Evaluating the adaptive potential of the European eel: is the immunogenetic status recovering?

Genetic estimates of diversity, population differentiation and demography

All 683 glass eels were sequenced using Sanger sequencing for the mitochondrial NADH dehydrogenase 5 (ND5) exactly replicating (Baltazar-Soares et al., 2014). Haplotype diversity (Hd) and nucleotide diversity (π) were calculated for each sampling location in DnaSP v5 (Librado & Rozas, 2009). Genetic structure was estimated using Arlequin v3.5 with 10.000 permutations (Excoffier & Lischer, 2009). Moment-based demographic parameters that test for changes in effective population size were calculated for each sampling location in DnaSP v5 under the assumption of mutation-drift equilibrium. Tajima’s D (Tajima, 1989) and raggedness’ r (Rogers & Harpending, 1992) were also calculated in DnaSP v5. Ninety-five percent confidence intervals were estimated through coalescence simulations using 1.000 permutations. We evaluated the nucleotide mismatch pairwise distributions within each geographical location (Rogers & Harpending, 1992). These distributions were compared to expected distributions under a constant population size and sudden population expansion (Librado & Rozas, 2009).

Genetic estimates of diversity, differentiation and demography amongst freshwater locations

All samples were genotyped for twenty-two microsatellite loci optimized from previous studies (Als et al., 2011; Pujolar et al., 2009; Wielgoss, Wirth & Meyer, 2008). Amplification took place in four PCR multiplexes of four to six loci each. Specifically: multiplex A–55 °C annealing temperature—included CT77, CT87, CA55, CA58, CT68, and AJTR-37; multiplex B—55 °C annealing temperature—included CT82, CT76, CT89, CT59, CA80, and CT53; multiplex C—60 °C annealing temperature—included C01, M23, AJTR-45, AJTR27, I14, and O08; multiplex D—60 °C annealing temperature—included AJTR-42, B09, B22, and N13. All reactions were performed in a total volume of 10 µl and followed the QIAGEN^© Multiplex PCR kit’s recommendations. Genotyping was performed on an ABI©3100 Genetic Analyzer. Alleles were called in GENEMARKER©v. 1.91 (Softgenetics LLC, State College, PA, USA).

Nei’s unbiased heterozygosity (He), observed heterozygosity (Ho) and F_IS were calculated for each sampling location in GENETIX (1.000 bootstrap, Belkir et al., 1999). Rarefied allelic richness (Ar) was calculated for each sampling location in HP-RARE v1.0 (Kalinowski, 2005). Genetic structure amongst sampling locations was inferred through pairwise comparisons in Arlequin v3.5 and Bayesian clustering in STRUCTURE v2.3.3 (Pritchard, Stephens & Donnelly, 2000). STRUCTURE was run assuming a maximum number of possible groups of K = 26, i.e., representing the sum of all spatial and temporal partitions of our sample, with 10,000 MCMC repeats after a 1,000 burn-in while assuming an admixture model with correlated allele frequencies. Three iterations were performed for each K.

Genetic signatures of a bottleneck were tested for each location using the tests available in BOTTLENECK (Cornuet & Luikart, 1996). These methods are sensitive to recent and severe reductions on effective population size (N_e) (Cornuet & Luikart, 1996). A two-phase mutation model was assumed with 10% of the loci allowed to evolve through stepwise mutation (Kimura & Ohta, 1978). Allele frequency distributions were also calculated for each location.

Genetic estimates of diversity, differentiation and demography— inter-generation level

In order to compare genetic diversity and demographic histories between the two cohorts and to avoid sampling bias from disproportionate number of samples in the two eel age groups (“glass eels” n = 481 and “silver eels” n = 202), we performed 10 rounds of re-sampling of the data without replacement using PopTools (Hood, 2010), hereafter referred to as “replicates.” Replicates were performed based on 50 individuals. This standardization is critical to validate future comparisons, as it has been long acknowledged that sample size affects the detection of genetic signatures of recent bottlenecks (Luikart et al., 1998) and the estimation of effective population size (Waples & Do, 2010).

Deviations from Hardy-Weinberg equilibrium (HWE) were calculated for each replicate in Arlequin v3.5 (10,000 permutations). Nei’s unbiased heterozygosity (He), observed heterozygosity (Ho), allelic richness (Ar) and F_IS were calculated and compared between groups of replicates, i.e., “glass eels” and “silver eels,” with two-sided t-tests in FSTAT (1,000 permutations) (Goudet, 1995). The distribution of genetic variance between “glass eels” and “silver eels” was assessed with an analysis of molecular variance (AMOVA, Arlequin v3.5) amongst groups of replicates. Demographic history was inferred using two approaches. First, we evaluated the possible genetic signature of the recent population decline using BOTTLENECK (1,000 iterations) for each age group as previously described. Second, we estimated the effective population size (N_e) of each replicate of “silver eels” and “glass eels” using the linkage-disequilibrium method implemented in NeEstimator V2.01 (Do et al., 2014). We utilized P_crit = 0.05, since lower P_crit can overestimate Ne (Waples & Do, 2008). All estimates were obtained with the composite Burrows method (Weir, 1990). The unweighted harmonic mean was calculated for each group according to the following equation: ${\hat{N}}_e=\frac{j}{{\sum }_{i=1}^j\left(1∕N_{e\left(i\right)}\mathrm{)}\right)}$ where j is the number of replicates, i is a given replicate and N_e(i) is the N_e estimate of the ith replicate (Waples & Do, 2010).

Molecular indices, population structure and demography amongst sampling locations

Analyses of 355 bp of the mtDNA ND5 in 683 European eels revealed 102 haplotypes including 73 singletons (Data S1). Forty-eight randomly picked singletons were verified by independent extraction, amplification and re-sequencing to eliminate possible risks of sequencing errors. After we eliminated the possibility of sequencing errors (48 out of 48 singletons were verified by independent sequencing) we included all 73 singletons in the subsequent analyses. Amongst sampling locations, haplotype diversity ranged between 0.575 (BU) and 0.934 (GL). Nucleotide diversity ranged between 0.003 (G_SPA) and 0.008 (GL), with an average of 0.005 (±0.001) amongst sampling locations (Table 1). Pairwise F_ST comparisons computed from haplotype frequencies amongst the 26 geographically confined groups revealed 17 significant pairwise differentiations however none passed corrections for multiple tests following the false discovery rate (threshold p = 0.013 for 26 tests (Narum, 2006)). All results are shown in Table S1.

Tajima’s D values were negative amongst almost all sampled locations, suggestive of population expansion or of population subdivision (Tajima, 1989). The three exceptions were GER (Germany), which belongs to a closed system, the Schwentine river, where recruitment is solely mediated by stocking (Prigge et al., 2013), as well as G_TITA (Italy) and G_WENG (England), where the values might reflect artificial stochasticity due to low sample sizes. Mismatch distribution analyses performed at the population level showed the typical pattern of a historical population expansion where the peak differs from zero (Rogers & Harpending, 1992) (Fig. 1A).

Molecular indices, population structure and demography between generations

Haplotype diversity and genetic diversity between “silver eels” and “glass eels” were very similar: Hd_silvereels = 0.821, Hd_glasseels = 0.842; π_silvereels = 0.0048, π_glasseels = 0.0049. No evidence for temporal genetic structure based on haplotype frequency distributions was detected between these groups, F_ST = 0.000, p = 0.138. Both groups also had negative and significant Tajima indices: D_silvereels = − 2.053, D_glasseels = − 2.357, both p < 0.05 (Table 2). Mismatch distribution analyses revealed that both “silver eels” and “glass eels” display the distribution of expanding populations, suggesting that the overall pattern is not driven by a single generation and has a true biological origin visible in both cohorts (Fig. 2).

Molecular indices, population structure and demography amongst locations

Across populations, He ranged between 0.687 (G_NIRL) and 0.758 (PT), Ho between 0.557 (G_TITA) and 0.667 (DK) and the average number of alleles per locus varied between 3.546 (G_VFRA) and 15.773 (G_AD2012). Allelic frequencies are reported in Data S2. F_IS varied between 0.0380 (G_NIRL) and 0.254 (Q) (Table S1). Even though F_ST estimates are very low, pairwise comparisons revealed 7 statistically significant pairwise comparisons after correcting for multiple testing (Table S1). Since six of those included comparisons between locations with low sample size, i .e. G_NIRL, G_WENG, G_TITA or G_BNIRL, the significance could be attributed to stochasticity due to low sample sizes. STRUCTURE analyses did not show any signs of population clustering as expected under the weak observed differentiation (Fig. S3).

None of the sampled locations showed either heterozygote excess or a mode shift in allele frequencies, genetic signatures characteristic of population bottlenecks (Fig. 1B).

Molecular indices, structure and demography between generations

Twenty-one loci were used to analyze “silver eels” and “glass eels” as locus AjTr-45 consistently deviated from Hardy-Weinberg equilibrium in all “silver eels” replicates. No significant differences between generations were apparent for He (He_silvereels = 0.744, He_glasseels = 0.743, p = 0.54), Ar (Ar_silvereels = 11.74, Ar_glasseels = 11.83, p = 0.46) and F_IS(F_ISsilvereels = 0.165, F_ISglasseels = 0.162, p = 0.07). Ho, however, was significantly higher in the “silver eel” group (Ho_silvereels = 0.621, Ho_glasseels = 0.612, p < 0.01). The AMOVA between “silver eel” and “glass eel” groups of replicates revealed a pattern of isolation by time (F_CT = 0.002, p < 0.001), supporting our a-priori assumption that those groups represent clear age structured cohorts.

None of the replicates showed evidence of heterozygote excess or deficiency. Averaged allele frequencies of neither “silver eels” nor “glass eels” deviated from an expected L-shape distribution (Fig. S3). However, we found that the averaged allele frequencies distribution observed in the “silver eels” group showed the signature of a 5-generations-old bottleneck identified from computer simulations (Luikart et al., 1998). This is particularly evident in the distribution of the two most common allele classes, 0.8–0.9 and 0.9–1.0. This signature was not visible anymore in the “glass eel” group (Fig. S4).

Estimates of effective population size (N_e) amongst replicates ranged between 0–625.2 for “silver eels” and 0–2708.9 for “glass eels”. The harmonic mean of effective population size estimates amongst replicates, ${\hat{N}}_e$, resulted in 480.9 $<{\hat{N}}_{e\mathrm{silver}\mathrm{eels}}<2941.7$ and $1380.2<{\hat{N}}_{e\mathrm{glass}\mathrm{eels}}<3506.0$ (Table S2).

Molecular indices and population structure

We sequenced a 247 bp fragment of the exon 2 of the MHC class II region (91% of the total size of the exon) in 327 individuals using 454 sequencing technology. We detected 229 different amino acid coding variants. Among those, 226 (98%) were found to be unique in the dataset but present in both independent replicated reactions and therefore kept as true variants (Data S3). A total of 116.276 sequence reads were used in this study. Amongst locations, MHC nucleotide diversity ranged between 0.102 (LL) and 0.138 (BT). The mean number of alleles per individual ranged between 2 (Q, SE = 0.298) and 4 (G_BU, SE = 0.392) (Table S3) and revealed to overall significantly differ amongst sampled locations (F₁₇ = 1.674, p = 0.046). However, post-hoc pairwise comparisons showed no significant differences between pairs of populations after correction for multiple testing (all p > 0.05). The mean nucleotide divergence (p-distance) ranged between 0.078 (BL) and 0.141 (FI) (Table S4) and was significantly different among sampled locations (F₁₇ = 1.860, p = 0.021). Post-hoc pairwise comparisons revealed two significant comparisons after correction for multiple testing (GER vs FI, t = − 3.551, p = 0.045, GER vs G_AD2011, t = − 3.961, p = 0.010), suggesting a reduced MHC diversity the stocked freshwater system of the Schwentine river, in Germany.

The ANOSIM showed no significant differences in MHC allele pools amongst populations (R = 0.001, p = 0.98). Overall, no correlation was found between Bray-Curtis similarity matrices on MHC and pairwise F_ST for both mtDNA (R² < 0.0001, p = 0.58) and microsatellites (R² < 0.0001, p = 0.62).

Between generations, no difference in MHC allele pools were observed (R = − 0.011, p = 0.87). Interestingly, “glass eels” had a significantly higher individual mean number of alleles (“glass eels“=3.423, SE = 0.166; “silver eels “= 2.856, SE = 0.101; F₁ = 8.819, p = 0.003) and a significantly higher individual mean nucleotide p-distance (“glass eels” = 0.117, SE = 0.006; “silver eels” = 0.101, SE =0.004; F₁ = 4.577, p = 0.032) (Table 3). Both the nucleotide diversity (π) and the number of minimum recombination events (Rm) detected between “silver eels” and “glass eels” were similar (π_glasseels = 0.118, π_silvereels = 0.123; Rm_silvereels = 11; Rm_glasseels = 10) while the R∕θ ratio was slightly higher in “silver eels” (R∕θ_silvereels = 2.174; R∕θ_glasseels = 2.089).

Screening for novel genetic diversity through events of gene conversion

Gene conversion segments with an average nucleotide length of 4 bp were detected within the “glass eels” group but not in the “silver eel” group. The average ψ of the whole segment was found to be 0.0002 (Fig. 3). This value is high enough to ascertain the occurrence of conversion events, but not robust enough to determine the exact length of the observed tracts (Betran et al., 1997).

Testing for positive selection

Model-based tests using CODEML revealed 11 sites under positive selection while MEME identified 27 sites that have experienced episodic events of positive selection (Table S4). The discrepancies between the two methods reflect the different assumptions underlying the fixed effect models implemented in CODEML and the mixed effect models of MEME. Positively selected sites detected by both methods matched 10 out of 19 antigen binding sites identified in humans by X-ray crystallography (Reche & Reinherz, 2003) (Fig. 3). Due to the functional role of the MHC, all amino acid sites that have experienced at least episodic events of selection were selected for further analyses (Fig. 3)

Recruitment decline of the 1980s did not affected genetic estimates of neutral evolving markers

The major objective of this study was to compare patterns of temporal variation of genetic diversity post-recruitment collapse observed in the 1980s. Here as well, our mitochondrial DNA results showed (i) no evidence for a genetic bottleneck, (ii) no differences in haplotype and nucleotide diversities between “silver eels” and “glass eels,” and (iii) no signature of bottleneck in the frequency distribution of pairwise mismatches. The later rather points towards an historical population expansion (Rogers & Harpending, 1992). Given the low variability of the mtDNA marker—in comparison with the set of highly polymorphic microsatellites and MHC gene—we suggest that such an event extends back in time to a scenario of expansion related to ice-sheet retreat after the last glacial maxima, as previously proposed for such a pattern (Jacobsen et al., 2014).

Investigating the distribution of the genetic variance observed at microsatellites, we found significant differentiation between “silver eels” and “glass eels” replicates. This pattern of genetic variance distribution is in line with previous reports that also attributed higher genetic variance amongst temporal, rather than spatial, partitions of A. anguilla along the European coasts (Dannewitz et al., 2005). It also confirmed our a-priori assumption of each group representing a distinct generation, and therefore excludes possible confounding factors associated with overlapping generations from the interpretation of demographic estimates (Cornuet & Luikart, 1996; Waples & Do, 2010). Note, even though less likely, we cannot exclude that this observed structure also relates to a spatially structured spawning area of this species (Baltazar-Soares et al., 2014; Dannewitz et al., 2005).

Using microsatellites, we detected no evidence of heterozygote excess in any of the replicates, nor any differences between allelic richness of “silver eels” and “glass eels.” This supports previous reports that the recruitment collapse and subsequent low abundance of the eel population did not leave the expected genetic signatures of reduced genetic diversity (Pujolar et al., 2011) suggestive of a system which replenishes genetic diversity rapidly.

However, in-depth demographic analyses suggest that the eel effective population size might actually not be stable. Several lines of evidence support this interpretation: firstly, we estimated ∼20% higher effective population size in “glass eel” replicates (harmonic mean N_e = 3506.0) compared to “silver eels” (N_e = 2941.7). These contemporary estimates are near the lower confidence intervals of historic effective population sizes previously reported (5,000 < N_e < 10,000; (Wirth & Bernatchez, 2003), but within contemporary estimates of 3,000 < N_e < 12,000 (Pujolar et al., 2011)). Noteworthy, the confidence intervals calculated in this study for each generation overlap, raising the need to cautiously interpret those results. Secondly, we observed fewer alleles in the most frequent class of allele frequencies in “silver eels”. The apparent reduction of the most frequent allele class may suggest that the species demography is experiencing a transitory stage from a severe bottleneck, partly detected in “silver eels.” It is important to mention that such a signature would only be detected under a severe population reduction a few generations in the past. Hence, we could speculate that the hypothetical bottleneck detected only in “silver eels” may relate to the drops in European eel recruitment that occurred in the beginning of the 1960s (EIFAAC/ICES, 2011). It is possible that the 1960s low recruitment had a major impact on the overall genetic diversity of the species. By the crash in the 1980s, the population would have already been depleted from its original genetic diversity, at least for neutrally evolving markers. Such a scenario requires further studies to be confirmed and would rely on historical samples to be analyzed.

MHC reveals signatures of selection

Extending the evaluation of genetic diversity to the evolutionary analysis of the adaptive immune genes of the MHC was motivated by two main reasons. The first relates to studies suggesting MHC diversity to be more sensitive than neutrally evolving markers in the detection of demographic shifts (Sommer, 2005; Sutton et al., 2011). The second relates to the invasion of European freshwater systems by the nematode parasite, Anguillicola crassus, for which the MHC was found to respond to in the paratenic host, the three-spined stickleback (Eizaguirre et al., 2012b). Using the exon 2 of the MHC class II β gene, we evaluated (1) genetic diversity, which might have been affected by the recruitment collapse and subsequent population reduction and (2) allele frequency shifts between generations which would be a signature consistent with directional parasite-mediated selection.

Using next-generation sequencing, we identified a total of 229 MHC alleles amongst 327 individuals. This indicates that the diversity within this species is not low and directly compares to observations made in wild populations of other fishes that are not qualified as endangered, such as for instances, the half-smooth tongue sole (88 MHC class II alleles amongst 160 individuals) (Du et al., 2011). Of note, the characterization of the MHC class II genes in this species revealed that up to six different alleles may exist per individual, suggesting the presence of at least three loci (Bracamonte, Baltazar-Soares & Eizaguirre, 2015; Bracamonte, 2013). Because next generation sequencing is thought to generally overestimate the number of MHC alleles detected (Babik et al., 2009; Lighten, Oosterhout & Bentzen, 2014; Sommer, Courtiol & Mazzoni, 2013), we took multiple precautions to avoid artifacts (reconditioning steps, reduced numbers of PCR cycles, duplicates). Despite such precautions and sequence confirmation using cloning (performed in Bracamonte, Baltazar-Soares & Eizaguirre, 2015), we cannot exclude that some variants were called alleles even though artifactual (Sommer, Courtiol & Mazzoni, 2013). Again, as mtDNA sequencing revealed a large number of haplotypes (N = 102), the large diversity at the MHC displayed in this species may not be surprising.

Generally, our results are suggestive of a pattern of selective sweep at the MHC between the two generations examined here. We found “silver eels” to exhibit lower mean number of alleles and lower mean nucleotide distance than “glass eels,” suggesting that the “silver eel” generation was under a selective pressure that reduced its pool of MHC alleles to fewer and more similar alleles. This observation suggests that either the selective pressure was widespread amongst continental locations or that it acted when all eels experienced similar conditions; as for instance, during the fastening spawning migration. A hypothetical selective pressure imposed by A. crassus meets both criteria. Not only is this parasite ubiquitous in European freshwater systems (Wielgoss et al., 2008) but it also impairs the swimming performance of infected eels (Palstra et al., 2007).

Overall, the higher genetic diversity of “glass eels” together with the identification of two short gene conversions in this study suggest an ongoing regeneration of the species immune adaptive potential. Indeed, gene conversation is a mechanism capable of generating novel diversity within the MHC region, and seems to be a predominant mechanism in genetically depauperate populations (Spurgin et al., 2011).

Contemporary loss of MHC diversity: evidence of selection?

The parasite A. crassus was presumably introduced in the European freshwater systems at the beginning of the 1980s (Taraschewski et al., 1987), quickly spreading across continental water bodies. The European eel is particularly susceptible to A. crassus infection (Knopf, 2006) and therefore its introduction provides an excellent biological calibration to evaluate its impact on the evolution of diversity of the MHC. More specifically, we expected a signature of selection by the parasite to be reflected in positively selected sites of the MHC variants. In total, we detected 27 sites to be under or that have experienced positive selection along their evolutionary history.

Bayesian skyline plots showed a steep decline in MHC genetic diversity as time approaches present. This pattern is visible independently of the substitution rates and is reproducible with independent runs, i.e., overlap of the probability density distributions of the posterior, skyline and likelihood. Together, this suggests a real pattern and not an artifact. This decline in genetic diversity of the adaptive gene is similar to those detected in genealogies exposed to events of episodic positive selection (Bedford, Cobey & Pascual, 2011), but also in functional regions involved in adaptive responses (Padhi & Verghese, 2008). Two main factors may explain it. Firstly, it can be attributed to the long terminal branching typical of phylogenies of genes evolving under balancing selection (Richman, 2000), amongst which the MHC is a classic example (Klein, Sato & Nikolaidis, 2007). MHC class II genes are also classical examples of genes evolving through recombination (Reusch & Langefors, 2005) and gene conversion (Spurgin et al., 2011)—two mechanisms that together with a relatively high copy number variation generates rapid genetic novelty (Chain et al., 2014). Therefore, a null expectation for balancing selection and generation of rapid genetic diversity—as predicted for MHC—would rather be that of either an expanding or stable population, which was not what we observed here.

Conversely, it can be attributed to an event of selection speculatively associated with the spread of A. crassus across the European freshwater systems. A. crassus was unknown to A. anguilla before its recent introduction, however it is naturally present in the A. japonica population (Wielgoss et al., 2008). Hence, the frequency of the MHC alleles, or group of functionally similar MHC alleles, that confer resistance against this parasite would either be low or even absent in the European eel population (Eizaguirre et al., 2012b). The selection for those rare variants could have triggered the major loss of diversity we observed in the Bayesian plots and confirmed by the lower diversity indices of the “silver eels.” Interestingly, the allelic lineage diversity of the MHC showed a constant increase with a particular acceleration approaching the contemporary period, as indicated by the lineages-through-time-plots of both generations. While we are unable to provide a direct functional link between such diversification and our observations of gene conversion and recombination within this specific MHC region, these mechanisms have been associated with signatures of recovery after a genetic bottleneck in genes under balancing selection (Richman, 2000). Therefore, the inferred recent steep decline followed by a very recent burst of lineage diversification upholds the occurrence of a selective sweep in the MHC genealogy that pre-dated both our sampling points, as suggested by the comparison between “silver eels” and “glass eels,” with an ongoing recovery of the MHC diversity.

From our results we can hypothesize two scenarios. A first scenario involves a link between a sudden reduction in population size, a loss of genetic diversity and a constant selective pressure extending after the bottleneck. In this scenario, genetic drift would affect overall genetic diversity but since selection would continue to act, genetic diversity of positively selected regions would remain low (Eimes et al., 2011). A second scenario relates to multiple MHC loci carrying similar alleles due to recent duplications and to the hypothesis that a population would experience a size reduction and an event of selection within the same time frame. In this scenario, the selective pressure through the bottleneck would lead to a faster fixation of resistant alleles (Eimes et al., 2011).

Limits of the study

Evaluating demography in the European eel is complex particularly due to the difficulty of sampling them at their mating ground and the reliance on indirect genetic evidence. Even though patterns of selection and recent recovery of the MHC diversity seem robust, due to the high allelic variation reported in this gene, we acknowledge that other coalescence models could revealed refined patterns (Árnason & Halldórsdóttir, 2015; Wakeley, 2013). It is indeed possible that multiple loci analysed together as one locus—although with same functional basis—could create a pattern similar to multiple coalescence. To the best of our knowledge, multiple merger models such as those described by (Árnason & Halldórsdóttir (2015) have not yet been applied to investigate the evolutionary signature that linked copies of the same (functional) gene produce on the evaluation of genetic diversity and demography. This thus limits working hypotheses and would be difficult to interpret. Nonetheless, while further investigations are obviously needed to clarify fluctuations in genetic diversity in such a complex but evolutionary relevant immune gene, we argue that our work represents a first empirical step along this line of research.

Conclusions

In summary, our work reveals signatures of recent reduction in MHC genetic diversity and suggests signs of ongoing recovery of this gene’s diversity contributing to the immunogenetic adaptive potential of this endangered species (Radwan, Biedrzycka & Babik, 2010; Sommer, 2005; Stiebens et al., 2013b). Future research will be needed to provide conclusive evidence as to which scenario holds, accommodating also newer theories to further verify the validity of our findings. A future perspective would be to extend the time-series analyses by incorporating screening of MHC diversity in ongoing monitoring practices. This would be a valuable approach to access the evolution of the species adaptive potential.