Climate change and nutrient enrichment altering sedimentary diatom assemblages since pre-industrial time: evidence from Canada’s most populated ecozone

Lakes worldwide are under threat by a myriad of environmental stressors that have been increasing in number and magnitude. These stressors can be regional such as climate change, or local such as nutrient-rich runoff, invasive species, and road salt contamination, to name but a few. To protect lake ecosystems from further deterioration, we need long-term data to define pre-disturbance baselines and to identify stressors that are causing the greatest ecological changes. Paleolimnology is an effective approach to reconstruct limnological history, providing an important window into past changes. Here, we applied paleolimnological tools to explore the pre-industrial and contemporary diatom assemblage changes of 27 lakes located in the most populated ecozone in Canada, the Mixedwood Plains. We also examined a full sediment core for Lac des Chicots (Southern Québec), aiming to disentangle the impacts of natural versus anthropogenic interactions and to assess their relative effects on the lake’s biotic structure. Our ordination analysis suggests that the Mixedwood Plains lakes have experienced varying lake-specific ecological changes over the past ~150 years, with two major trends across most study lakes: (1) a prevalent increase in planktic species, and (2) a rise in mesotrophic/eutrophic taxa in lakes receiving high human impacts. Our case study of Lac des Chicots identifies ecological impacts from both historical natural events and recent human activities, such as cultural eutrophication and climate warming. Overall, our study demonstrates that lakes in the Mixedwood Plains ecozone have experienced marked ecological changes that are mainly associated with human impacts.


Introduction
Over the past two centuries, many lakes have been affected by the growing impacts of multiple environmental stressors, including acidification, metal contamination, cultural eutrophication, invasive species, and climate change (Dixit et al. 2002;Rühland et al. 2008;Enache et al. 2011;Walsh et al. 2016;Griffiths et al. 2017). To assess the effects of these stressors, long-term limnological data are required to define the range of the natural variability of lakes and to quantify the magnitude of change that has occurred. Since long-term limnological monitoring records are usually rare, alternative methods such as the analyses of natural archives (e.g., lake sediments) are needed to obtain these missing data (Smol 2019).
Paleolimnology is an effective approach to reconstruct past ecological and environmental changes in lake ecosystems (Smol 2008;Gregory-Eaves and Beisner 2011), with a wide spectrum of indicators preserved in lake sediments. Subfossil diatoms are typically well preserved in sediments and are responsive to a number of key limnological gradients (e.g., nutrients, salinity, thermal stability), thereby serving as excellent biological indicators of ecological changes (Smol 2008). Additionally, geochemical analyses can also provide useful information on erosion and metals entering the lake system from atmospheric or local watershed sources (Last and Smol 2001).
In recent years, the concept of landscape-scale paleolimnology has developed along with the increasing recognition of spatial heterogeneity both within and among watersheds (Anderson 2014;Moorhouse et al. 2018). Many elements can affect the variability among lakes, including regional (e.g., regional landscape characteristics) and local factors (e.g., watershed, land use, and geomorphic setting; Soranno et al. 1999). When conducting analyses from a landscape-scale perspective, it is important to consider the similarities and differences between lake regions. Landscape-scale considerations are of particular importance for Canada, as this country is home to the greatest number of lakes in the world (Messager et al. 2016), while also characterized by diverse ecozones that encompass distinct geological, climatic, and vegetational features.
The Mixedwood Plains is the most densely populated ecozone in Canada with an extensive water network. It is bounded by three Great Lakes in southern Ontario and extends along the St. Lawrence River shoreline to Québec City. It has also been the industrial and commercial heartland of Canada since the establishment of European settlements in the seventeenth century, with recent development (especially urbanization) as well as population growth accelerating since the ~1850s. Unfortunately, settlement construction and resource extraction have altered the natural landscapes substantially, leading to heightened environmental degradation. Further, the Mixedwood Plains has been identified as one of the most important and threatened regions for biodiversity conservation in Canada (Kraus and Hebb 2020). Yet, the region has not been systematically investigated regarding how lake ecosystems have changed during this period of accelerated anthropogenic influences. Here, we aim to: (1) compare pre-1850 and recent sediments to broadly assess the ecological changes experienced by lakes in the Mixedwood Plains ecozone, and (2) conduct a case study on Lac des Chicots to better understand the relative impacts of multistressors on lake biotic structure over the past two centuries.

Study area
Our study lakes (n = 27) in the Mixedwood Plains ecozone are distributed across northeastern Ontario and southern Québec (Fig. 1). Lake sampling was conducted as part of the LakePulse program during the summer of 2017, following the protocols described in Huot et al. (2019) and detailed in NSERC Lake Pulse Network (2021). For each lake, detailed physiographic information (geographic location and lake morphology) as well as water chemistry data were collected and are presented in Table 1. Additionally, information on land-use composition for the studied watersheds is provided in Table S1.
Human impact (HI) scores were assigned to each study lake based on weighted differences in the land-use composition of the watershed, ranging from 0 (no disturbance) to 1 (high degree of disturbance; Huot et al. 2019). The lakes were split into low (HI index < 0.15) and high human impact (HI index ≥ 0.15; max score is 1) classes; the threshold in HI scores was calculated on the full Lake Pulse dataset using a univariate regression tree, with the water chemistry data as response variables and HI as the predictor (Griffiths et al. 2022). The lakes were also split into three size classes: small (< 0.5 ha), medium (≥ 0.5 and < 5 ha), and large (≥ 5 ha). Thermal stratification class was assigned based on the stratification status of the lake at the time of sampling. If the lake thermal profile was not available (for 5 of 27 lakes), the stratification status was assigned based upon a univariate conditional inference tree of stratification class as a function of maximum lake depth. This was run on the entire set of LakePulse lakes with thermal profiles. Based on this analysis, lakes without thermal profiles but that were > 5.7 m deep were considered as likely stratified (Griffiths et al. 2021).
Historical (1841 CE-2017 CE) monthly records of air temperature were obtained for each of the 27 study sites from Environment and Climate Change Canada using the R rclimateca package (Dunnington 2018). Only climate stations with at least a complete year of monthly data were selected to create the mean annual temperature estimates. Rather than assigning air temperature and precipitation estimates based on the nearest station (which can be hundreds of km away), estimates from stations within 75 km of the study site were spatially interpolated, forming rings of raster values (estimated temperature) around the input station. This step ensured that the estimated temperatures for any site were reflective of station input data.

Case study: Lac des Chicots
Lac des Chicots was chosen as our case study lake to conduct detailed diatom and geochemical analyses due to regional conservation concerns. The lake has been subjected to a series of environmental issues over the past few decades, including cyanobacterial blooms and a biological invasion of Myriophyllum spicatum ("Municipalité de Sainte-Thècle" n.d.). Additionally, the lake likely experienced natural events that pre-date the European expansion into this region. Several parts of the lake were littered with submerged tree trunks (resulting in the translated name "Stubs Lake"), likely due to a potential subsidence or flooding of the surrounding terrain prior to European settlement (Veillette 1973).
Lac des Chicots (46.8017° N, 72.5175° W), located at 143 m a.s.l. in the municipality of Sainte-Thècle (Québec, Canada), has a surface area of 71.1 ha and a maximum measured depth of 21 m. The lake measures 2 km in length and 0.8 km in width, with an elongated form running along a north east/south west axis. It can be subdivided into three sections, separated by narrows, with several embayments. The lake is supplied by four inlets, including the discharges of Lac Auguste-Leblanc, Lac à la Peinture, and Lac Rose, as well as Vandal Stream, and it has one outlet to the Rivière des Envies.
Lac des Chicots was classified as mesotrophic to eutrophic on the four occasions of summer sampling from 2008 to 2017 (Table S3; Tremblay et al. 2014;Huot et al. 2019;"Gouvernement du Québec" n.d.). It has a shallow Secchi depth (1.5 ± 0.4 m) and high chlorophyll a (chl a) content (14.3 ± 6.6 μg L -1 ) in comparison to other Mixedwood Plains lakes (Table 1), suggesting relatively high primary productivity. The lake was hypoxic or anoxic below 4 m depth in the summer of 2008 (G Cabana pers. commun.). Land-use analysis (Fig. 2) of the watershed of Lac des Chicots using ArcGIS indicates that most (~44%) of its watershed is composed of natural landscape, with ~23% used for agricultural production and ~19% for pasture. Urban landscapes account for ~8% of the watershed area.

History of local settlement in Sainte-Thècle
At the beginning of European settlement and development in the Sainte-Thècle region, Lac des Chicots was of considerable importance for local transportation, fishing, drinking water, and agriculture. In the 1860s, the first permanent settlements were founded around the lake ("Municipalité de Sainte-Thècle" n.d.). In the 1870s, land clearing and logging took place, and public buildings (a parish and a school) were built to serve the growing population. Meanwhile, infrastructure was installed, with roads constructed in the 1870s and railways in the 1880s. By the end of the nineteenth century, more than 1,000 inhabitants lived in this region. Developments and population growth further accelerated in the twentieth century in the town of Sainte-Thècle. The local population reached almost three thousand residents by 1986 (Statistics Canada 1986), which dropped to ~2,500 starting from the twenty-first century (Statistics Canada 2006).

Coring and dating
Sediment cores were extracted from approximately the deepest point of each of the 27 lakes in the summer of 2017. To retrieve cores that reached preindustrial times, we aimed to collect cores that were at least 32 cm in length when possible, a length targeted based on a survey of published sedimentation rates across Canada (Griffiths et al. 2021). To depict the general trend of ecological change that had occurred in the Mixedwood Plains ecozone, we used the "top-bottom" approach (Smol 2008). Specifically, we collected the surface sediment (0-1 cm interval) and a sediment interval from the bottom of the core (X-4 to X-3 cm interval, where X is the total length of the core) from each lake. Indicators preserved in the surface samples represent modern conditions (the "tops"), and those deposited in the bottom samples reveal historical conditions (ideally before the industrialization of North America; the "bottoms"). The bottom 2 cm of each sediment core was discarded due to potential smearing by the core plug. All cores for the top-bottom analyses were extruded in the field using a vertical extruder.
The full sediment core collected at Lac des Chicots was stabilized onsite with Zorbitrol (Tomkins et al. 2008) for further analyses at research facilities in Québec and Ontario. This core was scanned using Siemens SOMATOM Definition AS + 128 computedtomography (CT) core scanner at the Institut National de la Recherche Scientifique (INRS) in Québec City. The full sediment core was then sent to the University of Laval and split longitudinally in two. The working half was scanned on the non-destructive Cox Analytics micro-X-ray Fluorescence (µ-XRF) ITRAX core scanner located at INRS and subsampled at 1-cm intervals. Subsamples were subsequently freeze-dried for dating and diatom analyses. 210 Pb analyses using gamma spectrometry was completed at the Paleoecological Environmental Assessment and Research Laboratory (PEARL) at Queen's University. 210 Pb activities were measured for both the top (0-1 cm) and bottom (X-3 to X-2 cm interval, a slightly lower interval than used for diatoms due to the sharing of sample across multiple researchers and projects) samples for the 27 study lakes, while the Lac des Chicots full core was dated using a Constant Rate of Supply Model (Appleby 2001). One of the regional samples (lake 08-163) was excluded from the top-bottom analyses due to high unsupported 210 Pb levels in the bottom sample (greater 210 Pb activities than 214 Bi), suggesting that the bottom sample may not have reached the pre-industrial age (all the other study lakes met this requirement; Griffiths et al. 2022). The surface sample of lake 08-163 was used in ordination and other analyses requiring just the top samples. The dating model of our case study lake is provided as Figure S1, and its decay curve is within our expectations (i.e., unsupported 210 Pb generally decreased exponentially with depth).

Diatom processing
The top and bottom sediment intervals of all study lakes were processed for diatom analyses. For our case study lake, Lac des Chicots, 23 intervals spanning the full length of the core were examined. Diatoms were prepared for microscopic analysis following standard techniques described in Battarbee et al. (2002b). In brief, dried sediment subsamples of 0.01-0.02 g were taken from each interval. A 10% hydrochloric acid solution was used to dissolve carbonates, and the process of dilution, settling, and aspiration were repeated at least five times until the samples reached a neutral pH. Then, a 30% hydrogen peroxide solution was added to digest organic matter. Samples were put into a water bath and heated at around 80 °C for 8 h for three consecutive days, and the solution was diluted again until they reached a neutral pH. The cleaned diatom suspensions were pipetted onto coverslips and were mounted in Zrax onto glass microscope slides. For each sample, a minimum of 500 diatom valves were enumerated with a Leica DM 2300 microscope (1000 × magnification). Diatoms were counted to species, and variety level when possible, following Krammer and Lange-Bertalot (1986, 1988, 1991a, 1991b and Lavoie et al. (2008).

Numerical analysis
All numerical analyses were conducted in the R platform (R Core Team 2021). First, to visualize the relationships between environmental factors and diatom assemblages, we conducted a Principal Component Analysis (PCA) on the modern diatom species data using the vegan package (Oksanen et al. 2020), and passively plotted the environmental factors on the ordination plot. We divided the environmental factors into two groups: (1) physiographic and water quality variables, and (2) human-impact factors. Physiographic variables include geographic location and morphological information, while water quality variables describe the physical (e.g., Secchi) and chemical variables (e.g., nutrients) that may affect lake conditions for diatom communities. Human-impact factors include human impact (HI) index, population in the watershed, and land-use variables (e.g., percent agricultural land use in the catchment). To meet an assumption of the PCA (normal distributions of environmental variables), we tested the normality of the distribution for all environmental variables and transformed those that were not normally distributed, using Box-Cox, logarithmic, or square-root transformations (the type of transformation and transformation coefficient for each parameter are provided in Table S2). When necessary, variables with zero values were offset to positive values by adding 0.00001.
We investigated the positioning of the major diatom functional groups along the first two PCA axes by plotting them passively on the ordination. The major functional groups were divided based on their TP optima into oligotrophic (TP optima < 12 μg L −1 ; Tremblay et al., 2014) or mesotrophic/eutrophic taxa (TP optima ≥ 12 μg L −1 ; Tremblay et al., 2014), as well as their habitats into benthic, benthic-planktic (primarily benthic species that sometimes can be entrained in the plankton such as benthic fragilarioids), planktic, or tychoplanktic taxa (with the habitat assignations based mainly on descriptions in Round et al. 1990). Additionally, the dominant species, defined as those with a strong influence on the ordination (having a loading ≥ 0.45 on either PC axis 1 or axis 2), were also passively plotted on the ordination to examine the major drivers of the principal components of variation. We applied Hellinger transformation to all diatom data using the vegan package (Oksanen et al. 2020).
Then, to investigate how the diatom assemblages of Mixedwood Plains lakes have changed over time, we plotted the site scores of the bottom samples (preindustrial conditions) passively on the ordination using a weighted-averaging model. For each top-bottom pair of diatom assemblages we calculated the Bray-Curtis (BC) dissimilarity index to quantify the magnitude of temporal beta-diversity. To assess how these sites have shifted between their pre-industrial and modern conditions with respect to the two major axes of variation in the diatom assemblages across the landscape (PC axis 1 and 2), we subtracted the site score of each bottom sample on the PCA ordination from its respective top and plotted their differences onto a biplot of the changes across PC axis 1 and PC axis 2.
Lastly, we explored the distributions of a series of biotic and abiotic variables, including changes in diatom functional groups, Bray-Curtis dissimilarity index values, and water chemistry parameters (e.g., nutrients, ions), across three categorical variables: human impact class (high vs. low), stratification class (mixed vs. stratified), and lake surface area size class (large, medium, or small). Differences of these biotic and abiotic variables between categorical groups were tested for statistical significance (p-values) using Wilcoxon signed-rank test or Kruskal-Wallis one-way analysis of variance.

Down-core analyses
To track the limnological history of Lac des Chicots and to interpret its changes in the context of the Mixedwood Plains ecozone, we passively plotted its trajectory on the PCA ordination. Then, to draw a more complete inference of its past conditions, we developed a diatom stratigraphic diagram, calculated the planktic to benthic diatom ratio and Hill's N2 diversity value, as well as analyzed other biological indicators (chrysophyte scales and cysts, sedimentary chlorophyll a). We also investigated several geochemical indicators, including sediment density, Calcium (Ca), Titanium (Ti), Total Organic Carbon (TOC), and Total Nitrogen (TN).
The diatom stratigraphic diagram was produced using R package analogue (Simpson and Oksanen 2020), and only species that occurred at relative abundances of > 5% in at least one sample were plotted. Stratigraphic zones were identified by applying a stratigraphically constrained incremental sum-ofsquares (CONISS) analysis (Grimm 1987), and the significance of biostratigraphic zones was determined by a broken stick model (Bennett 1996). When calculating the planktic to benthic diatom ratio, we divided the planktic diatom counts by the combined counts of benthic-planktic and benthic taxa (together as benthic group), while excluding those tychoplanktic taxa (i.e., Aulacoseira spp.). Hill's N2 diversity was also plotted for the examined intervals in our stratigraphic diagram (Hill 1973).
Chrysophyte cysts and scales were counted whenever they were encountered in the diatom samples. The ratio of cysts and scales to diatom valves can be useful as a measure of lake production, as chrysophytes are typically found in higher relative abundances in oligotrophic lakes (Smol 1985). The relative abundances of scaled chrysophytes can also be used to indicate the water mixing regime since most of them are planktic species. Sedimentary chl a data was inferred using visible reflectance spectroscopy (VRS), and it tracks trends in whole-lake primary production in lakes (Michelutti et al. 2010). We used a log-transformed regression equation of the data in Michelutti et al. (2010) to calculate the sedimentary chl a from the spectral data due to improved model performance.

Mixedwood plains analysis
Among the Mixedwood Plains contemporary and pre-industrial diatom assemblages, we recorded 93 common taxa (present in greater than 2% relative abundance in two samples). Based on our PCA of the contemporary diatom assemblages, PC axis 1 (PC1) explains 18.9% of the diatom species variance and is positively correlated with dissolved oxygen concentrations in the bottom 1 m of the lake, chloride concentrations, water-column chl a content, and portion of pasture in the watershed, while negatively correlated with the maximum lake depth and Secchi depth (Fig. 3a, b). PC axis 2 (PC2) explains 13.5% of the variance and is negatively correlated with ion concentrations (Na + , K + ), human population in the watershed, proportion of urban land use, and the human impact (HI) index.
The PCA ordination indicates a clear differentiation of major diatom functional groups (Fig. 3c) and several dominant species along the key environmental gradients (Fig. 3d). In general, benthic and benthic-planktic taxa point towards the direction that is associated with higher bottom-water dissolved oxygen (DO) content and greater circularity of lake shape (more rounded), while in the opposite direction planktic species were associated with deeper sites. Mesotrophic/eutrophic taxa were associated with environmental parameters indicative of human impacts (e.g., pasture and urban land uses) as well as lakewater nutrient variables (TN and TP). Down to the species-level, many small cyclotelloid taxa such as Pantocsekiella comensis, Cyclotella cretica var. cyclopuncta, and Lindavia michiganiana (not shown on Fig. 3d) are associated with clear waters and low nutrients, whereas planktic, eutrophic diatoms (e.g., Stephanodiscus spp. and Fragilaria crotonensis) plotted negatively along PC2.
To examine how the diatom assemblages of the 26 lakes (excluding 08-163) have changed over the past ~150 years, we plotted their bottom samples passively on the ordination and compared these values to those obtained for their top samples (Fig. 4a). Although no coherent pattern in the diatom assemblage changes was identified across all the 26 study lakes along the major PCA axes, some trends are still apparent by examining the relative changes in PCA site scores (Fig. 4b). We found that the majority of lakes have shifted towards Quadrant III (n = 11) and Quadrant II (n = 7), with only a few moving towards Quadrant I (n = 4) or Quadrant IV (n = 4). Additionally, up to 81% (21 out of 26) of our study lakes increased in relative abundances of planktic taxa, including all lakes plotting in Quadrant III (Fig. 4c).
Regarding changes in mesotrophic/eutrophic taxa (Fig. 4d), all lakes plotting in Quadrant IV increased in mesotrophic/eutrophic taxa (average + 15.7%), while all lakes plotting in Quadrant II decreased (average − 16.3%). Most lakes plotting in the other two quadrants experienced an increase in mesotrophic/eutrophic taxa since the pre-industrial: three out of four in Quadrant I (average + 9.3%), and 7 out of 11 in Quadrant III (average + 11.0%).
Several biotic and abiotic variables were significantly associated with the categorical variables, including human impact class, stratification state, and size class (Fig. 5). The percent planktic taxa increased on average in every categorical class (Fig. 5a), with stratified lakes experiencing a larger increase in the percent planktic taxa than mixed lakes (p = 0.06), and medium-sized lakes having the greatest increases in planktic taxa among all lake sizes (p = 0.03). Lakes in the high human impact class were found to have a significantly greater average increase in mesotrophic/eutrophic taxa than lakes in the low human impact class (Fig. 5b). Additionally, large lakes experienced a lower degree of temporal species turnover (lower BC dissimilarity) between their pre-industrial and modern diatom assemblages in comparison to small diatom functional groups (grouped by nutrient status and by habitat preference) are passively plotted: oligotrophic vs. mesotrophic/eutrophic (in yellow); benthic vs. benthic-planktic vs. planktic vs. tychoplanktic (in green). Panel d) dominant diatom taxa (defined as taxa with a greater than 0.45 loading on PCA axis 1 or 2) and medium-sized lakes, while no significant differences were identified across the various human impact and stratification classes (Fig. 5c). Lastly, TN, Na + , and Cl − concentrations varied across both human impact class and stratification status, whereby lakes with high human impacts as well as mixed lakes had higher nutrient and ion concentrations ( Fig. 5d-f).

Ecological changes in Lac des Chicots
We assessed the changes in diatom assemblage composition of our case study lake by passively plotting its downcore samples on the PCA ordination (Fig. 6). The oldest section of the sediment record was located near the center of the ordination and shifted towards more negative PC axis 1 and 2 scores in modern  sediments. Based on the associations between diatom assemblages and major functional groups as well as environmental parameters, this trajectory shift may suggest an increase in planktic taxa and an overall increase in ion concentrations over time.
A CONISS analysis of the stratigraphy of Lac des Chicots identified three main stages spanning the past ~200-year history of this lake (Fig. 7): (1) relatively stable pre-industrial conditions before the nineteenth century (34-21 cm), (2) a phase of marked ecological change between the ~1800s and the ~1850s (20-15 cm), and (3) a period of increasing human impacts from the ~1870s to recent years (14-0 cm).
Prior to the nineteenth century, Lac des Chicots was dominated by planktic and tychoplanktic species (Discostella stelligera and Aulacoseira subarctica) as well as several small benthic species (Achnanthidium minutissimum, Staurosirella pinnata, and Pseudostaurosira brevistriata). At the beginning of the nineteenth century (20-19 cm), there was a major shift in diatom assemblage composition, with A. minutissimum increasing markedly in relative abundance, becoming the dominant species in the assemblage (reaching a peak of ~72%). A. minutissimum rapidly decreased in its relative abundance to around 20% in the subsequent 18-17 cm interval (the ~1820s) and continued declining until reaching a stable abundance of less than 10% after the ~1850s.
Starting from the ~1870s (14-13 cm), several planktic taxa increased in relative abundance, including Fragilaria crotonensis, Asterionella formosa, Ulnaria ulna, and Fragilaria nanana, driving the rise in the planktic to benthic taxa ratio. Starting from the ~1920s (10-9 cm), Fragilaria mesolepta appeared in the lake, reaching a relative abundance of ~10% of the diatom assemblage. Between the ~1930s and the ~1950s (9-7 cm), two abrupt and temporary spikes in the tychoplanktic Aulacoseira granulata and A. granulata var. angustissima were recorded, coincident with a slump of the planktic to benthic taxa ratio. Lastly, Lindavia michiganiana, a taxon that used to be found in the record historically but disappeared after the ~1800s, increased in the recent sediments, comprising ~10% of the diatom assemblage in the top intervals (2-0 cm). The planktic to benthic diatom ratio has remained relatively stable over the past few decades, approximately two times higher than the pre-disturbance condition.
When assessing the Hill's N2 diversity index, Lac des Chicots demonstrated several marked changes over the course of the sediment record (Fig. 7). The diversity index attained its minimum at the interval of 20-19 cm (the ~1800s), coincident with an overwhelming dominance of A. minutissimum. Later, it experienced two additional declines between 16-9 cm (ca. 1850s-1920s) and 8-4 cm (ca. 1940s-1970s), corresponding to the dominating periods of F. crotonensis. The highest Hill's N2 value was recorded at the 16-15 cm interval (the ~1850s), coincident with the emergence of several mesotrophic to eutrophic planktic taxa.

Other indicators
The chrysophyte cysts to diatoms (C:D) ratio (Smol 1985) remained relatively stable over the record, except for a spike at the interval of 9-8 cm (the ~1930s) as well as a recent increase in the top intervals (Fig. 7). The chrysophyte scales to diatoms (S:D) ratio remained low but showed a slight increase in the recent sediments (Fig. 7). Sedimentary chl a (log-transformed) displayed an increasing trend over time, with the concentrations in the most recent sediments having increased substantially as compared to the pre-industrial period (Fig. 8).
The TOC:TN ratio decreased until the mid-nineteenth century, remaining relatively low for a century before returning to its historical level after the 1980s (Fig. 8). The reported ITRAX proxies, Ti and Ca, both decreased from the bottom of the core to a depth of ~24 cm, suggesting a change in the sedimentary material entering Lac des Chicots during this period. These proxies remained relatively stable for the rest of the core, echoing the laminated pattern observable from the CT scan of the sediment core, which also suggests a decline in density from ~30 to ~24 cm followed by a period of stability.

Discussion
Overall, we observed two major trends regarding ecological changes of Mixedwood Plains lakes over the past ~150 years: (1) a prevalent increase in planktic Fig. 7 Stratigraphic diagram for Lac des Chicots' diatom and biotic changes. Relative abundances of common (> 5% in at least one interval) diatom taxa, planktic to benthic diatom ratio (P:B), Hill's N2 diversity index (N2), and chrysophyte cysts and scales to diatom frustules ratios (C:D and S:D respec-tively). Red horizontal lines represent divisions of stratigraphic zones, identified by applying the constrained incremental sumof-squares analysis (CONISS). Start dates for the relevant decades as calculated using a constant-rate-of-supply (CRS) model of the 210 Pb activities are provided along the depth axis taxa, potentially as an indirect reflection of climate warming, and (2) an overall increase in mesotrophic/ eutrophic taxa, associated with human impacts.

Ecological changes of Mixedwood Plains lakes
Based on our ordination analysis of the changes in the ordination space between the top-bottom sediment pairs (Fig. 4b), we found that most lakes have moved towards Quadrant III, associated with deeper water levels (Fig. 3a) and/or an increase in planktic taxa (Fig. 3c). This finding is supported by the fact that most of our study lakes (81%) now have more planktic taxa than in their pre-industrial periods. The increase in planktic taxa might be related to several factors such as climate warming and/or water depth changes due to lake level modification (e.g., installation of a dam). Given the regional ubiquity of this trend, we propose that a regional change in climatic condition is likely the principal stressor causing such a shift in diatom assemblage composition. By analyzing the climate data, we found a general trend of increasing air temperatures in the Mixedwood Plains ecozone, especially over the past four decades (Figure S2). A warming climate can extend the ice-free season and enhance thermal stratification in summers, thereby creating more habitats for planktic diatoms to proliferate and increase their relative abundances in the community (Rühland et al. 2008(Rühland et al. , 2015. This is further supported by the fact that deep stratified lakes have greater increases in planktic taxa than shallow lakes that do not stratify ( Fig. 5a; Figure S3). Additionally, medium-sized lakes exhibited a more pronounced increase in planktic taxa in comparison with small lakes (usually shallower and more likely to be mixed) and large lakes, which tend to have greater thermal inertia and thus are less susceptible to changes in ice-free season and thermal regime. Overall, climate warming is likely the main contributor to the prevalent increase in planktic taxa across Mixedwood Plains lakes (Fig. 3c). Our finding is in accordance with those of many other paleolimnological studies across the world, which have recorded a widespread diatom community shift from tychoplanktic and benthic taxa to planktic species, likely due to longer ice-free periods and enhanced thermal stratification related to climate warming (Battarbee et al. 2002a;Rühland et al. 2008Rühland et al. , 2015Enache et al. 2011;Griffiths et al. 2022).
More than half of our lakes have higher proportions of mesotrophic/eutrophic taxa in their modern diatom assemblages compared to their pre-industrial conditions, suggesting an overall increase in lakewater nutrient concentrations across the Mixedwood Plains ecozone. We found that all lakes that shifted towards Quadrant IV had rising percentages of mesotrophic/eutrophic taxa, whereas all lakes that shifted towards Quadrant II had declining relative abundances of high nutrient indicators (Fig. 4d). Considering that Quadrant IV is associated with lakes with presently high nutrient concentrations (Fig. 3a) as well as a suite of water quality and land-use factors indicative of human impacts (Fig. 3b), while Quadrant II is related to clear waters and more naturalized catchments, we suggest that human activities are likely the main drivers causing rises and declines in mesotrophic/eutrophic taxa. Additionally, lakes in the high human impact class were found to have significantly higher TN values (Fig. 5d) and greater increases in nutrient-rich taxa (Fig. 5b) than lakes in the low human impact class. A number of human activities can contribute to elevated nutrient concentrations of surface waters, including urbanization, agricultural intensification, and pasture expansion. For instance, sewage deposition from human settlements as well as runoff over agricultural and pasture lands can elevate the nutrient content of lake waters, creating more favorable conditions for species with high nutrient optima. Pienitz et al. (2006) conducted a downcore analysis on Lac Saint-Augustin, one of our study lakes, and observed dramatic increases in mesotrophic/eutrophic species over the past century, including Stephanodiscus hantzschii, A. formosa, F. crotonensis, and F. mesolepta. They attributed these increases to anthropogenic impacts, such as expansion of agriculture within the watershed of Lac Saint-Augustin, industrialization of the nearby Québec City and its surroundings, as well as the rapid population growth in the local municipality.
Additionally, lakes in the high human impact class also have significantly higher Na + and Cl − concentrations ( Fig. 5e and Fig. 5f), which on the ordination were associated with urban developments in the watersheds (Fig. 3b). Road salt application in urban areas can be a major stressor causing elevated Na + and Cl − concentrations in freshwater ecosystems, potentially inducing negative influences on local biotic structures. For example, Hintz et al. (2022) found that salt pollution has triggered a massive loss of important zooplankton taxa, leading to increased phytoplankton biomass in many lakes across North America and Europe. Further, since most freshwater organisms have no recent evolutionary history with high salinity, many salt-tolerant invasive species could outcompete those more sensitive native taxa and change the local biotic composition dramatically (Dugan et al. 2020).
Surprisingly, lakes under varying human impact levels do not differ significantly in terms of the BC dissimilarity index (Fig. 5c). It is possible that some lakes in the Mixedwood Plains were affected by human activities (e.g., settlement establishment and logging) that pre-date the ~1850s, our target date of the bottom samples (Griffiths et al. 2021). Additionally, some lakes may also have been subject to recent remediation efforts such that, based on the current land-use compositions of their watersheds, these lakes are classified into the low human impact class, but their diatom assemblages have not yet returned to their pre-disturbance conditions and therefore have a high BC dissimilarity index.

Lac des Chicots
The sediment record for Lac des Chicots depicts a dynamic lake history, covering more than two hundred years, reflecting changes induced by both natural and anthropogenic sources (Fig. 7). The lower part of the sediment core (below the 19-20 cm interval; i.e., before the nineteenth century) reveals the presettlement conditions of Lac des Chicots. At that time, the lake was dominated by not only benthic/ benthic-planktic (S. pinnata, P. brevistriata, and A. minutissimum), but also tychoplanktic (A. subarctica) and planktic (D. stelligera) species, suggesting that Lac des Chicots had a substantial littoral zone that supported the benthic community as well as a deep region that allowed the flourishing of tychoplanktic and planktic taxa. These historically dominant taxa are indicators of oligotrophic and mesotrophic waters (Cumming et al. 1995;Gibson et al. 2003;Tremblay et al. 2014), suggesting that the lake was relatively low in nutrients in this pre-settlement period.
At the beginning of the nineteenth century, a spike of A. minutissimum marks a dramatic limnological change. A. minutissimum is a benthic cosmopolitan species (tolerant to various types of stressors, such as hydrologic disturbances), and is also an early colonizer (Ponader and Potapova 2007). Its dominance indicates that there might have been a major shift in the limnological condition of Lac des Chicots, with A. minutissimum being able to recover more quickly than the other taxa and thus prospering in the lake for a while after the disturbance. Interestingly, there were no coincident changes in the sediment density or geochemical markers (Fig. 8) that can illuminate the source of this potential disturbance. Nonetheless, this event pre-dates European settlement in the region based on our dating model.
With the decline of A. minutissimum (18-17 cm interval; the ~1820s), a new group of planktic taxa that are characteristic of mesotrophic waters increased, including A. formosa, U. ulna, and F. nanana (Reavie et al. 1995;Ramstack et al. 2003), potentially suggesting rises in both water level and nutrient content. Starting from the ~1870s, a further increase in the relative abundance of A. formosa and the emergence of F. crotonensis (a species with even higher nutrient optimum than A. formosa; Saros et al. 2005) likely indicated increasingly elevated nutrient concentrations in the lake. These increases correspond with the growing local population and accelerated developments in the adjacent town of Sainte-Thècle, such as the constructions of roads and buildings ("Municipalité de Sainte-Thècle" n.d.).
According to a member from the MRC Mékinac organization (É Piché pers. commun.), a dam was constructed at the outlet of the lake around eight decades ago (around the 1930s-1940s), which might have increased the water depth and caused disturbances in the water column. The sediment deposited during this period (from 9 to 7 cm; the 1930s and 1940s) includes modest spikes in the relative abundance of A. granulata and A. granulata var. angustissima, with corresponding declines in F. crotonensis and A. formosa. Considering that the two Aulacoseira species are tychoplanktic (found in mixing waters) and are also indicative of nutrient-rich environments (Laperrière et al. 2007;Cumming et al. 2015), the lake likely experienced greater water mixing and turbulence around the 1930s-40s, while nutrient concentrations remained high.
Since the ~1940s, the dominances of mesotrophic/ eutrophic taxa (i.e., F. crotonensis, F. mesolepta, A. ambigua, and A. formosa) suggest a further increase in the nutrient concentrations of Lac des Chicots (Saros et al. 2005;Cumming et al. 2015). The increased nutrient loads likely occurred in response to local development and human activities. Our sedimentary chl a profile (Fig. 8) also tracks increasing primary production starting from the twentieth century, further supporting this inference.
Since the ~1980s (4-0 cm interval), a slight increase in small planktic cyclotelloid species (e.g., L. michiganiana and D. stelligera) occurred, which might be in response to climate warming. Changes in climate condition can alter diatom dynamics through a chain of linked processes (Winder et al. 2009). Climate warming, specifically, can reduce the mixing intensity in water column and enhance thermal stratification, favoring species that are able to maintain their vertical position in the euphotic zone, such as small-celled planktic species (Winder et al. 2009). The historical climate data near Lac des Chicots captured an accelerated rising trend in annual air temperature since ~1975 ( Figure S4), along with the recent increase in cyclotelloid taxa. Thus, we suspect that climate warming is potentially the main driver causing changes in the local diatom assemblage composition over the past few decades. Overall, our case study lake, Lac des Chicots, experienced a marked increase in mesotrophic and/or eutrophic taxa over the past century, as well as a general increase in planktic taxa, reflecting the dominant trends in the Mixedwood Plains ecozone.

Implications for conservation policy in Sainte-Thècle
In 2009, the municipality of Sainte-Thècle inaugurated its water conservation committee, with a mandate towards monitoring and reducing the occurrence of recently identified cyanobacterial blooms ("Municipalité de Sainte-Thècle" n.d.). Based on the information provided by long-time residents in the municipality, however, cyanobacterial blooms have been occurring in the lake for more than 30 years (SAMBBA pers. commun.). Our diatom analysis suggests that the lake-water nutrients have been steadily increasing since the beginning of the twentieth century. The apparent increase in cyanobacterial blooms over the past few decades, along with increasing primary production (as shown by our sedimentary chl a records), may be reflecting the legacy of a sustained period of cultural eutrophication. Meanwhile, climate warming may be further influencing the algal assemblage and the overall primary production synergistically, as enhanced thermal stratification can often trigger cyanobacterial blooms, even when the nutrient concentrations remain steady or are decreasing (Paerl and Paul 2012;Paterson et al. 2017;Smol 2019;Simmatis et al. 2020).

Conclusions
Our regional analysis suggests that lakes in the Mixedwood Plains have experienced various ecological changes over the past ~150 years, with two major trends: (1) a prevalent increase in planktic taxa, potentially driven by climate warming (decreased ice cover, increased thermal stratification), as well as (2) an overall increase in mesotrophic/eutrophic taxa, associated with human impacts. Additionally, our reconstruction of the ecological history of Lac des Chicots identifies the influences from both historical events and recent human activities on the diatom assemblages and the inferred lake water quality. Based on the observed shifts in diatom assemblage compositions, cultural eutrophication at Lac des Chicots commenced in the mid-nineteenth century, becoming more pronounced since the ~1940s. The limnological effects of recent climate warming were also reflected in the diatom assemblage changes of Lac des Chicots.