Are the spatial and temporal scales biologically meaningful?
Millette et al . (2020) examined IGD temporal trends across 909 animal species where COI sequences were available for ≥4 years. Sequences were grouped across ≤1,000 km to avoid “conflating spatial and temporal effects” . This scale far exceeds the dispersal capabilities of many included species, the scale of habitat change affecting them, and thus the scale at which population genetic processes influencing IGD operate. Additionally, the clustering algorithm used can ‘daisy chain’ locations together so that sites >1,000 km apart are grouped together (Appendix A1). Grouping sequences into biologically-implausible “populations” likely obscures any anthropogenic effects on IGD, especially when combined with the small sample sizes (<10 sequences/year for 77% of time-series overall) and the large number of locations sampled yearly (e.g. mean of 3 locations sampled per year for fish; Appendix A1). We reexamined 104 Inland and Coastal Bony Fish time-series from Millette et al. (2020), and found that the sequences included were sampled in inland freshwaters for most time-series (96/104), with sequences from multiple demographically-independent locations (i.e. from disconnected drainage basins; median 3 [range 0-15], see Figure 1 and Appendix A1 for examples) erroneously pooled into “populations”. By pooling sequences from independent locations, changes in IGD attributable to anthropogenic pressures would be lost in the noise, with uneven sampling across space and time compounding the issue.
Additionally, the median span of included time-series is only seven years overall, and represents an average of just 2.2 generations for fish (Appendix A1). Thus for many taxa, the data covers an insufficient time-span for most measurable evolutionary changes in IGD. Of course, such analysis also neglects any pre-1980s impacts. DNA from museums, herbaria and fossil archives are needed for this.