Long-term isolation of European steppe outposts boosts the biome's conservation value

The Eurasian steppes form the second-largest continuous biome on Earth, covering up to 7% of land surface and serving key ecological roles such as carbon sequestration. Outside the macroclimatically determined zonal steppes that extend from Europe to China, Europe hosts highly disjunct extrazonal steppe outposts embedded within forest matrices from the Carpathians to the Iberian Peninsula. Two competing hypotheses exist for the origin of European extrazonal steppes and their biota: (1) they are young remnants of the Pleistocene steppe belt with shallow or no divergence from zonal steppe populations; or (2) they represent long-term persisting refugial populations with deep divergence due to drift/adaptation. Resolving these hypotheses has direct implications for conservation prioritization, particularly given the imbalance between the area protected in zonal steppes (~1.4 million km²) versus EU-protected extrazonal steppic grasslands (~6841 km²). The study asks: Are European extrazonal steppes conservation-relevant independent entities or merely outposts of the zonal steppe belt? And, if conservation-relevant, should efforts target the entire extrazonal system or specific subregions? The authors define six biome-specific criteria (deepest phylogenetic splits, admixture patterns, divergence timing, refugial stability, phylogenetic diversity, and PD complementarity distribution) to test these questions using cross-phyla genomic and ecological analyses across six widespread steppe species.

Background literature indicates exceptionally high small-scale plant diversity in steppes, threats from agricultural intensification, and extensive historical loss of steppe habitats. Classical biogeographic hypotheses posited that European extrazonal steppes are young relicts from Pleistocene cold stages (e.g., Braun-Blanquet; Jännicke; de Soo). Conversely, phylogeographic studies suggest potential long-term persistence in refugia leading to deep lineage divergence. Conservation frameworks emphasize phylogenetic diversity and complementarity as key metrics for prioritization. Regional syntheses document the limited extent of extrazonal steppes (e.g., Pannonian basin potential extent ~37,000 km²) and uneven protection coverage between zonal and extrazonal steppes. Prior work on focal taxa (e.g., Euphorbia seguieriana) suggested Pleistocene diversification. Collectively, these studies motivate a comparative, multi-taxon, range-wide approach integrating genomics and ecological niche modeling to reassess the age and independence of extrazonal steppes and to inform conservation.

Study system and sampling: Six characteristic steppe species were selected: plants Astragalus onobrychis (Fabaceae), Euphorbia seguieriana (Euphorbiaceae), Stipa capillata (Poaceae); arthropods Omocestus petraeus and Stenobothrus nigromaculatus (Acrididae), Plagiolepis taurica (Formicidae). Sampling spanned 2014–2017 across full ranges with denser coverage in extrazonal Europe. Total: 456 populations from 320 localities; 370 populations from 266 localities sequenced via RADseq. Iberian S. capillata was sparsely sampled outside the Pyrenees due to evidence of a cryptic species. Genomic data generation: DNA extraction employed CTAB-based protocols for plants and Qiagen kits for animals. Genome sizes were estimated via flow cytometry to guide restriction enzyme choice. RADseq libraries followed published protocols with minor modifications. RADseq processing and SNP calling: Reads were demultiplexed and quality filtered (Picard BamIndexDecoder; Stacks process_radtags). For grasshoppers with large, repetitive genomes (O. petraeus, S. nigromaculatus), reads were mapped to a RepeatMasker-masked Locusta migratoria reference genome using Stampy; mapped reads proceeded to Stacks ref_map. For other taxa, de novo assembly and SNP calling used Stacks denovo_map with species-optimized clustering parameters to minimize paralogy. Filters removed loci with evidence of >2 alleles and excessive deleveraged tags. Datasets: (i) STRUCTURE input with one random SNP per locus; (ii) concatenated RAD loci for phylogenetic analyses (phylip format). Population structure analysis: STRUCTURE v2.3.4 ran admixture model with uncorrelated allele frequencies. Ten replicates for K=1–10, burn-in 200,000 and 2,000,000 MCMC iterations. Optimal K determined by likelihood plateau, cluster stability, non-empty clusters, and ΔK criterion. Phylogenetic inference: Alignments imported into R (phrynomics), filtered to remove loci/individuals with >75% missing or ambiguous at polymorphic sites (>85% for O. petraeus). ML phylogenies inferred with RAxML v8.2.8 under ASC_GTRCAT with ascertainment bias correction; model selection via SMS. Bootstrap with frequency-based stopping. Outgroups used for rooting (per Supplementary Table 7). Additional population-level trees (one random individual per population) were inferred similarly for visualization. Ecological niche modeling (ENM): Presence-only data were curated to reduce sampling bias: excluded sparsely sampled Eurasia east of Crimea; thinned occurrences at 5 km using PASSAGE2. Final dataset: 380 records. Climatic predictors from WorldClim v1.4 at 30 arc-sec; improved precipitation layers for Alps using 23 stations. Variables with Pearson r>0.9 removed; final 11 bioclim variables: bio2, bio3, bio4, bio5, bio9, bio10, bio11, bio12, bio15, bio16, bio17. MAXENT v3.3.3.k used with species-specific tuning and jackknife tests; evaluated via AUC and omission rates. Projections to Last Glacial Maximum using MIROC3 and CCSM3 with clamping; MESS analyses assessed analogue climates. Consensus maps averaged MIROC and CCSM projections. Areas of Stability defined as cells above the Maximum Training Sensitivity Plus Specificity (MTSS) threshold and ice-free during Pleistocene cold stages; average suitability maps computed post-thresholding. Mitochondrial DNA sequencing and dating: COI sequenced for O. petraeus and P. taurica. To avoid COI pseudogene amplification in grasshoppers, species-specific primers were designed from COI cDNA templates; Sanger sequencing performed. Substitution models selected via jModelTest (TrN+G for P. taurica; GTR+I+G for O. petraeus). Divergence dating used BEAST v1.8 with an uncorrelated lognormal relaxed clock, applying an arthropod COI rate of 0.01615 substitutions/site/Myr (3.2% divergence/Myr). Two runs of 10 million generations; parameter convergence assessed in TRACER (ESS>200); maximum clade credibility trees summarized with TreeAnnotator. Phylogenetic diversity (PD) and PD complementarity: PD computed on complete ML phylogenies; rarefaction via R package phylolrare to account for uneven sampling between extrazonal and zonal regions. Spatial PD complementarity assessed using 1-degree grid cells across extrazonal steppes, comparing each cell’s branch lengths against all branches present in zonal steppes as reference (PDA v1.0.3). To standardize for sampling intensity, datasets were repeatedly downsampled to one branch per grid cell (100 replicates), and median PD complementarity per cell was scaled 0–1 by each taxon’s maximum. Aggregate PD complementarity also computed for all extrazonal vs all zonal steppes. Data/code availability: RADseq and mtDNA sequences deposited in NCBI (accessions in supplements). Source data for figures provided. Custom code for PD downsampling available at https://github.com/philippkirschner/PD_downsampler.

• Genomic divergence and structure: Across six taxa, ML phylogenies recovered well-supported clades composed exclusively of extrazonal lineages. Except for Euphorbia seguieriana (where extrazonal lineages are derived and nested within zonal lineages), the deepest phylogenetic splits separated extrazonal from zonal lineages, indicating long-term independence. STRUCTURE analyses consistently supported K=2, largely mirroring extrazonal vs zonal lineages with minimal admixture. Notable exceptions include Stipa capillata (one Alpine cluster vs all other populations with local admixture) and E. seguieriana (broad admixture zone linking zonal and easternmost extrazonal populations, plus divergent Caucasian populations).
• Divergence timing: COI-based dating placed splits between zonal and extrazonal lineages in the early to mid-Pleistocene: Plagiolepis taurica ~1.50 Mya (95% HPD 0.92–2.08 Mya) and Omocestus petraeus ~2.46 Mya (95% HPD 1.00–4.52 Mya), congruent with RADseq topology.
• Ecological niches and refugia: ENMs performed well (high AUC, low omission) and predicted suitable conditions in present extrazonal areas, with some additional high-suitability regions (e.g., southern Denmark for multiple species). Zonal steppe areas (e.g., western Ukraine) showed lower suitability relative to extrazonal conditions, driven by temperature and precipitation seasonality differences; background tests confirmed significant niche divergence between zonal and extrazonal occurrences for all six species. MESS indicated moderate to high transferability to LGM climates. Areas of climatic stability (refugia) persisted throughout Quaternary cycles in extrazonal Europe, especially along Alpine margins outside LGM ice extent, and were never connected to larger Balkan or Pannonian refugia; small isolated refugia were also inferred north and west of the Alps for most species.
• Phylogenetic diversity: Extrazonal steppes harbored substantially higher downsampled PD than zonal steppes across taxa. PD complementarity was greater in extrazonal than zonal regions for all six taxa, indicating a larger amount of unique evolutionary history. Within extrazonal Europe, PD complementarity was broadly and relatively evenly distributed rather than concentrated, with recurrent hotspots by species in the Alps, southern France, Apennine Peninsula, and steppe relicts north of the Alps; species-specific maxima varied (e.g., O. petraeus in southern France; S. nigromaculatus in the Apennines; P. taurica in central Germany; E. seguieriana and S. capillata in the Western Alps; A. onobrychis in the southern Balkans).
• Conservation implications: European extrazonal steppe outposts contain endemic lineages, deep phylogenetic splits from zonal lineages, multiple stable refugia, and elevated unique PD. These features collectively satisfy predefined conservation criteria and demonstrate that conserving extrazonal steppes is essential to preserve the genetic legacy of the Eurasian steppe biome.

The study resolves long-standing biogeographic hypotheses by demonstrating that European extrazonal steppe biota are not recent offshoots of the zonal steppe belt but largely represent independently evolving lineages that have been isolated since the onset of Pleistocene glaciations. Minimal admixture zones and deep splits refute a simple recent recolonization model and instead support vicariance and long-term persistence in scattered refugia. In contrast to many temperate forest taxa that suffered diversity losses during Pleistocene cycles, cold-tolerant steppe taxa experienced range expansions during glacial stages and contractions during interglacials, fostering genetic diversification within extrazonal Europe. ENMs identify multiple, often small and isolated, long-term refugia outside the Pannonian Basin and Pontic plains, highlighting the role of extra-Pannonian refugia in steppe diversification. Elevated and widely distributed PD complementarity across extrazonal regions means no single area captures the evolutionary uniqueness of the extrazonal biota. Consequently, conservation strategies must recognize each European steppe outpost as a significant, largely independent conservation unit and prioritize the extrazonal steppe system as a whole. The cross-phyla congruence observed underscores the value of multi-taxon approaches in evaluating biome-level conservation priorities and calls for re-evaluation of classical biogeographic assumptions on a case-by-case basis.

By integrating genome-wide SNP data, mitochondrial dating, and ecological niche modeling across six plant and arthropod taxa, the study shows that European extrazonal steppe outposts have been long-term isolated from the Eurasian zonal steppes and harbor endemic, deeply divergent lineages, multiple stable refugia, and elevated unique phylogenetic diversity. These findings overturn the notion of extrazonal steppes as merely recent remnants and establish them as disproportionately valuable for conserving the evolutionary heritage of the Eurasian steppe biome. Conservation should target the extrazonal European steppes comprehensively rather than focusing on isolated hotspots, treating each outpost as a largely independent unit. Future research should extend multi-taxon, cross-phyla comparative frameworks to additional steppe species and regions to refine spatial conservation priorities and reassess historical hypotheses for other biomes.

• Sampling and modeling scope: ENMs excluded sparsely sampled areas east of Crimea, potentially limiting inference for eastern Eurasia. Iberian Stipa capillata was not exhaustively sampled outside the Pyrenees due to evidence of a cryptic species. Within-species support in subclades varied and was not explored in detail.
• Methodological constraints: Grasshopper RADseq reads required mapping to a masked reference genome to mitigate paralogy; residual biases may persist. MESS analyses indicated variable climatic analogue quality between LGM models (CCSM3 vs MIROC), introducing projection uncertainty mitigated by consensus averaging. STRUCTURE-based K selection and admixture inference, while robust, may miss finer hierarchical structure.
• Generalizability: Conclusions are based on six representative taxa; while cross-phyla congruence is strong, additional taxa may reveal further complexity in lineage histories and refugial patterns.