Uncovering state-dependent relationships in shallow lakes using Bayesian latent variable regression.

Ecosystems sometimes undergo dramatic shifts between contrasting regimes. Shallow lakes, for instance, can transition between two alternative stable states: a clear state dominated by submerged aquatic vegetation and a turbid state dominated by phytoplankton. Theoretical models suggest that critical nutrient thresholds differentiate three lake types: highly resilient clear lakes, lakes that may switch between clear and turbid states following perturbations, and highly resilient turbid lakes. For effective and efficient management of shallow lakes and other systems, managers need tools to identify critical thresholds and state-dependent relationships between driving variables and key system features. Using shallow lakes as a model system for which alternative stable states have been demonstrated, we developed an integrated framework using Bayesian latent variable regression (BLR) to classify lake states, identify critical total phosphorus (TP) thresholds, and estimate steady state relationships between TP and chlorophyll a (chl a) using cross-sectional data. We evaluated the method using data simulated from a stochastic differential equation model and compared its performance to k-means clustering with regression (KMR). We also applied the framework to data comprising 130 shallow lakes. For simulated data sets, BLR had high state classification rates (median/mean accuracy >97%) and accurately estimated TP thresholds and state-dependent TP-chl a relationships. Classification and estimation improved with increasing sample size and decreasing noise levels. Compared to KMR, BLR had higher classification rates and better approximated the TP-chl a steady state relationships and TP thresholds. We fit the BLR model to three different years of empirical shallow lake data, and managers can use the estimated bifurcation diagrams to prioritize lakes for management according to their proximity to thresholds and chance of successful rehabilitation. Our model improves upon previous methods for shallow lakes because it allows classification and regression to occur simultaneously and inform one another, directly estimates TP thresholds and the uncertainty associated with thresholds and state classifications, and enables meaningful constraints to be built into models. The BLR framework is broadly applicable to other ecosystems known to exhibit alternative stable states in which regression can be used to establish relationships between driving variables and state variables.

Watershed vs. within-lake drivers of nitrogen: phosphorus dynamics in shallow lakes.

Research on lake eutrophication often identifies variables affecting amounts of phosphorus (P) and nitrogen (N) in lakes, but understanding factors influencing N:P ratios is important given its influence on species composition and toxin production by cyanobacteria. We sampled 80 shallow lakes in Minnesota (USA) for three years to assess effects of watershed size, proportion of watershed as both row crop and natural area, fish biomass, and lake alternative state (turbid vs. clear) on total N : total P (TN : TP), ammonium, total dissolved phosphorus (TDP), and seston stoichiometry. We also examined N:P stoichiometry in 20 additional lakes that shifted states during the study. Last, we assessed the importance of denitrification by measuring denitrification rates in sediment cores from a subset of 34 lakes, and by measuring seston δ15 N in four additional experimental lakes before and after they were experimentally manipulated from turbid to clear states. Results showed alternative state had the largest influence on overall N:P stoichiometry in these systems, as it had the strongest relationship with TN : TP, seston C:N:P, ammonium, and TDP. Turbid lakes had higher N at given levels of P than clear lakes, with TN and ammonium 2-fold and 1.4-fold higher in turbid lakes, respectively. In lakes that shifted states, TN was 3-fold higher in turbid lakes, while TP was only 2-fold higher, supporting the notion N is more responsive to state shifts than is P. Seston δ15 N increased after lakes shifted to clear states, suggesting higher denitrification rates may be important for reducing N levels in clear states, and potential denitrification rates in sediment cores were among the highest recorded in the literature. Overall, our results indicate lake state was a primary driver of N:P dynamics in shallow lakes, and lakes in clear states had much lower N at a given level of P relative to turbid lakes, likely due to higher denitrification rates. Shallow lakes are often managed for the clear-water state due to increased value as wildlife habitat. However, our results indicate lake state also influences N biogeochemistry, such that managing shallow lakes for the clear-water state may also mitigate excess N levels at a landscape scale.

The legacy of large regime shifts in shallow lakes.

Ecological shifts in shallow lakes from clear-water macrophyte-dominated to turbid-water phytoplankton-dominated are generally thought of as rapid short-term transitions. Diatom remains in sediment records from shallow lakes in the Prairie Pothole Region of North America provide new evidence that the long-term ecological stability of these lakes is defined by the legacy of large regime shifts. We examine the modern and historical stability of 11 shallow lakes. Currently, four of the lakes are in a clear-water state, three are consistently turbid-water, and four have been observed to change state from year to year (transitional). Lake sediment records spanning the past 150-200 yr suggest that (1) the diatom assemblage is characteristic of either clear or turbid lakes, (2) prior to significant landscape alteration, all of the lakes existed in a regime of a stable clear-water state, (3) lakes that are currently classified as turbid or transitional have experienced one strong regime shift over the past 150-200 yr and have since remained in a regime where turbid-water predominates, and (4) top-down impacts to the lake food-web from fish introductions appear to be the dominant driver of strong regime shifts and not increased nutrient availability. Based on our findings we demonstrate a method that could be used by lake managers to identify lakes that have an ecological history close to the clear-turbid regime threshold; such lakes might more easily be returned to a clear-water state through biomanipulation. The unfortunate reality is that many of these lakes are now part of a managed landscape and will likely require continued intervention.

A 200-year perspective on alternative stable state theory and lake management from a biomanipulated shallow lake.

Multiple stressors to a shallow lake ecosystem have the ability to control the relative stability of alternative states (clear, macrophyte-dominated or turbid, algal-dominated). As a consequence, the use of remedial biomanipulations to induce trophic cascades and shift a turbid lake to a clear state is often only a temporary solution. Here we show the instability of short-term manipulations in the shallow Lake Christina (Minnesota, USA) is governed by the long-term state following a regime shift in the lake. During the modern, managed period of the lake, three top-down manipulations (fish kills) were undertaken inducing temporary (5-10 years) unstable clear-water states. Paleoecological remains of diatoms, along with proxies of primary production (total chlorophyll a and total organic carbon accumulation rate) and trophic state (total P) from sediment records clearly show a single regime shift in the lake during the early 1950s; following this shift, the functioning of the lake ecosystem is dominated by a persistent turbid state. We find that multiple stressors contributed to the regime shift. First, the lake began to eutrophy (from agricultural land use and/or increased waterfowl populations), leading to a dramatic increase in primary production. Soon after, the construction of a dam in 1936 effectively doubled the depth of the lake, compounded by increases in regional humidity; this resulted in an increase in planktivorous and benthivorous fish reducing phytoplankton grazers. These factors further conspired to increase the stability of a turbid regime during the modern managed period, such that switches to a clear-water state were inherently unstable and the lake consistently returned to a turbid state. We conclude that while top-down manipulations have had measurable impacts on the lake state, they have not been effective in providing a return to an ecosystem similar to the stable historical period. Our work offers an example of a well-studied ecosystem forced by multiple stressors into a new long-term managed period, where manipulated clear-water states are temporary, managed features.

Exploring watershed effects on nutrient concentrations in shallow lakes through stable isotope analysis.

Biogeochemistry patterns in shallow lakes are influenced by both in-lake factors such as ecosystem state as well as watershed-level factors such as land use, but the relative importance of in-lake versus watershed factors is poorly known. This knowledge gap makes it difficult for lake mangers to prioritize efforts on watershed versus in-lake strategies for stabilizing the clear-water state. We studied 48 shallow lakes in Minnesota, USA to assess the relative influence of lake size, land use in watersheds, and ecosystem state (turbid versus clear) on water column total nitrogen (TN) and total phosphorus (TP), as well as δ15N and δ13C in three species of fish. Our land use categories included natural areas, row crop agriculture, and all agriculture (row crops plus alfalfa). A model selection approach revealed different control mechanisms on the behavior of stable isotopes and nutrients. δ13C ratios in fish were most strongly influenced by lake size, while δ15N ratios were influenced by all agriculture in watersheds. In contrast, water column TN and TP concentrations were influenced by the in-lake factor of ecosystem state, with both nutrients lower in the clear state. We detected no effects of land use on TN or TP concentrations, likely due to strong effects of ecosystem state masking watershed effects. However, the strong relationship between agriculture and δ15N in fish indicated that watersheds did influence nutrient processing in shallow lakes, and that effects are not a legacy from past watershed events. Collectively, these observations indicate that lake managers should minimize agricultural intensity in shallow lake watersheds to facilitate the clear-water state, which will, in turn reduce water-column TN and TP relative to the turbid state.

Relative importance of phosphorus, fish biomass, and watershed land use as drivers of phytoplankton abundance in shallow lakes.

Phytoplankton abundance in shallow lakes is potentially influenced by ambient phosphorus concentrations, nutrient loading accentuated by human activities in lake watersheds, and abundance of planktivorous and benthivorous fish. However, few studies have simultaneously assessed the relative importance of these factors influencing phytoplankton abundance over large spatial scales. We assessed relative influences of watershed characteristics, total phosphorus concentrations, and fish biomass on phytoplankton abundance in 70 shallow lakes in western Minnesota (USA) during summer 2005 and 2006. Our independent variables included total phosphorus (TP), benthivore biomass, planktivore biomass, summed planktivore and benthivore biomass (summed fish), areal extent of agriculture in the watershed, region (prairie versus parkland lakes), and year. Predictive models containing from one to three independent variables were compared using an information theoretic approach. The most parsimonious model consisted of TP and summed fish, and had over 10,000-fold greater support compared to models using just TP or summed fish, or models comprised of other variables. We also found no evidence that relative importance of predictor variables differed between regions or years, and parameter estimates of TP and summed fish were temporally and spatially consistent. TP and summed fish were only weakly correlated, and the model using both variables was a large improvement over using either variable alone. This indicates these two variables can independently increase phytoplankton abundance, which emphasizes the importance of managing both nutrients and fish when trying to control phytoplankton abundance in shallow lakes.

The altered ecology of Lake Christina: a record of regime shifts, land-use change, and management from a temperate shallow lake.

We collected two sediment cores and modern submerged aquatic plants and phytoplankton from two sub-basins of Lake Christina, a large shallow lake in west-central Minnesota, and used stable isotopic and elemental proxies from sedimentary organic matter to explore questions about the pre- and post-settlement ecology of the lake. The two morphologically distinct sub-basins vary in their sensitivities to internal and external perturbations offering different paleoecological information. The record from the shallower and much larger western sub-basin reflects its strong response to internal processes, while the smaller and deeper eastern sub-basin record primarily reflects external processes including important post-settlement land-use changes in the area. A significant increase in organic carbon accumulation (3-4 times pre-settlement rates) and long-term trends in δ(13)C, organic carbon to nitrogen ratios (C/N), and biogenic silica concentrations shows that primary production has increased and the lake has become increasingly phytoplankton-dominated in the post-settlement period. Significant shifts in δ(15)N values reflect land-clearing and agricultural practices in the region and support the idea that nutrient inputs have played an important role in triggering changes in the trophic status of the lake. Our examination of hydroclimatic data for the region over the last century suggests that natural forcings on lake ecology have diminished in their importance as human management of the lake increased in the mid-1900s. In the last 50 years, three chemical biomanipulations have temporarily shifted the lake from the turbid, algal-dominated condition into a desired clear water regime. Two of our proxies (δ(13)C and BSi) measured from the higher resolution eastern basin record responded significantly to these known regime shifts.

Comparing effects of lake- and watershed-scale influences on communities of aquatic invertebrates in shallow lakes.

Constraints on lake communities are complex and are usually studied by using limited combinations of variables derived from measurements within or adjacent to study waters. While informative, results often provide limited insight about magnitude of simultaneous influences operating at multiple scales, such as lake- vs. watershed-scale. To formulate comparisons of such contrasting influences, we explored factors controlling the abundance of predominant aquatic invertebrates in 75 shallow lakes in western Minnesota, USA. Using robust regression techniques, we modeled relative abundance of Amphipoda, small and large cladocera, Corixidae, aquatic Diptera, and an aggregate taxon that combined Ephemeroptera-Trichoptera-Odonata (ETO) in response to lake- and watershed-scale characteristics. Predictor variables included fish and submerged plant abundance, linear distance to the nearest wetland or lake, watershed size, and proportion of the watershed in agricultural production. Among-lake variability in invertebrate abundance was more often explained by lake-scale predictors than by variables based on watershed characteristics. For example, we identified significant associations between fish presence and community type and abundance of small and large cladocera, Amphipoda, Diptera, and ETO. Abundance of Amphipoda, Diptera, and Corixidae were also positively correlated with submerged plant abundance. We observed no associations between lake-watershed variables and abundance of our invertebrate taxa. Broadly, our results seem to indicate preeminence of lake-level influences on aquatic invertebrates in shallow lakes, but historical land-use legacies may mask important relationships.