The temperature of surface and epilimnetic waters, closely related to regional air temperatures, responds quickly and directly to climatic changes. As a result, lake surface temperature (LSWT) can be considered an effective indicator of climate change. In this study, we reconstructed and investigated historical and future LSWT across different scenarios for over 80 major lakes in mainland Southeast Asia (SEA), an ecologically diverse region vulnerable to climate impacts. Five different predicting models, incorporating statistical, machine and deep learning approaches, were trained and validated using ERA5 and CHIRPS climatic feature datasets as predictors and 8-day MODIS-derived LSWT from 2000 to 2020 as reference dataset. Best performing model was then applied to predict both historical (1986-2020) and future (2020-2100) LSWT for SEA lakes, utilizing downscaled climatic CORDEX-SEA feature data and multiple Representative Concentration Pathway (RCP). The analysis uncovered historical and future thermal dynamics and long-term trends for both daytime and nighttime LSWT. Among 5 models, XGboost results the most performant (NSE 0.85, RMSE 1.14 °C, MAE 0.69 °C, MBE -0.08 °C) and it has been used for historical reconstruction and future LSWT prediction. The historical analysis revealed a warming trend in SEA lakes, with daytime LSWT increasing at a rate of +0.18 °C/decade and nighttime LSWT at +0.13 °C/decade over the past three decades. These trends appeared of smaller magnitude compared to global estimates of LSWT change rates and less pronounced than concurrent air temperature and LSWT increases in neighbouring regions. Projections under various RCP scenarios indicated continued LSWT warming. Daytime LSWT is projected to increase at varying rates per decade: +0.03 °C under RCP2.6, +0.14 °C under RCP4.5, and +0.29 °C under RCP8.5. Similarly, nighttime LSWT projections under these scenarios are: +0.03 °C, +0.10 °C, and +0.16 °C per decade, respectively. The most optimistic scenario predicted marginal increases of +0.38 °C on average, while the most pessimistic scenario indicated an average LSWT increase of +2.29 °C by end of the century. This study highlights the relevance of LSWT as a climate change indicator in major SEA's freshwater ecosystems. The integration of satellite-derived LSWT, historical and projected climate data into data-driven modelling has enabled new and a more nuanced understanding of LSWT dynamics in relation to climate throughout the entire SEA region.
Deforestation, climate change and invasive species constitute three global threats to biodiversity that act synergistically. However, drivers and rates of loss of freshwater biodiversity now and in the future are poorly understood. Here we focus on the potential impacts of global change on freshwater mussels (Order Unionida) in Sundaland (SE Asia), a vulnerable group facing global declines and recognized indicators of overall freshwater biodiversity. We used an ensemble of distribution models to identify habitats potentially suitable for freshwater mussels and their change under a range of climate, deforestation and invasion scenarios. Our data and models revealed that, at present, Sundaland features 47 and 32 Mha of habitat that can be considered environmentally suitable for native and invasive freshwater mussels, respectively. We anticipate that by 2050, the area suitable for palm oil cultivation may expand between 8 and 44 Mha, representing an annual increase of 2-11%. This is expected to result in a 20% decrease in suitable habitat for native mussels, a drop that reaches 30% by 2050 when considering concomitant climate change. In contrast, the habitat potentially suitable for invasive mussels may increase by 44-56% under 2050 future scenarios. Consequently, native mussels may compete for habitat, food resources and fish hosts with invasive mussels across approximately 60% of their suitable range. Our projections can be used to guide future expeditions to monitor the conservation status of freshwater biodiversity, and potentially reveal populations of endemic species on the brink of extinction. Future conservation measures-most importantly the designation of nature reserves-should take into account trends in freshwater biodiversity generally, and particularly species such as freshwater mussels, vital to safeguard fundamental ecosystem services.