Computational protocol: Population Genetics of an Ecosystem-Defining Reef Coral Pocillopora damicornis in the Tropical Eastern Pacific

Similar protocols

Protocol publication

[…] Population genetic statistics, including allele frequencies, clonality, probability of identity, F-statistics and Analyses of Molecular Variance (AMOVA) were calculated using GenAlEx 6.1 . The program Micro-Checker 2.3 was used to test for stutter bands, large allele dropout and null alleles. Linkage disequilibrium among loci was assessed using a Markov chain method in GENEPOP 4.0.10 , followed by sequential Bonferroni correction to account for multiple comparisons . Clones were identified in GenAlEx and because they introduce biases into population genetic analyses, only unique genets were used unless otherwise noted.Population genetic structure was calculated as FST , which assumes an infinite allele model , and RST , which is based on a stepwise mutation model of allele evolution . Hierarchical AMOVA were used to estimate genetic structure among regions, among populations and among populations within regions . Statistical significance of RST and FST is based on 999 permutations. RST recovered more genetic structure, especially among regions, whereas FST had a lower resolution and higher intra-regional variation (FRT = 15% of FST compared to RRT = 50% of RST). As a result, we preferentially focus on RST (e.g. & ). Principal Coordinate Analyses , implemented in GenAlEx, were used to detect and visualize the major patterns of pairwise RST population comparisons. Isolation by distance among populations was tested using a Mantel test, implemented in GenAlEx 6.1, to test for an association between pairwise population genetic differences and geographic distances.The Bayesian clustering method implemented in STRUCTURE 2.3 was used to infer the likely number of genetic clusters (K) in the dataset. STRUCTURE estimates the posterior probability P (X | K) that the data fit the hypothesis of K clusters using a Markov chain Monte Carlo approach to optimize genotypic equilibrium (linkage equilibrium and HWE) within each cluster. Five independent runs for each K from 2 to 19 were conducted using the admixture model and independent allele frequencies with a burn-in period of 105 steps and 106 Markov chain Monte Carlo replications for data collection (). STRUCTURE runs were aligned using Clumpp1.1 and the bar plot was generated with Distruct1.1 . To verify the optimal value for K, we compared the P (X | K) results with ΔK, the second order rate of change in the likelihood of K , which corresponds to the strength of genetic subdivision among clusters in the data ().Fine-scale Spatial Genetic Structure (SGS) within populations was analyzed using the program SPAGeDi 1.2 . To distinguish between spatial genetic structure (SGS) due to clonal aggregations and SGS among unique genets, we analyzed and compared two different dataset: one dataset consisted of all 207 samples including clones (i.e. the ramet dataset) and one dataset contained only the 194 unique genotypes excluding clones (i.e. the genet dataset). SGS was estimated for each of the three regions separately and for all populations combined using the pairwise kinship coefficient Fij , . Other kinship coefficients including Ritland and Moran's I produced similar results (not shown). 95% confidence intervals and standard errors were estimated by 10,000 permutations of the genetic and the spatial datasets. Kinship values outside the 95% confidence intervals were interpreted as significant SGS at the spatial distance. Three parameters were used to describe and compare SGS among regions and studies: F(1), genetic patch size and Sp. F(1) is the average kinship among individuals within the smallest distance interval (10 m). Its statistical significance (p) was obtained by comparing the slope b(F) of F(r) on ln(r) to 9999 random permutations of individuals among locations within populations using a Mantel test. The genetic patch size is the distance that corresponds to the first x-intercept of the kinship correlogram [cf. 148]. Within the genetic patch, individuals are more closely related than the population average, i.e. the association between genotypes is positive, while individuals outside of the patch are genetically independent, i.e. their correlation is negative. The Sp statistic (Vekemans and Hardy is based on F(1) and the decrease of SGS with distance (b(F)). […]

Pipeline specifications

Software tools GenAlEx, Genepop, CLUMPP, DISTRUCT, SPAGeDi
Applications Phylogenetics, Population genetic analysis
Organisms Pocillopora damicornis, Caenorhabditis elegans, Drosophila melanogaster