Computational protocol: Population Genetic Insights Into the Social Organization of Guinea Baboons (Papio papio): Evidence for Female-Biased Dispersal

Similar protocols

Protocol publication

[…] Obtaining accurate microsatellite genotypes from fecal samples can be difficult due to low DNA quality and quantity or poor extract quality (PCR inhibitors) [Taberlet et al., ]. We, therefore, rigorously evaluated genotyping errors and only included samples that passed our quality control (further details can be found in the supporting information). Genotype matching was performed using Gimlet 1.3.3 [Valiere et al., ] allowing one mismatch. Every duplicate genotype was excluded from the final dataset. The probability that a single genotype actually represents one single individual was calculated with the probability of identity P(ID) [Paetkau & Strobeck, ] and the more conservative estimator Probability of Identity between sibs P(ID) sib [Evett & Weir, ; Taberlet & Luikart, ] as implemented in Gimlet. The final dataset was converted to the specific input file formats of each software program using Create 1.3 [Coombs et al., ].Departures from Hardy–Weinberg equilibrium (HWE) were tested with exact tests using the program Genepop 4.0.11 (default settings: dememorization number: 10,000; number of batches: 20; iterations per batch: 5,000) [Rousset, ; Raymond & Rousset, ]. Expected heterozygosity HE and observed heterozygosity HO were calculated in Arlequin [Excoffier & Lischer, ]. Allelic richness and FIS were calculated in Fstat [Goudet, ].Population genetic parameters were calculated to investigate whether there is any population structuring despite the fact that there are no obvious barriers for gene flow between the sampling sites. First, the program Structure 2.3.3 [Pritchard et al., ] was used, which is based on a Bayesian approach. It identifies the most likely number of populations (K) in a data set and the likelihood of an individual to belong to this population. Program settings were set to a total run length of 1,000,000 iterations, a burnin of 100,000, and values of K from 1 through 6. The analysis was repeated 10 times to assure the consistency of the results. We chose the admixture model as ancestry model and the correlated frequency model as allele frequency model [Falush et al., ]. Furthermore, we used the Locprior model that takes into account the sampling location of individuals as a prior information to assist the clustering if the signal is relatively weak [Hubisz et al., ]. All other settings were left at their default value. To evaluate the most probable number of clusters, we employed the method suggested by Evanno et al. [] as implemented in Structure Harvester Web v0.6.92 [Earl & VonHoldt, ]. To further investigate population structuring, Weir & Cockerham's fixation index FST [Weir & Cockerham, ] among the sampling sites was calculated in Fstat and the relationship between geographic and genetic distances among sampling sites (isolation by distance; IBD) was tested with a Mantel test in Genepop using 1,000 permutations.We tested for sex-bias in dispersal by comparing several parameters between males and females. To begin with, population structure and IBD of females and males were examined with the same settings as in the analysis of the total population. To quantify the degree of population structuring, FST values were calculated for each sex separately and tested two-sided predicting males being philopatric with 1,000 permutations using Fstat. Sampling sites Gue Damantan, Simenti, Camp Du Lion, and Lingue Kountou were grouped together as one cluster and Niokolo constituted a second cluster following the results from the population structure analysis. Allelic frequencies of the dispersing sex should be more homogeneous, and therefore FST should be lower for the dispersing than for the philopatric sex. We refrained from testing other parameters available in the sex-biased dispersal test in Fstat, on the one hand, to avoid multiple testing and on the other hand, because these parameters have been shown to perform poorly under certain conditions, whereas the FST statistic is the most powerful measure to detect sex-bias in dispersal, regardless of sampling scheme and magnitude [Goudet et al., ]. Sex-biased dispersal should also influence the distribution of relatedness in a population. Pairwise relatedness coefficients R were calculated using the regression estimator derived by Queller & Goodnight [Queller & Goodnight, ] as implemented in Coancestry 1.0 [Wang, ]. The average relatedness of males and females within a gang, among gangs, and among communities, respectively, was compared (for within gang comparisons only dyads in the Simenti community were included). We tested for significance using a permutation test as implemented in the R package coin [Hothorn et al., ] in R 3.1.1 [R Development Core Team, ] with 99,999 Monte Carlo resamplings. A set of 14 microsatellites does not suffice to infer kinship reliably without any additional information and putative misclassification would lead to erroneous conclusions [Van Horn et al., ]. With the absence of pedigree (e.g. known mother-offspring pairs) and demographic information [Arora et al., ; Harris et al., ], we therefore refrained from analyzing dyadic relatedness.To visualize the genetic distances and frequencies of HVRI haplotypes, we generated a haplotype network in HapStar 0.6 [Teacher & Griffiths, ] based on pairwise distances output from Arlequin. In order to assess the diversity of HVRI haplotypes, we calculated the levels of nucleotide and haplotype diversity for males and females, respectively, using DNaSP version 5.10.1 [Librado & Rozas, ], both for the whole study population and for every community separately, as well as for females and males, respectively. We tested for significance using the difference test in Statistica (StatSoft® (Europe) GmbH, Hamburg, German).To investigate the temporal stability of gangs, we examined whether individuals that were sampled multiple times on different days were repeatedly sampled with the same individuals in the same gang. […]

Pipeline specifications

Software tools GIMLET, Genepop, Arlequin, Structure Harvester, HapStar, DnaSP, Statistica
Applications Miscellaneous, Population genetic analysis
Organisms Papio papio, Homo sapiens, Papio hamadryas, Sesamum indicum