ResearchPDF Available

Conservation biology of the Giant Bullfrog, Pyxicephalus adspersus. PhD thesis

Authors:
Conservation biology of the giant bullfrog, Pyxicephalus adspersus
(Tschudi, 1838)
by
Caroline Angela Yetman
Submitted in partial fulfilment of the requirements for the degree
Doctor of Philosophy
(Zoology)
in the
Faculty of Natural and Agricultural Sciences
Department of Zoology and Entomology
University of Pretoria, Pretoria
South Africa
April 2012
i
Conservation biology of the giant bullfrog, Pyxicephalus adspersus
(Tschudi, 1838)
Student: Caroline A. Yetman
1
Supervisor: J. Willem H. Ferguson
2
Department:
2
Centre For Environmental Studies,
1,2
Department of Zoology and
Entomology, University of Pretoria, Pretoria, 0002, South Africa
Degree: Doctor of Philosophy (Zoology)
Abstract
The giant bullfrog, Pyxicephalus adspersus, is a large, explosive-breeding anuran from
southern Africa, which spends most of the year buried in a state of torpor. In South Africa
this species is considered to be Near-Threatened by habitat loss and other factors,
especially in the densely human populated Gauteng Province. The aim of this thesis was
to obtain essential outstanding information about the ecology of P. adspersus to
contribute towards improved conservation management of this species.
A model was used to predict the geographic range of P. adspersus in southern Africa, and
recent land cover data were used to determine the amount of suitable habitat remaining
for this species in Gauteng. As a step towards identifying P. adspersus conservation
management units, genetic structure and gene flow for populations from 23 localities in
Gauteng and seven additional localities in the north-eastern interior of South Africa was
ii
quantified using 708 base pairs of the mitochondrial gene cytochrome b. To investigate
the unpredictable activity and unknown spatial habitat requirements of P. adspersus, a
population’s spawning and non-breeding activity was monitored, and the movements of
70 adult frogs were radio- or spool-tracked during five summers at a site in Diepsloot,
Gauteng. Using skeletohronology, the age distribution of breeding P. adspersus at this
and two other peri-urban sites near Johannesburg, Gauteng, was examined.
Bioclimatic conditions were predicted to be suitable for P. adspersus in the temperate to
semi-arid interior, but not the low-lying eastern subtropical and arid western sides of
southern Africa. Limited genetic data suggested that P. adspersus was common in the
north-eastern interior of South Africa, and that populations in the Free State Province
represent an evolutionary significant unit of this species. In central Gauteng, where P.
adspersus may have declined by > 90%, populations < 20 km apart exhibited significant
genetic differentiation, possibly as a result of genetic drift. At Diepsloot, both annual
numbers of spawning events and numbers of spawning males were positively correlated
with rainfall, although other meteorological variables also affected the activity of P.
adspersus. Radio- or spool-tracked frogs showed high fidelity to their breeding site and
burrows, which were situated up to 1 km away from the water. Male P. adspersus
probably live 20 years in the wild, but at some peri-urban breeding sites adult life
expectancy and body size may be declining.
The geographic range of P. adspersus was predicted to be slightly smaller than that
reported by other authors, and deserves phylogeographic validation. The main
iii
conservation priority for P. adspersus in South Africa should be the protection of
terrestrial habitat for adult foraging and aestivation around, and for juvenile dispersal and
gene flow between, breeding sites. In Gauteng, the conservation of a P. adspersus meta-
population is critical, and could most likely be achieved in the northern region of this
province. Populations in the Free State Province deserve improved protection given their
reported genetic uniqueness. At local spatial scales specific threats (e.g. pollution) should
be ameliorated, and long-term monitoring should be implemented to detect real
population trends.
Keywords amphibian, Anura, biology, Gauteng, grassland, Pyxicephalidae, southern
Africa, threatened, wetland
iv
Acknowledgements
I am eternally indebted to Jimmy my father, Alice my mother, Johan Lötter my fiancé
and Mark my brother for their immeasurable support and assistance, morally, financially
and in the field. I have yet to find the caliber of your character in other mortals. I regret
and apologize deeply that this project took almost a decade to complete. There are
infinitely more profitable ways that your money and my time could have been spent.
I sincerely thank Wendy Duncan for assisting with field work in Diepsloot and Peter
Mokonoto for helping with the histology component of this project. I also thank others
who made a small but significant contribution to this research.
The Endangered Wildlife Trust sourced funding for a large portion of this project, which
was gratefully received from Rand Merchant Bank and the Pretoria East branch of the
South African Hunter’s and Game Conservation Association. Arrow Bulk Marketing is
thanked for funding fuel during the first three seasons of field work, and Cellar Rats
Wine Club, Diaz Films, and the late Bill Flynn are kindly acknowledged for collectively
sponsoring several radio-transmitters.
v
Disclaimer
This dissertation was written as a collection of five manuscripts to simplify the process of
publication of the results, as recommended by the Department of Zoology and
Entomology. There has consequently been duplication of some information between the
chapters. Chapters 2, 3 and 4 have been published and are, respectively, cited below.
Chapters 5 and 6 might be submitted for peer-review in the future.
Chapter 2
Yetman, C.A. & J.W.H. Ferguson (2011a). Spawning and non-breeding activity of adult
giant bullfrogs (Pyxicephalus adspersus). African Journal of Herpetology 60: 13-
29.
Chapter 3
Yetman, C.A. & J.W.H. Ferguson (2011b). Conservation implications of spatial habitat
use by adult giant bullfrogs (Pyxicephalus adspersus). Journal of Herpetology 45:
56-62.
Chapter 4
Yetman, C.A., P. Mokonoto & J.W.H. Ferguson (2012). Conservation implications of the
age/size distribution of giant bullfrogs (Pyxicephalus adspersus) at three peri-urban
breeding sites. Herpetological Journal 22: 23-32.
vi
Declaration
I, Caroline Angela Yetman declare that the thesis, which I hereby submit for the degree
Ph.D. (Zoology) at the University of Pretoria, is my own work and has not been
submitted by me for a degree at this or any other tertiary institution.
Caroline A. Yetman: _________________________ Date: _______________________
vii
Table of contents
Abstract .............................................................................................................................. i
Acknowledgements .......................................................................................................... iv
Disclaimer ......................................................................................................................... v
Declaration ....................................................................................................................... vi
Table of contents ............................................................................................................. vii
Chapter 1 Introduction
Global decline of amphibians .................................................................... 1
Conservation status of South African anurans .......................................... 2
Biology of Pyxicephalus adspersus .......................................................... 3
Relevance and objectives of this thesis ..................................................... 5
Chapter 2 Spawning and non-breeding activity of adult giant bullfrogs
(Pyxicephalus adspersus)
Abstract ..................................................................................................... 9
Introduction ............................................................................................. 11
Materials and Methods ............................................................................ 13
Results ..................................................................................................... 19
Discussion ............................................................................................... 23
Tables ...................................................................................................... 30
Figures ..................................................................................................... 34
viii
Chapter 3 Conservation implications of spatial habitat use by adult giant
bullfrogs (Pyxicephalus adspersus)
Abstract ................................................................................................... 37
Introduction ............................................................................................. 39
Materials and Methods ............................................................................ 41
Results ..................................................................................................... 45
Discussion ............................................................................................... 49
Tables ...................................................................................................... 55
Figures ..................................................................................................... 56
Chapter 4 Conservation implications of the age/size distribution of giant
bullfrogs (Pyxicephalus adspersus) at three peri-urban breeding sites
Abstract ................................................................................................... 62
Introduction ............................................................................................. 64
Materials and Methods ............................................................................ 66
Results ..................................................................................................... 70
Discussion ............................................................................................... 73
Tables ...................................................................................................... 81
Figures ..................................................................................................... 82
Chapter 5 Conservation implications of giant bullfrog (Pyxicephalus adspersus)
population genetic structure in Gauteng Province, South Africa
Abstract ................................................................................................... 86
ix
Introduction ............................................................................................. 88
Materials and Methods ............................................................................ 90
Results ..................................................................................................... 95
Discussion ............................................................................................... 98
Tables .................................................................................................... 104
Figures ................................................................................................... 105
Chapter 6 Conservation implications of habitat preference and geographic range
of the giant bullfrog (Pyxicephalus adspersus) at two spatial scales
Abstract ................................................................................................. 111
Introduction ........................................................................................... 113
Materials and Methods .......................................................................... 116
Results ................................................................................................... 122
Discussion ............................................................................................. 125
Tables .................................................................................................... 132
Figures ................................................................................................... 136
Chapter 7 Conclusion
Summary of results ................................................................................ 140
Concluding remarks .............................................................................. 145
Literature Cited .......................................................................................................... 148
1
Chapter 1
Introduction
Global decline of amphibians
Concern about widespread amphibian population declines emerged in 1989 at the First
World Congress of Herpetology (Stuart et al. 2004). Research has shown, however, that
declines had already begun in Australia, and in north and central America during the
1970s (Czechura & Ingram 1990; Drost & Fellers 1996; Burrowes et al. 2004). Although
amphibian populations are known to fluctuate dramatically in response to variable
weather conditions, individual survival and reproductive success (Pechmann & Wilbur
1994; Marsh 2001), probalistic models indicate that the declines have been significantly
more severe and extensive than expected under normal fluctuationg demographic
conditions (Pounds et al. 1997).
To properly assess the situation, the Global Amphibian Assessment was driven by the
IUCN (World Conservation Union) to obtain data on the distribution, abundance,
population trends and threats of the 5 743 amphibian species that were recognized at that
time (Stuart et al. 2004). The Assessment revealed that almost one third (32.5%) of the
evaluated species was globally threatened and that amphibians were, therefore, the most
threatened of the five Vertebrate classes (Stuart et al. 2004). Moreover, since 22.5% of
2
the species were considered Data Deficient, the actual proportion of threatened
amphibians was almost certainly larger. The most common and widespread threat was
habitat loss (Stuart et al. 2008), although many amphibians suffered severe enigmatic
declines possibly related to the virulent fungal disease chytridiomycosis, climate change,
increased UV-B radiation or synergistic effects of different threats (Pounds et al. 2006).
Conservation status of South African anurans
South Africa boasts an amazing variety of anurans of which 43% are endemic (Measey
2011). In 2004, 117 anuran species were recognized in the country of which 20 (17%)
and 8 (7%) were considered globally threatened or Data Deficient, respectively (Minter et
al. 2004). In 2010, the number of South African anurans recognized increased to 118, of
which 17 (14%) are considered globally threatened and none are Data Deficient (Measey
2011). Habitat loss is currently the greatest threat to South African anurans, of which
50% are affected by agriculture and aquaculture, 37% by invasive organisms (mainly
alien plants), 33% by residential and commercial development, 26% by modification of
natural systems (e.g. fire) and 15% by pollution (Measey 2011).
Although considered Least Concern globally (Measey 2011; IUCN 2011), since 2001 the
giant bullfrog, Pyxicephalus adspersus, has been regarded as Near-Threatened in South
Africa (Minter et al. 2004) due to estimated population declines of between 50 and 80%
(Harrison et al. 2001). These declines were observed mainly in South Africa’s Gauteng
Province, and habitat loss has been considered the greatest threat to this species, although
3
road traffic and harvesting for human consumption have also significantly affected
certain populations (Harrison et al. 2001; Cook 2002; Minter et al. 2004). In Swaziland,
where P. adspersus appeared to occur marginally, populations have alledgedly gone
extinct (Boycott 2001).
The aim of this thesis was to obtain essential outstanding information about the ecology
of P. adspersus to contribute towards improved conservation management of this species.
A brief description of what was known about P. adspersus prior to this research follows.
Biology of Pyxicephalus adspersus
Pyxicephalus adspersus (Tschudi, 1838) is the largest amphibian in southern Africa
(Channing 2001) and one of the largest anurans in the world (Mattison 1992). Adult
Pyxicephalus adspersus are also known for their aggressive disposition and tendancy to
bite using the canine-like projections on their lower jaw. An additional unique feature of
P. adspersus is the pronounced, reversed sexual size dimorphism of adults (males: ~ 400-
1000 g; females: ~ 90-300 g; Cook 1996), which otherwise have a uniform olive-green
dorsum with elevated longitudinal skin ridges, and a white to yellow ventral surface
(Carruthers 2001).
Pyxicephalus adspersus inhabits mainly grassland and open savanna in parts of east and
across most of southern Africa, including all nine provinces of South Africa (Channing
2001; Clauss & Clauss 2002; Minter et al. 2004). Individuals spend most of the year in a
4
state of torpor buried underground, where they produce a cocoon comprising multiple
layers of sloughed skin (Parry & Cavill 1978). Cocoon formation may reduce evaporative
water loss by 50% (Loveridge & Crayé 1979), and reduction in metabolic rate can
facilitate a 75% decline in the rate of oxygen consumption of P. adspersus in torpor
(Loveridge & Withers 1981). This enables individual P. adspersus to survive
underground for typically 6-8 months of the year where conditions are dry during winter.
During summer, adult P. adspersus emerge after heavy rainfall to breed explosively in
shallow, seasonal wetlands with emergent grassy vegetation (Balinsky & Balinsky 1954;
Grobler 1972; Cook 1996). Males produce a soft, low-pitched “whoop” (Channing et al.
1994) and exhibit different size-related breeding behaviours (Cook 1996). The largest
males aggressively defend territories where conditions appear to be most favourable for
oviposition and tadpole development. Medium-size males congregate in groups around
territorial males, jostling continuously for territorial space and females. The smallest
(satellite) males wait in anticipation of opportunistic matings on the periphery of the
spawning activity.
Adult females show preference for mating with territorial males (Cook 1996), moving
towards them with only their eyes above the water (Channing 2001). During amplexus,
females stretch their legs to raise their vent above the water, thereby facilitating
fertilization of their ova from the sperm of their mate (Balinsky & Balinsky 1954).
Couples spend approximately 3-29 minutes in amplexus (Channing et al. 1994), during
which time a female may deposit 1 000-6 000 eggs (Cook 1996). The fertilized eggs are
5
deposited with those of other females in a dominant male’s territory, and within 36 hours
small, black tadpoles hatch from the eggs (Du Preez & Cook 2004).
Some territorial males remain with offspring to defend them against predators, and may
excavate a channel with their legs to regulate the depth and temperature of water for
offspring (Kok et al. 1989; Cook et al. 2001). P. adspersus tadpoles congregate in
massive schools at the water’s edge, and become increasingly grey in colour as they
reach up to 71 mm in length (Channing 2001). Tadpole metamorphosis is usually
completed in 18-33 days depending on water temperature and food availability (Van Wyk
et al. 1992; Channing 2001). Juvenile P. adspersus have a white ventral surface and a
mottled green and black dorsum with a conspicuous bright-green vertebral stripe
(Carruthers 2001).
Relevance and objectives of this thesis
This thesis was designed to investigate several poorly understood aspects of the ecology
of P. adspersus, which would contribute towards improved conservation management of
this species in South Africa, and in particular, Gauteng Province. The main motivation
and objectives of the different research chapters comprising this thesis follow.
6
Chapter 2
The activity of P. adspersus is sporadic and difficult to predict, and consequently,
populations have been poorly monitored. To better understand the activity of this species
for improved monitoring and conservation management of populations, the objectives of
this study were to:
quantify the spawning and non-breeding activity of adult P. adspersus during > 1
season.
determine whether adult male and female P. adspersus exhibit different patterns
of diurnal or nocturnal activity.
examine the activity of adult P. adspersus in relation to meteorological variables.
infer consequences of the results on the conservation management of P.
adspersus.
Chapter 3
Nothing is known about terrestrial habitat use by P. adspersus, which has important
implications for in situ conservation management of populations. To better understand
the spatial habitat requirements of this species, the objectives of this study were to:
investigate the movements of adult P. adspersus within and between seasons.
determine whether adult male and female P. adspersus exhibit different spatial
patterns of habitat use.
establish whether there is any relationship between spatial habitat use and the
body size or body condition of same-sex adult P. adspersus..
7
infer consequences of the results on the conservation management of P.
adspersus.
Chapter 4
Nothing is known about the age of wild P. adspersus although age at maturity, and
longevity, have important conservation implications for threatened species. To better
understand this fundamental life history aspect of P. adspersus, the objectives of this
study were to:
estimate the age range of wild, breeding P. adspersus males and females using
skeletochronology.
examine the relationship between body size and age in adult P. adspersus.
compare the age, body size and body condition of adult P. adspersus at different
breeding sites.
infer consequences of the results on the conservation management of P.
adspersus.
Chapter 5
Habitat loss has fragmented and isolated many P. adspersus populations in Gauteng
Province and elsewhere in South Africa. Yet nothing is known about this species’
population genetic structure and gene flow. To gain a better understanding of the past
connectedness, present viability and appropriate future conservation management of P.
adspersus, the objectives of this study were to:
8
quantify P. adspersus population genetic structure and gene flow in Gauteng
Province.
quantify P. adspersus population genetic structure and gene flow in other parts of
South Africa, where possible.
identify genetically unique and, therefore, important P. adspersus populations in
the study area.
infer consequences of the results on the conservation management of P.
adspersus.
Chapter 6
The geographic range of P. adspersus has been difficult to accurately assess due to the
sporadic activity and superficial morphological similarity of species in this genus. In
South Africa’s Gauteng Province, where considerable habitat transformation has
occurred, there is an urgent need to identify remaining patches of natural habitat for
proactive and effective conservation of P. adspersus. The objectives of this study were,
therefore, to:
predict the potential geographic range of P. adspersus in southern Africa using a
species distribution model.
estimate the amount of suitable habitat for P. adspersus remaining in Gauteng
Province, South Africa using recent high resolution land cover data.
infer consequences of the results on the conservation management of P.
adspersus.
9
Chapter 2
Spawning and non-breeding activity of adult giant bullfrogs
(Pyxicephalus adspersus) *
Caroline A. Yetman
1
& J. Willem H. Ferguson
2
2
Centre For Environmental Studies,
1,2
Department of Zoology and Entomology,
University of Pretoria, Pretoria, 0002, South Africa
Abstract.Populations of the Near-Threatened giant bullfrog (Pyxicephalus adspersus)
have been poorly monitored due to the unpredictable appearance of this species
aboveground. To better understand the activity of P. adspersus we quantified spawning
by a population during five summers, and the activity of twenty adult frogs radio-tracked
at the same site ca. twice weekly during the first three summers. In addition we examined
animal activity, and population spawning in relation to meteorological variables, day of
season, and moon light. During the six-month summer period, males and females,
respectively, spent 10 ± 7 and 3 ± 2 days at water, and 22 ± 5 and 13 ± 5 nights active.
Greater proportions of radio-tracked animals moved overland, and/or foraged at night,
around full moon, after heavy rainfall, when cooler, and less windy conditions prevailed.
More animals were found at water, or on land during the day, and population spawning
was more likely, earlier in summer, following heavier rainfall. Spawning occurred most
10
frequently, in descending order, during December, January, and November, and was
triggered by 40 ± 16 mm rain in 24 h. Spawning events lasted 2 ± 1 days, but were
prolonged around full moon. Numbers of males at spawning events varied between 30
and 500 males, and were positively correlated with the previous day’s rainfall. Annually
6 ± 1 spawning events occurred, and numbers of annual spawning events were positively
correlated with total summer rainfall. Results of this study show that single counts of
spawning adults will often result in gross underestimates of population size, and
therefore, long-term adult counts are necessary to detect real population trends. Due to
the unpredictable activity of adults, however, it may be more practical to monitor, within
certain areas, the number of aquatic sites where breeding occurs, for improved
conservation management of P. adspersus.
* Published as:
Yetman, C.A. & J.W.H. Ferguson (2011a). Spawning and non-breeding activity of adult
giant bullfrogs (Pyxicephalus adspersus). African Journal of Herpetology 60: 13-
29.
11
Introduction
Although photoperiod and the lunar cycle provide predictable cues that amphibians may
use to synchronize their behaviour (Fitzgerald & Bider 1974; Both et al. 2008; Canavero
& Arim 2009; Grant et al. 2009), the activity of many species is ultimately dictated by
unpredictable climatic variables because of their permeable skin and ectothermic bodies
(Duellman & Trueb 1994). Among these, precipitation tends to exert the greatest
influence (Bulger et al. 2003; Lemckert & Brassil 2003), but temperature (Howard 1978;
Van Gelder et al. 1986), humidity (Bellis 1965; Bartelt et al. 2004), wind (Penman et al.
2006; Philips et al. 2007), light intensity/cloud cover (Jameson 1955; Blankenhorn 1972),
and barometric pressure (Blankenhorn 1972) can also significantly affect amphibian
activity. Amphibians in dry and/or cold environments are most limited by suitable warm,
wet conditions, making it difficult to predict their behaviour or even their presence (e.g.,
Bulger et al. 2003; Bartelt et al. 2004; Goldberg & Schwalbe 2004). This can represent a
serious challenge for effective conservation management of threatened amphibian taxa
(e.g., Penman et al. 2006).
Giant bullfrogs (Pyxicephalus adspersus) are large frogs (males: ~ 400-1 000 g; females:
~ 90-300 g; Cook 1996) inhabiting arid to subtropical grassland and savannah in southern
Africa (Channing 2001). Individuals remain underground in a state of torpor except in
summer when triggered by heavy rain to spawn explosively in shallow, ephemeral water
bodies (Balinsky & Balinsky 1954; Grobler 1972). Little is known about the non-
breeding biology of this species apart from its diet, and burrowing physiology (Loveridge
& Withers 1981; Van Aardt 1992; Du Preez & Cook 2004; Secor 2005). Currently, P.
12
adspersus is listed as Near-Threatened in South Africa (Minter et al. 2004; IUCN 2008),
where habitat loss and other factors have led to local population declines of 50-80%,
particularly in Gauteng Province (Harrison et al. 2001). Unfortunately, due to the highly
unpredictable activity of P. adspersus, the species has been subject to limited field-based
research and monitoring (Jacobsen 1989; Kok et al. 1989; Van Wyk et al. 1992; Cook
1996). This has hindered accurate assessment of the size, and breeding success, of
important historical populations (e.g., at Glen Austin and Bullfrog pans near
Johannesburg). The resulting lack of demographic information for the species has
impeded assessment of its conservation status, and provides a weak baseline for
evaluation of efforts to protect threatened populations.
In the present study we aimed to examine the spawning and non-breeding activity of
adult P. adspersus in relation to meteorological variables, to contribute towards improved
detection, monitoring, and conservation management of this species. Specific objectives
of the study were to:
Quantify the annual spawning, and non-breeding activity of adult P. adspersus.
Determine whether males and females exhibit different patterns of diurnal or
nocturnal activity.
Examine the diurnal, nocturnal, and spawning activity of adult P. adspersus in
relation to meteorological variables, day of season, and moon light.
Infer consequences of the results on the conservation management of P.
adspersus.
13
Materials and Methods
Study site
The study site was located in Diepsloot, Gauteng Province, South Africa (2528CC:
25º56'23.59"S, 28º01'21.88"E; 1 427-72 m a.s.l.), and comprised ~ 100 ha of degraded
Egoli Granite Grassland (Mucina et al. 2005). Local bullfrogs bred in three small (~ 0.2-
0.9 ha), shallow (< 2 m deep), human-created, seasonal dams (named Dams 1, 2 and 3),
situated in the southern section of the site. The site is representative of many areas where
P. adspersus faces encroaching urbanization on remaining fragments of undeveloped
habitat. We collected data on P. adspersus at the site between 1 October and 31 March in
the seasons 2003/2004, 2004/2005, 2005/2006, 2006/2007, and 2007/2008 - hereafter
referred to as Summers 1-5.
Radio-tracking
In Summers 1-3, we haphazardly selected 10 males and 20 females for radio-tracking
from adult P. adspersus caught for a separate mark-recapture study. We captured frogs
by hand or with a hand-held net during spawning events in Dams 1 and 2 (P. adspersus
did not favour Dam 3). Animals were each implanted off-site with a radio-transmitter by
a veterinary surgeon, monitored overnight, and released the next day at their capture sites.
The transmitters and surgical protocol are described in Yetman & Ferguson (2011b,
[Chapter 3]).
14
Tracking was conducted between 06h00 and 18h00 on average twice per week, with
additional surveys undertaken in response to rainfall events. A four-element Yagi antenna
and AVM LA12 receiver (AVM Instruments, Champagne, Illinois) were used to detect
radio-tagged frogs. Between three and 16 animals were tracked on a day. When a frog
was located, we recorded its geographic coordinates (with a global positioning system,
accurate to 4 m), and the date, time, and behaviour observations. An animal’s behaviour
was classified as: buried (inside an open or closed burrow); present at a spawning event
(regardless of whether an animal mated); guarding offspring (resting < 1 m from or
aggressively defending eggs or tadpoles); resting at water (when an animal was found in
or < 1 m from a dam or other inundated depression, and spawning or parental care of
offspring was not observed); or resting on land (resting aboveground > 1 m from water,
usually under vegetation). Data on the diurnal activity of radio-tracked animals were
obtained for 129 days during Summers 1, 2, and 3.
Evidence of nocturnal frog activity
No nocturnal radio-tracking was performed because the area was considered unsafe at
night. However, in Summer 3 we deduced the nocturnal emergence of animals from their
burrows when the radio-tracked location of individuals changed overnight, or when a
small stone placed on the plug sealing an animal’s burrow had disappeared or been
displaced off the plug between consecutive days of tracking. Pearson (1955) used chalk
dust for the same purpose, but in our study stones were used as less conspicuous markers
to prevent animals from being harvested for human consumption (Du Preez & Cook
15
2004). Data on the nocturnal activity of radio-tracked animals were obtained for 31 nights
in Summer 3.
Monitoring of spawning
All aspects of breeding by P. adspersus take place at the edge of ephemeral wetlands
where the water is most shallow and warm, and thus favourable for tadpole development
(Cook 1996). We therefore inspected the edge of the seasonal dams on foot, usually every
day that animals with radio-transmitters were tracked in Summers 1-3, and after rainfall
events of 20 mm or more rain at the study site in Summers 4-5. Evidence of spawning by
P. adspersus included observation of calling, fighting and/or mating adults, eggs,
tadpoles, and/or newly-metamorphosed froglets – which were clearly visible at the
water’s edge and are highly diagnostic (Du Preez & Carruthers 2009).
When a spawning event was observed, males at Dams 1-3 were counted in groups of
approximately 10 individuals, if possible at different times of the day for each day that
adults were seen calling, fighting and/or mating. Thus the largest count of males, and the
duration of each observed spawning event was determined. Females were not counted
because their smaller size and inconspicuous behaviour (Channing et al. 1994; Cook
1996) made them difficult to detect.
When eggs, tadpoles or newly-metamorphosed froglets were found without spawning
having been observed, we estimated the date of spawning from the development of
offspring (Gosner 1960; Van Wyk et al. 1992; Haas 1999) and measurements of daily
16
rainfall from Diepsloot Nature Reserve. Hence, for Summers 2-5, we estimated the
annual number of spawning events as the sum of all spawning events that were observed
or revealed by eggs, tadpoles or froglets. The total number of spawning events in
Summer 1 could not be determined since the study site was found in late mid-summer
(January 2004).
Meteorological data
For periods when P. adspersus spawned at the study site, we obtained measurements of
daily rainfall from a local resident who lived < 100 m from the dams. These data were
supplemented with rainfall measurements taken at 07h00 daily in gauges at five different
localities in Diepsloot Nature Reserve, which lies ~ 6 km west of the study site
(25°55'36.35"S, 27°57'56.09"E). We treated 1 October as the first day of season (DS),
and used a lunar calendar (South African Astronomical Observatory) to determine the
number of days since full moon on specific dates. Days since full moon were assigned to
one of eight lunar phases, ranked from 0-4 depending on the amount of moon light (ML;
Table 1, adapted from Grant et al. 2009).
We obtained daily measurements of maximum and minimum atmospheric temperature
(MaxT and MinT in °C), relative humidity (RH in %), wind speed (WS in m s
-1
), and
atmospheric pressure (P in hPa) from a weather station in Irene, situated ~ 19 km east of
the study site (25°54'36.00"S, 28°12'36.00"E). Missing temperature data in November
and December 2005, and October 2006 were substituted with daily MaxT and MinT
measurements from a weather station in Kempton Park (situated ~ 31 km south-east of
17
the study site; 26°08'60.00"S, 28°13'48.00"E). These measurements were corrected by
0.4-3.0 °C depending on the mean monthly difference in each variable between the two
weather stations during Summers 1-3.
Statistical analyses
Statistical tests were performed in Statistica 7.0 (© StatSoft, Inc. 1984-2004 Tulsa).
Animals found present at a spawning event, guarding offspring, resting at water, or
resting on land during radio-tracking, were considered active by day. Nocturnal activity
was deduced from overnight change in an animal’s radio-tracked location, or overnight
displacement of the stone placed on an animal’s burrow. Mann-Whitney U tests were
performed to assess whether the sexes differed in the mean number of dates during
summer that individuals were buried or active by day, and/or buried or active at night.
Behavioural data were used from six males and seven females that were successfully
radio-tracked in the day through Summers 2 and/or 3; and three males and seven females
for which data on their nocturnal activity were obtained through Summer 3. Data
presented are mean ± standard deviation.
We used generalized linear models (GLZs) to investigate animal activity, and population
spawning in relation to meteorological variables. Diurnal or nocturnal animal activity was
measured as the daily diurnal or nocturnal “active” (arc sine transformed) proportion of
animals that were successfully radio-tracked on 129 days in Summers 1-3, or 31 days in
Summer 3, respectively. Population spawning was measured as a binomial variable on
743 days in Summers 1-5. Diurnal animal activity and population spawning were
18
examined in relation to study year, the amount of previous day’s rainfall (PDR), MaxT,
MinT, mean day time RH, WS, and P (measured at 08h00 and 14h00), and DS. Nocturnal
animal activity was examined in relation to the amount of daily rainfall (DR), MaxT,
MinT, night time RH, WS, and P (measured at 20h00), DS, and ML.
We tested the predictor variables for multicollinearity using multiple regressions, and
excluded predictors with a Variance Inflation Factor (VIF) > 2, and/or a Tolerance (T)
value < 0.5. We therefore excluded RH (VIF = 3.1; T = 0.3) from the models of diurnal
animal activity, DS (VIF = 2.4; T = 0.4), and MaxT (VIF = 2.3; T = 0.4) from the models
of nocturnal animal activity; and RH (VIF = 3.6; T = 0.3) from the models of population
spawning. VIF and T values for the remaining predictor variables in the models are
shown in Table 2. We assessed the utility of models of animal activity or population
spawning based on a normal or binomial distribution, and an identity or logit link
function, respectively. We corrected values of Akaike’s Information Criterion (AIC) that
were generated in Statistica, for over-dispersion of the data in all models (QAIC), and for
small sample size in the models of diurnal or nocturnal animal activity (QAIC
c
; Symonds
& Moussalli 2010). For model averaging we used models with a corrected AIC value that
differed by < 2 from that of the “best” model of population spawning, or diurnal or
nocturnal animal activity (Symonds & Moussalli 2010). The importance of a specific
predictor was estimated as the sum of the Akaike weights (Σw
i
) of all models of diurnal
or nocturnal animal activity, or population spawning, which included that specific
predictor (Symonds & Moussalli 2010).
19
We used a logistic regression with binomial errors to more accurately describe the
relationship between PDR and population spawning, measured on 743 days in Summers
1-5. The duration of spawning events, or numbers of spawning males could not be
examined in relation to meteorological variables using GLZs or step-wise multiple
regressions, due to a small sample size (n = 10 observed spawning events),
multicollinearity, and skewed variable distributions. Therefore, Spearman Rank
correlations were used instead. Due to a small sample size (n = 4 summers), a Kendall
Tau correlation was performed to determine whether annual numbers of spawning events
were related to total summer rainfall through Summers 2-5. This method is more
sensitive to some types of dependence, and deals better with ties between variables than
the Spearman method (Kendall 1938).
Results
Tracking
Of the 30 transmitter-implanted animals, 10 males and 12 females were tracked post-
release. Eight females were lost from the start, and one female died and another was lost
for uncertain reasons shortly after release. The 10 males and 10 remaining females were
each tracked to at least one burrow. Of these 20 animals, three females and two males
were tracked for several days or weeks. Seven females and four males were tracked for a
complete summer, and four males were tracked during two or three successive summers.
We removed transmitters from five animals, but most animals were lost when their
transmitters expired.
20
Male versus female activity
Radio-tracked males were found on significantly more days in summer at water (10 ± 7
days, range: 4-18 days, n = 6 males tracked through Summer 2 and/or 3), than females (3
± 2 days, range: 2-6 days, n = 7 females tracked through Summer 3, Mann-Whitney U =
5.0, P = 0.02). More specifically, males were found in summer on 4 ± 2 days at a
spawning event (range: 3-6 days, n = 6 males), 1 ± 1 days guarding offspring (range: 0-2
days, n = 4 males), and 5 ± 5 days resting at water (range: 0-11 days, n = 5 males).
Females were found in summer on significantly fewer days at a spawning event (2 ± 1
days, range: 2-3 days, n = 7 females, U = 6.0, P = 0.03), and on 1 ± 1 days resting at
water (range: 0-3 days, n = 2 females, U = 8.5, P = 0.07). Only males were found in
summer, on 2 ± 1 days resting on land (range: 0-4 days, n = 5). In Summer 3, males were
active on more nights (22 ± 5 nights, range: 16-26 nights, n = 3 males) than females (13 ±
5 nights, range: 9-20 nights, n = 7 females, U = 2.0, P = 0.05). Overall, males were active
in the day and/or night on a greater cumulative number of dates in Summer 3 (27 ± 10
dates, range: 17-37 dates, n = 3 males), than females (14 ± 5 dates, range: 9-23 dates, n =
7 females, U = 1.5, P = 0.04).
Activity in relation to meteorological variables
The relative importance (Σw
i
) of each predictor of diurnal or nocturnal animal activity is
shown in Table 3. Model averaging involved 62 models of diurnal animal activity, and 14
models of nocturnal animal activity. The coefficient estimate (ß) and standard error (SE)
obtained from model averaging, for the relationship between each predictor variable, and
21
daily diurnal or nocturnal active” proportions of successfully radio-tracked animals, is
shown in Table 3.
PDR and DS were the first and second most important predictors in all 127 models of
diurnal animal activity, and were included in 100% and 52% of the 62 “best candidate”
models used for model averaging (Table 3). The coefficient estimates (Table 3) indicate
that more animals were diurnally active (at water or on land) earlier in summer, and
following heavier rainfall. MinT, night time WS, DR, and ML were, in descending order,
the most important predictors in all 31 models of nocturnal animal activity, and were,
respectively, included in 100%, 86%, 71%, and 57% of the 14 “best candidate” models
used for model averaging (Table 3). The coefficient estimates (Table 3) indicate that
more animals were nocturnally active (i.e., moved overland, and/or foraged) around full
moon, after heavy rainfall, when cooler, and less windy conditions prevailed.
Spawning in relation to meteorological variables
Daily rainfall in Diepsloot, and P. adspersus spawning at the study site in Summers 1-5,
are depicted in Figures 1a-e, respectively. The logistic regression between PDR and
population spawning, measured on 743 days in Summers 1-5, is shown in Fig. 1f. The
coefficient estimate, SE, and P value was, respectively, 0.121, 0.014, and < 0.001 for
PDR, and -4.254, 0.301, and < 0.001 for the regression intercept. The mean amount of
rain recorded 24, 48 or 72 h prior to 25 spawning events that were observed, or revealed
by offspring, was 40 ± 16 mm rain (min. = 20 mm), 45 ± 20 mm (min. = 22 mm), and 51
± 25 mm (min. = 22 mm), respectively. Spawning occurred most frequently, in
22
decreasing order, during December, January, and November (Fig. 2). The mean
maximum count of males at 10 observed spawning events was 162 ± 148 males (range:
30-500 males), and the mean duration of 15 spawning events, of which 12 were observed
and three were revealed by observation of one- or two-day old eggs, was 2 ± 1 days
(range: 1-4 days).
The relative importance (Σw
i
), coefficient estimate (ß), and standard error (SE) of each
predictor of population spawning, is shown in Table 3. Model averaging was based on 64
models. PDR and DS were, respectively, the first and second most important predictors in
all 743 models of population spawning, and were included in 100% and 50% of the 64
“best candidate” models used for model averaging (Table 3). The coefficient estimates
(Table 3) indicate that population spawning was more likely earlier in summer, and
following heavier rain.
The Spearman Rank correlation between the duration of spawning events, or numbers of
spawning males, and each of 11 different predictor variables, is shown in Table 4. The
duration of spawning events was positively correlated with ML (P = 0.05), and the
number of males at spawning events was positively correlated with the amount of
previous day’s rainfall (P = 0.04). Annually, there were 6 ± 1 spawning events (range: 4-
7 spawning events, n = 4 summers), and estimated annual numbers of spawning events
during Summers 2-5 were positively correlated with total summer rainfall (Kendall Τ =
1.0, P = 0.04, Fig. 3).
23
Discussion
Male versus female activity
Pyxicephalus adspersus is a difficult species to observe. Study animals cumulatively
spent a month or less active during the six-month summer period. Other burrowing
anurans inhabiting mesic or arid habitats exhibit similarly brief activity. For example,
eastern spadefoots (Scaphiopus holbrooki) in Gainesville, Florida (Pearson 1955), and
shoemaker frogs (Neobatrachus sutor) and orange-crowned toadlets (Pseudophryne
occidentalis) in Western Australia (Thompson et al. 2003) spent, respectively, ca. 29
nights, and up to four days on 9 and 17 occasions, active in summer. During their brief
activity, male P. adspersus that were radio-tracked for more than one season, evidently
caught sufficient prey to sustain their large bulk through winter. This was possible,
despite the degraded habitat, because P. adspersus preyed on termites, which were
abundant on the site (CAY pers. obs.).
Certain activity differences between male and female P. adspersus might be due to their
pronounced dimorphism in body size (Hayes & Licht 1992). Females avoid being
terrestrially active during the day possibly because their greater surface-to-volume ratio
increases their vulnerability to desiccation (Peters 1983). Males spend a greater number
of nights active in summer because they visit breeding habitat more often, and because
they probably need to spend more time foraging than females to satisfy their greater
absolute metabolic requirements (Branch 1976; Paukstis & Reinbold 1984). Similarly,
Ovaska (1992) claimed that females of the burrowing frog Eleutherodactylus johnstonei,
24
moved longer distances than males overland, due to their larger size and greater energy
requirements. In contrast, similar-sized (~ 50 g) male and female Hoplobatrachus
occipitalis in a West African savanna, exhibited no difference in their timing of
migrations to breeding sites, activity and home range areas, and growth during summer
(Spieler & Linsenmair 1998).
Due to the more costly and limited production of ova relative to sperm (Van Beurden
1979; Reading 1988; Blem 1992), females of many anurans only spawn 1-3 times in a
season, whereas males visit most spawning events to increase their reproductive success
(e.g., Perrill & Daniel 1983; Loman & Madsen 1986; Sinsch 1988a; Barandun et al.
1997). P. adspersus appears to fit the same trend. If adult P. adspersus are present for
only one day of a spawning event, females and males would, respectively, attend two
(33%) and four (67%) of six mean spawning events per annum. Thus, assuming P.
adspersus populations have an adult sex ratio of 1:1, spawning events will usually appear
male-biased in this species, as observed by Cook (1996).
Human disturbance at the Diepsloot dams accounts for why at least one radio-tracked
male abandoned offspring (CAY pers. obs.), and likely explains why males spent one day
on average in summer guarding offspring. In contrast, 109 males at Glen Austin Pan
remained with their offspring for 21 ± 9.8 days (Cook et al. 2001).
25
Activity in relation to meteorological variables
Radio-tracked males and females, respectively, spent 40 and 67% of their diurnal summer
activity attending spawning events. Therefore, as with population spawning, a greater
number of animals were active in the day after heavy rain, earlier in summer. Adult P.
adspersus visit breeding sites after heavier rain because the risk of desiccation during
overland migration between their burrows and breeding sites (Sjögren-Gulve 1998;
Schwarzkopf & Alford 2002), and risk of evaporation and mortality of offspring at
breeding sites (Kok et al. 1989; Cook et al. 2001) is probably reduced. Adults visit
breeding sites, and spawn earlier in summer because adults may attend more spawning
events per annum, adults and newly-metamorphosed froglets may accumulate larger
energy reserves prior to winter, and/or the abundance of predators at breeding sites may
increase over summer.
On nights with less wind, and/or increased moon light, greater proportions of animals
were active because it was possibly easier to catch prey, and/or detect predators
(Fitzgerald & Bider 1974; Penman et al. 2006). The influence of the moon or wind on
anuran activity has been considered in only a few studies. For example, Penman et al.
(2006) showed that the activity of Heleioporus australiacus was related to rainfall, as
well as temperature, humidity, and wind strength. A meta-analysis by Grant et al. (2009)
revealed that various European amphibians exhibit mass migration and spawning around
full moon. Some anurans are, however, terrestrially less active during full moon possibly
because the risk of detection by predators is greater, and/or important prey is less active
26
at this time (Church 1960; Ferguson 1960; Fitzgerald & Bider 1974; Goldberg &
Schwalbe 2004).
Spawning in relation to meteorological variables
Pyxicephalus adspersus often spawns after ca. 50 mm or more rainfall (Grobler 1972;
Cook 1996; Du Preez & Cook 2004), but not always. At the Diepsloot site a minimum of
20 mm rain in 24 h could trigger spawning, which is equal to that reported for P.
adspersus in the Free State Province (Kok et al. 1989), and for other explosive-breeding,
burrowing anurans in mesic or arid habitats, e.g., S. holbrooki (Gosner & Black 1955).
Although P. adspersus spawns readily in October, population spawning appears to be
most common in November, December, and January (e.g., Balinsky 1969; Kok et al.
1989; Van Wyk et al. 1992; Cook 1996). This occurred at the study site because in
October the artificial dams often remained dry, and by February and March, females, at
least, had probably spent most or all of their ova.
Greater moon light prolonged spawning at the study site even though P. adspersus
spawns predominantly during day light (Balinsky 1969; Channing et al. 1994). This was
possibly because moon light facilitated foraging at night around the artificial dams, which
enabled males to call, fight, and/or spawn over a longer period. We noticed that spawning
was also protracted when the weather was cool, windy, and/or overcast (CAY pers. obs.).
However, due to a small sample size perhaps, no significant relationship emerged
between spawning duration, and temperature, or WS. Greater numbers of males attended
spawning events triggered by heavier rain because this possibly reduced the risk of
27
desiccation during overland migration (Sjögren-Gulve 1998; Schwarzkopf & Alford
2002), and/or improved the probability of successful reproduction, particularly for young,
weak, less experienced, and/or distantly buried males.
Cook et al. (2001) reported that P. adspersus breeds, on average, once in four years at a
particular site. In contrast, P. adspersus at the Diepsloot study site spawned, on average,
six times each summer. The reproductive success of P. adspersus may differ between
large, natural pans, and small, artificial dams, but many spawning events probably go
unnoticed due to the unpredictable, sporadic, and brief activity of this species. Therefore,
P. adspersus could experience greater reproductive success than might be assumed.
However, cohorts of larval P. adspersus often suffer high mortality caused by
desiccation, and predation (Cook et al. 2001), and at high larval densities there is
increased competition for food (Cook 1996), cannibalism (Grobler 1972), and possibly
spread of disease.
Conservation implications
Significant differences in the temporal and spatial use of aquatic and terrestrial habitat by
adult male and female P. adspersus have important conservation implications for the
species. Males spend significantly more time at breeding sites than females, and
consequently, are more vulnerable to harvesting for human consumption, which is
common in Limpopo Province (Du Preez & Cook 2004). Increased mortality of males at
breeding sites could adversely affect tadpole survival and juvenile recruitment due to
paternal care of offspring in P. adspersus (Cook et al. 2001). In contrast, females spend
28
virtually their entire lives in burrows situated significantly farther from breeding sites
than those of males (Yetman & Ferguson 2011b [Chapter 3]). Females are consequently
more vulnerable to encroaching land transformation (Du Preez & Cook 2004).
Pyxicephalus adspersus therefore requires effective protection of both aquatic and
terrestrial habitat.
For estimation of P. adspersus population sizes, single counts of spawning adults are
grossly inadequate because they will frequently result in severe underestimates. This was
evident at our study site, where only a small proportion (10-50%) of the highest male
count (i.e., 500 males in January of Summer 3) was present at most spawning events.
Adult male P. adspersus remain relatively close to their breeding site within and among
years (Yetman & Ferguson 2011b, [Chapter 3]); therefore, frog movements are unlikely
to account for the observed variation in numbers of males at spawning events. The great
variation in numbers of spawning males at the Diepsloot dams obscured any trend in
population size. Long term adult counts (exceeding 10 or 20 years; see Meyer et al. 2010)
will, therefore, be necessary to detect real trends in the size of P. adspersus breeding
populations. However, due to the unpredictable and brief activity of adults, it may be
more practical to monitor, within certain areas, the number of aquatic sites where P.
adspersus breeding occurs (Joseph et al. 2006). In this way large-scale trends can be
detected for improved conservation assessment and management of P. adspersus at
regional, provincial and national scales (Lindenmayer & Likens 2010).
29
Acknowledgements
We sincerely thank Jimmy, Alice and Mark Yetman, and Johan Lötter for their patient
and generous assistance with field work; Wendy Duncan for measuring local rainfall,
reporting frog activity at the dams, and kindly assisting with field work; Roger Wood and
Diepsloot Nature Reserve for providing additional rainfall data; the South African
Weather Service for providing the bulk of the meteorological data; the Gauteng
Department of Agriculture and Rural Development for issuing permits 1 240 and 1 296 to
CAY; and Mike Perry, who introduced us to the site. The study was funded through the
Endangered Wildlife Trust by Rand Merchant Bank, the Pretoria East branch of the South
African Hunter’s and Game Conservation Association, Arrow Bulk Marketing, Cellar
Rats Wine Club, Bill Flynn and Diaz Films. We thank Michael Bates, Frank Lemckert,
and three anonymous reviewers for valuable comments on previous drafts of this
manuscript.
30
Table 1. Assignment of days since full moon (DFM) to one of eight lunar phases, ranked
from 0-4 depending on the amount of moon light (adapted from Grant et al. 2009).
Full
moon
Waning
gibbous
3
rd
quarter
Waning
crescent
New
moon
Waxing
crescent
1st
quarter
Waxing
gibbous
Lunar 1 2 3 4 5 6 7 8
phase
DFM 28, 29,
0, 1
2, 3,
4, 5
6, 7,
8
9, 10,
11, 12
13, 14,
15, 16
17, 18,
19, 20
21, 22,
23
24, 25,
26, 27
Rank 4 3 2 1 0 1 2 3
31
Table 2. Variance Inflation Factor (VIF) and Tolerance (T) values for
predictor variables used in generalized linear models of the diurnal,
nocturnal, or spawning activity of Pyxicephalus adspersus at the
Diepsloot study site. The values were obtained from multiple
regressions performed after certain variables with a VIF > 2 and/or T <
0.5, were excluded (excl.) to reduce multicollinearity.
Diurnal activity Nocturnal activity
Spawning
VIF
T VIF T VIF
T
Study year 1.2 0.8 - - 1.0 1.0
Day of season 1.4 0.7 excl. excl. 1.2 0.8
Previous day’s rain 1.3 0.8 - - 1.2 0.8
Daily rain - - 1.2 0.8 - -
Max. temperature 1.3 0.8 excl. excl. 1.5 0.7
Min. temperature 1.3 0.8 1.7 0.6 1.4 0.7
Wind speed 1.1 0.9 1.1 0.9 1.2 0.9
Relative humidity excl.
excl. 1.4 0.7 excl.
excl.
Atmospheric pressure
1.1 0.9 1.5 0.7 1.2 0.8
Moon light - - 1.2 0.8 1.0 1.0
32
Table 3. Percent (%) inclusion, relative importance (Σw
i
), coefficient estimate (ß), and standard
error (SE) of predictor variables used in generalized linear models of the diurnal, nocturnal, or
spawning activity of Pyxicephalus adspersus at the Diepsloot study site. Percent inclusion
pertains to the 62, 14, or 64 “best” models of diurnal, nocturnal, or spawning activity used for
model averaging. Σw
i
represents the sum of the Akaike weights of every diurnal, nocturnal, or
spawning activity model (max. n = 127, 31, or 743 models, respectively) that involved a specific
predictor.
Diurnal animal activity
Nocturnal animal activity
Population spawning
% Σw
i
ß
SE % Σw
i
ß
SE % Σw
i
ß
SE
Intercept - - 0.672 0.003
- - 3.687 0.078 - - 54.446
0.122
Study year 48 0.46
-0.012
4x10
-
5
- - - - 50 0.49
-0.056
2x10
-
4
Day of season 52 0.58
-0.001
2x10
-
6
- - - - 50 0.60
-0.012
2x10
-
5
Previous day’s rain 100
0.87
0.009 5x10
-
6
- - - - 100
0.93
0.122 6x10
-
5
Daily rain - - - - 71 0.63
0.006 9x10
-
5
- - - -
Max. temperature 50 0.51
-0.005
1x10
-
5
- - - - 50 0.50
-0.007
7x10
-
5
Min. temperature 50 0.54
-0.016
3x10
-
5
100
0.76
-0.061
5x10
-
4
50 0.50
0.054 2x10
-
4
Wind speed 48 0.45
3x10
-
5
7x10
-
6
86 0.63
-0.061
7x10
-
4
50 0.49
-0.014
9x10
-
5
Relative humidity - - - - 43 0.47
7x10
-
4
2x10
-
5
- - - -
Atmospheric pressure
48 0.45
-4x10
-
5
3x10
-
6
36 0.47
-0.003
9x10
-
5
50 0.51
-0.068
1x10
-
4
Moon light - - - - 57 0.54
0.043 9x10
-
4
- - - -
33
Table 4. Spearman Rank correlation between the duration of,
or maximum number of males counted at, 10 observed
Pyxicephalus adspersus spawning events, and various
meteorological variables in Diepsloot. Not significant = n.s.,
* P < 0.05.
Spawning duration
Max. no. of males
r
s
P r
s
P
Study year -0.52 n.s. 0.12 n.s.
Day of season
1
-0.10 n.s. 0.19 n.s.
Previous day’s rain
1
0.03 n.s. 0.65 *
Total rain
2
0.43 n.s. -0.11 n.s.
Max. temperature
3
0.02 n.s. -0.18 n.s.
Min. temperature
3
0.04 n.s. 0.04 n.s.
Wind speed
3
0.28 n.s. 0.21 n.s.
Relative humidity
3
-0.27 n.s. -0.02 n.s.
Atmospheric pressure
3
-0.08 n.s. 0.41 n.s.
Moon light
1
0.64 * 0.06 n.s.
Max. no. of males
2
0.34 n.s. - -
Duration of spawning
- - 0.34 n.s.
1
For the first day of spawning.
2
Over the full duration of spawning.
3
Mean of daily values during spawning.
34
Figure 1. Daily rainfall in Diepsloot, Gauteng Province, South Africa, and known
Pyxicephalus adspersus spawning activity at the study site during Summers 1-5, shown in
A-E, respectively. Spawning events represented by vertical bars were observed. Those
indicated by were revealed by the presence of eggs, tadpoles, and/or froglets. Х
indicates when spawning did not occur after heavy rain. (F) Logistic regression between
the amount of previous day’s rainfall, and P. adspersus spawning at the site, measured on
743 days in Summers 1-5.
35
Figure 2. The number of recorded Pyxicephalus adspersus spawning
events at the Diepsloot study site in different months during Summers 1-
5. Thirteen spawning events were observed, and 12 were revealed by the
presence of eggs, tadpoles and/or froglets.
36
Figure 3. Kendall Tau correlation between total summer rainfall, and
annual estimated numbers of Pyxicephalus adspersus spawning events
during four summers at the Diepsloot study site.
37
Chapter 3
Conservation implications of spatial habitat use by adult giant bullfrogs
(Pyxicephalus adspersus) *
Caroline A. Yetman
1
& J. Willem H. Ferguson
2
2
Centre For Environmental Studies,
1,2
Department of Zoology and Entomology,
University of Pretoria, Pretoria, 0002, South Africa
Abstract.—In South Africa, particularly Gauteng Province, populations of the large,
explosive-breeding giant bullfrog (Pyxicephalus adspersus) are suffering increasing
habitat loss due to encroaching urbanization. To investigate the spatial habitat
requirements of this regionally threatened species, 70 adult frogs were radio- or spool-
tracked during five summers around a peri-urban breeding site. Male and female P.
adspersus moved a maximum overnight distance of 350 m when returning to their
burrows post-spawning. On average animals of either sex used one long-term burrow
(LTB) in a summer. Four males each used a single LTB or burrowing area for two or
three consecutive summers. The LTBs of females were situated almost four times further
(mean = 446.8 m) from the seasonal dams where spawning occurred, than those of males
(mean = 131.0 m). Female body condition was significantly positively correlated with
distance of their burrows from the seasonal dams (r
s
= 0.77). Limited evidence indicated
38
that adult P. adspersus probably forage mostly within 20 m of their burrows. To protect
the LTBs of all radio-tracked animals a 950-1 000 m wide buffer would be necessary
around the seasonal dams. Since adult P. adspersus appear philopatric, juvenile dispersal
is predicted to be largely responsible for gene flow among populations.
* Published as:
Yetman, C.A. & J.W.H. Ferguson (2011b). Conservation implications of spatial habitat
use by adult giant bullfrogs (Pyxicephalus adspersus). Journal of Herpetology 45:
56-62.
39
Introduction
The growing number of threatened amphibian species (Houlahan et al. 2000; Stuart et al.
2004) and greater sophistication of equipment and methods to track wildlife (e.g.,
Sampson & Delgiudice 2006; Wikelski et al. 2007) have led to increased research on
amphibian terrestrial habitat use. Several authors (Semlitsch 2000; Semlitsch & Bodie
2003; Lemckert 2004; Smith & Green 2005) have reviewed this expanding body of
information to derive general recommendations for improved in situ conservation of
amphibians and their habitat. The majority of studies upon which these recommendations
have been based were performed on species living in temperate areas within the Nearctic
and Palearctic Ecozones (e.g., Appendix 1 of Semlitsch & Bodie 2003; Tables 1-6 of
Lemckert 2004; Table 4 of Smith & Green 2005). Relatively few analogous studies have
been conducted in more tropical or arid locations and/or within the southern hemisphere
(e.g., Woolbright 1985; Miaud et al. 2000; Goldberg & Schwalbe 2004). Spieler &
Linsenmair (1998) performed the only published study on movements and habitat use of
an amphibian in Africa to date. There is therefore, a clear need for additional information
on terrestrial habitat use by amphibians living in hot environments and/or in Africa.
The giant bullfrog (Pyxicephalus adspersus) is a large, explosive-breeding anuran that
inhabits arid to sub-tropical grassland and savanna throughout most of southern Africa,
and further north into southern Angola and Kenya (Channing 2001; Minter et al. 2004).
Although listed as Least Concern globally (IUCN 2008), since 2001 P. adspersus has
been considered Near-Threatened in South Africa (Harrison et al. 2001; Minter et al.
2004). Here, the most severe and widespread threat to the species is habitat loss caused
40
by urban development and agriculture (Harrison et al. 2001; Minter et al. 2004). Nothing
is known about terrestrial habitat use by P. adspersus, except that individuals spend most
of the year buried underground in a state of torpor (Parry & Cavill 1978; Loveridge &
Withers 1981). In areas where there is increasing loss of terrestrial habitat, however,
population declines suggest that P. adspersus excavate burrows at significant distances
from breeding sites (e.g., Glen Austin and Bullfrog pans in suburban Johannesburg; Cook
2002; Slater-Jones, unpubl. data).
Many amphibians, especially anurans, travel hundreds of meters or several kilometers
between breeding sites and terrestrial habitats used for foraging and/or overwintering,
and/or among breeding sites (e.g., Table 4 of Smith & Green 2005). Without access to
suitable terrestrial habitats individuals would be unable to complete their life cycles, and
populations would eventually fail to persist (Semlitsch & Bodie 2003; Trenham &
Shaffer 2005). Unfortunately, general recommendations for in situ protection of anurans
are unlikely to adequately protect all populations, because there is great variation in
distances moved by individuals within and among species (e.g., Fig. 2 of Lemckert 2004;
Fig. 3 of Smith & Green 2005; Fig. 2 of Smith & Green 2006). Therefore, studies on the
spatial habitat requirements of threatened taxa remain crucial (Lemckert 2004).
We examined spatial habitat use by adult P. adspersus within and among seasons, and in
relation to their sex, size, or body condition. Adult males may exceed one kilogram but
typically weigh between 400 and 800 g, whereas adult females usually weigh between 90
and 300 g (Cook 1996). Therefore differences in spatial habitat use between the sexes
41
could be expected (e.g., Ovaska 1992). Adult male P. adspersus also exhibit three distinct
size-related mating tactics, including territorial, non-territorial, and satellite behavior
(Cook 1996). In some species same-sex but different-sized individuals have been found
to exhibit different patterns of spatial habitat use, apparently reducing their intra-specific
and intra-sexual competition for resources (e.g., Martof 1953; Berven & Grudzien 1990;
Ponséro & Joly 1998). Where there is less competition especially for food, animals are
likely to exhibit improved body condition, which could in turn increase their reproductive
success (e.g., through clutch size). Therefore relationships between habitat use and body
condition might have implications for in situ protection of threatened amphibian
populations (e.g., Ponséro & Joly 1998).
Materials and Methods
Study Site
We concentrated field work in a ~ 100 ha triangular tract of degraded Egoli Granite
Grassland (Mucina et al. 2005) located in Diepsloot, Gauteng Province, South Africa
(25º56'23.59"S, 28º01'21.88"E, 1 427-72 m a.s.l., Fig. 3). The site is representative of
many areas in Gauteng, and elsewhere in South Africa, where P. adspersus face
encroaching urbanization on remaining fragments of undeveloped habitat. A busy tarred
road and a dirt road bordered the site to the west and east, respectively. West of the tarred
road was low-cost housing. Agricultural small-holdings and vacant grassland surrounded
the rest of the site. Three small (~ 0.2-0.9 ha), shallow (< 2 m deep), seasonal, human-
42
created dams (named Dams 1, 2, and 3) near the south-eastern edge of the site provided
habitat used by > 500 Giant bullfrogs for breeding (Yetman, unpubl. data).
General sampling
We caught study animals during spawning events by hand or with a hand held net at or
near Dam 1 or 2. Dam 3 was much less favored by P. adspersus. An electronic balance
and steel tape-measure were used to measure each animal’s body mass (to the nearest
gram) and snout-vent length (SVL; to the nearest millimeter), respectively. During mark-
recapture research male P. adspersus weighing under 200 g have not been encountered at
spawning events (Cook 1996; Yetman, unpubl. data). We therefore classified animals in
this study lighter than 200 g and shorter than 120 mm as females.
Radio-tracking
We performed radio-tracking between 1 October and 31 March in each of three seasons
2003/2004, 2004/2005, and 2005/2006 (hereafter referred to as Summers 1-3). A total of
10 males and 20 females was released with radio-transmitters over these periods. A
veterinary surgeon anaesthetized and implanted animals with transmitters. A 15-25 mm
longitudinal incision was cut through the ventral abdominal skin and muscle, roughly 10-
30 mm left or right of the ventral abdominal vein. We used different types and sizes of
radio-transmitters (Biotrack, Ltd, United Kingdom) depending on frog body mass. TW31
(15 g or 18 g) transmitters were implanted into males; PIP22 (5 g) and PIP21 (1-4 g)
transmitters were inserted into females. A transmitter never exceeded 5% of an animal’s
body mass. Abdominal musculature was closed in a simple continuous pattern; skin was
43
closed with several horizontal mattress sutures. Study animals awoke within 15-40 min
and resumed fully alert behavior 1-3 h post-surgery. Animals were kept in their
containers overnight to allow further recovery before being released where they had been
captured.
We used a locally-designed four-element Yagi antenna mounted on a 4.3 m pole, and an
AVM LA12 receiver (AVM Instruments, Champagne, Illinois) to track animals when
weather conditions promoted Giant bullfrog activity, or otherwise on average twice per
week, between 06h00 and 18h00. We did not perform nocturnal radio-tracking because
the area was considered unsafe at night. When a radio-tracked (RT) animal was located,
we recorded the date, time, the frog’s behavior, and geographic co-ordinates measured
with a global positioning system (GPS) unit. We lost eight females from the start, but
successfully tracked all 10 males and 10 of the 12 remaining females to at least one
burrow (Table 1). We tracked seven females and four males through a complete summer,
and four males during two or three successive summers. We lost most animals when their
transmitters expired.
Spool-tracking
To investigate the movements and habitat use of adult P. adspersus at a finer spatial
resolution, we performed spool-tracking (Heyer, 1994) between 1 October and 31 March
in each of two seasons 2006/2007 and 2007/2008 (hereafter referred to as Summers 4 and
5). A total of 26 males and 14 females was released with spools over these periods.
44
The white, cocoon-shaped polyester spools (Nm120/2 10W: ~ 4.5 g and ~ 330 m;
Danfield, Ltd, United Kingdom) weighed less than 5% of the body mass of each animal.
Because attempts to fit animals with a waistband or harness failed, and because adult P.
adspersus are strong enough to break a spool’s thread and could be lost without having a
waistband or harness removed, we attached spools with cyanoacrylate glue (Henkel
International, Düsseldorf, Germany). Prior to field work we allowed a strip of silicone
(Permoseal, Ltd, South Africa), applied along the length of each spool, to set. We cleaned
an animal’s dorsum with water, and dabbed it dry with disposable toweling before
applying a small patch of glue behind the animal’s head. The patch was left to dry for a
few minutes to create a suitable attachment site. For attachment, we applied glue to the
silicone strip of a spool, which was held in place on the dorsum of an animal for a few
minutes. Each animal was kept in a separate container for 30 min prior to release at its
capture point. Here, we tied the loose end of an animal’s thread to a flagged stake.
Released animals exhibited no or very limited escape behavior (at most a few hops,
usually < 1 m, across land or into water). We therefore measured thread trails from the
flagged stakes onwards. We recorded the movements of spool-tracked (ST) animals daily,
post-release, using a GPS unit to save the co-ordinates along trails every 2-6 m, and
whenever the direction of a trail changed markedly. Thread was collected as it was
followed. We lost three males from the start, but tracked 37 animals for 1-2 days, of
which five males and four females were each tracked to one burrow (Table 1). Animals
were lost when their spools detached prematurely, or when their thread trails snapped,
ended, or became badly entangled with other thread.
45
Data analysis
We measured animal movements and spatial habitat use using Hawth’s Analysis Tools
3.26 2002-06) in ArcMap 9.2 ESRI, Inc. 1999-2006), and performed statistical
tests in Statistica 7.0 StatSoft, Inc. 1984-2004 Tulsa). All shapefiles were projected
using transverse Mercator; spheroid: WGS 1984, with a central meridian of 29. Numbers
of relocation points obtained for RT males during more than one summer were sufficient
to construct minimum convex polygons (MCPs) representing individual seasonal “home
ranges.” The proportion of each male’s unique relocation points contained within each
season’s MCP was calculated to assess individual spatial use of habitat among summers.
We performed Spearman Rank correlations to investigate linear relationships with body
mass, SVL, or body condition (= body mass/SVL) of study animals. Mann-Whitney U
tests were performed to investigate differences between the sexes. Data presented are
mean ± standard deviation (SD), except where otherwise indicated. The level of
significance for all tests was P = 0.05. Multiple tests were subjected to Bonferroni
correction.
Results
Overnight movement
Study animals moved mostly at night. During daytime radio-tracking study animals were
almost always buried or at the seasonal dams. Only 19 (2%) of 976 location fixes
obtained for all RT animals represented instances where males were found aboveground
during daytime away from water (resting under vegetation). The only daytime overland
46
movements observed were on 6 December 2004 and 6 November 2005 when
respectively, 53 mm and 56 mm of rain fell in under an hour during the afternoon,
prompting P. adspersus to emerge from their burrows and move rapidly to the seasonal
dams.
Animals spent a day or two at spawning events, and four RT males were found guarding
offspring on one or two days each in summer. Females, as well as those males that did
not remain at the seasonal dams to guard offspring, returned directly to their burrows
(Fig. 1, 2). ST males and females exhibited little difference in the distances calculated
between their actual post-mating movement away from the seasonal dams and the
measured straight line distances between the end points (males: 21.5 ± 24.9%, range =
0.9-65.7%, n = 6; females: 18.4 ± 11.8%, range = 4.7-42.7%, n = 9; Mann Whitney U =
23.00, df = 4, P = 0.64). The maximum known distance moved overnight post-spawning
was similar for males (350.5 m) and females (350.4 m).
Spool-tracking revealed that animals readily: used human-created foot paths when
returning to their burrows; travelled along the gravel road on the eastern boundary of the
site, particularly in furrows on the road side where rain water was channeled; and moved
through an electrified fence if it was necessary to reach their burrow (Fig. 1, 2).
Since spools were found to only last a night or two, and because nocturnal radio-tracking
was not done, limited data on overnight foraging of adult male and female P. adspersus
were obtained. We directly observed adult P. adspersus foraging in the evening at the
47
water’s edge and within 20 m of the seasonal dams after several spawning events. An ST
male was observed foraging under house lights within 10 m of his burrow (Fig. 1), and a
concentration of short, criss-crossing tracks left by another ST male indirectly indicate
that this frog foraged within 12 m of his burrow before retiring underground (Fig. 1). The
recorded movement of an ST female around a termite mound is perhaps indicative of
foraging (Fig. 2).
Seasonal movement
A burrow used by an animal for less than two weeks in a summer was classified as
“temporary.” Burrows used for longer periods were classified as “long-term” burrows
(LTBs). Temporary burrows were used briefly, for example, by females en route to LTBs
(Fig. 3). Empty burrows of either type had an elliptical entrance, and a volume just large
enough to contain the inflated body of an adult male or female P. adspersus ca. 150 mm
beneath the ground surface. Burrows retained their form, and animals exhibited high
burrow fidelity. The number of LTBs used per summer by males (1.1 ± 0.2 burrows,
range = 1.0-1.5 burrows, n = 9) and females (1.1 ± 0.3 burrows, range = 1.0-2.0 burrows,
n = 9) did not differ significantly (U = 33.00, df = 7, P = 0.51). An electrified fence and
vehicle traffic on the tarred road on the western boundary of the site also did not deter
two RT males from repeatedly returning to their burrows (Fig. 4).
LTBs of animals were measured from the center (not the edge) of the nearest seasonal
dam (CNSD) because each dam’s surface area was small (and changed continually in
response to weather) relative to large, natural breeding sites (e.g., Glen Austin [~ 9 ha]
48
and Bullfrog [~ 81 ha] pans). LTBs of females were situated almost four times further
from the seasonal dams (446.8 ± 295.4 m, range = 119.1-902.7 m, n = 9) than those of
males (131.0 ± 105.2 m, range = 33.3-329.8 m, n = 9, U = 11.00, df = 7, P = 0.009). The
latter result remained significant following a Bonferroni correction.
Movement among years
Spatial use of habitat by four RT males during two or three successive summers suggests
animals had a strong preference and good memory for specific areas (Fig. 5). The
proportion of unique relocation points contained within the minimum convex polygons
constructed around each summer’s relocation points was 83.1 ± 14.0% (range of 68.8-
100.0%) for all four males across all summers. These animals used between one and two
(1.3 ± 0.2, range = 1.0-1.5, n = 4) burrows per summer, and LTBs used consecutively by
three of the four males within and/or among summers, were separated by 28.9 ± 21.5 m
(range = 16.4-53.7 m, n = 3) (Fig. 5a, b, d). One male used the same burrow for three
summers (Fig. 5c) and appeared to use the same burrow for the following two seasons
(the burrow was occupied, but we could not determine for certain if it was the same
male). Only one animal was known to return (after a period of 53 days) to a LTB (Fig.
5b). During winter, males with functional transmitters (n = 7) did not change position.
Body size and condition
The Spearman rank correlation between body mass, SVL or body condition, and the
number of LTBs used in a summer by males (n = 9) was r
s
= 0.51 (P = 0.17), r
s
= 0.63 (P
= 0.07), and r
s
= 0.34 (P = 0.37), respectively. The correlation between body mass, SVL
49
or body condition, and the number of LTBs used in a summer by females (n = 9) was r
s
=
0.55 (P = 0.13), r
s
= 0.55 (P = 0.13), and r
s
= 0.41 (P = 0.27), respectively. There were no
significant correlations between male (n = 9) body mass (r
s
= -0.13, P = 0.74), SVL (r
s
= -
0.08, P = 0.84), or body condition (r
s
= 0.10, P = 0.79), or female (n = 9) body mass (r
s
=
0.52, P = 0.16) or SVL (r
s
= -0.02, P = 0.97), with distance of their LTBs measured from
the CNSD. Only female body condition and distance of their LTBs measured from the
CNSD were significantly correlated (r
s
= 0.77, P = 0.02, n = 9).
Discussion
Overnight movement
Like many pond-breeding anurans (e.g., Sinsch 1988b, 1990; Sjögren-Gulve 1998;
Schwarzkopf & Alford 2002), P. adspersus are largely nocturnal, and move directly to
and from a breeding site. This reduces both their risk of desiccation, and predation by
diurnally-active birds that are major predators on this species (Channing 2001). Footpaths
and the nearby dirt road were likely used by study animals because there was less
vegetation to hamper their movement; channeled rain water run-off provided more
moisture; and/or prey and predators would have been more visible. Similar behavior has
been reported for other anurans (Moore 1954; Sinsch 1988b; Mazerolle 2005).
Pyxicephalus adspersus moved a maximum known overnight distance that is similar to
that reported for other pond-breeding anurans: 200-500 m (e.g., Gittins et al. 1980;
Carpenter & Gillingham 1987; Miaud et al. 2000; Bulger et al. 2003). It was, however,
50
shorter than expected considering that animal size and movement capacity are predicted
to scale positively (Peters 1983), and that a smaller, pond-breeding anuran species
(Hoplobatrachus occipitalis) inhabiting savanna habitat in West Africa, covers 1.4 km in
a night (Spieler & Linsenmair 1998).
Limited evidence suggested that adult P. adspersus forage up to 20 m from their burrows.
This behavior is analogous to the concentrated foraging of many anurans around burrows
or “forms” (i.e., small patches of ground cleared for resting) situated within a terrestrial
summer home range or non-breeding “activity area” (e.g., Dole 1965; Lamoureux et al.
2002; Penman et al. 2008).
Movement within and among seasons
Pyxicephalus adspersus shows strong site fidelity to their non-breeding habitats with both
adult males and females being highly faithful to their LTBs over time. Strong fidelity to
specific terrestrial sites has been reported for various anurans (e.g., Kelleher & Tester
1969; Haapanen 1970; Penman et al. 2008; Pitman et al. 2008), and may increase
familiarization of individual anurans with surrounding habitat features for improved
predator avoidance and foraging efficiency (Bellis 1965). Pyxicephalus adspersus
commonly prey on termites (Channing 2001), therefore, familiarity of individuals with
the location of termite colonies around their burrows could be beneficial.
As predicted by their extreme dimorphism in size, adult males and females exhibited a
clear difference in spatial habitat use. Results were, however, contrary to the expectation
51
that males would migrate further to satisfy their greater absolute energy needs. The LTBs
of females were situated almost four times further from the seasonal dams compared to
those of males. Greater spatial concentration of males around breeding sites appears to be
common among pond-breeding anurans (e.g., Pilliod et al. 2002; Regosin et al. 2005;
Bull 2006; Johnson et al. 2007), although the trend is not ubiquitous (e.g., Lemckert &
Brassil 2003; Smith & Green 2006; Kovar et al. 2009). We argue that resource-
competition (Austin et al. 2003; Palo et al. 2004) provides the best explanation for the
biased spatial distribution of non-breeding adult male and female P. adspersus around
breeding sites. By remaining close to breeding habitat males might increase their chances
of securing a territory, obtaining mates, and rearing offspring where aquatic habitat
conditions are most favorable for tadpole development. By moving further from breeding
habitat females might reduce their competition for food (e.g., termites and other
invertebrates) with the high abundance of much larger males that are also able to
consume larger prey items (e.g., birds, anurans and snakes; Channing 2001) near the
water.
Body size and condition
Body mass and SVL appear to be poor predictors of spatial habitat use by P. adspersus,
as is the case for other anuran species (e.g., Bellis 1965; Miaud et al. 2000; Bulger et al.
2003; Lemckert 2004). Body condition instead might be a more useful predictor (e.g.,
Sztatecsny & Schabetsberger 2005). In this study, a positive relationship was found
between female body condition and distance of their LTBs from the seasonal dams. Not
52
only may individuals move further if they have a better body condition; in doing so they
might benefit from reduced interspecific competition for food.
Conservation implications
Semlitsch & Bodie (2003) determined that 32 amphibian (including 19 anuran) species
utilized terrestrial habitat within a mean range of 159-290 m from aquatic breeding
habitat. Lemckert (2004) established that the mean distance moved away from breeding
habitat to terrestrial sites was approximately 300 m for bufonids, hylids, and ranids,
treated separately or combined (n = 28). Therefore, a 300 m wide buffer could be
expected to protect roughly half the members of an “average” pond-breeding anuran
population (Lemckert 2004). At the Diepsloot site a 300 m wide buffer measured from
the center of the three seasonal dams would include the LTBs of 67% (n = 12) of RT
animals that used one or more LTBs (n = 18) (Fig. 6). However, since the LTBs of
females were almost four times further than those of males, a 300 m wide buffer would
include the LTBs of 89% of males (n = 9) but only 44% of females (n = 9) that used one
or more LTBs. Moreover, LTBs of females in the best body condition were situated
furthest from the seasonal dams. Since body condition can influence survival and
reproductive success in anurans (Reading 2007), a 300 m or even a 500 m wide buffer
around the seasonal dams would exclude the LTBs of females that were potentially most
fecund.
To protect the LTBs of all study animals, a 950-1 000 m wide buffer would be necessary
around the Diepsloot seasonal dams. Assuming that results of this study are
53
representative of spatial habitat use by P. adspersus at other localities, buffer zones 500-1
000 m wide will be necessary to protect most adults buried around breeding sites. If these
are larger than the Diepsloot seasonal dams, buffer zones should be measured from the
periphery (not the center) of spawning sites. Similarly large protective buffers have been
proposed for other pond-breeding anurans (e.g., Richter et al. 2001).
If there was more natural habitat around the Diepsloot site, would animals have moved
further from the seasonal dams? Walls around properties clearly constrained animal
movements, and P. adspersus probably cannot bury into ferricrete (i.e., sand and gravel
cemented into a hard mass by iron oxide), exposed by topsoil erosion immediately west
of Dam 1, and between Dams 1 and 3. P. adspersus crossing nearby roads were likely to
get hit by traffic, and were therefore perhaps less likely to be caught or successfully
tracked. Hence animal movements extended mainly north of the seasonal dams. Most
adult P. adspersus on the site, however, probably remained within a kilometer of the
seasonal dams. Although the species could move up to ~ 1.5-2 km north and north-east of
the dams, the furthest animal relocation was 903 m north-east of the Dam 1. Yet a
population genetics study (Yetman& Ferguson, unpubl. data, [Chapter 5]) based on P.
adspersus sampled from Diepsloot and other localities mostly in Gauteng Province
revealed significant (historical) gene flow between populations up to 20-100 km apart.
We therefore suggest that like other pond-breeding anurans (e.g., Dole 1971; Schroeder
1976; Sinsch 1997; Berven & Grudzien 1990), P. adspersus adults are generally
philopatric to their breeding sites, and utilize terrestrial habitat most often within a
54
kilometer radius thereof; whereas immatures typically disperse, facilitating gene flow
between neighbouring populations.
Acknowledgements
Jimmy, Alice and Mark Yetman, Johan Lötter, Wendy Duncan and others are sincerely
thanked for their invaluable assistance with field work. Dr Dorianne Elliot and staff at the
Bird and Exotic Animal Clinic, Onderstepoort Veterinary Institute are kindly
acknowledged for surgically implanting the study animals with transmitters. The study
was strongly supported by the Endangered Wildlife Trust, and generously funded by
Rand Merchant Bank, the Pretoria East branch of the South African Hunter’s and Game
Conservation Association, Arrow Bulk Marketing, Cellar Rats Wine Club, Bill Flynn and
Diaz Films. The Gauteng Department of Agriculture, Conservation and Environment is
thanked for issuing collection, handling and transport permits 1 240 and 1 296 for the
radio- and spool-tracking study animals, respectively. Study animals were handled in a
manner complying with the “Guidelines for Use of Live Amphibians and Reptiles in
Field Research” (Society for the Study of Amphibians and Reptiles, the American
Society of Ichthyologists and Herpetologists, and The Herpetologists’ League). Mark
Robertson and Luke Verburgt are thanked for their comments on a draft of this
manuscript.
55
Table 1. Measures of success with radio- and spool-tracking
of adult Pyxicephalus adspersus in Diepsloot, Gauteng
Province, South Africa.
Radio-tracking
Spool-tracking
No. of animals
Released 10 20 26 14
Tracked: 10 12 23 14
through < 1 summer
2 5 23 14
through 1 summer 4 7 - -
through > 1 summer
4 - - -
to 1 burrows 10 10 5 4
Lost 6 8 15 10
Died 1 2 0 0
Device removed or shed 3 2 8 4
56
Figure 1.
Overnight movements by spool-tracked adult male
Pyxicephalus adspersus (n = 8) that returned to their burrows ()
after spawning activity at three seasonal dams (D1, 2, and 3). The
black arrow indicates where an animal moved though an electrified
fence. = where foraging by an animal was observed, or inferred
from its movements.
57
Figure 2. Overnight movements by spool-tracked adult female
Pyxicephalus adspersus (n = 11) that returned to their burrows ()
after spawning activity at three seasonal dams (D1, 2, and 3). The
black arrow indicates where an animal moved though an electrified
fence. = where foraging by an animal was inferred from its
movement around a termite mound. AS = agricultural small-
holding.
58
Figure 3. Within-season movements by radio-tracked adult female
Pyxicephalus adspersus (n = 11) between seasonal dams (D1, 2 and
3) and their burrows. = “temporary” burrows used for less than
two weeks. = “long-term” burrows used for more than two
weeks. = where an animal died for uncertain reasons. AS =
agricultural small-holding. LH = low-cost housing.
59
Figure 4. Within-season movements by adult male Pyxicephalus
adspersus (n = 6) radio-tracked (for a maximum of one full
summer) between seasonal dams (D1, 2 and 3) and their burrows.
= “temporary” burrows used for less than two weeks. =
“long-term” burrows used for more than two weeks. The black
arrow indicates where an animal moved though an electrified fence.
AS = agricultural small-holding.
60
Figure 5. Between-season movements by adult male Pyxicephalus
adspersus radio-tracked between seasonal dams (D1, 2 and 3) and
their burrows, during: a-b) two summers (n = 2 males); or c-d)
three summers (n = 2 males). = “temporary burrows used for
less than two weeks. = “long-term” burrows used for more than
two weeks. AS = agricultural small-holding.
61
Figure 6. Number of “long-term” burrows (LTBs; i.e., used for more than
two weeks) of radio-tracked adult male or female Pyxicephalus adspersus as
a function of distance from the center of nearest seasonal dam (CNSD) on
the site. The LTBs of most males (i.e., five of n = 9 males) were located
within 100 m, whereas those of all females (n = 9) were found beyond 200
m from the CNSD.
62
Chapter 4
Conservation implications of the age/size distribution of giant bullfrogs
(Pyxicephalus adspersus) at three peri-urban breeding sites *
Caroline A. Yetman
1
, Peter Mokonoto
2
, J. Willem H. Ferguson
1,3
1
Department of Zoology and Entomology,
2
Section of Anatomical Pathology,
Department of Paraclinical Sciences, Faculty of Veterinary Sciences, and
3
Centre for Environmental Studies, University of Pretoria, Pretoria, 0002, South Africa
Abstract.Nothing is known about the age of wild giant bullfrogs (Pyxicephalus
adspersus); yet this information has important conservation implications for this
regionally threatened species. We quantified and compared the age, body size, and body
condition of adult male and female P. adspersus caught during spawning events at peri-
urban breeding sites in Diepsloot, and at Glen Austin and Bullfrog pans in Gauteng
Province, South Africa. Age was estimated from lines of arrested growth (LAG) counted
in cross-sections of animal phalanges. Males and females from all three sites possessed 6
± 2 (max. 16) and 4 ± 1 (max. 11) LAG, respectively, suggesting shorter female
longevity. Individuals with < 3 LAG were not encountered at the breeding sites, implying
that newly metamorphosed P. adspersus require at least three years to reach sexual
maturity. There was no significant difference in the LAG counts of same-sex animals
63
between the three sites. However, mean male snout-vent length, mass, and body
condition was greatest at Glen Austin Pan, and lowest at Bullfrog Pan. The latter is
possibly explained by chemical contamination of Bullfrog Pan from an adjacent disused
landfill. At Glen Austin Pan males and females sampled in 2004-06 for this study were
significantly shorter than those sampled at the same site in 1992-93 for a different study.
Our results suggest that male P. adspersus may live for 20 years or more in the wild, but
at some peri-urban breeding sites adult life expectancy is declining. Juvenile P. adspersus
are most threatened by terrestrial habitat transformation because they take 3 years to
mature, during which period they may move great distances from their natal site.
Differences in the size and condition of P. adspersus between the study sites, suggests
that the species requires site-specific management in addition to conservation at larger
spatial scales.
* Accepted for publication as:
Yetman, C.A., P. Mokonoto & J.W.H. Ferguson (In press). Conservation implications of
the age/size distribution of giant bullfrogs (Pyxicephalus adspersus) at three peri-
urban breeding sites. Herpetological Journal
64
Introduction
In temperate, mid-latitude environments anurans require from several months to three
years to reach sexual maturity (e.g. Zug & Zug 1979; Tsiora & Kyriakopoulou-
Sklavounou 2002) while at high latitudes or altitudes anurans may take four or more
years to mature (e.g. Metter 1967; Matthews & Miaud 2007). Many anurans live less than
10 years (e.g. Measey 2001; Esteban et al. 2004; Guarino & Erismis 2008), but larger
species and/or captive individuals can exceed this, e.g. wild and captive Rana
catesbeiana reached 10 and 16 years, respectively (Oliver 1955; Goin & Goin 1962) and
wild and captive Bufo marinus reached 15 and 40 years, respectively (Tyler 1975). Age at
maturity and longevity have important conservation implications for threatened anurans.
As age at maturation increases, generation time increases and population resilience
decreases (Duellman & Trueb 1994). Older females are generally larger than younger
females, and therefore produce larger and more clutches during a breeding season
(Howard 1978; Reading & Clarke 1995; Reading 2007). Although there have been
numerous studies on anuran age (see Tables 2-9 in Duellman & Trueb 1994; Table 1 in
Monnet & Cherry 2002), very limited data have been obtained for mainland Africa
species, e.g. Amietophrynus pardalis (Cherry & Françillon-Vieillot 1992), R. saharica
(Esteban et al. 1999; Meddeb et al. 2007) and Xenopus laevis (Measey 2001).
The giant bullfrog (Pyxicephalus adspersus) is one of the largest extant anurans (Du
Preez & Cook 2004) and is exceptional in that males (~ 400-1 000 g) can weigh up to ten
times more than females (~ 90-300 g; Cook 1996). The species is widely distributed
across the grassland and savanna regions of southern Africa (Channing 2001) but is
65
regarded as Near-Threatened in South Africa due to habitat loss and other factors
(Harrison et al. 2001; Minter et al. 2004). Due to the very brief and unpredictable
appearance of P. adspersus aboveground during summer (Yetman & Ferguson 2011a,
[Chapter 2]), little systematic field-based research and monitoring has been performed on
this species (Jacobsen 1989; Kok et al. 1989; Van Wyk et al. 1992; Cook 1996). The
resulting lack of demographic information has impeded the assessment of the
conservation status of P. adspersus, and provides a weak baseline for evaluation of
efforts to protect threatened populations. Captive P. adspersus have reportedly reached an
estimated 45 years of age (Channing 2001) but nothing is known about the age of wild
specimens. Juvenile P. adspersus are expected to require several years to mature given
the large body size of adults (Peters 1983); therefore populations may have slow
generation turnover and low resilience to perturbations in density.
Skeletochronology is a commonly-used technique for aging anurans (Halliday & Verrell
1988) and has been applied to a wide variety of species from arid (Rogers & Harvey
1994; Tessa et al. 2007), temperate (Miaud et al. 1999; Guarino et al. 2003), subtropical,
and even tropical environments (Morrison et al. 2004; Lai et al. 2005). The technique
involves analysis of concentric lines of arrested growth (LAG) visible in cross-sections of
animal long bones and phalanges (Castanet & Smirina 1990). LAG are deposited during
periods of animal inactivity and are therefore pronounced in species with highly seasonal
activity. Adult P. adspersus are active after heavy rainfall for cumulatively less than a
month between October and March (Yetman & Ferguson 2011a, [Chapter 2]), and were
therefore considered appropriate subjects for phalangeal skeletochronology, which
66
precludes the need to sacrifice whole animals. Using this method we aimed to estimate
the age of wild, adult P. adspersus to provide important life history and demographic
information for improved conservation management of this species. In particular we
wanted to determine whether P. adspersus at different breeding sites differed
significantly in age, body size or body condition, and therefore, require site-specific
conservation management. Specific objectives of this study were to:
estimate the age range of wild, breeding P. adspersus males and females using
skeletochronology.
examine the relationship between body size and age in adult P. adspersus.
compare the age, body size and body condition of adult P. adspersus at different
breeding sites.
infer consequences of the results on the conservation management of P.
adspersus.
Materials and Methods
Study sites
Adult P. adspersus were caught by hand or with a hand-held net at spawning events
between 1 October and 31 March in the 2003/2004, 2004/2005 and 2005/2006 summer
seasons at three peri-urban, seasonal breeding sites in Gauteng Province, South Africa. At
the Diepsloot site (25º56'23.59"S, 28º01'21.88"E) P. adspersus bred in three small (~ 0.2-
0.9 ha) artificial dams where there was a relatively high level of human activity. Glen
Austin Pan (25º58'36.00"S, 28º09'58.33"E; ~ 9 ha) and Bullfrog Pan (26º08'22.35"S,
67
28º18'51.10"E; ~ 81 ha) were proclaimed bird sanctuaries that gave limited protection for
the historically large resident populations of P. adspersus (Cook 1996; Slater-Jones
1996). The three sites (separated by 15-36 km) had very similar climates and remaining
natural grassland (Mucina et al. 2005) relative to the variety of habitats that P. adspersus
inhabits in southern Africa. The agricultural small-holdings surrounding each site were
becoming increasingly urbanized.
Field work
Study animals were handled in a manner complying with the “Guidelines for Use of Live
Amphibians and Reptiles in Field Research” (Society for the Study of Amphibians and
Reptiles, the American Society of Ichthyologists and Herpetologists, and The
Herpetologists’ League). We used an electronic balance (accurate to 1 gram) and steel
tape-measure (accurate to 1 millimeter) to measure each animal’s body mass and snout-
vent length (SVL), respectively. Animal body condition was estimated as mass/SVL
(Schulte-Hostedde et al. 2005). A sterilized bone cutter or wire clipper was used to clip
the two most distal phalanges from the second (i.e. longest) toe on the right hind limb of
each animal. Toe clips were stored in separate vials containing 70% ethanol, and animals
were released near their point of capture.
Histology
To analyze animal body size in relation to age, we selected toe clips of five small (90 mm
< SVL 110 mm) and five large (110 mm < SVL 130 mm) females, as well as five
small (130 mm < SVL 150 mm), five medium (150 mm < SVL 170 mm) and five
68
large (SVL > 170 mm) males from each site except Bullfrog Pan, where only three large
males were caught. Toe clips of the three largest males caught during the study (which
included two males from Diepsloot and one from Glen Austin Pan with a SVL 190
mm) were added to the largest size category. Data for the size-age analysis were therefore
obtained from 76 animals, including 30 females and 46 males (hereon referred to as the
“size selected” animals). To compare the age of same-sex animals between the three
locations we used the complete sample set again to randomly select toe clips of 10
females and 15 males from each site. Hence data for comparisons of animal age between
the populations were obtained for 75 animals, including 30 females and 45 males (hereon
referred to as the “randomly selected” animals).
Toe clips were decalcified in 8% formic acid for 48 h, and embedded in paraffin wax. A
rotary microtome was used to cut 6 µm thick sections which were then stained with
Ehrlich’s haematoxylin. This method was adapted from the method used by Castanet &
Smirina (1990), which involved nitric acid for decalcification, and a freezing microtome.
We examined mounted sections under a standard light microscope at 4x, 10x, or 20x
magnification, and selected for each sampled animal one or two sections with the
smallest medullar cavity. We counted lines of arrested growth (LAG) from the endosteal
bone outwards and treated the perimeter of the periosteal bone as a LAG if the animal
had been sampled during the first spawning event of a season. False and double LAG
were respectively, distinguished as incomplete or complete feint LAG within bands of
summertime bone growth (Hemelaar & Van Gelder 1980; Sagor et al. 1998; Cvetković et
al. 2005).
69
Data analysis
Statistical tests were performed in Statistica 7.0 StatSoft, Inc. 1984-2004 Tulsa). We
used least squares regression analyses to examine relationships between the body mass,
SVL, and LAG counts of size selected males or females from the three sites combined.
We used ANOVA to compare LAG counts between small and large females, and
between small, medium and large males. The SVL, body mass, body condition, or LAG
counts between randomly selected males and females from the three sites treated
separately, or combined, were compared using t-tests (when n < 30) or ANOVA (when n
> 30). We used ANOVA to compare the SVL, body mass, body condition, or LAG
counts of randomly selected same-sex animals between the three sites. In addition, we
calculated the two-sample z- and Smirnov d-statistics to, respectively, compare the mean
values and cumulative frequency distributions of the SVL or body mass of all males or
females sampled during spawning events at Glen Austin Pan in 2004-06 (for this study)
and in 1992-93 (for a separate study; Cook 1996). The data presented are mean ±
standard deviation (SD) and the level of significance for all tests was P = 0.05. We
applied sequential Bonferroni corrections to comparisons of SVL, body mass, body
condition, or LAG counts between randomly selected males and females from the same
site or same-sex animals from different sites, or all same-sex animals sampled at Glen
Austin Pan for the two separate studies.
70
Results
Interpretation of sections
The most hematoxylinophilic LAG observed in the sections (Fig. 1) were arranged in a
consistent geometric pattern among the samples (e.g. Fig. 1a and b in Bastien & Leclair
1992; Fig. 1c and d in Guarino et al. 2003; Fig. 1 in Tessa et al. 2007). We assumed that
each of these LAG was deposited while an animal experienced torpor underground
during a six- to eight-month winter period (Loveridge & Withers 1981). Animals
experienced the most bone growth during their second summer. Growth declined slightly
during the third summer and decreased dramatically thereafter. The number of complete
and partially remaining LAG prior to this transition in bone growth revealed that
complete or partial resorption of the first LAG and partial resorption of the second LAG
was common among the samples. Within bands of summertime bone growth multiple
false and double LAG were observed, which did not exhibit any consistent pattern among
the samples.
Body size and age of size-selected animals
The females (n = 30) weighed 99-285 g, measured 92-136 mm in SVL and possessed 3-
11 LAG. The males (n = 46) weighed 213-872 g, measured 130-198 mm in SVL and
possessed 3-16 LAG. No animal possessed < 3 LAG. The oldest female (with 11 LAG)
weighed 285 g, measured 136 mm in SVL and was sampled at Bullfrog Pan. The oldest
male (with 16 LAG) weighed 704 g, measured 178 mm and was sampled at Glen Austin
Pan. The three males with a SVL 190 mm possessed 7 or 8 LAG each.
71
Body mass increased exponentially with SVL in both males (y = 42.10 x
2.465
, t = 11.7, P
< 0.001, n = 46) and females (y = 26.05 x
2.317
, t = 8.0, P < 0.001, n = 30; Fig. 2a). There
was a weak linear increase in body mass with LAG counts in both males (y = 29.92x +
289.63, r
2
= 0.2, P < 0.001, n = 46) and females (y = 18.52x + 17.63, r
2
= 0.6, P < 0.001,
n = 30). SVL increased logarithmically with LAG counts in males (y = 31.96Ln(x) +
103.27, r
2
= 0.3, P < 0.001, n = 46) and linearly with LAG counts in females (y = 4.09x +
90.36, r
2
= 0.5, P < 0.001, n = 30; Fig. 2b). LAG counts differed significantly between
small, medium and large males (F
2, 43
= 7.0, P = 0.002, n = 46; Fig. 2c), and between
small and large females (F
1, 28
= 6.9, P = 0.01, n = 30; Fig. 2d). The lack of a strong
sigmoid relationship between LAG counts and SVL precluded growth analysis to
estimate maximal body size of males and females.
Males versus females
The SVL, body mass and body condition of all sampled P. adspersus, and LAG counts of
the 45 males and 30 females whose toe clips were randomly selected for sectioning are
shown in Table 1. At each of the three sites males were significantly longer and heavier
than the females (Table 1). These differences, and the difference in LAG counts between
males and females from the Diepsloot dams, or from all three sites combined, remained
significant following Bonferroni corrections. The frequency distribution of LAG counts
of the randomly selected males and females from the three sites combined is shown in
Fig. 3. Eighty percent of the males possessed 4-7 LAG, and 83% of the females
possessed 3-5 LAG. Only 13% of the males and 17% of the females possessed 8-11 LAG
and 6-7 LAG, respectively.
72
Study site comparisons
There was no significant difference in the LAG counts of randomly selected same-sex
animals between the three sites (Table 1). However, mean male SVL, body mass and
condition was greatest at Glen Austin Pan and lowest at Bullfrog Pan. Mean female SVL
was longest at Glen Austin Pan and shortest at Diepsloot.
Male (n = 510) and female (n = 204) P. adspersus sampled by Cook (1996) during 1992
or 1993 at Glen Austin Pan had a body mass of 561 ± 87 g (range: 320-970 g) and 173 ±
51 g (range: 60-400 g), and a SVL of 184 ± 13 mm (range: 137-227 mm) and 116 ± 10
mm (range: 95-141 mm), respectively. The mean SVL of males and females sampled at
this site in 2004-06 for our study measured, respectively, 19 mm (z = 15.9, P < 0.001)
and 3 mm (z = 2.1, P = 0.02) less than in 1992-93. These differences in mean SVL
remained significant following sequential Bonferroni corrections. Male and female mean
body mass measured 6 g (z = 0.5) and 4 g (z = 0.4) less, respectively, in 2004-06 than in
1992-93, but these differences in mean body mass were non-significant (P = 0.3 for both
tests). Frequency distributions of the SVL or body mass of all male or female P.
adspersus sampled in 1992-93 or 2004-06, are shown in Figures 4a-d. The cumulative
frequency distributions of SVL and body mass differed significantly between the two
studies for males (Smirnov d = 0.41 and 0.18, respectively, P < 0.05), but not for females
(d = 0.20 and 0.24, respectively).
73
Discussion
Interpretation of sections
The reliability of skeletochronological age estimates can be affected by endosteal
resorption of periosteal bone, deposition of false and double LAG, and rapprochement of
LAG near the perimeter of the periosteal bone (Hemelaar & Van Gelder 1980; Castanet
& Smirina 1990). Skeletochronological age estimates are also less reliable when
phalanges (not long bones) are used or study species are long-lived (Wagner et al. 2011).
Therefore the application of skeletochronology ideally requires reference to LAG in
bones of known-age individuals. Known-age, wild P. adspersus were not available;
therefore we attempted to sample captive specimens. However, most owners would not
allow toe-clipping, accurate age records did not exist, and many specimens were fed and
remained active throughout the year (CAY pers. obs.).
Nevertheless, the progressive increase in numbers of LAG with the body size of sampled
P. adspersus (Fig. 2b-d) indicates that the use of LAG was effective. The geometric
pattern of LAG deposition and the complete or partial resorption of the first and/or
second LAG observed in the P. adspersus sections is also common among anurans
(Smirina 1972; Hemelaar 1985; Patón et al. 1991). False and double LAG were however,
unusually abundant among the sampled P. adspersus (Fig. 1). False LAG have been
associated with periods of drought during the active season of other anurans (e.g. Rogers
& Harvey 1994), and were probably deposited when sampled P. adspersus remained
inactive for extended periods during summer between activity episodes following heavy
rainfall. This was evident at the Diepsloot site where radio-tracked P. adspersus were
74
buried for approximately 22-24 weeks cumulatively during the six-month (~ 26 week)
summer period (Yetman & Ferguson 2011a, [Chapter 2]).
Body size and age
Pyxicephalus adspersus has a voracious appetite (Branch 1976; Paukstis & Reinbold
1984) and juveniles grow rapidly post-metamorphosis (Van Wyk et al. 1992; Douglas
1995). After nine months, which included winter torpor, newly metamorphosed P.
adspersus kept under semi-natural conditions had grown 1 326% heavier and twice as
long (Conradie et al. 2010). The consistently widest band of bone growth between the
second and third LAG (Fig. 1) among the sections in this study suggests that sampled P.
adspersus experienced the greatest somatic growth during their second active season.
Subsequent reduction in bone growth appeared typical of a shift in resource allocation
from growth to reproduction at sexual maturation (Caetano & Castanet 1993; Smirina
1994), suggesting that at our study sites, female and at least some male P. adspersus
attained sexual maturity following their third active season. Other large anurans mature at
an earlier or similar age, e.g. Bufo marinus, 1-2 years (Easteal 1982) and R. catesbeiana,
1-4 years (Howard 1981; Harding 1997).
We suspect, considering that the oldest aged male and female in this study possessed 16
and 11 LAG respectively, that at undisturbed sites male and female P. adspersus can
reach 20 and 15 years of age respectively. Similar ages have been reached by other
anurans in the wild, e.g. > 20 years for Bombina variegata (Płytycz & Bigaj 1993) and 15
years for Bufo marinus (Tyler 1975). In more arid areas such as the Kalahari, where P.
adspersus can reportedly spend several successive years in torpor underground (Du Preez
75
& Carruthers 2009), individuals are likely to live much longer than 20 years. This is
because, across different habitats, P. adspersus may have a similar mean physiological
longevity calculated as, e.g. mean age in years multiplied by the mean number of summer
rainfall days at a specific locality (modified from Bastien & Leclair 1992).
Males versus females
Male P. adspersus fight aggressively during spawning events, and since the largest males
secure more matings and perform parental care of offspring (Cook 1996; Cook et al.
2001), there is probably strong selection for large male body size in P. adspersus (Shine
1979; Arak 1988). At our three study sites male P. adspersus were on average
approximately three times (360 g) heavier, 1.5 times (54 mm) longer and possessed two
more LAG than females suggesting shorter female longevity. There have been few
similar studies on other anurans with reversed sexual size dimorphism (e.g. Briggs &
Storm 1970; Sinsch et al. 2001). Although greater body size may be attained through
greater longevity of the larger sex in anurans (Monnet & Cherry 2002), male P.
adspersus were much larger than females of comparable LAG count (Fig. 2b). This was
probably due to slower growth in females compared to males, which was observed in
captive P. edulis (mistaken as P. adspersus, it seems) following the appearance of
differentiated gonads in both sexes from approximately 60 days post-metamorphosis
(Hayes & Licht 1992). Gonadectomy of a portion of these animals did not affect their
subsequent growth, suggesting that sex hormones are not related to the divergence in the
growth rates of male and female Pyxicephalus.
76
Study site comparisons
Differences in the size and condition of animals between the three study sites were not
matched by differences in their age (Table 1). This was probably because the aged
animals represented a small subset of all the sampled animals (n = 45 of 211 males; n =
30 of 68 females). When we compared the mass, SVL or condition of only the aged
males or females (not shown here), there was no significant difference in their size or
condition between the three sites. Therefore animals at Glen Austin Pan were larger
because they were probably also older.
A plausible explanation for the smaller body size and poorer condition of males at
Bullfrog Pan is the contamination of this site by toxic leachate from an adjacent disused
landfill. Since 1993 the leachate was channelled by a plastic drain to a pump-house; but
in 1996 a leak in the drain was discovered. At this time a study (Slater-Jones 1996) was
performed when two newly metamorphosed P. adspersus with facial deformities were
found in the pan. The study revealed an elevated concentration of lead (6.4 parts per
million) in the tissue of newly metamorphosed P. adspersus, and a significant reduction
in their body size with decreasing distance from the leak. Assuming in our study that
males with 7-10 LAG were born between 1994 and 1999, those from Bullfrog Pan were
possibly exposed to significant contamination during metamorphosis, which could have
reduced their size and condition as froglets and later as adults. In 1997 the leak was fixed,
which could explain why males with < 7 LAG (born after 1997) were larger than males
with 7 LAG at this site. All but two females at this site possessed < 7 LAG. The one
female with 7 LAG appeared to be small for her age (Fig. 2b). The other female with 11
77
LAG was large (Fig. 2b), possibly because she was born in 1993 when leachate was
successfully diverted from the pan. Deformities and reduced growth, body size, condition
and/or longevity of amphibians from chemically contaminated sites have been reported in
various studies (e.g. Carey & Bryant 1995; Rowe et al. 2001; Spear et al. 2009; Brodeur
et al. 2011).
The modest decrease in SVL of male and female P. adspersus between 1992-93 and
2004-06 at Glen Austin Pan was unlikely due to differences in sampling or measurement
because these were performed in the same way (Cook 1996). Considering the increasing
loss of habitat and mortality of P. adspersus with urban encroachment in Gauteng (CAY
pers. obs.), the reduction in SVL of adults at Glen Austin Pan possibly indicates that very
large males (SVL > 190 mm) and females (SVL > 130 mm) have become increasingly
rare. This is suggested by all four graphs in Fig. 4. The significance of the size decrease
of P. adspersus at Glen Austin Pan is emphasized by considering that the mean SVL and
mass of adults at this site during 1992-93 was greater than at any of the three sites during
our study. Mean SVL of males at Glen Austin Pan in 1992-93 was also greater than the
maximum SVL of males at Bullfrog Pan in 2004-06. Although we cannot completely
discount variation in ecological conditions (e.g., rainfall, food availability or predation;
Reaser 2000) as a cause of the temporal or spatial differences in the body size or
condition of P. adspersus in this study, indirect evidence suggests that large P. adspersus
have become scarcer at several peri-urban breeding sites.
78
Conservation implications
Since some anurans take four or more years to mature (e.g. Metter 1967; Matthews &
Miaud 2007), the generation time of P. adspersus is less than may have been predicted
from the large adult size of this species (Peters 1983). Therefore P. adspersus populations
could have greater resilience than has been assumed. However, population breeding
success is likely to fluctuate dramatically because P. adspersus spawning and larval
survival is strongly related to rainfall which varies greatly between years (Read 1990;
Yetman & Ferguson 2001a, [Chapter 2]). In addition, juvenile P. adspersus probably
experience high mortality prior to their sexual maturation when they spend 3 years
moving large distances overland, as revealed by the philopatric behaviour of adults
(Yetman & Ferguson 2011b, [Chapter 3]) and gene flow between populations 20 km
apart (Yetman & Ferguson, unpubl. data, [Chapter 5]). Terrestrial habitat conservation is
therefore critical for P. adspersus juvenile survival, dispersal and recruitment, which
probably exerts a stronger influence than adult survival on the growth of populations
(Conroy & Brook 2003; Grafe et al. 2004).
The oldest aged animal in this study was less than half the maximum age of 45 years
reported for a captive P. adspersus (Channing 2001). Moreover, < 20% of males and
females in this study had > 7 or > 5 LAG respectively (Fig. 3). This suggests that the life
expectancy of P. adspersus at peri-urban sites is low, which would explain why very
large males and females appear to have become increasingly rare, such as at Glen Austin
and Bullfrog pans. This is of conservation concern because the largest male P. adspersus
are reproductively most successful, partly because usually only they perform parental
79
care of tadpoles, which are highly vulnerable to desiccation and predation (Cook 1996;
Cook et al. 2001). As in other anurans, the largest female P. adspersus probably produce
the greatest number of eggs and clutches in a season (Howard 1978; Reading & Clarke
1995; Reading 2007). Therefore efforts to reduce unnatural mortality of P. adspersus at
threatened sites are strongly recommended. Habitat loss and road traffic generally
represent the greatest threats to P. adspersus (Du Preez & Cook 2004). Therefore the
destruction of undeveloped terrestrial habitat connected to breeding sites should be
prevented (Yetman & Ferguson 2011b, [Chapter 3]), and safer movement of P. adspersus
across roads near breeding sites should be ensured (Langton 1989; Puky 2005).
Differences in the size and condition of P. adspersus at different breeding sites, suggests
that the species requires conservation management at site-specific, as well as broader
spatial scales (Boyd et al. 2008). An important local concern is the poorer condition of P.
adspersus at Bullfrog Pan, which is the largest known historical breeding site for this
species in Gauteng, and perhaps South Africa. Research is warranted to determine if P.
adspersus at Bullfrog Pan is being adversely affected by pollution or some other factor.
Studies have shown that the body size and/or age of anurans can give useful indications
of habitat quality. For example, SVL and age of R. catesbeiana was significantly lower at
sites with high pesticide contamination (Spear et al. 2009), and age at maturity in males
and longevity in females of Bufo viridis was negatively related to the intensity of human
land use within a 1 km radius of breeding sites (Sinsch et al. 2007). Presuming that the
spatio-temporal differences in body size and condition of the sampled P. adspersus
reflect anthropogenic transformation of habitat, P. adspersus could serve as a valuable
80
indicator of degradation of seasonal wetlands and grasslands, which are both highly
threatened in South Africa (Low & Rebelo 1996; Le Roux 2002).
Acknowledgements
We kindly thank Jimmy, Alice and Mark Yetman, Johan Lötter, and Wendy Duncan, as
well as the Twine, Du Toit, Muller, Morris, and Rabie families for their invaluable
assistance with field work. We gratefully acknowledge the assistance of Leon Prozesky
and colleagues in the Pathology Laboratory at Onderstepoort Veterinary Institute, where
many samples were sectioned and stained. Michel Laurin, Claude Miaud, Gary Matson,
and Louis Du Preez are acknowledged for providing technical histological advice. Luke
Verburgt is thanked for commenting on a draft of this manuscript. The study was
supported by the Endangered Wildlife Trust, and funded by Rand Merchant Bank, the
Pretoria East branch of the South African Hunter’s and Game Conservation Association,
and Arrow Bulk Marketing. The Gauteng Department of Agriculture and Rural
Development is thanked for issuing permit 1 240 to CAY for the field work.
81
Table 1. Mean ± standard deviation, and range, of the snout-vent length (SVL), body mass, body condition or
age (estimated from LAG = lines of arrested growth in phalanges) of male or female Pyxicephalus adspersus
from three peri-urban breeding sites. ANOVA F, t-, or z-test, and P values (*** P < 0.001) pertain to
comparisons of variables for same-sex animals between the three sites (column 3), or between males and
females from the three sites combined (column 4) or treated separately (columns 5, 6 and 7). Values of P >
0.01 were non-significant following sequential Bonferroni correction.
Column 2
3 4 5 6 7
Sex
F, P All three sites Diepsloot dams Bullfrog Pan Glen Austin Pan
SVL (mm)
F
2, 208
= 8.5
***
163 ± 13
130-198
n = 211
160 ± 14
131-190
n = 56
155 ± 12
130-181
n = 23
165 ± 12
130-198
n = 132
F
2, 65
= 5.8
P = 0.004
109 ± 9
92-136
n = 68
105 ± 8
93-129
n = 30
110 ± 12
92-136
n = 16
113 ± 6
100-123
n = 22
F
1, 277
= 978.2**
*
F
1, 84
= 376.0*** t = -11.6*** F
1, 152
= 386.5***
Body mass (g)
F
2, 208
=
25.2
***
512 ± 131
204-872
n = 211
454 ± 108
204-699
n = 56
405 ± 102
222-643
n = 23
555 ± 124
239-872
n = 132
F
2, 65
= 3.4
P = 0.04
152 ± 43
90-294
n = 68
139 ± 33
95-233
n = 30
154 ± 53
90-285
n = 16
169 ± 42
117-294
n = 22
F
1, 277
= 498.7***
F
1, 84
= 242.1*** t
= -9.0*** F
1, 152
= 209.2***
Body condition
F
2, 208
=
30.2
***
3.1 ± 0.6
1.5-4.4
n = 211
2.8 ± 0.5
1.5-3.7
n = 56
2.6 ± 0.5
1.7-3.7
n = 23
3.3 ± 0.6
1.8-4.4
n = 132
F
2, 65
= 2.5
P = 0.09
1.4 ± 0.3
0.9-2.5
n = 68
1.3 ± 0.2
1.0-1.8
n = 30
1.4 ± 0.3
0.9-2.1
n = 16
1.5 ± 0.3
1.0-2.5
n = 22
F
1, 277
= 504.4***
F
1, 84
= 242.9*** t
= -8.9*** F
1, 152
= 218.6***
LAG
F
2, 42
= 1.3
P = 0.3
6 ± 2
3-11
n = 45
6 ± 1
3-7
n = 15
6 ± 1
3-8
n = 15
7 ± 2
3-11
n = 15
F
2, 27
= 1.2
P = 0.3
4 ± 1
3-7
n = 30
4 ± 1
3-5
n = 10
5 ± 2
3-7
n = 10
5 ± 1
3-6
n = 10
F
1, 73
= 18.2*** t = -3.7, P = 0.001 t = -1.6, P = 0.1
t = -2.5, P = 0.02
82
Figure 1. Black arrows indicate eight lines of arrested growth (LAG) visible in a
phalangeal cross-section of an adult Pyxicephalus adspersus. These LAG, we assumed,
were each deposited during 6-8 months of winter torpor and are numbered consecutively
from the endosteal bone outwards. The first LAG has been partially resorbed. White
arrows point to comparatively feint, incomplete LAG within bands of summer bone
growth. These LAG were likely deposited during extended periods of rest in summer
between bouts of animal activity following rainfall.
83
Figure 2. Scatterplots showing the relationship between A) snout-vent length (SVL) and
body mass, B) the estimated age, and SVL; or the size category, and estimated age of C)
male (n = 46), and D) female (n = 30) adult Pyxicephalus adspersus from three peri-
urban breeding sites. Age was estimated as the number of lines of arrested growth (LAG)
counted in cross-sections of animal phalanges. Circular, triangular, or square data points
represent animals sampled at Diepsloot, Glen Austin or Bullfrog Pan, respectively. White
or light grey data points represent small or large females, respectively. Black, white with
a shadow, or dark grey data points represent small, medium, or large males, respectively.
Data points with an extra outline represent more than one individual. Solid regression
lines pertain to males () or females () from all three sites combined.
84
Figure 3. Frequency distribution of the estimated age of randomly selected adult male (n
= 45) and female (n = 30) Pyxicephalus adspersus from three peri-urban breeding sites.
Age was estimated as the number of annual lines of arrested growth (LAG) counted in
cross-sections of animal phalanges.
85
Figure 4. Frequency distribution of the snout-vent length (SVL) or body mass of male or female Pyxicephalus adspersus
caught during spawning events at Glen Austin Pan in 1992-93 by Cook (1996) or in 2004-06 for this study.
86
Chapter 5
Conservation implications of giant bullfrog (Pyxicephalus adspersus)
population genetic structure in Gauteng Province, South Africa
Caroline A. Yetman
1
& J. Willem H. Ferguson
2
2
Centre For Environmental Studies,
1,2
Department of Zoology and Entomology,
University of Pretoria, Pretoria, 0002, South Africa
Abstract
.
The giant bullfrog (Pyxicephalus adspersus) is regarded as Near-Threatened
in South Africa where many populations in Gauteng Province have been destroyed or
remain increasingly threatened by habitat loss and other threats. As a first step towards
identifying conservation management units for P. adspersus, we quantified genetic
structure and gene flow for populations from 23 localities in Gauteng and seven
additional localities in the north-eastern interior of South Africa, using 708 base pairs of
the mitochondrial cytochrome b gene. Gene flow was limited between populations > 200
km apart in the north-eastern interior of South Africa, reflecting genetic differentiation at
this scale (F
ST
= 0.60). Populations in the Free State Province may represent an
evolutionary significant unit of P. adspersus. In Gauteng, substantial gene flow between
populations < 20 km apart was detected, and effective population size estimates were
high. However, given recent male counts at remaining breeding sites in central Gauteng,
87
it appears that P. adspersus has declined by > 90 % in this area. The lack of correlation
between genetic and geographic distance of samples suggested that the genetic
differentiation in P. adspersus between the central, eastern and northern regions of
Gauteng (F
ST
= 0.26), and between Diepsloot, Glen Austin and Monavoni in central
Gauteng (F
ST
= 0.15) was due to genetic drift. The latter is possibly explained by reduced
gene flow between P. adspersus populations with expansion of the Pretoria and
Johannesburg metropolitan areas since the early 1900s. In northern Gauteng, where there
has been considerably less human habitat transformation, no genetic differentiation
between P. adspersus populations was found. Conservation of P. adspersus in South
Africa’s highly threatened Grassland biome is considered a priority and should involve
separate protection of populations in Gauteng and the Free State Province, where P.
adspersus is, respectively, highly threatened and genetically unique.
88
Introduction
Species populations lose potential to adapt to changing environmental conditions and
may suffer from inbreeding depression when they become small and isolated (Reed &
Frankham 2003; Willi et al. 2006). Inbreeding depression results because small, isolated
populations lose genetic variation as the relative influence of genetic drift exceeds that of
natural selection (Allendorf & Leary 1986; Rowe & Beebee 2003; Frankham 1995). The
rate at which genetic variation is lost depends on a population’s effective size, which is
equal or less than the total number of breeding individuals. A population’s effective size
can be considerably smaller than its actual size if breeding individuals exhibit a skewed
sex ratio or high variance in progeny production, or if population size varies greatly
between generations (Lande 1993; Charlesworth 2009). It is therefore desirable to
maintain high genetic variation in species by maintaining a number of large, well-
connected breeding populations (Lande & Barrowclough 1987; Lande 1988).
Of the five Vertebrate classes the Amphibia has the greatest proportion (32.5%) of
species that are globally threatened (Stuart et al. 2004; Beebee & Griffiths 2005).
Relative to other vertebrates that are less moisture dependent and more vagile,
amphibians are particularly vulnerable to habitat loss and climate change (Cushman
2006; Wake & Vredenburg 2008). Due to their permeable bodies and aquatic larval life
stages amphibians may also be especially sensitive to environmental contamination and
UV-B radiation (e.g. Blaustein et al. 1994; Relyea 2005; but see Crump et al. 1999 and
Kerby et al. 2010). Amphibian survival and breeding are known to fluctuate greatly in
response to e.g. weather variation (Seppä & Laurila 1999; Marsh 2001). Amphibian
89
populations are susceptible to population bottlenecks and genetic drift (Lande 1993),
caused by skewed sex ratios (Alho et al. 2008), short life expectancies (Monnet & Cherry
2002) and small effective population sizes (Funk et al. 1999). As a result of low genetic
variation, reduced fitness has been detected in various amphibian populations (Rowe &
Beebee 2003; Andersen et al. 2004; Pearman & Garner 2005).
The giant bullfrog (Pyxicephalus adspersus) is a large, aggressive anuran that breeds
explosively in shallow, seasonal wetlands following heavy rainfall in summer (Balinsky
& Balinsky 1954; Cook 1996). The species is widespread in southern Africa (Channing
2001; Du Preez & Carruthers 2010) but is considered to be Near-Threatened in South
Africa (Minter et al. 2004), where estimated population declines of between 50 and 80%
have been reported (Harrison et al. 2001). During the past two decades many P.
adspersus populations have been destroyed or remain increasingly threatened by habitat
loss and other factors in Gauteng Province (Carruthers 2007), South Africa’s economic
centre. Although Gauteng covers only 1.4% of the land area in South Africa, it has the
highest provincial human population growth rate and is experiencing rapid urban
development even outside the formal urban edge (GDACE 2004).
Unfortunately, in situ conservation of P. adspersus in Gauteng has been limited to
physical protection of the populations concentrated around Glen Austin and Bullfrog pans
near Johannesburg. Bullfrog Pan has a history of chemical contamination from a nearby
landfill, which may explain why juvenile and adult P. adspersus at this site have,
respectively, exhibited deformities and poor body condition (Slater-Jones 1996; Yetman
90
et al., in press, [Chapter 4]). At Glen Austin Pan there has been a significant decline in
the body size of adult P. adspersus since 1992-93 (Yetman et al., in press, [Chapter 4]).
No doubt, protection of P. adspersus at these two breeding sites alone is unlikely to
ensure the long term persistence of this species in Gauteng (Hitchings & Beebee 1998;
Hamer & McDonnell 2008). With increasing isolation of these populations and loss of
remaining natural habitat in Gauteng, there is an urgent need to investigate the genetic
structure of remaining P. adspersus to evaluate the past connectedness, present viability
and future conservation management of populations in this province (Nunney &
Campbell 1993; Crandall et al. 2000; Sutherland et al. 2004). The objectives of this study
were, therefore, to:
quantify P. adspersus population genetic structure and gene flow in Gauteng
Province.
quantify P. adspersus population genetic structure and gene flow in other parts of
South Africa, where possible.
identify important, genetically unique P. adspersus populations in the study area.
infer consequences of the results on the conservation management of P.
adspersus.
Materials and Methods
Sampling
Sampling was performed during the 2003/2004-2005/2006 summer seasons but
opportunities to encounter specimens were limited due to the unpredictable and sporadic
91
activity of P. adspersus (Yetman & Ferguson 2011a, [Chapter 2]). A total of 129 samples
were collected (Fig. 1), of which 107 were obtained from 23 localities in Gauteng
Province. Of the remaining 22 samples we obtained nine from three localities in the Free
State Province, five from three localities in Mpumalanga Province, and eight from one
locality in Limpopo Province (Table 1). The mean minimum distance between all 30
localities was 32 km (range: 2-186 km). Samples constituted tail clips from tadpole (n =
7) or toe clips from froglet (n = 12) or adult (n = 110) specimens found live or dead.
Adult specimens were targeted as far as possible to minimize the likelihood of sampling
siblings at a site. Live specimens were handled in a manner complying with the
“Guidelines for Use of Live Amphibians and Reptiles in Field Research” (Society for the
Study of Amphibians and Reptiles, the American Society of Ichthyologists and
Herpetologists, and The Herpetologists’ League). Samples were preserved in 70-99%
ethanol or frozen dry at -70º C prior to DNA extraction.
DNA extraction, amplification and sequencing
Total genomic DNA was extracted from each sample using the High Pure PCR Template
Preparation Kit (Roche Diagnostics). To increase the concentration of eluted DNA, half
(100 µ l) of the standard volume of elution buffer was used. A 708 base pair segment of
the mitochondrial gene cytochrome b was amplified by polymerase chain reaction (PCR)
using the protocol and primers L14841 and CB3-H developed by Mausfeld et al. (2000)
to analyze Scincid lizard populations. For amplification, 100 ng extracted DNA was
added to a 50 µl reaction containing 20 pmol of each primer (Integrated DNA
Technologies), 10 mМ dNTPs, and 10X buffer in the presence of 1 unit of BIOTOOLS
92
DNA polymerase (BIOTOOLS B&M Labs). The thermo-cycling profile entered into a
Perkin Elmer Gene Amp 2 400 thermocycler (Applied Biosystems) included an initial
denaturing step at 96°C for 30 s, followed by 34 cycles of denaturing at 95°C for 90 s,
annealing at 50°C for 60 s, and extension at 72°C for 90 s, with a final elongation step at
72°C for 50 s.
PCR product (5 µl) was combined with 2 µl loading dye, electrophoresed on a 1.5%
agarose gel containing ethidium bromide or GoldView Nucleic Acid Stain (SBS
Genetech), and the quality of the product visually assessed under a UV light. PCR
product was purified using the High Pure PCR Product Purification Kit (Roche
Diagnostics). For cycle sequencing (in the forward and reverse directions), depending on
the quality of purified PCR product, 2-6 µ l was added to a 10 µl reaction containing 1 µl
5X buffer, 3.2 pmol primer, 2 µl BigDye v.3.1.1 (Applied Biosystems), and 0-4 µl water.
The cycle sequencing thermo-cycling profile included initial denaturation at 96°C
followed by 25 cycles of denaturing at 96°C for 10 s, annealing at 50°C for 5 s, and
extension at 60°C for 4 min. Cycle sequencing product was precipitated using a 70%
ethanol sodium acetate method, and sequenced on an Applied Biosystems 3 130xl
Genetic Analyzer.
Sequence chromatograms were edited, aligned, and the forward and reverse sequences for
each sample converted into one contiguous sequence (contig) using MEGA 4.0.1
(Tamura et al. 2007). The contigs were deposited in the GenBank database under
accession numbers FJ613 658-786. Sequences for two outgroup species were obtained
93
from GenBank for Rana perezi and Rana nigromaculata (accession numbers DQ902146
and DQ006267). (Comparative sequence data for suitable African anuran species were
not available at the time).
Spatial scales of analyses
Population genetic structure and gene flow were estimated at various spatial scales as
encouraged by Boyd et al. (2008). These scales were: i) the north-eastern interior of
South Africa; ii) Gauteng Province; and iii) two local spatial scales in Gauteng (Fig. 1;
Table 1). For the north-eastern interior of South Africa localities from the Free State,
Gauteng plus Mpumalanga, and Limpopo provinces were separately grouped for analysis
given the large (200-400 km) distances separating these groups. In Gauteng Province
localities from the central, eastern and northern regions of this province (20-100 km
apart) were separately grouped for analysis based on differences in climate and habitat
between these regions (Table 1). While northern Gauteng is characterized by warmer
savanna, the rest is characterized by cooler grassland (Mucina et al. 2005), and in eastern
Gauteng there is an unusually high density of pans and other wetlands where P.
adspersus may breed (GDACE 2004). Samples from Diepsloot, Glen Austin and
Monavoni (10-15 km apart; Table 1; Fig. 1) were used for the local spatial scale analysis
in central Gauteng, and samples from Buffelsdrif, Wallmannsthal and Hammanskraal (7-
19 km apart; Table 1; Fig. 1) were used for the local spatial scale analysis in northern
Gauteng (Table 1). Samples from other localities in these two sub-regions of Gauteng
were too few for inclusion in these local spatial scale analyses.
94
Analysis methods
An unrooted haplotype network of the 129 P. adspersus contigs (hereon referred to as the
dataset) was generated using TCS 1.21 (Clement et al. 2000). A neighbour-joining
dendrogram based on p-distances among all haplotypes in the dataset plus the two
outgroup species was constructed from 1 000 bootstrap replicates in MEGA 4.0.1
(Tamura et al. 2007).
At each spatial scale, comprising a separate analysis, haplotype frequency variation and
differentiation within and among populations was quantified in Arlequin 3.11 (Excoffier
et al. 2005) from 1 000 permutations. Since the indirect estimates of migration rate from
F
ST
involve often-unrealistic assumptions (e.g. equal-sized populations, equal rates of
inter-population gene flow, etc.; Whitlock & McCauley 1999), migration rates and
effective population sizes here were estimated directly using the maximum likelihood
estimate (MLE) approach of Beerli & Felsenstein (1999, 2001), implemented in Migrate
3.0.3.
To increase the probability of estimating
θ and M more accurately, for each spatial scale
five analyses with different settings were tested. The first analysis was based strictly on
the default MLE “search” settings in Migrate 3.0.3. The second analysis matched the first
except that the number of short and long sampling chains was doubled to 20 and 6,
respectively. The third and fourth analyses matched the second except that an adaptive
heating scheme was implemented, which involved four chains with starting temperatures
ranging logarithmically from 1 to 10, and 1 to 10 000, respectively. The fifth analysis
95
matched the fourth except that the heating scheme used 10 (not four) chains. Each
analysis was repeated five times using different automatically-generated seed numbers.
From the 25 analyses for each spatial scale the mean value of the five most consistent
estimates of each parameter was used. Effective population size was calculated from θ
using a mitochondrial nucleotide mutation rate of µ = 15.43 x 10
-9
base substitutions per
site per year as reported for amphibians and reptiles by Lynch et al. (2006).
A Mantel test (of 1 000 permutations) was performed in Arlequin 3.11 to establish
whether sampled P. adspersus populations showed isolation by distance. The 30 sampled
localities were split into five geographical groups including Limpopo Province, Free
State Province, central Gauteng, northern Gauteng, and eastern Gauteng plus
Mpumalanga Province. Geographic distance between group centroids was calculated
using Hawth’s Analysis Tools 3.26 (© 2002-06) in ArcMap 9.2 (© ESRI, Inc. 1999-
2006). Genetic distance was calculated in Arlequin as F
ST
/(1 - F
ST
).
Results
Eighteen single-base pair (bp) polymorphisms were found in the 708 bp region of
cytochrome b sequenced for all 129 samples. The mean base composition of the 129 P.
adspersus contigs was A = 23.8%, C = 34.0%, G = 16.1% and T = 26.1%, with a
transition/transversion ratio of 8.5.
96
Of the 15 haplotypes in the unrooted network (Fig. 2), 10 (67%) were defined by only
one bp polymorphism. The most common haplotype among samples in the study is
labelled “Haplotype 1,” which was found in Gauteng, Mpumalanga and Limpopo
provinces, and in all three sub-regions in Gauteng. It represents 83 (64%) of the 129
samples in the dataset. Except for haplotypes 1 and 13 the remaining 13 (87% of)
haplotypes were each unique to a particular province (Fig. 2). Haplotypes 2-9, 10, 11-12
and 14-15 were respectively found exclusively in Gauteng, Mpumalanga, Limpopo and
Free State Province.
The neighbour-joining dendrogram comprises two main