2.4.1 Threshold values for indicators
Threshold values for indicators are regarded as helpful in management to evaluate rates of change (Maria Jansson, SwAM, pers. comm.). Prosed threshold values for the ΔH indicator relate to the recommendation that in 100 years, a population should retain at least 95% of its heterozygosity (Allendorf & Ryman, 2002). The proposed thresholds values for this indicator are: less than 0.05% reduction per year, assuming constant rate of change (reflecting retention of c. 95% heterozygosity after 100 years); this rate of reduction is suggested to reflect status of color green for “Good” (Figure 3). A rate of reduction between 0.06-0.3% per year (reflecting expected retention of c. 75-94% heterozygosity over 100 years) is proposed to reflect status “Warning”/yellow where further investigation of the reason for reduction is warranted. Finally, a reduction rate of more than 0.3% per year (resulting in ≤ 75% of genetic variation expected to be retained after 100 years) reflects status red alert where prompt measures are called for to understand the reason for decline and thereafter taking immediate steps to halt the reduction and restore genetically safe conditions.
Here, we apply indicator ΔH in each of the identified populations (that occur in samples at both points in time) as well as to the metapopulations that they belong to. Genetic diversity was measured as expected heterozygosity (H E), observed heterozygosity (H O), allelic richness (A R), number of alleles per locus (N A), and proportion of polymorphic loci (P L), and testing for potential changes was done by t-tests and non-parametric Wilcoxon matched pairs tests. In cases with statistically significant change, we translate the difference between the two points in time approximately 40 years apart (details on time span between samples in Table S1) into an annual change. Depending on the observed rate of change we translate it into either of the three indicator signals – green, yellow, or red (i.e., “Good”, “Warning”, or “Alarm”). If genetic diversity within sampling localities does not change (no statistically significant change) or with a statistically significant increase over time we consider the ΔH indicator as green/“Good”. We apply the same threshold values for annual genetic reduction (i.e., ≤0.05%; 0.06-0.3%; ≤0.3%; Figure 3) for not onlyH E but also for the other measures of genetic diversity (H O, A R,N A, P L).
Suggested thresholds for the N e indicator are based on the conservation genetic rule of thumb thatN e≥50 and N e ≥500 is necessary for a population’s short term and long-term survival, respectively (Franklin, 1980; Jamieson & Allendorf, 2012). The proposed thresholds are: N e≥500 (“Good”), 50<N e<500 (“Warning”), andN e<50 (“Alarm”), and should apply to single isolated populations as well as to metapopulations. TheN e of local subpopulations of metapopulation cannot be ignored, however. Rather, it is important that gene flow is of a magnitude that assures that sufficient levels of genetic diversity reaches the population so that the adaptive potential is maintained. Laikre et al. (2016) suggested that the realized, local effective sizes of meta­populations should also reflect inbreeding rates that are so low that realized local N e≥500 for long term viability to be attained.
In practice, however, it is not straightforward to estimate theN e that reflect the actual rate of inbreeding (N eI) and/or loss of additive genetic variance (N eAV) in substructured populations (Hössjer et al., 2016; Ryman et al., 2014, 2019). Here, we useN e estimates from both the temporal (N eV) and linkage disequilibrium methods (N eLD; Section 3.2) and base classifications on the estimate of these two that generally is the largest, in line with recommendations for non-isolated populations (Ryman et al., 2019). We apply this indicator to metapopulations as well as to separate subpopulations.
For the ΔF ST indicator we are aware of no previously suggested guideline or rule of thumb. We apply and extend the proposal from Johannesson and Laikre (2020) that without detectable change of F ST among populations over time, the status of this indicator is classified as “Good” (Figure 3). With an increase of F ST between the two points in time that reflect a c. 25% decrease of genetic exchange between populations (number of migrants is reduced by 25%) this is classified as “Warning”. A decrease of F ST is expected with increase of gene flow. Such increase can be warranted following management activities to connect previously fragmented populations. However, decrease of F ST can also be an effect of homogenization following e.g. release activities. Such activities are not expected to be carried out in the present case since all monitored lakes are located in protected areas. However, large scale release activities resulting in genetic homogenization have been documented in e.g. Baltic salmon populations (Östergren et al., 2021). Thus, because decreased divergence can also reflect a genetic threat, we propose (in line with Johannesson & Laikre, 2020) that anF ST reduction representing c. 50% increase of gene flow should classify as a “Warning” in cases where unwanted gene flow can have occurred. With a ΔF ST reflecting a 50% decrease of the number of migrants or a 100% increase in genetic exchange (number of migrants) this indicator is classified as “Alarm”. Further, if one or more local population goes extinct over the monitoring period this indicator is also classified as “Alarm”. We note that these proposed limits are highly subjective and “forgiving” with respect to changes of connectivity.
We apply the following approach (described in detail in Appendix S1) for converting a statistically significant ΔF ST into an indicator of change in genetic exchange (migration) between subpopulations. Thus, we translate the observedF ST among subpopulations at the first point in time (here denoted “past” and referring to the 1970-80s samples) to the expected number of migrants by assuming an island model in migration-drift equilibrium. This hypothetical island model has the same number of subpopulations as the metapopulation considered, and the subpopulation N e is set to the harmonic averageN e over those subpopulations. In the next step we calculate limiting values for change of migration rates. Here we consider a reduction of migration by 25% or 50% to correspond to yellow/warning or red/alarm, respectively. Similarly, we consider an increased migration rate of 50% or 100% to reflect yellow/warning or red/alarm. Finally, we translate the limiting values of migration rates into F ST values and compare them to the empirically observed F ST at the second time point (here denoted “present” and referring to the 2010 samples).