A century of decline: loss of genetic diversity in a southern African lion-conservation stronghold

Aim There is a dearth of evidence that determines the genetic diversity of populations contained within present-day protected areas compared with their historic state prior to large-scale species declines, making inferences about a species’ conservation genetic status difficult to assess. The aim of this paper is to demonstrate the use of historic specimens to assess the change in genetic diversity over a defined spatial area. Location Like other species, African lion populations (Panthera leo) are undergoing dramatic contractions in range and declines in numbers, motivating the identification of a number of lion conservation strongholds across East and southern Africa. We focus on one such stronghold, the Kavango-Zambezi transfrontier conservation area (KAZA) of Botswana, Namibia, Zambia and Zimbabwe. Methods We compare genetic diversity between historical museum specimens, collected during the late 19th and early 20th century, with samples from the modern extant population. We use 16 microsatellite markers and sequence 337 base pairs of the hypervariable control region (HVR1) of the mitochondrial genome. We use bootstrap resampling to allow for comparisons between the historic and modern data. Results We show that the genetic diversity of the modern population was reduced by 12% to 17%, with a reduction in allelic diversity of approximately 15%, compared to historic populations, in addition to having lost a number of mitochondrial haplotypes. We also identify reduced allelic diversity and a number of ‘ghost alleles’ in the historical samples no longer present in the extant population. Main Conclusions We argue a rapid decline in allelic richness after 1895 suggests the erosion of genetic diversity coincides with the rise of a European colonial presence and the outbreak of rinderpest in the region. Our results support the need to improved connectivity between protected areas in order to prevent further loss of genetic diversity in the region.


Introduction
6 historical lion samples taken from museum specimens in order to compare historic levels of genetic 110 diversity against modern levels from the same region. We used a suite of microsatellite markers as 111 well as sequencing of part of the hypervariable control region (HVR1) of the mitochondrial genome 112 (mtDNA) to assess the degree to which genetic diversity in this population has been lost as a result 113 of regional declines in lion numbers and distribution. 114

Methods 115
Samples 116 The Natural History Museum of London's collections contain large numbers of lion skins and skulls 117 from across the species range. The labelling of the collection data was of varying quality so 118 specimens were cross-referenced with collector catalogues wherever possible. Twenty-seven lion 119 specimens were sampled, originally collected from within the study region between 1879 until 1935 120 (Table 1, Fig. 1). Scrapes of any tissue remaining on the skulls or skin, or fragments of detached 121 maxilloturbinal bone (thin bones inside the nasal cavities) were collected from each specimen. 122 Modern samples were collected from 204 free ranging wild lions between 2010 until 2013 ( Fig. 1) in 123 the form of blood (n=23), fresh tissue (n=113), dry tissue (n=13), faecal (n=14) and hair-pulls (n=41). 124 Fresh tissue samples were collected using a remote biopsy dart system (Karesh et al., 1987). Hair 125 pulls and blood were taken from immobilised animals. Dry tissue samples were taken from animals 126 shot by the trophy hunting industry. 127 128 Ancient DNA precautions 129 All pre-PCR work was performed in a laboratory exclusively devoted to ancient DNA, situated on a 130 different floor from the PCR amplification laboratory and with an independent air handling system. 131 To avoid sample cross-contamination a different set of equipment was used for each extraction (e.g. 132 mortar and pestle, scalpel blades etc). Single-use equipment was immersed in sodium hypochlorite 7 and removed from the working area after use. The working area was cleaned with sodium 134 hypochlorite solution before work on the next sample commenced. All equipment was UV-irradiated 135 overnight prior to further use. Filter tips were used to reduce cross contamination (Rohland & 136 Hofreiter, 2007). Two blank extractions containing no tissue or bone were included during both 137 extraction protocols to serve as negative extraction and PCR controls. Each fragment was 138 independently amplified by PCR at least three times following the multi-tube approach (Taberlet et 139 al., 1999) in an attempt to detect contamination and genotyping errors. 140

DNA extraction 142
Total genomic DNA was extracted from each museum skin sample using approximately 25mg of 143 tissue using DNeasy ® Blood and Tissue kits (Qiagen). We followed the manufacturer's instructions 144 but added a second incubation. To increase tissue lysis the first incubation was run overnight, for the 145 second digestion we added a further 180µl Buffer ATL and 20µl proteinase K (600mAU/ml) and then 146 incubated for a further 3 hours at 56°C. 147 DNA from bone samples was extracted using approximately 100mg of bone powder previously 148 ground in a pestle and mortar. A master mix was prepared which, for each sample, comprised of 149 0.2ml 10% SDS (Invitrogen), 0.15ml proteinase K at 15mg/µl, a 1x1mm piece of DTT at 10mM and 150 1.65ml EDTA of pH 8.0 at 0.5M. This was warmed at 56°C until all ingredients dissolved and added to 151 each bone-powder sample. Samples were incubated on a rotator at 56°C for 48 hours. Following 152 digestion, tubes were centrifuged for one minute at 1300rpm and supernatant transferred to an 153 Amicon ® Millipore Ultra Centrifuge filter which was centrifuged for 30 minutes at 1300rpm. A 154 MinElute purification kit (Qiagen) was used to purify 100µl of extract following the manufacturer's 155 instructions, washing three times with PE buffer.

8
Modern DNA was extracted using approximately 25mg of tissue, 100μl of raw blood or 5-6 hair 157 follicles using DNeasy ® Blood and Tissue kits (Qiagen) according to the manufacturer's instructions. There is an inherent inability to control the sampling design when using museum collections, 198 including sample size, date and location of their collection. To allow comparisons between modern 199 and historic nucleic diversity we used a bootstrapping procedure. When analysing the more rapidly 200 mutating nuclear microsatellite data, we progressively restricted; i) the spatial extent of the historic 201 samples, to match with more certainty the extent of the modern samples; ii) the time period over 202 which the historic samples were collected, to restrict the possible influence of genetic drift with time 203 within the sample set. Thus, we divided our historic data into three spatial zones representing; I) the 204 samples within the modern sampling area; II) the samples likely to be within male dispersing 205 distance of the modern sampling area, taken as 200km; III) all remaining samples across the region 206 (Table 1). We also divided the historic data into two time periods, 1874-1895 (A) and 1929-1935 (B) 207 (Table 1). The results from the historic samples sets were compared against our modern dataset 208 using a bootstrapping procedure implemented in POPTOOLS (Hood, 2011). We created 100 209 populations of equal size to the historic data being used. Furthermore, to account for an apparent Mitochondrial 'ghost' alleles 220 Following the identification of all haplotypes present in the combined modern and historic data set, 221 we were able to assess private haplotypes only present in one or other time period. Due to the much 222 poorer quality of the museum sample data many sequences were considerably shorter than the 223 modern counterparts, making direct comparisons of diversity difficult and lacking power. However, 224 we were able to identify haplotypes only present in the historic data, likely to have been lost from Across the data we identified 29 alleles present only in the historic samples and 54 private alleles 249 only found in the modern data, however the latter come from a much larger data set. The mean 250 number of private alleles is consistently higher in the historic data than in the modern data when 251 controlling for sample size (Fig. 2) The mtDNA data (Table 3) indicates six haplotypes present within the historic dataset (H = 0.6993, π 266 = 0.00065), but three of these appear to be missing from the extant lions (H = 0.3257, π = 0.0007). 267 Tajima's D for both the historic (D = -1.09629; p < 0.1) and modern (D = -0.53568, p < 0.1) population 268 are negative but not significant, suggesting no deviation from neutrality. Aside from the three 'ghost' 269 haplotypes identified, there may be others present within the same mtDNA region that due to the 270 degradation of the historic DNA remain unidentified. Since two of the 'ghost' haplotypes were 271 identified from single individuals, each only with a single nucleotide insertion, we must caution that 272 they may be false haplotypes caused by DNA degradation (Wandeler et al., 2007). Even following a 273 more conservative approach, one previously common haplotype remains unrepresented in the 274 modern samples. 275

Discussion 277
The value of genetic diversity is increasingly recognized for contributing to individual fitness, species' 278 evolutionary potential, and ecosystem function and resilience (Whitham et al. 2008). There is 1897 (Parsons, 1993). Furthermore, modern firearms became more prevalent following European 314 settlement and predators were often persecuted as vermin (Woodroffe, 2000), which likely 315 contributed to the earlier decline of lions in the study region. Whilst the timing of genetic decline 316 and colonial settlement is compelling enough to suggest causation, the evidence is not conclusive. imperative that efforts are made to conserve genetic diversity. Without such genetic diversity, a 325 species resilience and ability to adapt to future stochastic events becomes greatly compromised 326 (Whitham et al. 2008). This study provides quantitative data on temporal genetic monitoring that is 327 urgently needed to optimize conservation and management efforts. Since KAZA is considered one of 328 the more stable lion populations in Africa, the work presented here should provide motivation for 329    Table 3 Mitochondrial DNA control region haplotypes from historical specimens and the extant lion population of the KAZA 496 region. "-" and "N/A" represent a deletion or missing sequence data, respectively, at the specified nucleotide position.  Historic  221  343  367  368  378   31  5  i  --T  A  -9 ii