Projected climatic refugia for Taxus wallichiana in Lam Dong Province, Southern Vietnam

MAI-PHUONG PHAM (Joint Vietnam-Russia Tropical Science and Technology Research Center. 63 Nguyen Van Huyen, Nghia Do 10000, Hanoi, Vietnam)
DINH DUY VU (Joint Vietnam-Russia Tropical Science and Technology Research Center. 63 Nguyen Van Huyen, Nghia Do 10000, Hanoi, Vietnam)
THI HANH TONG (Le Quy Don University. 236 Hoang Quoc Viet, Nghia Do 11307, Hanoi, Vietnam)
QUOC KHANH NGUYEN (Joint Vietnam-Russia Tropical Science and Technology Research Center. 63 Nguyen Van Huyen, Nghia Do 10000, Hanoi, Vietnam)
PLUGATAR YURI VLADIMIROVICH (The Nikita Botanical Gardens, National Scientific Center of the RAS. 52 Nikitsky Spusk, Nikita, Yalta 298648, Republic of the Crimea, Russia)
SHEVCHUK OKSANA MIKHAILOVNA (The Nikita Botanical Gardens, National Scientific Center of the RAS. 52 Nikitsky Spusk, Nikita, Yalta 298648, Republic of the Crimea, Russia)
PAPELBU VLADIMIR VLADIMIROVICH (The Nikita Botanical Gardens, National Scientific Center of the RAS. 52 Nikitsky Spusk, Nikita, Yalta 298648, Republic of the Crimea, Russia)
SAKHNO TATIANA MIKHAILOVNA (The Nikita Botanical Gardens, National Scientific Center of the RAS. 52 Nikitsky Spusk, Nikita, Yalta 298648, Republic of the Crimea, Russia)
SMIRNOV VASILY ANDREEVICH (The Nikita Botanical Gardens, National Scientific Center of the RAS. 52 Nikitsky Spusk, Nikita, Yalta 298648, Republic of the Crimea, Russia)
GERASIMCHUK VOLODYMYR NIKOLAEVICH (The Nikita Botanical Gardens, National Scientific Center of the RAS. 52 Nikitsky Spusk, Nikita, Yalta 298648, Republic of the Crimea, Russia)

Biodiversitas Journal of Biological Diversity
Vol. 27 No. 9 (2026)

Abstract. Pham M-P, Vu DD, Tong TH, Nguyen QK, Vladimirovich PY, Mikhailovna SO, Vladimirovich PV, Mikhailovna ST, Andreevich SV, Nikolaevich GV. 2026. Projected climatic refugia for Taxus wallichiana in Lam Dong Province, Southern Vietnam. Biodiversitas 27 (9): d270910. https://doi.org/10.13057/biodiv/d270910. Taxus wallichiana (Taxaceae) is an endangered montane conifer for which climate-resilient conservation areas need to be identified. This study evaluated long-term climatic refugia in Lam Dong Province, Vietnam, by integrating a field-survey dataset of 758 individuals with ecological niche modeling and an exploratory overlay of previously published population-genetic parameters. Long-Term Climatic Refugia (LTCR) were delineated from habitat-suitability projections based on 52 spatially thinned wild occurrence records, climatic and topographic predictors, and five CMIP6 GCMs under SSP2-4.5. Previously published He and Fis values from four populations were used only as an exploratory post-hoc overlay to support conservation interpretation rather than as SDM predictors or validated conservation-zonation data. The findings highlighted ecologically suitable areas located within BiDoup-Nui Ba National Park and Dalat Pine Forest, warranting priority attention for conservation and future genetic-resource management under five CMIP6 GCMs SSP2-4.5 (ACCESS-CM2, CMCC-ESM2, EC-Earth3-Veg, HadGEM3-GC31-LL, and MIROC6). Ecological niche modeling revealed that elevation and annual mean temperature were the dominant variables influencing model performance. Under random bootstrap partitioning, the presence-only workflow (M2) yielded higher internal discrimination metrics (AUC: 0.987±0.004) than the pseudo-absence workflow (M1; AUC: 0.724±0.021). Because the workflows differ in sampling design and random splits may retain spatial autocorrelation, these statistics are interpreted as internal, workflow-specific validation rather than independent external predictive accuracy. M2 was therefore used as the operational basis for LTCR mapping. Across the five GCMs, high-suitability habitat declined from 612.0 km2 (2.53% of Lam Dong Province) to an ensemble mean of 549.6 km2 (2.27%), a mean decline of 10.19%. The stricter intersection of current suitability with all five future GCMs delineated 250.0 km2 (1.03%) of persistent consensus refugia, 59.15% smaller than the current high-suitability baseline; this 59.15% value represents the strict persistence filter, not the mean projected habitat loss.