Demographic and conservation genomic assessment of the threatened marbled teal (Marmaronetta angustirostris)

Abstract

Genetic assessment of species that have experienced dramatic population declines provides critical information that is instrumental for the design of conservation recovery programs. Here, we use different sources of molecular data (mtDNA and ddRAD-seq) to evaluate the genetic status of wild and captive populations of marbled teal (Marmaronetta angustirostris), a duck species classified as critically endangered in Spain and near threatened at a global scale. First, we determined the evolutionary and demographic trajectories of the wild population from Spain and the currently much larger population from Iraq, which is also the documented source of European zoo stocks. Second, we evaluated the suitability of the different captive populations for ongoing restocking programs in Spain and assessed their potential impact on the genetic composition of wild populations. Populations from Spain and Iraq were assigned to distinct genetic clusters, albeit with an overall low level of genetic differentiation, in line with their recent divergence (<8000 years ago) and lack of phylogeographic structure in the species. Demogenomic inferences revealed that the two populations have experienced parallel demographic trajectories, with a marked bottleneck during the last glacial period followed by a sudden demographic expansion and stability since the onset of the Holocene. The wild population from Spain presented high levels of inbreeding, but we found no evidence of recent genetic bottlenecks compatible with the human-driven decline of the species during the past century. The captive populations from the two Spanish centers involved in restocking programs showed genetic introgression from European zoos; however, we found limited evidence of introgression from the zoo genetic stock into the wild population from Spain, suggesting captive-bred birds have limited breeding success in the wild. Our study illustrates how ex situ conservation programs should consider the genetic distinctiveness of populations when establishing breeding stocks and highlights the importance of genetically assessing captive populations prior to reinforcement actions.

Keywords: Marmaronetta angustirostris; conservation genetics; conservation‐breeding programs; demographic bottleneck; genetic diversity; introgression; marbled teal.

FIGURE 1

Demographic models tested using fastsimcoal2 for wild populations of marbled teal (M. angustirostris) from Spain (ES, in blue) and Iraq (IQ, in brown). Parameters include mutation‐scaled ancestral (θ _ANC, θ _BOT1, and θ _BOT2) and contemporary (θ _ES and θ _IQ) effective population sizes, migration rates per generation (m _S), timing of divergence (T _DIV), and timing of population size change (T _BOT1 and T _BOT2). The full set of tested models (i.e., considering alternative gene flow scenarios) is illustrated in Figure S1.

FIGURE 2

Haplotype network (TCS algorithm) based on mtDNA control region sequences for captive and wild populations of marbled teal (M. angustirostris). Circle size is proportional to the number of samples with each haplotype (see Table S1), and mutational steps are marked with hashes. ^aAll zoos from continental Europe present the same haplotype (haplotype I) and were grouped under the code cZOOS; ^bContemporary samples from Spain (wGUAD: 2005–2020; wVALE: 1991–2020); ^cHistorical samples from Spain predating the massive release of captive‐bred birds under reinforcement programs (wGUAD: 1967–1978; wCAST: 1970). All other population codes as described in Table 1.

FIGURE 3

Heatmap showing co‐ancestry (above the diagonal) and relatedness (below the diagonal) values between each pair of genotyped individuals of marbled teal (M. angustirostris). Simple co‐ancestry was estimated with fineradstructure. Relatedness was calculated using the kinship coefficient (φ _ij) developed by Manichaikul et al. (2010), which ranges from 0 (or negative values) for unrelated individuals to 0.5 for individual‐self. All relatedness values hidden by the inset picture are φ _ij  ≤ 0 (i.e., light yellow). Picture of marbled teal by Francis C. Franklin (licensed under CC BY‐SA 2.0). Population codes as described in Table 1.

FIGURE 4

Genetic diversity (a, b) and structure (c) for captive and wild populations of marbled teal (M. angustirostris). (a) Violin plots show estimates of heterozygosity (H _O) for each individual (small coloured dots) and mean and confidence intervals (black dots and vertical bars, respectively) for each population (n ≥ 4). (b) Heterozygosity (H _O) for each individual, with black dots and white triangles indicating estimates below and above the minimum value (black square) observed across all samples from Iraq, respectively. (c) Genetic assignments based on the program structure for K = 6. Each individual is represented by a vertical bar partitioned into K coloured segments showing the individual's probability (q) of belonging to the cluster with that colour. Symbols on top of the structure bar plot indicate the haplotype of each individual based on mtDNA control region sequences (legend on the right); note this information is missing for one individual from cSALE. Analyses are based on a dataset only including unrelated individuals (φ _ij  ≤ 0; see Figure 3). Vertical black lines separate different populations. Population codes as described in Table 1.

FIGURE 5

PCAs of genetic variation for captive and wild populations of marbled teal (M. angustirostris). Analyses are based on datasets only including unrelated individuals (φ _ij  ≤ 0; see Figure 3) and were run for all populations (a) and separately for wild populations (b). Colours indicate the main genetic cluster to which each individual was assigned according to structure analyses for K = 6 (see Figure 4c). Population codes as described in Table 1.

FIGURE 6

Demographic history of wild and captive populations of marbled teal (M. angustirostris) inferred using stairway plot. Analyses were performed for wild populations from Spain (a) and Iraq (b), captive populations from Spain involved in reinforcement programs (c), and European zoos (d). Panels show median (solid lines) and 2.5 and 97.5 percentiles (shaded areas) of effective population size (N _e) through time, estimated assuming a mutation rate per site per generation of 4.83 × 10⁻⁹ and 4‐year generation time for the species (both axes in a logarithmic scale). The number of unlinked SNPs used to calculate the SFS in each analysis is indicated in parentheses. Vertical dashed lines indicate the last glacial maximum (LGM, ~ 21 ka BP).