Effect of Fire on the Herpetofauna of the Koanaka Hills , Ngamiland , Botswana

Ngamiland is one of the most remote regions in Botswana, and its herpetofauna is largely under-surveyed. This study documents the herpetofauna of the Koanaka Hills (KH) in Ngamiland in 2009 following extensive fire destruction and compares it to the pre-fire herpetofauna collected in 2008. We also provide new records for the region for three amphibian and six reptile species, and document vouchers for two taxa that were sighted but not collected in 2008. During 2009, 14 reptile and three amphibian species were collected, bringing the total number of confirmed herpetofaunal taxa near the KH to three amphibian and 19 reptile species. For seven species this is the first published occurrence in quarter degree square 2021 Aa. Analyses measuring changes in the KH herpetofauna following the fire are inconclusive due to differences in collection effort and weather conditions. However, these data suggest that fire impact was minimal.


Introduction
The herpetofauna of Botswana is poorly studied when compared to that of adjacent South Africa, Namibia, and Zimbabwe.In 1856, Wahlberg was the first to comment on the country's herpetofauna, noting the presence of a puff adder (Bitis arietans) near Lake Ngami (Wahlberg 1994).However, the herpetological literature pertaining to Botswana as a whole is limited.Botswana is included in several guides and general publications about the herpetology of southern Africa (e.g., FitzSimons 1943;Visser 1984;Broadley 1990;Branch 1998;Channing 2001;Alexander and Marais 2007;Du Preez and Carruthers 2009), but the only books specific to the herpetofauna of the country are Auerbach (1987) and Clauss and Clauss (2002).In addition, a relatively small number of technical publications have addressed the herpetofauna of selected regions or dealt with selected taxonomic groups (e.g., FitzSimons 1935a, b;Haacke 1966;McLachlan 1981;Auerbach 1984Auerbach , 1985Auerbach , 1986;;Butchart 2000;Bauer et al. 2009).The paucity of published data on the herpetofauna of Botswana makes any new contributions especially critical in gaining a better understanding the country's reptile and amphibian fauna.
In 2008, the then-recent discovery of microvertebrate fossils in a Pleistocene cave system in the Koanaka Hills of northwestern Botswana (Kennedy et al. 2010) in addition to the poor understanding of the region's modern fauna prompted a collecting trip to document the area's amphibians and reptiles and to create a preliminary comparative collection for the interpretation of the fossil herpetofauna (Bauer et al. 2009;Ferguson et al. 2010).Three months following the initial collecting trip in the Koanaka Hills, the region was consumed by a concentrated bush-fire (M.Gabadirwe, pers. com.).A follow-up survey of the area was undertaken 13-29 July 2009 to document potential changes in the fauna following the fire.An additional objective of the second survey was to provide Abstract: Ngamiland is one of the most remote regions in Botswana, and its herpetofauna is largely under-surveyed.This study documents the herpetofauna of the Koanaka Hills (KH) in Ngamiland in 2009 following extensive fire destruction and compares it to the pre-fire herpetofauna collected in 2008.We also provide new records for the region for three amphibian and six reptile species, and document vouchers for two taxa that were sighted but not collected in 2008.During 2009, 14 reptile and three amphibian species were collected, bringing the total number of confirmed herpetofaunal taxa near the KH to three amphibian and 19 reptile species.For seven species this is the first published occurrence in quarter degree square 2021 Aa.Analyses measuring changes in the KH herpetofauna following the fire are inconclusive due to differences in collection effort and weather conditions.However, these data suggest that fire impact was minimal.

Data Collection
New collections were made primarily within a 5 km radius of Koanaka South and a 1 km radius of Koanaka North.However, one additional collecting day was spent at the nearby Gcwihaba Hills.All collections were made between 881 and 993 m asl.Collecting teams of two to four people concentrated on rocky outcrops of the hills and in the dry pans west of Koanaka South.Searches were made during both day and night, although rapid temperature drops after sunset precluded most nocturnal reptile activity.As a result, most specimens were caught from 9:00 h to 16:30 h.
Most specimens were collected opportunistically by hand.However, a few specimens were collected using pitfall and funnel traps.Two arrays of drift fences with pitfall traps were set near the base of Koanaka South in the scrub brush.One array was Y-shaped with a pitfall trap on the end of each arm and one in the center.Funnel traps were place in the center of each arm on the Y-shaped array.The second line was a straight fence with three pit-fall traps.Pitfall traps within each array were placed approximately three meters apart.Each array was built with 0.5 m tall drift fence made out of black plastic sheeting and wooden stakes, and five liter buckets were used for the pitfall traps.Double-sided funnel traps were made out of fine mesh wire sheets.The arrays were checked twice a day, once just after sunrise and once just before sunset.Unless otherwise noted below, all vouchered specimens were collected by hand.
Specimens were euthanized with intraperitoneal injections of sodium pentabarbatol.Samples of either liver or tail tissue were taken from each specimen for future DNA sequencing.When present, stomach contents and parasites also were collected.The first four specimens of each taxon collected were prepared as osteological specimens and associated skins were preserved in 70% ethanol.Every fifth specimen of a taxon collected was fixed in 10% formalin in the field and later transferred to 70% ethanol.All material was deposited in either the Texas Natural History Collections (TNHC) at The University of Texas at Austin, Austin, Texas, USA or the Botswana National Museum (BNM), Gaborone, Botswana.Specimens deposited at BNM are still awaiting catalog numbers so have been listed here with field numbers (KHH).
Our 2009 records presented here are supplemented by data from the 2008 collections and a small collection made by Wulf D. Haacke at the Koanaka Hills in 1965 as previously published by Bauer et al. (2009).Our list of vouchered species includes all species represented by specimens actually collected near the Koanaka Hills.Species not collected but expected to occur in the Koanaka Hills (based on literature and museum records) are listed in Table 1.
Specimens were collected under permits CHA 1/17/2 IX (64) and CYSC 1/17/21 (81) issued by the Botswana Ministry of Youth, Sport and Culture to P. J. Lewis and imported into the United States under permits from the U.S. Centers for Disease Control and Prevention, U.S. Fish and Wildlife Service, and U.S. Department of Agriculture APHIS Permit 105705 issued to M.L. Thies.Species identifications followed standard regional references (e.g., Auerbach 1987;Branch 1998;Alexander and Marais 2007;Du Preez and Carruthers 2009).Amphibian taxonomy follows du Preez and Carruthers (2009).Reptile taxonomy follows Uetz ( 2012) and the primary sources cited therein, except for the recognition of family level taxa within the Elapoidea, for which we follow Kelly et al. (2008Kelly et al. ( , 2009)).

Data Analysis
To test whether the herpetofauna near the Koanaka Hills changed following the concentrated bush-fire in the region in late 2008, species richness, composition, and diversity were compared between the 2008 and 2009 collections.However, as described in detail in the discussion, sampling effort during the 2009 collecting season was greater than that in 2008.Therefore, these comparisons must be evaluated in this light.
Species Richness Estimation: Rarefaction curves estimate true species richness through the random resampling of the given data (Lyman 2008).To detect a change in species richness after the fire, we compared Chao 2 rarefaction curves for each year.Chao 2 is particularly useful for calculating species richness based on small sample sizes (Ugland et al. 2003).Chao 2 values were calculated in EstimateS and graphed in Excel.The resulting rarefaction curves were then visually analyzed for overlap.Any overlap in rarefaction curves suggests a non-significant difference in the faunas, whereas no overlap indicates a significant difference.
Species Composition: Because the goal of this study was to measure the differences between years, we calculated the Jaccard dissimilarity index (J = C / (A + B -C); where: A = number of taxa in fauna A; B = number of taxa in fauna B; C = number of taxa common to both faunas).That index emphasizes differences in taxonomic composition (Lyman 2008).The Jaccard index results in a number ranging 0-1, with 1 implying that the taxon compositions are completely different and 0 implying that they are identical (Lyman 2008).
Taxonomic Diversity: For analyzing the differences in taxonomic diversity, the mean Shannon Diversity index (S = -∑ P i (ln P i ); where: P i = proportion (P) of taxon i in the fauna) for each year was calculated and statistically compared using a Student's t-test in the statistics program JMP.To establish the variation of the Shannon indices within years, a Shannon index was calculated for every three collecting days in EstimateS.

Species Accounts
The species observed at the Koanaka Hills are listed alphabetically within families and genera.Specimens collected during the 2009 survey are noted under the heading 'Material.''Other records' refers to specimens collected in 2008 and 1965 (Bauer et al. 2009).Comments under the heading 'Location and Habitat' relate to the specific habitat where the specimens were collected, as well as the precise GPS coordinate data for that capture.Additional comments on the specimen(s) or collection is noted under heading 'Comments'.

Ptychadenidae
Ptychadena subpunctata (Bocage 1866) Material.Thirteen specimens: TNHC 85033-45.Location and Habitat.All 13 specimens were collected by digging into the dry pan west of Koanaka South at 20°09.20'S, 21°11.36'E.These specimens were buried less than 10 cm below the surface of the pan.The location was chosen when one individual was first noticed emerging from the soil after a slightly dewy morning.
Comments.These specimens represent the first published record of this species in QDS 2021 Aa.

Pyxicephalidae
Cacosternum boettgeri (Boulenger 1892) Material.Nine specimens: TNHC 84956-64.Location and Habitat.All nine specimens were collected by digging into the dry pan W of Koanaka South at 20°09.20'S, 21°11.36'E.These specimens were buried less than 10 cm below the surface of the dry pan at the same location where the Ptychadena subpunctata were collected.
Comments.These specimens represent the first published record of this species in QDS 2021 Aa.
Pyxicephalus adspersus (Tschudi 1838; Figure 3) Material.One specimen: TNHC 85046.Location and Habitat.This specimen was collected from the dry pan W of Koanaka South at 20°09.20'S, 21°11.36'E.It was buried approximately 20 cm below the surface in a state of torpor just below the same location that the other anurans were collected.

Comments.
This specimen represents the first published record of this species in QDS 2021 Aa.
Location and Habitat.TNHC 84949-52 were collected in a dry pan at 20°09.20'S, 21°11.36'E.TNHC 84953 was collected on a dirt track between the Koanaka and Gcwihaba Hills.
Location and Habitat.The remains of this specimen were collected from the stomach of a honey badger, (Mellivora capensis), which was trapped near Koanaka South.Due to the fossorial tendencies of both the prey and predator, it is presumed that this specimen was captured underground.
Location and Habitat.This species was found on dolomitic rocks at the entrance to the cave system in Koanaka South (Bauer et al. 2009).
Comments.This specimen represents the first published record of this species from QDS 2021 Aa.
Other records.This highly vocal species was heard calling sporadically during the 2008 survey, but was not collected (Bauer et al. 2009).
Location and Habitat.This specimen was collected in a burrow in tall grass, on loose sand substrate, 50 m west of Koanaka South at 20°09.30'S, 21°11.34'E.
Comments.Although previously published by Bauer et al. (2009) as present in the area based on sound records, this specimen represents the first vouchered record of this species in QDS 2021 Aa.

Ichnotropis squamulosa Peters 1854
Material.No specimens were collected in the 2009 survey.
Location and Habitat.In 2008 this taxon was found darting in and out from under bushes in the sandveld area surrounding the dry pan near Koanaka South (Bauer et al. 2009).
Location and Habitat.All but one (TNHC 85015) P. namaquensis were active in and around the bases of shrubs in dry pans at 20°09.20'S, 21°11.36'Eand 20°08.34'S,21°11.41'E.TNHC 85015 was found in a sandy patch west of one of the pans at 20°09.03'S, 21°10.30'E.
Comments.These specimens represent the first published record of this species in QDS 2021 Aa.
Comments.These specimens represent the first published record of this species in QDS 2021 Aa.
Location and Habitat.All specimens collected in 2009 were found on Koanaka South running in and out of crevices between the dolomitic rocks.
Other records.B. Williams (pers.com.) reported observing this species in a cave entrance on Koanaka South in 1996 (Bauer et al. 2009).
Comments.These specimens represent the first published record of this species in QDS 2021 Aa.
Other records.This species was listed as confirmed in the region by a sight record in 2008 (Bauer et al. 2009).
Location and Habitat.In 2008 this species was spotted moving down the slope of Koanaka South (Bauer et al. 2009).
Habitat.In 2008 one specimen was found basking in a dry pan near Koanaka South (Bauer et al. 2009).

Faunal Dynamics Analyses
Species Richness.A graphical representation of the Chao 2 rarefaction curves shows that they do not overlap with increasing samples (Figure 10).Rarefaction curves are most informative at the point of the highest sample.Therefore, a non-overlap of the upper bound 95% confidence interval in 2008 with the lower bound 95% confidence interval in 2009 suggests significantly higher potential species richness in 2009.Species Composition.A Jaccard dissimilarity index value of 0.3 indicates that in 2009 the faunal data were similar to those of 2008 that were collected before the Koanaka Hills were subjected to extensive bush fire.
Diversity Dynamics.The differences in the mean Shannon indices were not significant (p=0.49),indicating a similar diversity of herpetofauna between 2008 and 2009.
Ngamiland is one of the most remote regions in Botswana, and has been severely under-collected in general.One of only two full-length references on Botswanan herpetofauna (Auerbach 1987) references no amphibians and only four reptiles within QDS 2021 Aa.The paucity of collections may be attributed chiefly to the difficulty of access, although it was suggested by Bauer et al. (2009) that the herpetofauna of the Koanaka Hills is naturally less diverse than the herpetofauna of many other southern African localities.This is attributed to the region's relative lack of significant rocky habitat, which is favored by many southern Africa reptile taxa (Bauer 1999;Bauer et al. 2009).Furthermore, the arid nature of the Kalahari presents a challenge for any amphibians in the area due to their dependence on water for reproduction.
Our new collection in the Koanaka Hills supplements the previously published inventory by Bauer et al. (2009) who collected nine reptile species.Bauer et al. (2009) mentioned four additional species based on previous collections in 1965 (Heliobolus lugubris), sound records (Ptenopus garrulus), and sight records (Dendroaspis polylepis, and Psammophis trinasalis).We collected three additional amphibian and six additional reptile species in 2009, including two that were reported but not collected in 2008 (Heliobolus lugubris and Ptenopus garrulus).Of the nine reptiles collected in 2008, all but Chondrodactylus turneri, Ichnotropis squamulosa, and Bitis arietans were also collected in 2009.
In 2009, 109 specimens of 14 reptile species and three amphibian species were collected.This brings the confirmed herpetofauna near the Koanaka Hills to 22 species (three amphibians, 19 reptiles), including the two snakes identified by sight in 2008 (Bauer et al. 2009).For seven of those species, this report constitutes the first published record of occurrence in quarter degree square 2021 Aa.Despite the low diversity in the region, the probable species list for QDS 2021 Aa includes 15 frogs (up from an earlier estimate of six, Bauer et al. 2009) and 46 reptiles (Table 1).
Superficially, the analyses measuring the changes in the herpetofauna of the Koanaka Hills following a fire appear inconclusive.For example, comparisons of the species composition and diversity between years both suggest that the faunas were similar (using Shannon and Jaccard indices).Conversely, the species richness analysis suggests higher species richness in 2009 following the fire (using Chao 2 rarefaction curves).While collection technique and location were the same both years, differences in collection effort and weather conditions between years might account for the species richness results.For example, in 2008 fewer people focused on collecting the herpetofauna (1-2 on most days in 2008, versus 2-4 on most days in  2009), and in 2009 there were six additional trapping days.Additionally, the weather was slightly warmer and wetter in 2009.During the 2008 collection daytime temperatures averaged 26°C, night time temperatures averaged 2°C, and there was no precipitation, whereas in 2009 daytime temperatures averaged 28°C, night time temperatures averaged 5°C, and light precipitation occurred one day during the collecting trip.Both higher temperatures and rainfall greatly increase the probability of encountering reptile and amphibian taxa in the dry, winter months.Since collection effort and weather conditions may account for the significant result in the species richness analysis, the fire impact on the herpetofauna of the Koanaka Hills was likely minimal.While the frequency of fires in the Koanaka Hills is currently unknown, given the prevailing weather conditions in the region it would not be surprising if such events were frequent.It is possible that the fossorial lifestyle of some taxa and the use of rocky or subterranean retreat sites by many species among the Koanaka herpetofauna offers some protection from several local threats and environmental extremes, including fire.

Figure 1 .
Figure 1.Map of Botswana showing the position of the Koanaka Hills in Ngamiland (Bauer et al. 2009).

Figure 2 .
Figure 2. Map of a portion of northwestern Botswana showing landmarks and place names mentioned in the text.Inset shows Koanaka North and Koanaka South in greater detail (Bauer et al. 2009).

Figure 9 .
Figure 9. Burnt tree beneath which one specimen Psammophis subtaeniatus was found.

Figure 10 .
Figure 10.Chao 2 rarefaction curves for 2008 and 2009 predicting species richness with increasing samples.The 2008 upper bound 95% confidence interval curve does not overlap with the 2009 lower bound 95% confidence interval curve, which indicates a significant difference in the species richness of the herpetofauna in 2008 and the herpetofauna of 2009.

Table 1 .
Checklist of observed and expected taxa at the Koanaka Hills.Confirmed species include those recorded from the Koanaka Hills region (a maximum of ~25 km distant from the Koanaka Hills): 1 = collected in 2009, 2 = collected in 2008, 3 = collected in 1965, 4 = unvouchered record.Unless noted by an asterisk (*), taxa in the list of expected species were plotted with ranges including the Koanaka region by Branch (1998) or Du Preez and Carruthers (2009).Additional literature records including the Koanaka Hills in shaded range maps are also indicated: Visser 1984 (a), Auerbach 1987 (b), Broadley 1990 (c), Carruthers and Du Preez 2011.(d), Channing 2001 (e), Clauss and Clauss 2002 (f), and Wüster and Broadley 2007 (g).