Rapid climate change results in long-lasting spatial homogenization of phylogenetic diversity

The study investigates how rapid climatic changes, particularly since the Last Glacial Maximum (LGM), have shaped the spatial structure of phylogenetic beta diversity (phylogenetic turnover, phylo-β) among neighboring sites across Europe. Building on the idea that climatic stability promotes the evolution and persistence of species (and thus high diversity and turnover), while rapid climate change favors widespread generalists and can erode diversity, the authors focus on spatial patterns rather than solely temporal trends. Europe, with its strong Quaternary climate oscillations, glaciations, and documented southern and hypothesized northern refugia, provides an ideal system. The central hypotheses are that phylogenetic turnover corrected for environmental and geographic distance (phylo-βsimc) is lower (i) where past climate change velocity was high, (ii) farther from glacial refugia, and (iii) where species have larger range sizes. The purpose is to quantify spatial phylo-βsimc for European seed plants (angiosperms and gymnosperms), evaluate its drivers, and infer the enduring imprint of past climate dynamics on current biodiversity patterns, with implications for forecasting future impacts.

Prior work shows that climatically stable regions accumulate endemic and small-ranged species, increasing taxonomic and phylogenetic turnover, whereas rapid climatic shifts select against endemics and favor widespread generalists with strong dispersal (e.g., Dynesius & Jansson 2000; Sandel et al. 2011). Numerous studies project future losses of evolutionary heritage due to climate change and emphasize conserving phylogenetic diversity, but most focus on temporal change at single sites rather than spatial turnover among sites. European biogeography is profoundly influenced by Quaternary glaciations, with species persisting in southern refugia (Iberia, Italy, Balkans) and postglacial recolonization shaping distributions; potential northern refugia have been proposed for some taxa but remain debated. Previous research also documents limited postglacial range filling and the prevalence of large-ranged generalists in regions with rapid historical climatic change. These insights motivate examining how past climate change velocity, distance to refugia (or climate stability), and species’ range sizes structure spatial phylogenetic turnover independent of local environmental and geographic distance effects.

- Study area and data: The analysis spans most of Europe, excluding certain islands and eastern regions with limited data. Seed plant distributions (presence/absence) for all European gymnosperms and ~25% of European angiosperms were obtained from Atlas Florae Europaeae (AFE) at ~50 × 50 km resolution (1970 sampled cell centroids). Only native taxa were retained; names were standardized against The Plant List and Euro+Med. - Phylogenies: Species were matched to the PhytoPhylo megaphylogeny using S.PhyloMaker. Three tree-building approaches (A1–A3) were tested; main results use A1 (genera/species added as basal polytomies where needed), with 20 random trees sampled. Robustness was also assessed with two dated phylogenies from Smith & Brown (ALLOTB, ALLMB). After filtering, core datasets included 3904 angiosperm species (268 genera, 41 families) and 41 gymnosperm species (4 families). - Environmental distances: Current climate (19 WorldClim bioclim variables, 10 arcmin) was aggregated to 50 km; PCA on standardized variables retained 6 axes explaining 98% variance. Environmental distances among AFE points were Euclidean distances in PCA space. - Past climate metrics: Using 1000-year time slices back to 21 ka (Maiorano et al. paleoclimate), the gradient-based velocity of climate change (Vocc) for temperature and precipitation was calculated with VoCC, averaged across time, combined (maximum across variables), log-transformed, filtered (5 × 5 window), and outliers trimmed at 99.5th percentile. Climate stability (ClimStab) combined inverted variability of temperature and precipitation (product of rescaled stabilities) from LGM to present. - Refugia and distance to refugia (DistRef): For each species, logistic regression (GLM) of current occurrences on annual temperature and precipitation (linear and quadratic terms; AUC ≈ 0.89 angiosperms, 0.88 gymnosperms) was hindcast to LGM. Potential refugial cells were identified by simulating range expansions with KISSMig through 1000-year steps, selecting LGM-suitable cells whose accessibility improved model deviance explained (D²) with positive effect. Migration rates were explored (21–210 iterations since LGM; ~19–186 m/yr), choosing the rate maximizing explanatory power. For each assemblage cell, DistRef was computed as the mean minimum distance to the nearest refugial cell across its species. - Range size (RangeS): Mean range size per assemblage was computed as the mean number of occupied AFE cells across Europe for its species. - Phylogenetic turnover and correction: True phylogenetic turnover (Simpson’s phylo-βsim correcting for richness) between each focal cell and its 24 nearest neighbors was calculated (betapart). To remove effects of local environmental and geographic distance, a GLM with logit(phylo-βsim) as response and environmental and geographic distances (linear and quadratic terms) as predictors was fit; residuals define phylo-βsimc, representing spatial turnover independent of local heterogeneity. Residuals were mapped at 50 km resolution. - Explanatory models and variable importance: Standardized linear regressions with linear and quadratic terms for DistRef (or alternatively ClimStab), Vocc, and RangeS explained phylo-βsimc. Model fit (R²) and relative importance (lmg in relaimpo) were reported. Correlations among predictors were assessed; DistRef and ClimStab were highly correlated (R ≈ −0.77 to −0.79). Mapping used LAEA projection over European hillshade.

- Spatial patterns: Phylo-βsimc is strongly structured across Europe and generally higher in Southern Europe than in Northern Europe for both angiosperms and gymnosperms. Low phylo-βsimc occurs widely in Northern Europe and also within, north, and east of the Alps for angiosperms; higher values appear in Southern Europe, the Carpathians, Benelux, Northern Germany, and South England. Gymnosperms show similar gradients, with low values along much of the Northern Atlantic coast and the Tatra mountains. - Drivers and model performance: • Angiosperms: The model explained R² = 0.49 of phylo-βsimc variance. Relative variable importance: Distance to refugia (DistRef) 68%, Velocity of climate change since LGM (Vocc) 18%, Mean range size (RangeS) 14. • Gymnosperms: The model explained R² = 0.36. Relative importance: DistRef 43%, Vocc 25%, RangeS 32. • Using climate stability (ClimStab) instead of DistRef yielded similar explanatory power; DistRef and ClimStab were highly correlated (r = −0.77 angiosperms; r = −0.79 gymnosperms). - Relationship directions: Phylo-βsimc was negatively related to distance to refugia (closer to refugia → higher turnover) and negatively related to past climate change velocity; it was positively related to climate stability. Range size contributed to explaining patterns across both clades. After correcting for species richness, environmental and geographic distances still positively predicted true phylogenetic turnover (adjusted R²: 0.31 angiosperms; 0.13 gymnosperms), and removing these local distance effects (to obtain phylo-βsimc) had the strongest impact in mountainous and coastal areas. - Refugia signal: Areas of higher phylo-βsimc in Central and Northern Europe coincide with hypothesized northern refugia, supporting persistence of some lineages outside the well-known southern refugia. - Homogenization: High past climate change velocity and greater distances from refugia are associated with low phylo-βsimc (spatial homogenization) across Central and Northern Europe, indicating a lasting imprint of rapid Quaternary climate change on regional phylogenetic diversity.

The findings support the hypotheses that rapid past climate change and distance from refugia have imposed a strong, enduring legacy on the spatial structuring of Europe’s seed plant phylogenetic diversity, independent of local environmental and geographic heterogeneity. Low phylo-βsimc across Central and Northern Europe reflects regional homogenization plausibly driven by postglacial recolonization by closely related, widespread generalist species with higher dispersal capacities and broader niches, while many small-ranged specialists failed to fully recolonize newly suitable northern areas within the Holocene timeframe. The strong negative association with climate change velocity highlights the sensitivity of regional lineage turnover to the pace of climatic displacement, and the negative association with distance to refugia underscores the dominant role of historical range dynamics over contemporary environmental gradients in structuring present-day assemblages. Elevated phylo-βsimc in locations aligning with proposed northern refugia provides biogeographic support for their existence and influence. Robustness across phylogenetic sources (PhytoPhylo and Smith & Brown trees) and across both major seed plant clades suggests generality of the mechanisms. Together, these results imply that future climate change, projected to be of comparable or greater magnitude, may further homogenize regional phylogenetic diversity, compounding local losses and threatening evolutionary heritage.

This study demonstrates that rapid Quaternary climate change led to long-lasting spatial homogenization of phylogenetic diversity in European seed plants, with low phylo-βsimc predominating in Central and Northern Europe and higher turnover near glacial refugia. Past climate change velocity and distance to refugia emerged as primary drivers, with species’ range sizes also contributing, and signals were consistent across angiosperms and gymnosperms and robust to phylogenetic uncertainty. The work highlights the enduring imprint of historical climate dynamics on regional biodiversity patterns and warns that ongoing and future climate change may intensify spatial homogenization and loss of evolutionary heritage. Future research could refine estimates with broader taxonomic coverage and finer spatial resolution, incorporate additional ecological and functional traits, and integrate dynamic dispersal processes to improve forecasts of spatial phylogenetic responses under climate change.

- Taxonomic coverage: Only about 25% of European angiosperms are mapped in AFE (all gymnosperms included); 51 angiosperm genera absent from the backbone phylogeny were excluded. Results were tested for robustness with alternative phylogenies and methods but incomplete coverage remains a constraint. - Spatial resolution and extent: Analyses are at ~50 km resolution; finer-scale heterogeneity is intentionally removed to reveal large-scale legacies, which may obscure local processes. Some regions (e.g., Iceland, Balearics, Malta, parts of Eastern Europe) were excluded due to data limitations. - Model assumptions: Refugia identification relies on GLMs with only annual temperature and precipitation and on KISSMig’s simple neighborhood dispersal algorithm; true migration rates are unknown and were approximated by exploring a range of rates. Paleoclimate reconstructions (1000-year steps) and derived metrics (Vocc, climate stability) carry uncertainties. - Predictor collinearity: Distance to refugia and climate stability are highly correlated, precluding their simultaneous inclusion; interpretations rely on separate models. - Data processing choices: Subspecies mergers, synonym standardization, addition of taxa as polytomies, Vocc outlier trimming and smoothing may influence estimates. Environmental and geographic corrections depend on model form (linear/quadratic) and neighborhood size (24 nearest neighbors).