Computational protocol: Conservation Genetics of Threatened Hippocampus guttulatus in Vulnerable Habitats in NW Spain: Temporal and Spatial Stability of Wild Populations with Flexible Polygamous Mating System in Captivity

Similar protocols

Protocol publication

[…] Genetic diversity and demographyAllele frequencies were obtained from genotype data () using Fstat v2.9.3.2 []. Conformance to Hardy-Weinberg (HW) expectations and linkage disequilibrium were checked using exact tests implemented in Genepop v4.0 [], applying Bonferroni correction for multiple tests. The presence of null alleles and scoring errors (allele dropout, stuttering) was investigated using Micro-Checker v2.2.3 []. Allelic richness (AR) was obtained using Fstat. Allele number per locus (A), and observed and expected heterozygosity (H o, H e) were estimated using Cervus v3.0 [].Contemporary effective population size (N e) was computed using single point and temporal methods []. i) N e in single spatial samples caught in the same year from Cantabrian (CS06) and South-Atlantic (SA06) estuaries was estimated using ONeSAMP v1.2 [, ]. N e prior intervals of 2 to 500 were applied based on the small and unequal population density observed across locations during underwater surveys [], and assuming that sampling size (37–52) could represent around 10% of the population size []. ii) Starting from the most distant temporal samples from South-Atlantic estuaries (SA06 vs. SA11; ), a temporal moment-based estimate of N e was obtained using NeEstimator v1.3 [], assuming a generation time of 1.8 years [].Signatures of genetic bottlenecks were investigated using the M-value method implemented in M_P_Val []. Mean M-values over loci and their significance (10,000 replicates) were calculated assuming a proportion of 0.10 for multi-step mutations with mean size of 3.5. A wide range of possible scenarios was covered (θ = 4N e μ), from small (θ = 0.1) to large (θ = 10) pre-bottleneck N e in natural populations, and also a long-term N e estimate for each sample derived from H e (N e = [(1/(1-H e))2–1]/(8μ); μ: mutation rate []) [, ], assuming equilibrium and stepwise mutation model (SMM) [, , ].Population structureAllelic differentiation A ST between pairs of population samples was estimated using Metapop v2.0.a1 []. The statistics Jost’s [] D EST and the standardized Hedrick’s [] G” ST were estimated using GenAlEx v6.5 [] (999 permutations, 999 bootstraps), since they are not influenced by intrapopulation diversity. G” ST is further corrected for bias when number of populations is small []. Analysis of molecular variance (AMOVA; []) was performed using Arlequin v3.11 [] to assess the distribution of the genetic variation among and within estuaries, and between temporal samples under a priori grouping of samples according to their estuarine location (Betanzos, Arousa, Pontevedra and Vigo; see and ).Isolation-by-distance over the studied distribution area was evaluated using the correlation of Rousset’s distance measure [] based on G” ST against the logarithm of the geographical distances among the four estuaries of origin. Mantel test (30,000 replicates) was conducted to assess the relationship between genetic and geographical distances, using the software IBDWS [].Number of population units (K) was inferred using the Bayesian MCMC approach implemented in Structure v2.3.1 []. Analysis of the whole sampling data was carried out under the admixture ancestral model with correlated allele frequencies [], without prior population information and also using prior population information since it could aid to detect cryptic structure under small genetic differentiation [] (burn-in: 50,000; MCMC: 200,000). Ten independent runs were conducted for the Ks tested (1 to 13) and the mean of Ln probabilities of data (Ln Pr(X|K)) across runs were calculated. Posterior probability of each K was computed from Bayes’ Rule []. The most likely value of K was also estimated according to the Evanno’s ΔK statistic using Structure Harvester v0.6.94 [, ]. Finally, the Structure output for the best ΔK was summarized to correct the variance across runs using Clumpp v1.1.2 [], and graphically displayed using Distruct v1.1 [].Despite the departure from HW expectations at one locus and the possible presence of null alleles at very low frequency in another one (see ), all population structure analyses were based on all loci. The relative coefficients of genetic and allelic differentiation together with the Structure analyses were recalculated excluding these loci and rendered very similar results (see below; ; .).Broodstock genetic analysisThe conditions for the maintenance of the individuals in captivity were reported by Planas et al. [, ]. Briefly, adult seahorses were kept under temperature (15°C in winter to 19°C in summer) and natural photoperiod (16L:8D in June-July; 10L:14D in December-January) regimes and fed ad libitum twice daily on enriched adult Artemia (EG, Inve, Spain), supplemented with captured Mysidacea (Leptomysis sp. and Siriella sp.).Genetic diversity in the renewed broodstock since 2009 (Stock09) was compared with the Galician wild population based on the complete set of 13 loci, and with the stock founded in 2006 (Stock06; []) using the six most polymorphic loci []. Departure from HW expectations and linkage disequilibrium were checked using exact tests implemented in Genepop, applying Bonferroni correction for multiple tests. Genetic diversity estimators (A, A R, H e), together with theoretical probabilities for exclusion (Excl1 and Excl2, when the other parent is unknown and known, respectively) and for sibling identity (SI) were estimated using Cervus. D EST and G” ST between Stock09 and South-Atlantic wild population samples, as well as between Stock09 and Stock06 were estimated using GenAlEx (999 permutations, 999 bootstraps).Relatedness (r) between all pairs of breeders in the renewed Stock09 was computed using Wang estimator [] (SPAGeDi v1.2; []). The midpoints between the expected r-values for unrelated (UR; r = 0), half-sibs (HS; r = 0.25) and full-sibs (FS; r = 0.5) kinships were used as thresholds to classify individuals (UR≤0.125

Pipeline specifications