AbstractEcosystem states are often influenced by both concurrent and antecedent environmental drivers. However, the relative importance of antecedent conditions varies within and among ecosystems. Here, we analyzed long-term depth-profile data from 382 temperate lakes across 10 countries to assess how differential changes in spring vs. summer air temperature mediate summer water quality. We found that summer bottom-water conditions were more associated with spring air temperatures, while surface-water conditions were more associated with summer air temperatures. The relative influence of spring vs. summer temperature was mediated by lake morphometry, stratification, and latitude. Across these lakes, summer air temperatures have increased more rapidly than spring air temperatures, potentially contributing to a growing thermal difference between surface and bottom waters (median = +0.5 ºC/decade). Consequently, our results demonstrate that predicting the ecological impacts of climate change may require considering spatial differences in ecological memory within ecosystems.1. INTRODUCTIONWhile anthropogenic climate change has driven global increases in air temperature, the rate of warming varies seasonally throughout the year and spatially worldwide (Balling et al. 1998; IPCC 2023; Kharin et al. 2013; Santer et al. 2018; Stine et al. 2009). For example, winter (December, January, February) air temperatures have increased approximately four times as quickly as summer (June, July, August) air temperatures in the Arctic (Bintanja & van der Linden 2013), though the season with the most rapid change in temperature differs around the world (IPCC 2023; Santer et al. 2018). As the many components of an ecosystem often have varying sensitivity to air temperatures throughout a year, anticipating the effects of future climate change will require characterizing seasonally-distinct impacts of air temperature on ecological function. Diverging seasonal air temperature trends may contribute to phenological and spatial mismatches in the effects of climate change. Phenological mismatches occur as climate change alters the temporal dynamics of ecological interactions, including the timing of migration, breeding, and predation (Straile et al. 2015). Similarly, climate change is also altering the spatial dynamics of ecological interactions, creating spatial mismatches in habitat quality and connectivity within and among ecosystems (Schweiger et al. 2008). However, while phenological mismatches are well documented in populations and communities across multiple ecosystems (Straile et al. 2015), spatial mismatches in climate responses within ecosystems have received less attention.Freshwater lakes provide a compelling example of how spatial mismatches in climate responses may occur within one ecosystem (Del Giudice et al. 2018; Dugan 2021; Oleksy & Richardson 2021; Piccolroaz et al. 2021). While surface waters are directly impacted by concurrent air temperatures throughout the year, summer thermal stratification creates two distinct layers in many temperate lakes, limiting the transfer of heat and mass between surface and bottom waters (MacIntyre & Hamilton 2024; Figure 1). Consequently, in lakes that experience summer thermal stratification, summer bottom water conditions may be more responsive to climate change during the spring mixing period than during the concurrent summer period (Adrian et al. 2012; Dugan 2021; Gerten & Adrian 2001; Oleksy & Richardson 2021; Figure 1). If spring and summer air temperatures are changing at differing rates with climate change, this could potentially produce misaligned trends in surface vs. bottom water temperature and oxygen dynamics, with implications for habitat availability, biogeochemical cycling, and water quality. Here, we refer to the relatively greater influence of antecedent (i.e., spring) vs. concurrent (i.e., summer) conditions as ecological memory, following Ogle et al. (2015), Dugan (2021), Pilla et al. (2023), and others. Spatial mismatches in water quality trends between surface and bottom waters have both ecological and scientific implications. If surface temperatures warm faster than bottom-water temperatures, cooler bottom waters may serve as a refugia for cold-water organisms (Jane et al. 2024; Rose et al. 2016). Conversely, elevated bottom water temperature is expected to drive decreased dissolved oxygen (DO) solubility and increased rates of decomposition in bottom waters, both of which could decrease DO concentrations during the summer stratified period and degrade habitat quality for aerobic organisms (Pace & Prairie 2005; Yvon-Durocher et al. 2012; Figure 1). If bottom-water temperatures change at a different rate than surface temperatures (e.g., following Pilla et al. 2020), temperature-driven changes in bottom-water habitat quality would not necessarily be detected from summer surface-water measurements, which are much more common than bottom-water measurements across lakes (Pilla et al. 2020; Schneider & Hook 2010).As ecological memory underlies the potentially divergent effects of seasonal air temperature on surface and bottom water conditions in lakes, understanding the potential for spatial mismatch requires characterizing how ecological memory varies across lakes with differences in water chemistry, morphometry, and climate (Dugan 2021; Gerten & Adrian 2001). In bottom waters, cross-seasonal ecological memory is facilitated by summer thermal stratification (Figure 1), and will likely be strongest in lakes that exhibit strong, stable thermal stratification throughout the summer. However, these patterns may be further modulated by the morphometric, geographic, and biological characteristics of a given lake. For example, in lakes where high surface productivity results in variable inputs of phytoplankton biomass to bottom waters (Lewis et al. 2024), summer air temperature may be a relatively more important driver of summer bottom-water DO dynamics compared to lakes with low surface productivity. Similarly, latitudinal gradients regulate the degree of seasonality in air temperature, solar radiation, and other environmental drivers (Lewis, Jr. 1996) potentially resulting in latitudinal differences in cross-seasonal ecological memory. Ultimately, the relationships between lake characteristics and cross-seasonal ecological memory remain largely untested because they can only be identified through analysis of long-term depth profile data across many widespread lakes. While surface-water dynamics can be characterized across hundreds of thousands of widespread lakes using remote sensing (Khandelwal et al. 2022; Topp et al. 2021; Yang et al. 2022), bottom-water data collection is typically resource intensive, logistically challenging, and expensive, as it entails either physically traveling to the waterbody and conducting manual profiles from a boat or purchasing and maintaining in situ sensors (Pilla et al. 2020). Consequently, bottom-water data availability is much more limited across lakes.Here, we compiled and analyzed a large dataset of depth profile data from hundreds of stratified lakes to understand the influence of seasonal air temperature dynamics on surface and bottom-water temperature and DO. We focused on water temperature and DO due to the critical role that these environmental variables play in determining lake habitat quality and ecological function, and because these variables have been broadly monitored across many lakes over time (Jane et al. 2021; Pilla et al. 2020; Richardson et al. 2017). Our analysis addressed two primary research questions: Q1) What period of air temperature exerts the strongest influence over summer water temperature and DO dynamics in surface and bottom waters of temperate lakes? and Q2) Which lake characteristics control the relative influence of antecedent vs. concurrent air temperatures on surface and bottom-water temperature and DO? Through these analyses, we provide an assessment of how seasonal air temperature dynamics may contribute to spatial mismatches in water quality amidst a changing climate. 2. METHODS2.1 Dataset and data processingTo explore associations between seasonal air temperatures and water quality dynamics, we compiled and analyzed a dataset of temperature and DO profiles from 382 temperate lakes (Lewis et al. 2023). This dataset includes lakes across 10 countries and 4 continents (Figure 2), with a median depth of 15 m (Zmax range: 3–370 m) and median surface area of 1.19 km2 (range: 0.011–1269.09 km2). The median time series duration was 30 years, and all lakes had at least 15 years of data. Detailed metadata descriptions are provided by Lewis et al. (2023). All data analyses were performed in R, version 4.3 (R Core Team, 2023), and analysis code is archived on Zenodo (Lewis 2025). 2.1.1 Summer in-lake metricsWe calculated mean summer surface- and bottom-water temperature and DO for each lake-year using the top 0–1 m of the lake and the bottom 1 m of sampled depths, respectively. For these calculations, we interpolated summer temperature and DO profiles to a 1-m depth resolution, following Jane et al. (2021). We then calculated mean surface and bottom temperature and DO concentrations during the summer of each lake year, which was defined as July and August in the northern hemisphere and January and February in the southern hemisphere. Data availability differed among lakes and variables, with a median of two profiles per summer (Table S1, Figure S1). Changing the duration of time defined as “summer” or number of profiles included in the analysis did not qualitatively affect results (Figure S1; Figure S3). Changes in spring air temperature may affect bottom-water DO via multiple pathways, including altered stratification onset, oxygen solubility, and rates of oxygen demand throughout the summer stratified period (Figure 1). To assess whether spring air temperatures affected biogeochemical processing rates during the summer, we calculated bottom-water oxygen demand for each lake year. Specifically, we used bathymetric contours from Lewis et al. (2023) and interpolated DO profiles to calculate volume-weighted hypolimnetic oxygen concentrations on each measurement date in lake-years that exhibited a period of summer thermal stratification. We then calculated volume-weighted hypolimnetic oxygen demand (VHOD) as the linear rate of decline in volume-weighted DO concentrations during the stratified period (Lewis et al. 2024).We calculated the maximum buoyancy frequency throughout the water column in 1-m intervals to assess the strength of thermal stratification. Maximum buoyancy frequency was calculated for each profile using the rLakeAnalyzer R package (Winslow et al. 2019), then averaged throughout the summer of each lake-year. 2.1.2 Climate dataWe collated daily 2-m air temperature for 1980–2022 from the ERA5 climate reanalysis (Hersbach et al. 2019). ERA5 is a commonly used climate reanalysis product and provides meteorological data on a 0.25-degree global grid (Hersbach et al. 2019). Meteorological data were extracted from the nearest grid cell for each lake. 2.2 Interannual variabilityTo assess which sub-annual periods of air temperature are most strongly correlated with interannual variability in summer water temperature and DO dynamics in each lake, we calculated the relationship between mean air temperature in 30-day rolling windows (January–August in the Northern Hemisphere, offset by six months in the Southern Hemisphere) and four summer focal variables: surface-water temperature, bottom-water temperature, bottom-water oxygen demand, and bottom-water DO concentration (following e.g., Huber et al. 2010). For this analysis, we used partial Spearman’s correlations that accounted for measurement year to isolate the role of interannual variability, as opposed to long-term trends (see Section 2.3 and Figure S4). We included all lakes with at least 15 years of paired data for air temperature and any of the four focal lake variables from 1980–2022 (surface-water temperature n = 365; bottom-water temperature n = 365, bottom-water oxygen demand n = 157). For DO, we only included lakes that had a median DO >1 mg/L (n = 213), as the physical constraint that DO cannot decrease below 0 mg/L limited our ability to assess correlation between air temperature and summer DO in lakes that frequently experienced anoxia. Changing the duration of the rolling mean window did not substantially alter results (Figure S5). 2.2.1 Cross-Seasonal Ecological MemoryWe quantified the relative influence of summer vs. spring air temperatures on each summer focal lake variable (surface-water temperature, bottom-water temperature, bottom-water oxygen demand, and bottom-water DO concentration). First, we calculated the correlations between our summer focal variables and all possible 30-day rolling windows of mean spring air temperature to identify which window of air temperature exhibited the strongest correlation. For this analysis, we used spring air temperatures during March–May in the Northern Hemisphere, with Southern Hemisphere dates offset by six months. We then repeated this analysis using all possible 30-day rolling windows in the Northern Hemisphere’s July and August (Southern Hemisphere dates offset by six months). We subtracted the maximum summer correlation from the maximum spring correlation for each focal variable, to calculate a Cross-Seasonal Ecological Memory (CSEM) value for each lake (Figure S6). Consequently, positive values of CSEM indicate a stronger correlation with antecedent spring air temperatures and negative values indicate a stronger correlation with concurrent summer air temperatures (Figure S6). Because air temperature is mechanistically expected to be positively correlated with water temperature and oxygen demand but negatively correlated with bottom-water oxygen concentrations (Figure 1), we used the absolute value of the minimum (most negative) correlation for bottom-water DO and the absolute value of the maximum (most positive) correlation for all other variables. The period of air temperatures that were identified as “spring”—March through May in the Northern Hemisphere—encompass the typical onset of thermal stratification across the lakes in this dataset (Figure S7). Conversely, the period identified as “summer”—July and August in the Northern Hemisphere—match the timing of in-lake measurements used for summer water quality means (2.1.1 Summer in-lake metrics). Results did not qualitatively change if we restricted spring dates to March and April to match the two-month duration of summer air temperature dates (Figure S8). Using the maximum correlation from spring and summer rather than, e.g., the median correlation throughout these periods, allowed us to isolate the specific windows of air temperature that were most important in driving variation in each summer focal lake variable, as seasonal phenology (e.g., timing of spring thaw and stratification onset) varies with lake size, geographic location, wind dynamics, and other factors (Magee and Wu 2017; Woolway et al. 2021). However, results also did not differ qualitatively between median and maximum correlations (Figure S9). While our analysis primarily focused on DO concentrations, rather than DO percent saturation, we re-ran the CSEM calculations and found that the same patterns were observed for percent saturation (Figure S10). 2.2.2 Random Forest RegressionsWe assessed how the relative influence of spring vs. summer air temperature (i.e., CSEM) may vary in the surface and bottom waters of many lakes worldwide using random forest analyses. For these analyses, we predicted CSEM of our four focal variables based on a consistent set of three explanatory variables. The candidate explanatory variables we selected were variables that (1) could be assessed using available data and (2) were expected to potentially mediate the influence of spring air temperatures on summer water quality: lake maximum depth, stratification strength (median buoyancy frequency at the thermocline in summer), and latitude (absolute value). Correlations among explanatory variables are presented in Figure S11. We evaluated each random forest model using out-of-bounds R2 and Kling–Gupta Efficiency (KGE), then assessed the importance of each explanatory variable for each focal response by calculating the percent increase in mean squared error (MSE) that results from removing that variable from the random forest model. To assess the relationship between each driver and CSEM, we plotted the partial dependence of CSEM on each driver across the range of observed driver values. We repeated this analysis for lakes with 20+ years of data to assess the extent to which results depended on data availability (Figure S12). All random forest analyses were performed using the randomForest package in R (Breiman et al. 2024). 2.3 Multiannual trendsTo assess the relationship between decadal water temperature and seasonal air temperature trends, we calculated trends in each variable across all lakes, using Sen’s slope as a non-parametric trend statistic. We accounted for variation in seasonal phenology across lakes by using the spring and summer air temperature dates identified in section 2.2.1, and we limited the years of air temperature data to the duration of in-lake data available at each lake. Trend analyses were performed using the trend R package (Pohlert 2023). We used Spearman’s correlations to assess the relationship between multiannual spring and summer air temperature trends and bottom-water temperature trends, as well as the spatial mismatch in water temperature changes. We then calculated non-parametric trends (via Sen’s slope) in the magnitude of difference between surface and bottom water temperature, and we plotted the density distribution of these trends across all lakes. 2.4 ISIMIP-lakes model-based verificationWhile our study was focused on empirical data analysis across hundreds of temperate lakes, we sought additional mechanistic validation of the cross-seasonal ecological memory calculations by analyzing within-lake patterns over multiple decades with hydrodynamic modeling. We compiled daily water temperature estimates for the 43 lakes in our dataset that were modeled as part of the Inter-Sectoral Impact Model Intercomparison Project lakes sector (ISIMIP global lakes v3a). For this analysis, we used output from the widely-used General Ocean Turbulence Model (GOTM), which was also driven by ERA5 climate data (Golub et al. 2022). We limited our analysis to years ≥1980 to correspond with our empirical analyses above. Using the ISIMIP-modeled daily water temperatures of these 43 lakes, we re-ran the analyses described in sections 2.1 and 2.2 to assess whether the timing of stratification onset was related to the timing of peak correlation between spring air temperature and summer bottom-water temperature across lakes. Modeled daily water temperature output allowed us to precisely quantify the timing of summer stratification onset in each lake-year, which we defined as the last date before July 1 (Northern Hemisphere) or December 1 (Southern Hemisphere) with a water density threshold of <0.1 kg/m3. 3. RESULTSWe found strong support for a cross-seasonal ecological memory effect that may drive a spatial mismatch in water quality trends between surface and bottom waters of lakes worldwide (Figure 3). While summer surface-water temperature tended to be positively correlated with summer air temperatures, summer bottom-water temperature, bottom-water oxygen demand, and summer bottom-water DO (concentrations and percent saturation) tended to be more closely correlated with spring air temperatures (Figures 3, S10, S13, S14). The magnitude of cross-seasonal ecological memory in surface and bottom waters was mediated by lake characteristics. Across all surface and bottom water variables, cross-seasonal ecological memory increased with increasing thermal stratification (as buoyancy frequency). However, the effect of buoyancy frequency on ecological memory was stronger in bottom waters than surface waters (Figure 4). Omitting buoyancy frequency as a driver of cross-seasonal ecological memory led to a 37% increase in MSE for bottom-water temperature, compared to 20% in surface waters. Deeper lakes exhibited increased CSEM for bottom water temperature and decreased CSEM for surface water temperature, while high latitude lakes exhibited decreased CSEM for both surface-water temperature and bottom-water DO (Figure 4, S15). The amount of variation in CSEM that was explained by driver variables was higher for bottom-water temperature than surface-water temperature or bottom-water DO dynamics (Figure 4).The specific dates of spring air temperature that were most correlated with summer bottom-water temperature varied across lakes. Analyzing modeled water temperature from the 43 ISIMIP lakes, we found a strong positive association between median stratification onset date and the day of air temperature that was most correlated with bottom-water temperature (left-aligned 30-day rolling mean; Figures S7, S16). Likewise, in a subset of lakes where we were able to empirically estimate stratification onset (n = 27), the median stratification onset date corresponded with the window of air temperature that was most correlated with summer bottom-water temperature (Figure S13).Across the lakes in this dataset, spring air temperature trends were smaller and more variable than summer air temperature trends (mean±SD = 0.09±0.54 ºC/decade in spring vs. 0.23±0.3 ºC/decade in summer from 1980–2022; Figure 5). Decadal trends in summer bottom-water temperature were more correlated with spring air temperature trends than summer air temperature trends (Figure 5). Surface and bottom water temperature trends were positively correlated across the lakes in this dataset, but this correlation was relatively weak (Spearman’s rho = 0.17, p = 0.001; Figure 6a). Averaged across all lakes, the difference in water temperature between surface and bottom layers tended to increase over time. The median rate of change in temperature differential between surface and bottom waters was +0.5 ºC/decade (5–95% range = -1.5 to 4.0 ºC/decade; Figure 6b). Increases in thermal difference between surface and bottom waters were most commonly observed in lakes where summer air temperatures have increased more rapidly than spring air temperatures and CSEM > 0 (Figure 6c). However, increases in thermal difference over time were observed in a majority of lakes regardless of CSEM and seasonal differences in warming rates (Figure 6c). 4. DISCUSSIONThrough our analysis of 382 widespread temperate lakes, we identified that seasonal differences in climate warming may contribute to within-ecosystem spatial mismatches in climate responses. While surface-water temperatures are increasing in conjunction with increased summer air temperature, bottom-water temperatures are more strongly associated with spring air temperatures, which are not changing as rapidly across lakes (Figure 5). Our results help explain why previous studies have found little to no association between summer air temperature and summer bottom-water dynamics (Pilla et al. 2020; Winslow et al. 2017; Zhou et al. 2023), and highlight the important indirect role that climate change can play in driving changes in bottom-water temperature and DO. The magnitude of ecological memory in both surface and bottom waters varied substantially among lakes, complicating efforts to predict future climate-driven changes in water quality. In both surface and bottom waters, polymictic and weakly-stratified systems were comparatively more impacted by concurrent vs. antecedent temperatures, potentially because these systems experience more frequent mixing or entrainment during the summer period (Figures 4, S7). Conversely, antecedent spring air temperature was a more important driver of bottom-water temperature, oxygen demand, and DO in deeper lakes, likely as a result of greater thermal inertia and physical distance from surface-water processes (e.g., Gerten & Adrian 2001; Magee & Wu 2017; Figure 4). For surface water temperature and bottom-water DO, we documented lower cross-seasonal ecological memory in high-latitude lakes, potentially as a result of the greater importance of under-ice dynamics in regulating temperature change and oxygen replenishment in these systems (Vincent et al. 2008). Notably, this cross-lake analysis of which factors modulate the strength of cross-seasonal ecological memory would not have been possible without the compilation and analysis of long-term depth profile data from many lakes worldwide. Although we did not have available data to explore winter conditions, summer in-lake conditions may subsequently influence ecological states throughout the autumn and winter, with small (low surface area) lakes exhibiting higher levels of cross-seasonal ecological memory for DO from the summer to the winter under-ice period (Jansen et al. 2025). Our analysis identified a growing thermal difference between summer surface and bottom water temperatures across lakes worldwide (Figure 6), which supports similar results from previous cross-lake analyses (Kraemer et al. 2015; O’Reilly et al. 2015; Pilla et al. 2020; Richardson et al. 2017). This effect appears to be partially mediated by an interaction between seasonal differences in climate warming rates and ecological memory, as the percentage of lakes that show increasing thermal difference between surface and bottom waters was highest in cases when summer air temperature warmed faster than spring air temperature and CSEM for bottom-water temperature was greater than 0 (Figure 6; Figure S17). However, changes in bottom-water temperature are further regulated by changes in transparency, among other factors (e.g., Bartosiewicz et al. 2019; Rose et al. 2016), which is likely responsible for the fact that a majority of lakes showed increasing thermal differences between surface and bottom waters regardless of ecological memory or the relative difference in spring vs. summer warming rates (Figure 6). Over time, slower rates of bottom-water warming compared to surface-water warming may contribute to strengthened thermal stability (Crossman et al. 2016; Kraemer et al. 2015; Richardson et al. 2017; Figure 6), which could potentially increase the strength of the seasonal memory effect across lakes (e.g., Figure 4). Greater rates of surface vs. bottom warming may alter ecological dynamics across the whole lake, with potential consequences including increased cyanobacterial blooms in warm surface waters, decreased vertical entrainment of nutrients from deep waters, and altered habitat availability (Jane et al. 2024; Rose et al. 2016; Woolway et al. 2022; Pearl and Huisman 2008; Soranno et al. 1997). In a small subset of lakes, we observed the opposite pattern—a decreasing thermal difference between surface and bottom waters. These cases were most common in lakes where bottom-water temperature was more correlated with summer vs. spring air temperature. Decreasing thermal difference could be driven by increases in the frequency and intensity of storm events, which can drive thermal mixing of surface and bottom waters (Klug et al. 2012). Here, lakes with decreasing thermal difference tended to be shallow and weakly stratified, and therefore more susceptible to these storm-driven mixing events (Figure S17). Ecological implications of decreased thermal difference may include loss of cool-water refuge in bottom waters, thereby decreasing habitat for thermally-sensitive species (e.g., Kraemer et al. 2021).Variation in spring air temperature is likely associated with simultaneous changes in ice dynamics, mixing period duration, and stratification onset date, and the relative importance of these factors in shaping summer bottom-water dynamics may vary across lakes (e.g., Dugan 2021; Powers et al. 2022; Toffolon et al. 2020). In particular, the relevant window of spring air temperature is likely to shift earlier over time due to increasingly early onset of thermal stratification (Kraemer et al. 2015; Woolway et al. 2021), which could also result in changing mixing regimes (e.g., polymictic to monomictic). However, stratification onset dates will be further constrained by solar radiation, wind, and ice, among other factors (Flaim et al. 2020). Here, limited spring data availability prevented us from empirically examining these mechanisms. More broadly, it is unlikely that long-term spring data exist across many widespread lakes due to the logistical challenges of sampling during intermittently ice-covered periods. In the absence of spring observations, our analysis of daily modeled ISIMIP water temperature provided a complementary approach for verifying the relationship between stratification onset and cross-seasonal ecological memory and provided support for our empirical analysis. As stratification and ice dynamics continue to change across lakes worldwide, mechanistic characterization of the broad-scale patterns identified in this study will be helpful for predicting future changes in lake ecosystem function.The cross-seasonal ecological memory effect identified in our study likely operates in addition to, and in interaction with, numerous other controls over bottom-water temperature and DO dynamics. Notably, our analysis was focused on temperate lakes, and our conclusions may therefore not apply to tropical or boreal systems that exhibit differing seasonal phenology. Wind patterns, which are changing over time across many lakes, also exert a strong influence over stratification, ice, and mixing dynamics (Flaim et al. 2020; Janatian et al. 2020; Stetler et al. 2021; Woolway et al. 2019). However, substantial uncertainty in locally downscaled wind data across our widespread lakes (e.g., Wu et al. 2024) limited our ability to include wind as a driver of cross-seasonal ecological memory. Accounting for these additional impacts and characterizing cross-seasonal ecological memory across a broader latitudinal gradient would aid in predicting the timing and strength of cross-seasonal ecological memory across lakes. Across ecosystems and variables, ecological memory and legacy effects have gained increasing attention in recent decades to describe the influential role that antecedent conditions can play in driving changes in population, community, and ecosystem dynamics (e.g., Dugan 2021; Hanson et al. 2023; Liu et al. 2019; Ogle et al. 2015; Pilla et al. 2023; Van Meter et al. 2018; Wang et al. 2011). Similar to our findings in lake ecosystems, research across terrestrial ecosystems has found that antecedent effects of winter or spring air temperatures have the potential to exert a stronger role over summer terrestrial vegetation than concurrent summer air temperatures (Kreyling et al. 2019; Wang et al. 2011). Freshwater lakes provide an important example of a case where differences in ecological memory may contribute to a spatial mismatch in the ecological effects of climate change, with divergent summer water quality trends between surface and bottom waters. The mechanisms underlying a cross-seasonal ecological memory effect in temperate lakes are well-supported by decades of research (e.g., Dugan 2021; Oleksy & Richardson 2021; Jansen et al. 2025) and by our empirical analysis across many lakes. Ultimately, our results emphasize the importance of accounting for spatial and temporal heterogeneity in ecosystem responses to climate change.Statement of authorship
