Climate and environmental changes significantly impact lake thermal structure, affecting ecological functions like nutrient cycling and deepwater oxygen levels. Increased thermal stratification isolates deeper waters, reducing vertical mixing and influencing nutrient and oxygen availability, primary productivity, and fisheries. These deeper waters are critical habitats for temperature-sensitive organisms and sites of biogeochemical processes (phosphorus release, methane production). Understanding long-term changes in thermal stability and deepwater temperatures is crucial for assessing lake ecosystem health globally. However, most global studies focus on surface water temperature trends, leaving a knowledge gap regarding whole-lake thermal changes and their drivers.

Previous research, primarily focusing on surface water temperatures, has shown consistent and rapid warming in lakes worldwide. One global study examining 26 large lakes found that deepwater temperature trends averaged +0.04 °C decade⁻¹ but were highly variable. This inconsistency contrasts with the consistent surface water warming, highlighting the need to understand the differing drivers influencing surface and deepwater temperatures. The inconsistent warming rates between surface and deeper waters can lead to trophic mismatches and affect habitat availability for various organisms.

This study analyzed a long-term (1970–2009) dataset of summertime vertical lake temperature profiles from 102 lakes across five continents. Five thermal metrics were used: surface water temperature, deepwater temperature, mean water column temperature, density difference, and thermocline depth. The lakes varied widely in location, elevation, water quality, trophic status, and morphometry. Lakes were classified based on lake thermal region, a global classification system based on seasonal surface temperature dynamics. Two primary questions were addressed: (1) How has vertical lake thermal structure changed? (2) Do lake characteristics (thermal region, geography, morphometry, water quality) explain observed trends? A single temperature profile per year representing maximum summer stratification was selected for each lake. Sen's slope was used to calculate temporal trends. A random forest analysis with ten explanatory variables (thermal region, latitude, elevation, surface area, maximum depth, Secchi depth, chlorophyll-a, DOC, browning region, and mixing type) was used to determine the importance of lake characteristics in explaining trends in thermal metrics.

The study revealed strong surface water warming trends (+0.37 °C decade⁻¹ for 1970–2009) and significant increases in density difference (+0.08 kg m⁻³ decade⁻¹). Deepwater temperatures, however, showed no significant overall trend (+0.06 °C decade⁻¹), exhibiting high variability across lakes. There was no relationship between surface water temperature trends and deepwater temperature trends (r = 0.09, p = 0.12), or between density difference trends and deepwater temperature trends (r = −0.08, p = 0.17). Random forest analysis revealed that only a small percentage of the total variance in deepwater temperature trends (8.4%) could be explained by the tested lake characteristics. Surface area, thermal region, elevation, and DOC were the most important predictors, with small lakes showing decreasing deepwater temperatures and larger lakes showing slow warming. Northern Warm and Northern Hot lakes exhibited rapid deepwater cooling. Mean water column temperature trends showed 15.6% of the total variance explained by thermal region. Surface water temperature trends (3.5% variance explained) were best predicted by maximum depth, with shallower lakes showing more rapid warming. Density difference trends (16% variance explained) were also primarily predicted by maximum depth, with shallower lakes experiencing the most rapid increases in density difference. Thermocline depth trends were not explained by the tested variables.

The high variability and low explanatory power of lake characteristics in predicting deepwater temperature trends suggest that factors beyond those examined (geomorphometry, thermal region) are significant drivers. These could include climate variables (wind speeds, ice breakup, snowpack), groundwater flux, and changes in water transparency due to eutrophication or browning. Decreased water transparency leads to surface warming and deepwater cooling, increasing thermal stratification. Droughts can cause opposing effects. The under-representation of certain lake types (e.g., Northern Frigid lakes) in the dataset may limit generalizability of results. The consistent increase in thermal stratification strength, especially in small lakes, has predictable ecological implications.

Consistent surface water warming and increased thermal stratification are observed, but deepwater temperature trends show high variability. Tested lake characteristics only partially explain these trends; external drivers such as climate and water quality likely play major roles. Future research should focus on improving geographic coverage, integrating dynamic time series, and studying various processes affected by thermal structure changes (e.g., nutrient cycling, oxygen depletion, greenhouse gas production).

The dataset has geographic biases, with over-representation of lakes in North America and Europe. The under-representation of high-latitude lakes may limit understanding of their specific responses to warming. The focus on the summer stratified period may not fully capture the complete picture of annual thermal dynamics. The lack of continuous time series for water quality variables hinders a thorough investigation of their effects on thermal structure.