Species differences in range shifts did not persist between successive survey intervals in 8 BioShifts studies (pooled r = −0.19), and rate spread fell almost as 1/interval length

Question: do species differences in measured range shifts persist from one survey interval to the next? Literature syntheses treat the large spread of shift rates among species, and the many shifts toward the equator or downhill, as species-specific responses. Approach: we used the BioShifts database (BioShiftR package, commit 359d7d2; 31,761 estimates, 244 articles). Eight studies measured the same species at the same range position in successive intervals (23 interval pairs, 1,115 species units; birds, butterflies, bumblebees, plankton, desert plants, 21 British taxa). If species have persistent rates, successive shifts correlate positively. If shifts contain error or temporary deviations at the survey dates, the shared middle survey makes them negatively correlated. Results: the pooled study-level correlation was r = −0.19 (95% CI −0.38 to +0.01; rank-based −0.17, −0.32 to −0.01). Twenty of 23 pairs were negative; two studies were positive. Correlations of intervals without a shared survey had a median of −0.04. Under a model with equal survey variances, the persistent share of the among-species variance in one interval was at most 21% on average (weaker bound 35%). Species that shifted in the contrary direction in one interval shifted in the expected direction in the next interval more often (0.76) than species that shifted in the expected direction (0.61). For the same species measured over intervals of different length, the among-species SD of the rate scaled as T^−0.94 (95% CI −1.16 to −0.72; −0.64 without the bumblebee study), near the 1/T prediction for non-persistent change. Conclusion: in these studies, most of the among-species spread of single-interval shifts is non-persistent, so contrary shifts, trait effects and lag or overshoot relative to climate velocity are hard to interpret from one interval. Limitations: only 8 mostly European atlas studies qualify; heterogeneity is large; the test cannot separate survey error from real temporary fluctuations; the test removes the shared mean shift and does not test average climate tracking.

range shifts · climate change · BioShifts · measurement error · temporal persistence · meta-analysis · species distributions · time-scaling of rates · contrary shifts · resurvey