39:505-522.
Webb, R. G., and W. L. Roueche. 1971. Life history aspects of the tiger salamander
{Ambystoma tigrinum mavortium) in the Chihuahuan Desert. Great Basin
Naturalist 31:193-212.
163
Welsh, H. W., Jr. 1990. Relictual amphibians and old-growth forests. Conservation
Biology 4:309-319.
Werner, E. E. 1986. Amphibian metamorphosis: growth rate, predation risk, and the
optimal size at transformation. American Naturalist 128:319-341.
Wemer, E. E., and M. A. McPeek. 1994. Direct and indirect effects of predators on two
anuran species along an environmental gradient. Ecology 75:1368-1382.
Wiens, J. A., R. L. Schooley, and R. D. Weeks, Jr. 1997. Patchy landscapes and animal
movements: do beetles percolate? Oikos 78:257-264.
Wiest, J. A., Jr. 1982. Anuran succession at temporary ponds in a post oak-savanna
region of Texas. U. S. Fish and Wildlife Service Research Report 13:39-47.
164
Whitaker, J. O., Jr., D. Rubin, and J. R. Munsee. 1977. Observafions on food habits of
four species of spadefoot toads, genus Scaphiopus. Herpetologica 33:468^75.
Wilbur, H. M. 1977. Interactions of food level and population density in Rana sylvatica.
Ecology 5S:206-209.
Wilbur, H. M. 1980. Complex life cycles. Annual Review of Ecology and Systematics
11:67-93.
With, K. A., and T. O. Crist. 1995. Critical thresholds in species' responses to landscape
structure. Ecology 76:2446-2459.
Yanes, M., J. M. Velasco, and F. Suarez. 1995. Permeability of roads and railways to
vertebrates: the importance of culverts. Biological Conservation 71:217-222.
165
Yeargers, E. K., R. W. Shonkwiler, and J. V. Herod. 1996. An introduction to the
mathematics of biology. Birkhauser, Boston. Massachusetts, USA.
166
APPENDIX A
167
Table A.l. Percent coverage of cover types in landscapes^ surrounding 16 playa wetlands
between landuses on the Southem High Plains, Texas, 1999 and 2000.
Landuse"
Cover t> pe*^ Cropland Grassland
X SE X oh
Alfalfa Medicago sativa 0.12 0.08 0.02 0.01
Barley Hordeum spp. 0 0 0.04 0.04
Com Zea mays 2.11 1.31 1.07 0.47
Cotton Gossypium hirsutum 60.3 5.37 2.75 1.27
Ha> Phalaris arundinacea 0.43 0.19 1.17 0.67
Kleingrass Panicum coloratum 0.45 0.19 0.34 0.17
Millet Pennisetum glaucum 0.15 0.15 0 0
Oats Arena sativa 0.21 0.11 0.09 0.08
Rye Secale cereale 0.17 0.17 0 0
Sorghum Sorghum vulgare 4.78 1.39 3.21 0.64
Soybean Glycine max 1.14 0.81 0 0
Sunflower Helianthus annuus 0.79 0.24 0 0
Vegetables (various spp.) 0.04 0.02 0 0
Wheat Triticum aestivum 6.95 2.28 10.9 2.96
Fallow 2.55 0.66 1.65 0.52
Buildings 0.96 0.24 0.38 0.12
Maintained county roads 1.67 0.09 1.05 0.09
Interstate 0.32 0.32 0 0
Other anthropogenic structures 0.02 0.02 0.01 0.01
Canyon 0.96 0.96 0 0
Native grass 3.19 0.67 28.8 4.34
Replanted grass 8.47 1.29 46.1 6.78
Playa wetland 3.86 0.32 2.35 0.23
Water 0.29 0.14 0.02 0.02
^Landscapes were 2830-ha circular plots (i.e., 3-km radius) with their origins positioned
at the center of each study playa.
''Percent cover of cotton, county roads, and native and replanted grass differed (P<0.006)
between landuses by Wilcoxon 2-sample nonparametric tests (Conover 1980:216); no test
was performed between landuses when x and SE=0; « = 8 playas/landuse.
'^Other anthropogenic stmctures included cemeteries, quarries, and airstrips; replanted
grass was cultivated previously then revegetated by the Conservation Reserve Program.
168
APPENDIX B
169
Table B.l. Statistics for ANOVAs testing differences in relative daily abundance and
diversity of amphibians between landuse t>pes (cropland \s. grassland) and years and
among species at 16 playa wetlands on the Southem High Plains, Texas. May-October
1999 and April-August 2000.
Variable Tests and Effects^'' df^ F P
Abundance 3-wa> ANOVA
Landuse 1 11.97 <0.001
Year 1 10.54 0.002
Species 6 55.67 <0.001
Landuse x Year 1 0.80 0.375
Landuse x Species 6 8.36 <0.001
Year >: Species 6 2.52 0.051
Landuse x Year x Species 6 1.40 0.223
2-wa> ANOVA
NSF
Landuse 20.33 <0.001
Year 2.63 0.131
Landuse x Year 2.34 0.152
BTS
Landuse 1.26 0.283
Year 8.08 0.015
Landuse x Year 0.02 0.895
GPT
Landuse 0.01 0.913
Year 2.35 0.152
Landuse x Year 0.11 0.747
PSF
Landuse 7.94 0.016
Year 0.51 0489
Landuse x Year 3.30 0.094
SCF
Landuse 0.23 0.639
Year 1.99 0.184
Landuse x Year 0.11 0.751
PLF
Landuse 0.156 0.156
Year 0.289 0.289
Landuse x Year 0.151 0.151
WHT
Landuse 0.33 0.576
Year 1.00 0.338
Landuse x Year 0.97 0.345
170
Table B.l. Continued.
Variable Tests and Effects*^^ df F
Abundance 1-way ANOVA
Cropland
Species 6 34.20 <0.001
Grassland
Species 6 18.74 <0.001
1999
Species 6 20.06 <0.001
2000
Species 6 15.56 <0.001
Diversity' 2-way ANOVA
Landuse 1 0.91 0.359
Year 1 0.29 0.602
Landuse x Year 1 1.04 0.329
Overall 3-way ANOVA on abundance was separated b\ species to test for differences
between landuses and years (i.e., 2-way analyses) then by landuse and year (i.e.. l-wa>
analyses) to test the species effect, because species interacted with landuse and year in
the 3-wa>' analysis.
''NSF = New Mexico spadefoot {Spea multiplicata), BTS = barred tiger salamander
{Ambystoma tigrinum mavortium). GPT = Great Plains toad {Bufo cognatus), PSF =
plains spadefoot {S. bombifrons), SCF = spotted chorus frog {Pseudacris clarkii). PLF =
plains leopard frog {Rana blairi), and WHT = Woodhouse's toad {B. woodhousii).
'^Error (denominator) df for 3-way ANOVA, 2-way ANOVAs. and 1-way ANOVAs was
84 (2 landuses x 2 years x 7 species x 3 playas), 12 (2 landuses x 2 >ears x 3 playas). and
49 (7 species x 7 playas), respectively.
'^Diversity was calculated using the Shannon-Weaver algorithm (Hair 1980:273); no
species effect was included in analyses because all species were used in estimation of the
index.
171
APPENDIX C
172
Table C.l Stafistics for ANOVAs for species (by age class) that differed significantly in
relati\ e dail>' abundance between landuse types (cropland vs. grassland) or years (see
Table B.l) at 16 playa wetlands on the Southern High Plains, Texas, May-October 1999
and April-August 2000.
Species Tests and Effects df
NSF 2-way ANOVA
Metamorph
Landuse 12 6.79 0.023
Year 12 13.99 0.003
Landuse x Year 10.19 0.008
Subadult
Landuse 12 17.18 0.001
Year 12 50.06 <0.001
Landuse x Year 12 21.26 <0.001
Adult
Landuse 12 14.02 0.002
Year 12 0.10 0.758
Landuse x Year 12 048 0.499
PSF Metamorph
Landuse 12 1.93 0.189
Year 12 12.1 0.005
Landuse x Year 12 0.001 0.966
Subadult
Landuse 12 6.74 0.023
Year 12 11.69 0.005
Landuse x Year 12 8.66 0.012
Adult
Landuse 12 3.71 0.078
Year 12 7.88 0.016
Landuse x Year 12 2.79 0.121
BTS Metamorph
Landuse 12 2.07 0.176
Year 12 10.50 0.007
Landuse x Year 12 0.01 0.909
Adult
Landuse 12 0.22 0.648
Year 12 0.76 0.399
Landuse x Year 12 0.01 0.907
NSF 1-way ANOVA
Metamorph
Landuse in 1999 1,6 17.46 0.006
Landuse in 2000 1,6 0.17 0.698
Year in Cropland 1.6 34.84 0.001
Year in Grassland 1,6 0.11 0.747
173
T a b l e d . Continued.
Species Tests and Effects df F P
NSF 1-way ANOVA
Subadult
Landuse in 1999 1,6 0.36 0.570
Landuse in 2000 1,6 22.55 0.003
Year in Cropland 1,6 43.66 <0.001
Year in Grassland 1,6 6.96 0.039
PSF 1-way ANOVA
Subadult
Landuse in 1999 1,6 1.96 0.211
Landuse in 2000 1,6 7.79 0.032
Year in Cropland 1,6 10.59 0.017
Year in Grassland 1,6 1.27 0.302
"^NSF = New Mexico spadefoot {Spea multiplicata), PSF = plains spadefoot {S.
bombifrons), and BTS = barred tiger salamander {Ambystoma tigrinum mavortium).
Overall 2-way ANOVA on abundance of significant species was separated by landuse
and year for metamorph NSF and subadult NSF and PSF (i.e., 1-way analyses), because
landuse and year main effects interacted.
174
APPENDIX D
175
176
Figure D.l. Anthropogenic and natural cover types in landscapes (i.e., 2,830-ha [3-km
radius] circular plots) surrounding 16 playa wetlands on the Southem High Plains, 1999
and 2000. Each study playa is positioned at the landscape plot origin. Crop types
included alfalfa {Medicago sativa), barley {Hordeum spp.), com {Zea mays), cotton
{Gossypium hirsutum), domestic grass {Panicum coloratum), hay {Phalaris
arundinacea), millet {Pennisetum glaucum), oats {Avena sativa), sorghum {Sorghum
vulgare), soybean {Glycine max), sunflower {Helianthus annuus), various vegetable
species, and wheat {Triticum aestivum). Fallow was arable but unplanted land. Non-
playa water was permanent water resulting from anthropogenic modifications of the
landscape (e.g., tailwater pit, effluent treatment pond). Conservation Reserve Program
(CRP) grasslands were cultivated previously and replanted with native and exotic grasses
for erosion control and wildlife habitat.
Cropland Landscapes, 1999
Figure D. 1. Continued.
177
Grassland Landscapes, 1999
N
Cotton Buildings
^?^
Domestic Grass Maintained Roads
Fallow
178
Cropland Landscapes, 2000
179
Grassland Landscapes, 2000
t Oats
180