Climate change could erase the last strongholds of a vulnerable karst-dwelling tiger gecko

A new species distribution modeling study predicts that Goniurosaurus luii, the Lui tiger gecko listed as Vulnerable on the IUCN Red List, could lose essentially all of its highly suitable habitat by the end of the century under moderate-to-high greenhouse gas scenarios. Using the maximum entropy algorithm (MaxEnt) trained on 45 occurrence records and six bioclimatic variables, researchers projected the species’ current and future potential distribution across a 218,176 square kilometer study area spanning southern China and northern Vietnam. Under the ssp370 and ssp585 pathways, the area of highly suitable habitat within the karst zone drops to zero square kilometers by the 2090s, and under ssp370 it is already effectively gone by the 2070s, when only 13 square kilometers remain.

The Lui tiger gecko was first discovered in southern China and later found in the northern Vietnamese provinces of Cao Bang and Lang Son. It is a specialist of limestone karst forest mountains, a habitat that has been heavily fragmented by timber logging, quarrying for cement production, and road and infrastructure construction. The species has also suffered from unsustainable collection for the pet trade, both domestically and internationally, which has caused local extirpations. These pressures have earned G. luii a place on the IUCN Red List as Vulnerable, in CITES Appendix II, and in Vietnam’s Decree No. 84/2021/ND-CP (Group IIB). Like other tiger geckos of the genus, which occur in the Ryukyu Archipelago of Japan, southern China, and northern Vietnam, it combines high habitat specialization with low dispersal ability, small population sizes, and low levels of genetic diversity, a suite of traits that makes it particularly susceptible to a rapidly changing climate.

To build the model, the researchers compiled occurrence data from the Global Biodiversity Information Facility, supplemented with records from the literature, interviews with local people, and direct field observations. Two coordinates from Hainan Island were excluded because taxonomic re-assessments confirmed that all Goniurosaurus populations on that island belong to the G. lichtenfelderi group, not G. luii. After removing neighboring points within a one-kilometer radius, 45 representative records remained. Nineteen bioclimatic variables from Worldclim 2.1 at 30 arc-second resolution were screened for multicollinearity using pairwise Pearson correlation analysis, and variables with correlation coefficients above |0.75| were eliminated, leaving six predictors: Mean Diurnal Range, Temperature Seasonality, Maximum Temperature of the Warmest Month, Annual Precipitation, Precipitation of the Driest Month, and Precipitation Seasonality.

Future projections relied on the MRI-ESM2-0 global circulation model from the sixth Coupled Model Intercomparison Project (CMIP6), chosen for its ability to capture regional monsoon dynamics, ocean-atmosphere interactions, and tropical cyclone behavior. Three Shared Socioeconomic Pathways were evaluated over three 20-year periods: ssp245, the middle-of-the-road scenario with roughly 2.7 degrees Celsius of warming by century’s end; ssp370, the regional rivalry scenario with 3.2 to 4.1 degrees Celsius; and ssp585, the fossil-fueled development scenario with 4.7 to 5.1 degrees Celsius. To avoid overfitting with a small number of presence locations, the team used the ENMeval package to test 130 candidate model configurations, varying feature combinations and regularization multipliers from 0.5 to 5, evaluated with four-fold spatial block cross-validation. The optimal configuration combined the LQHPT feature classes with a regularization multiplier of 0.5, and the final model was run with 100 bootstrap replications.

The optimized model achieved a high average training AUC of 0.988, indicating strong discriminative performance. Temperature Seasonality emerged as the most important predictor, contributing nearly 50 percent of the model, followed by Precipitation of the Driest Month, Precipitation Seasonality, and Annual Precipitation; together these four variables accounted for 91 percent of model contribution and 88.2 percent of permutation importance. The response curves showed that G. luii is predicted to occur where temperature seasonality ranges from 451.5 to 625.7 (standard deviation multiplied by 100), precipitation of the driest month ranges from 7 to 34 millimeters, precipitation seasonality ranges from 73.3 to 92.1 (coefficient of variation), and annual precipitation ranges from 1,042 to 2,081 millimeters per year. These values characterize the tropical monsoon climate of the species’ range, with marked seasonal temperature differences and a pronounced dry winter.

Under current conditions, the model predicted highly suitable habitats almost entirely within the border regions of Guangxi Province in China and the Cao Bang and Lang Son provinces of Vietnam. Areas of lower climatic suitability were projected in southern Lang Son and in several other Vietnamese provinces, including Bac Kan, Ha Giang, Quang Ninh, Thai Nguyen, Tuyen Quang, Vinh Phuc, and Yen Bai, as well as some districts of Guangxi and Yunnan in China. The total area of suitable habitat under the minimum training presence threshold was calculated at approximately 24,250 square kilometers, of which more than 90 percent, or 21,987 square kilometers, fell within an 80-kilometer buffer around known occurrences. Because G. luii is a karst-adapted specialist, the predictions were overlaid with the karst layer from the World Karst Aquifer Map: within the karst buffer, suitable habitat covered 15,407 square kilometers, of which 3,175 square kilometers, about 20.6 percent, qualified as highly suitable under the stricter 10 percent training omission cloglog threshold.

The future projections revealed a striking pattern. Although the total climatically suitable area fluctuated among periods and pathways, and in some cases exceeded the current extent, the overlap between current and future suitable habitats, the region of stability, covered less than 35 percent, while the regions of loss and expansion each exceeded 90 percent of the respective layers. Multivariate environmental similarity surface (MESS) analyses documented a replacement of interpolated habitats, which are environmentally similar to the model’s training range, by novel extrapolated conditions in the future. Extracted climatic values at the occurrence sites differed significantly between the present and each future scenario, with t-tests yielding P-values below 0.01. Most critically, when the 10 percent training omission threshold was applied to identify highly suitable core habitat, all scenarios projected significantly smaller areas than the current prediction, and novel climatic conditions covered the entire study area by the 2090s under all three pathways.

The authors caution that the apparent expansion of total suitable area in some future scenarios should not be read as good news, because the new areas lie under novel climatic conditions outside the species’ known environmental tolerances. G. luii follows an evolutionary pattern of niche conservatism, retaining ancestral niche preferences rather than adapting to new conditions, a pattern identified in most species of the Chinese and Vietnamese tiger gecko group. By contrast, the Vietnamese species G. catbaensis and G. lichtenfelderi, which show niche divergence, were forecast to be less susceptible to climate change. Previous distribution modeling of G. catbaensis, G. lichtenfelderi, and G. huuliensis similarly projected significant shrinkage and poleward shifts of their current ranges. The loss of genetic diversity expected from population declines and habitat loss further reduces the likelihood that G. luii could adapt in place, and the pet trade has already depressed population sizes and genetic diversity, contributing to local extirpations in China.

Dispersal offers little escape. Range-restricted tiger geckos cannot migrate the large distances required to reach climatic refugia, both because of their habitat specialization and because forest corridors connecting karst blocks remain unsettled, and the refugia themselves may not persist under future climates. Karst ecosystems, with their complex topography and variable micro-climatic conditions, can create unique micro-refugia that buffer species against suboptimal regional climates, and high levels of diversity in gecko genera such as Cyrtodactylus and Cnemaspis have been documented in these formations. But this last survival chance is narrowing as deforestation and quarrying destroy the micro-refugia within karst formations. The study’s authors therefore highlight the green patches of karst habitat along the China-Vietnam border, corresponding to the modeled core refugia in southern Guangxi and northern Cao Bang and Lang Son, as the highest-priority areas for karst-forest protection, arguing that safeguarding these limestone landscapes is crucial for mitigating climate change impacts and securing the long-term survival of G. luii.

The findings place the Lui tiger gecko alongside a growing list of range-restricted tropical reptiles for which climate models forecast severe range loss. More than one-third of global lizard species are projected to be at risk of extinction by the 2080s because of limited capacity to adapt to rising temperatures or shift their ranges, with tropical species of restricted range being especially vulnerable. For G. luii, the combination of niche conservatism, karst specialization, low dispersal, and ongoing exploitation means that the modeled disappearance of highly suitable habitat by the 2070s under ssp370 and by the 2090s under both ssp370 and ssp585 represents a narrowing window for conservation action in the border mountains where the species clings on.

Subject of Research: Projected impacts of climate change on the distribution and conservation of the vulnerable karst-specialist tiger gecko Goniurosaurus luii in southern China and northern Vietnam

Article Title: Slipped through the narrow door: Effects of climate change on the distribution and conservation of a vulnerable tiger gecko, Goniurosaurus luii , from China and Vietnam

Article References: Ngo, H. N., Nguyen, T. T., Nguyen, H. H., Nguyen, T. Q., Nguyen, H. Q., Ngo, H. T. T., Nguyen, H. T. T., Ha, L. T. T., Grismer, L. L., Rödder, D., & Ziegler, T. (2025). Slipped through the narrow door: Effects of climate change on the distribution and conservation of a vulnerable tiger gecko, Goniurosaurus luii, from China and Vietnam. Nature Conservation, 59, 21-41. https://doi.org/10.3897/natureconservation.59.151689

Image Credits: AI Generated

DOI: 10.3897/natureconservation.59.151689

Keywords: Goniurosaurus luii, tiger gecko, climate change, species distribution modeling, MaxEnt, karst habitat, Vietnam, China, conservation, niche conservatism, Shared Socioeconomic Pathways, IUCN Vulnerable