3.4 Effective population size
Average estimates of variance effective size (N eV, per generation) per population (cluster) varied between 47 and infinity, and was below 500 in 20 of the 29 populations that were observed at both points in time and thus allowed assessment of N eV (Figure 4; Table S7). These estimates are expected to coincide with the inbreeding effective size of a subpopulation under isolation. In the face of migration, inbreeding rates are expected to be lower than what these estimates indicate (i.e., local N eI is expected to be larger than localN eV, cf. Ryman et al., 2019). MetapopulationN eV was below 500 in five of the eight metapopulations and in one of the two single lakes. In three metapopulations and in one of the single lakesN eV was above 500 (Figure 4; Table S7).
There was a significant positive correlation betweenN eV and expected heterozygosity (H E) (r =0.48 and p=0.008 for present estimates; Figure 6a). A similar relationship betweenN eV and the other diversity measures was also observed at both time points (Figure S8; Table S8).
There was no apparent link between N eV and predicted retention of expected heterozygosity (H E; section 2.4.1) over 100 years (r =0.39; p=0.16; Figure 6b).
N e estimates based on the linkage disequilibrium method (N eLD) where consistently lower than estimates based on the temporal method (N eV) (Table S7). Only four populations in the past and six present ones showN eLD estimates above 500. In the past (1970s), 11 populations have N eLD<50 while in the present (2010s) five populations showN eLD<50. N eLD of separate populations appear to have increased over the 40 years between sampling (p=0.048; Wilcoxon matched pairs test), although this trend was not observed for metapopulations. None of the metapopulations had anN eLD above 500 at any point in time (Table S7).