Sibship assignment to the founders of a Bangladeshi Catla catla breeding population

Abstract

Catla catla (Hamilton) fertilised spawn was collected from the Halda, Jamuna and Padma rivers in Bangladesh from which approximately 900 individuals were retained as 'candidate founders' of a breeding population. These fish were fin-clipped and genotyped using the DArTseq platform to obtain, 3048 single nucleotide polymorphisms (SNPs) and 4726 silicoDArT markers. Using SNP data, individuals that shared no putative parents were identified using the program COLONY, i.e. 140, 47 and 23 from the Halda, Jamuna and Padma rivers, respectively. Allele frequencies from these individuals were considered as representative of those of the river populations, and genomic relationship matrices were generated. Then, half-sibling and full-sibling relationships between individuals were assigned manually based on the genomic relationship matrices. Many putative half-sibling and full-sibling relationships were found between individuals from the Halda and Jamuna rivers, which suggests that catla sampled from rivers as spawn are not necessarily representative of river populations. This has implications for the interpretation of past population genetics studies, the sampling strategies to be adopted in future studies and the management of broodstock sourced as river spawn in commercial hatcheries. Using data from individuals that shared no putative parents, overall multi-locus pairwise estimates of Wright's fixation index (F_ST) were low (≤ 0.013) and the optimum number of clusters using unsupervised K-means clustering was equal to 1, which indicates little genetic divergence among the SNPs included in our study within and among river populations.

Fig. 1

Heatmaps of relationship matrices for individuals from the Halda river. a Genomic relationship matrix (G) generated by using observed allele frequencies in all individuals, b COLONY-derived additive relationship matrix (A), c G using observed allele frequencies in individuals with no putative parents in common, and d A derived from manual sibship assignment. For A matrices, black represents a full sibling relationship (i.e. 0.50), dark grey represents a half sibling relationship (i.e. 0.25) and light grey represents no relationship (i.e. 0.00). Individuals in a–d are ordered according to clustering, using the ‘Ward2’ algorithm, of genomic relationships in c. Insets of a and c contain histograms of observed genomic relationships

Fig. 2

Heatmaps of relationship matrices for individuals from the Jamuna river. a Genomic relationship matrix (G) generated by using observed allele frequencies in all individuals, b COLONY-derived additive relationship matrix (A), c G using observed allele frequencies in individuals with no putative parents in common, and d A derived from manual sibship assignment. For A matrices, black represents a full sibling relationship (i.e. 0.50), dark grey represents a half sibling relationship (i.e. 0.25) and light grey represents no relationship (i.e. 0.00). Individuals in a–d are ordered according to clustering, using the ‘Ward2’ algorithm, of genomic relationships in c. Insets of a and c contain histograms of observed genomic relationships

Fig. 3

Heatmaps of relationship matrices for individuals from the Padma river. a Genomic relationship matrix (G) generated by using observed allele frequencies in all individuals, b COLONY-derived additive relationship matrix (A), c G using observed allele frequencies in individuals with no putative parents in common, and d A derived from manual sibship assignment. For A matrices, black represents a full sibling relationship (i.e. 0.50), dark grey represents a half sibling relationship (i.e. 0.25) and light grey represents no relationship (i.e. 0.00). Individuals in a–d are ordered according to clustering, using the ‘Ward2’ algorithm, of genomic relationships in c. Insets of a and c contain histograms of observed genomic relationships