Multiproxy paleolimnological records provide evidence for a shift to a new ecosystem state in the Northern Great Plains, USA

Wetlands in the Prairie Pothole Region of the North American Northern Great Plains perform multiple ecosystem services and are biodiversity hotspots. However, climatological changes can result in sudden shifts in these important ecosystems. For example, marked increases in precipitation in the last few decades have resulted in a widespread shift in wetlands across the Prairie Pothole Region to a new ecohydrological state. We used multiproxy analyses (diatom community composition and invertebrate stable isotopes) of 210Pb‐dated sediment cores from two adjacent, but morphologically and hydrologically different, prairie‐pothole wetlands to assess the effects of hydroclimatic variability on these wetland ecosystems. Our results provide evidence that the recent ecohydrological shift in the region's wetlands is unprecedented over the past ca. 178 yr. Oxygen stable isotopes in chironomid head capsules provide a record of paleohydrology changes. The most recent sediments (i.e., those deposited after the state shift) from both wetlands revealed novel changes in diatom communities that differed greatly from earlier community compositions. In addition, a depleted signal in deuterium and 13C carbon stable isotopes observed in chironomid head capsules and Daphnia ephippia, respectively, after 1993 is likely related to an increase in methane production in these wetlands. Our study highlights the importance of considering basin morphometry including whether a wetland has an overflow point, and multiple biological indicators to study climate‐change influences on freshwater ecosystems. Research using these techniques can lead to an improved understanding of recent ecosystem shifts, an understanding that will be essential for future climate‐change adaptation and mitigation in this ecologically important region.

Wetlands play vital ecological and economic roles on the landscape by storing floodwater, sequestering carbon, cycling nutrients, facilitating pesticide degradation, and providing critical habitat and resources for plant communities and wildlife populations (Gleason et al. 2008;Dalcin Martins et al. 2017;Salimi et al. 2021). Wetland ecosystems are vulnerable to both anthropogenic and climate-change induced state shifts (Mckenna et al. 2017;Liu et al. 2018). More than 60% of wetlands have been lost globally since 1900 from the effects of climate change, and additional human impacts including drainage and filling, increased irrigation of croplands, and urbanization (Davidson 2014). Furthermore, extreme climate events (drought and deluge) add additional challenges to wetland conservation and restoration (McKenna et al. 2017;Donnelly et al. 2020). Thus, understanding climate-change impacts on wetlands is critical for future conservation.
The 750,000-km 2 Prairie Pothole Region (PPR) of the Northern Great Plains is characterized by abundant natural depressions formed as a result of the last major glaciation during the Pleistocene (Winter 2003). Wetlands that formed in these depressions (i.e., prairie-pothole wetlands) are reliant on rainfall and snowmelt as primary water sources (Leibowitz et al. 2016). The complex topography of the PPR greatly affects hydrological processes controlling surface and groundwater flows (Hayashi et al. 2016;Mushet et al. 2015). In addition, the climate across the PPR is highly variable with wet and dry cycles occurring at both short and long timescales, greatly influencing the physical, chemical, and biological conditions of these wetlands (Winter and Rosenberry 1998;Winter 2003). Thus, the spatial and temporal hydrodynamics of prairie-pothole wetlands can be complex and highly variable (Hayashi et al. 2016;Leibowitz et al. 2016) making it difficult to predict how these ecosystems may change in the future.
Since the early 1990s, specifically after 1993, the PPR has experienced increased precipitation resulting in a new ecohydrological state in the region's wetland ecosystems (Mushet et al. 2015;Mckenna et al. 2017). The resulting increases in water inputs to prairie-pothole wetlands across the PPR has significantly influenced their water depth, permanency, and chemistry; biological communities; and carbon cycling (Bansal et al. 2016;LaBaugh et al. 2016;Mckenna et al. 2017). Given that the frequency and intensity of extreme events (i.e., drought and deluge) have and will likely continue to increase (Pachauri et al. 2014), an expanded knowledge about wetland responses to past changes, including the recent novel ecohydrological state in the PPR, should reveal valuable information needed by wetland managers and conservation policy makers.
Biotic remains found in wetland sediments can provide a reliable temporal perspective on community and limnological changes over relatively long timescales, that is, hundreds to millions of years (Smol 2010). Among the different biological proxies, diatoms are the most common indicator used in paleolimnological studies because their silicified cell walls (frustule) are generally well preserved in sediments (Fritz et al. 1991;Smol 2010). In addition, they have short lifespans that make them reflective of prevailing environment conditions, they are abundant in most aquatic systems, and they have a well described taxonomy (Fritz et al. 1991;Smol 2010;Spaulding et al. 2022). Changes in environmental conditions can directly and indirectly alter diatom community composition (Hayashi 2011;Wigdahl-Perry et al. 2016). Consequently, diatoms have been widely used to track changes in the environmental conditions of aquatic systems (Fritz et al. 1991;Smol 2010;Hobbs et al. 2014).
Stable-isotope analyses of aquatic-invertebrate chitinous remains in sediments can be used to provide additional information about past environmental and ecological changes to freshwater systems (Leng 2006;Smol 2010;Belle et al. 2015). Midges (Diptera: Chironomidae) and water fleas (Cladocera: Daphniidae: Daphnia), hereafter referred to as chironomid and Daphnia, respectively, are abundant in many aquatic systems. The head capsules of chironomids and ephippia of Daphnia are often well preserved in sediments (Smol 2010). The ratio of the oxygen stable isotopes 16 O and 18 O in water (δ 18 O water ) are affected by climate processes, including variations in temperature, precipitation, and evaporation (Leng 2006;Verbruggen et al. 2011). Generally, areas with higher temperature have higher δ 18 O water values due to the relatively high δ 18 O precipitation they receive compared with cold areas (Leng and Marshall 2004;Leng 2006). Similarly, high temperatures lead to greater evaporation rates and higher values of δ 18 O water because 16 O evaporates at a higher rate than the heavier 18 O at the same temperature (Leng and Marshall 2004;Leng 2006). The fractionation of oxygen isotopes between chitin and water (28‰, Wooller et al. 2004) has been thought to be independent of temperature (Mayr et al. 2015). δ 18 O in Chironomidae is mainly affected by the ambient water in which they live, and to a lesser extent by diet (Wang et al. 2009;Verbruggen et al. 2011). Therefore, δ 18 O in chironomid head capsules can be used to reconstruct past changes in lake-water, inferring changes in temperature, precipitation, and ambient water condition (Wooller et al. 2004;Verbruggen et al. 2011;Chang et al. 2016).
The strong relationship between the composition of stable isotopes of carbon (δ 13 C) and nitrogen (δ 15 N) and diet allows the use of in Daphnia ephippia to infer organic carbon and nitrogen sources (Schilder et al. 2015a,b;Morlock et al. 2017). Similarly, deuterium (δD) in chironomid head capsules has been shown to be a reliable indicator of foods consumed by chironomid larvae while living, between 53% and 85% of total hydrogen has been shown to be contributed from diet (Wang et al. 2009;Soto et al. 2013;Belle et al. 2015). Moreover, both chironomid larvae and Daphnia are first-order consumers of algae, bacteria (including methane-oxidizing bacteria [MOB]), and detritus. Therefore, these two invertebrates are important links between primary producers and higher order consumers in aquatic food webs (Perga 2011;Schilder et al. 2015a). Consequently, δ 13 C and δ 15 N found in Daphnia ephippia and δD in chironomid head capsules both provide a means to gain insights into the past cycling of carbon and nitrogen cycling, and potential food-web changes.
In this study, we compare two hydrologically distinct (closed-basin and open-basin) wetlands in the Cottonwood Lake Study Area of the southern PPR. We analyzed multiple biotic proxy records in sediment records, which provide independent indicators of past change. By combining 60 yr of field-collected environmental data with biotic proxy of 210 Pbdated sediment cores from two representative wetland systems in PPR, we aimed to assess the effects of a highly variable climate on wetland ecosystem changes. First, we analyzed δ 18 O in chironomid head capsule to assess the main drivers that affect past wetland water changes in the context of climate change, in particular extreme events of drought and deluge. Second, we analyzed diatom assemblages to assess the patterns of respond to the extreme climate events. Third, we analyzed δ 13 C and δ 15 N from Daphnia ephippia and δD in chironomid head capsules to track shifts in organic matter source, especially carbon and methane derived carbon changes following environmental changes. Lastly, based on the information gained from the above proxy records, we identify a novel ecosystem state in recent decades, and provide potential implications for conservation management in prairie wetlands.

Study site
The 92-ha Cottonwood Lake Study Area is located in the southern PPR near the eastern edge of the Missouri Coteau in Stutsman County, North Dakota, USA. The study area has been owned by the U.S. Fish and Wildlife Service since 1963 and has been used as a study site by the U.S. Geological Survey's Northern Prairie Wildlife Research Center since 1966 (Winter 2003) (Fig. 1). Native prairie grasslands and wetlands cover over 80% of the site (Winter 2003). The bedrock in this area consists of Precambrian igneous and metamorphic rocks, early Paleozoic sandstones, limestones, dolomites, and Cretaceous shales and siltstones (Winter 2003). The glacial drift that covers the bedrock is mostly clayey, silty till in the upper 5-15 m. This upper till is often oxidized and fractured, resulting in higher hydraulic conductivity than the deeper glacial deposits that have extremely low hydrologic conductivity and therefore facilitate the ponding of surface waters (Hayashi et al. 2016). Our two adjacent study wetlands, P1 and P8, are both semipermanently ponded wetlands that contain water throughout the year except during exceptionally dry periods ( Fig. 1; Table 1).
The climate in the study area is semi-arid with periods of excess precipitation as well as drought. Average  annual precipitation at the study area is 440 mm and average annual evaporation is 810 mm (Winter 2003). The average annual temperature is 4 C (Winter 2003). Annual snowfall averages 865 mm and the snowpack commonly persists from early December to late March (Winter 2003). Due to frequent drying, annual freezing, and high salt concentrations, wetlands in this study area are typically fishless, which leads to a high abundance of invertebrates including chironomid and Daphnia (Winter 2003).
Wetland P1 is a turbid shallow system with a maximum and average water depth of 2.8 and 1.4 m, respectively ( Table 1). The closed-basin wetland (P1) had a wider range in historical water-level fluctuations due to its lack of any overland outflow. Currently, wetland P1 is hypereutrophic with high concentrations of total phosphorus (224.3 μg L À1 ), total Kjeidahl nitrogen (16.4 mg L À1 ), and chlorophyll a (Chl a, 29.5 μg L À1 ). Average Secchi depth was 0.2 m during the ice-free seasons of 2018 and 2019 (Table 1). Wetland P1 dried completely every summer during a drought from 1988 to 1992 (Winter 2003). The dominant major ion in wetland P1 was magnesium sulfate (Table 1). Submerged macrophytes and emergent vegetation were prevalent from June to September (Winter 2003).
Wetland P8 is smaller (0.02 km 2 ) and shallower (average depth = 0.9 m and max depth = 1.2 m, Table 1) than wetland P1. P8 has a natural outlet at its northern end that limited water levels (Fig. 1). In addition, wetland P8 is intersected by a large sand lens that can stabilize water-level changes by transmitting groundwater during drought (Winter 2003). Currently, it is a clear, eutrophic, shallow system with lower concentrations of total phosphorus (217.0 μg L À1 ), total Kjeidahl nitrogen (15.8 mg L À1 ), and Chl a (7.5 μg L À1 ) than wetland P1, and higher transparency, that is, visible to the bottom. In surveys from June to September in 2018 and 2019, submerged vegetation in Wetland P8 covered more than 95% of the wetland. Wetland P8 tends to have a magnesiumbicarbonate or sulfate-bicarbonate water type (Winter 2003). Anaerobic conditions occur under the cattail mats of wetland P8 allowing sulfate reduction that removes sulfur from groundwater inputs (Winter 2003). Wetland P8 only completely dried out during summer droughts in 1991 and 1992 (Winter 2003). Both wetlands are polymictic systems that are without stratification during the ice-free season.

Long-term archived data
The meteorological data (annual temperature, precipitation, and wind speed) used in this study were weighted averages of data from three nearby National Weather Service stations, Carrington (39-km north of the Cottonwood Lake Study Area), Jamestown (39-km southeast), and Pettibone (32-km west). Data were accessed from the National Oceanic and Atmospheric Administration website (https://www.noaa.gov/). Temperature and precipitation data are available from 1893 ( Fig. 2a,b). Palmer Drought Severity Index (PDSI) data were obtained from North Dakota State University Annual North Dakota Climatic Data (https://www.ndsu.edu/climate), which is available from 1895 to 2019 (Fig. 2c). Long-term monitoring data, including water levels (  (Mushet et al. , 2017a. Water chemistry parameters, including total phosphorus (TP), soluble reactive phosphorus, ortho phosphorus, total Kjeidahl nitrogen, Chl a, phaeophytin, total ammonia, alkalinity (CaCO 3 ), specific conductance, pH, and major ion concentrations (Ca 2+ , Cl À , K + , Na + , Mg 2+ , SO 2À 4 , SiO 2 ), were used to explore potential mechanisms of indicator changes (Supporting Information Fig. S2).

Coring and sampling and chronology
Sediment cores were collected from wetlands P1 (64-cm length) and P8 (40-cm length) on 19 March 2018 and 24 July 2018, respectively, from the deepest part of each basin using a Universal Percussion Corer (Aquatic Research Instruments) (Fig. 1). The sediment cores were sectioned in the field at 1-cm intervals for the top 5 cm and in the laboratory for depths below 5 cm. Samples were stored at 4 C until being freeze-  (PDSI, a PDSI value > 4 represents very wet conditions, while a PDSI < -4 represents an extreme drought), (d) wetlands P1 and P8 mean summer water depth, (e) and specific conductance. PDSI data were collected from NDSU Annual North Dakota Climatic Data, https://www.ndsu.edu/climate, water depth data from Mushet et al. (2016) and specific conductance data from Mushet et al. (2017c).
dried for subsequent analyses. Samples of each core segment were analyzed for 210 Pb, 137 Cs, organic matter (LOI550), and carbonate (LOI950) (Heiri et al. 2001). Sediment chronology was estimated based on analysis of excess 210 Pb activity in sediment core samples with validation from an independent analysis of 137 Cs. Details are provided as Supporting Information.

Diatom community analyses
Diatom frustules were prepared for enumeration following the methods of Battarbee et al. (2001). Briefly, hydrochloric acid (10%) was used to remove carbonates before processing. Hydrogen peroxide (30% concentration) was added at $ 80 C to remove organic material. A portion of the digested slurry was dried on cover slips and mounted with Naphrax ® . Diatom identification and enumeration were performed under oil immersion at Â1000 magnification using an Olympus model BX60 microscope with phase contrast. A minimum of 400 diatom frustules were identified and counted from each sample. Diatom taxonomy followed Krammer and Lange-Bertalot (1986-1991. Updated naming conventions were obtained from Diatoms of North America (https://diatoms.org/; Spaulding et al., 2022). Community data were expressed as percent abundance of the total diatom sum in each sample. Diatom species were categorized as either planktonic or benthic following their habitat preferences based on Spaulding et al. (2022; https://diatoms.org/).

Invertebrate stable-isotope analyses
Stable isotope analyses of chironomid head capsules and Daphnia ephippia were conducted following the methods of Wang et al. (2008). Briefly, 2-4 g of freeze-dried sediment were treated with 10% KOH in a warm (60-70 C) water bath for 15-20 min. After the sample cooled, sediments were rinsed with distilled water through a 100-μm sieve . Residual material was refrigerated at 4 C in a 50-mL centrifuge tube until sorted. For sorting, small aliquots of aqueous residual material were transferred into a Bogorov counting tray using a pipette and examined under a dissecting microscope at Â25 magnification. Chironomid heads capsules (0.05-2.5 mg) and Daphnia ephippia (0.07-0.6 mg) were separately transferred using fine forceps into preweighed, tin capsules filled with distilled water. After all invertebrate remains were transferred into tin capsules, the open tin capsules were allowed to air dry overnight. The tin capsules were then crimped to leave a small opening and freeze dried for > 6 d. The tin capsules were then reweighed, folded into a ball, and stored in a 96-well culture plate in a desiccator. Empty tin capsules were included every 5 samples in sample preparation (total 8 empty tin capsules for 43 samples) for the oxygen and hydrogen isotopes blank correction ). The mass of each sample was calculated by subtracting the tare weight (i.e., weight of the empty tin capsule). The chironomid samples were then loaded into a zero-blank auto sampler attached to an on-line pyrolysis thermochemical reactor (ThermoElectron TC/EA) coupled via a Conflo III with a thermoFinnigan Delta plus XP IRMS at the Alaska Stable Isotope Facility (University of Alaska Fairbanks). Chironomid headcapsule results are reported in units of δ 18 O and δD per mil (‰) relative to Vienna Standard Mean Ocean Water (V-SMOW). Daphnia ephippia stable isotopes were quantified using a Costech ECS4010 elemental analyzer (EA) attached via a conflo III to an IRMS (thermoFinnigan Delta plus XL). δ 13 C values are reported relative to Vienna Pee Dee Belemnite and δ 15 N values are expressed relative to atmospheric (atm.) nitrogen. Analytical precision for δ 18 O, δD, δ 13 C, and δ 15 N were 0.6‰, 4‰, 0.2‰, and 0.1‰, respectively. Invertebrate stable isotopes were quantified for wetland P1 and P8 to a core depth of 21 and 22 cm, respectively.

Numerical methods
Major zones of biostratigraphy were identified through the analyses of constrained incremental sum of squares (Juggins 2020). A broken-stick model was used to test for the significance of each break (Juggins 2020). We conducted ordination analyses to define historical trajectories of diatom changes. Preliminary investigation with detrended correspondence analysis of P1and P8 diatom assemblages revealed short gradient lengths of 1.7 and 1.1 SD, respectively, indicating that diatom taxa of each wetland responded in a linear fashion to environmental gradients (ter Braak and Prentice, Ter Braak and Prentice 1988). Thus, a principal components analysis (PCA) was performed to identify changes in diatom species compositions. Data analyses were performed using the R packages vegan (Oksanen et al. 2015) and rioja (Juggins 2020). Long-term archive data were calculated based on each wetland sedimental rate to explore the relationship between our multiple biological proxy and monitoring environmental data. All variables with more than six samples were included in the correlation analysis (Supporting Information Tables S2, S3). The collinearity matrix was built based on the function "ggpairs" in the GGally R package (Schloerke et al. 2018).

Long-term archived data
Mean annual temperature showed an increasing trend from an average of 4.5 C AE 1.2 SD from 1893 to 1992 to 5.0 C AE 1.4 C during the last three decades (Fig. 2a). Annual precipitation displayed greater variability over the past century. Annual precipitation from 1893 to 1912 averaged 476 AE 129 mm. By 1939, annual precipitation decreased to 411 AE 118 mm. From 1940 to 1992, the annual precipitation was relatively steady at 444 AE 84 mm. This was followed by a large increase between 1993 and 2019 to 489 AE 113 mm (Fig. 2b). The PDSI revealed two extreme drought periods, one from 1933 to 1940 and a second from 1988 to 1992 (Fig. 2c). In contrast, an extreme wet period occurred from 1993 to 2019 (Fig. 2c).
A total of 58 diatom taxa were identified in the P8 core, including 34 and 25 taxa with a maximum relative abundance ≥ 1% and 2%, respectively, in at least one sample (Fig. 4b).
The first two PCA axes of diatom communities were statistically significant in both wetlands. In wetland P1, PCA axis 1 explained 42% of the total variation and was strongly correlated with water-level (r = 0.79; p = 0.01; n = 10). Diatom benthic taxa Epithemia adnata, Sta. capucina, Am. libyca, Ps. brevistriata were associated with shallow water levels (i.e., the negative side of PCA axis 1), while Nitzschia palea, Ste. hantzschii, Navicula cryptocephala, Cy. meneghiniana were associated with deeper waters (i.e., the positive side of PCA axis 1) (Fig. 5a). PCA axis 1 site scores were stable until ca. 1966, decreased thereafter and exhibited a slightly reversed trend after ca. 2006 (Fig. 3c). PCA axis 2 explained an additional 10% of the total variance in wetland P1 diatom communities and was correlated with organic matter concentrations. Gomphonema parvulum was strongly and positively correlated with PCA axis 1 (Fig. 5a). The scores of PCA axis 2 were relative stable until ca. 1966, following by a steadily decreasing trend to recent (Fig. 3d).
The first two PCA axes from the PCA of wetland P8 diatoms were also significant and explained 28% and 12% of the total diatom community variation, respectively (Fig. 5b). G. parvulum, Gomphonema gracile, Ni. palea, and Tabularia fasciculata were also positively associated with water depth (PCA axis 1). Sta. capucina was negatively associated PCA axis 1. The scores of PCA axis 1 were stable before ca. 1994 and showed an increase recent in recent years (Fig. 3c). Co. placentula was positively correlated with organic matter (PCA axis 2). Navicula cincta was negatively correlated with the organic matter axis (Fig. 5b). PCA axis 2 site scores in wetland P8 were relative stable until 1994, decreased thereafter, and showed a reversed trend after 2008 (Fig. 3d). PCA axis 2 scores were highly significant and negatively correlated with the planktonic changes (r = À0.67; p < 0.001; n = 33) in wetland P8.

Discussion
The extended wet period that began in 1993 in our study portion of the PPR has resulted in a novel ecohydrological state in the region's wetlands (McKenna et al. 2017). Our analyses of multiple paleolimnological indicators provided additional evidence of a recent shift to a novel state and demonstrated that ecosystem responses at the Cottonwood Lake Study Area were unprecedented over the past 178 yr. This shift has resulted in a reorganization of the hydrology and freshwater communities with implications for ecosystem functioning and management. Chironomid δ 18 O from two hydrologically different wetlands revealed the changes of past hydrological condition. The two wetland ecosystems both showed similar patterns of changing diatom communities with a marked increase in planktonic species tied to the recent (past 30 yr) shift to wetter conditions. Over this same period, the D-depleted in chironomid head capsules and δ 13 C signal in Daphnia ephippia indicated a shift to increased reliance on methanotrophic bacteria as a carbon source, indicating a likely increase in overall CH 4 fluxes from these wetlands. The different morphometry (i.e., closed vs. open basin) of the two wetlands may have contributed to differences observed in the timing and magnitude of changes.

Past hydrological changes
The chironomid δ 18 O changes we observed may reflect hydrological changes, changes that were a combination of temperature, precipitation, evaporation, and wetland  morphometry processes. Even though the relative contribution of each environmental variable was different in our two wetlands, the strong correlation between δ 18 O chironomid and PDSI in the closed basin wetland showed the amalgamated effect of changes in temperature, precipitation, and evaporation. Closed basin systems in tropic or arid areas lose water (δ 15 N) stable isotopes of cladoceran Daphnia ephippia, (e) Daphnia ephippia total carbon content (TC) and (f) total nitrogen content (TN), and (g) ephippia mass C : N ratio from wetlands P1 and P8 at the Cottonwood Lake Study Area, Stutsman County, North Dakota. The years of 1993 and 2008 were selected based on both instrumental data (meteorological data and PDSI) and diatom community paleo record of statistical analysis (constrained cluster analysis) in wetland P8. mainly through evaporation compared to open basin systems, so the changes of δ 18 O water inferred from δ 18 O chironomid in wetland P1 could indicate long-term changes in precipitation/ evaporation ratios (Leng and Marshall 2004). Although δ 18 O chronomid of open-basin wetland P8 may represent mostly the influence of precipitation with evaporation playing a lesser role (Leng and Marshall 2004). In our study sites, δ 18 O chironomid changes were consistent with this typical pattern related to closed-basin vs. open-basin systems. In Northwest Greenland and North American Arctic areas, studies that combine contemporary and sediment-core data demonstrated that chironomid δ 18 O can provide insights into past changes in the δ 18 O of precipitation over long periods related to annual air temperature (Wooller et al. 2004;Lasher et al. 2017). Past summer temperature could be reconstructed in European boreal-forest lakes by using chironomid δ 18 O data (Verbruggen et al. 2010;Verbruggen et al. 2011). However, studies in Iceland and Australia suggested chironomid δ 18 O represented more paleoenvironmental information, like hydrological conditions, rather than solely temperature and precipitation changes Chang et al. 2016).
Until now, there have been no similar studies in the highly variable climate area of the PPR. Unlike high-latitude arctic ponds in which δ 18 O chironomid values are mainly affected by local precipitation, there was no strong correlation between δ 18 O chironomid values and mean annual temperature or precipitation in our two wetland systems. Even though, our 2-yr (2018 and 2019) seasonal survey showed δ 18 O of wetland water was significantly influenced by the air temperature changes in both wetlands (P1: r = 0.53, p < 0.001, n = 20; P8: r = 0.44, p < 0.01, n = 20) (K. Hu et al. unpubl.). In our paleolimnological records, chironomid δ 18 O more likely reflected overall hydrological changes rather than specific environmental drivers. However, we cannot exclude the vital effect (biogenic metabolism activity) during our study period, especially the 1ow δ 18 O value in both wetlands during the drought 1989 to 1992 (Fig. 6a) (Chang et al. 2016). Contemporary field surveys relating δ 18 O chironomid to parameters such as δ 18 O water , δ 18 O precipitation , water temperature, and air temperature will be helpful for examining the contribution of individual processes to the paleo-hydrological changes (Verbruggen et al. 2011;Heiri et al. 2012;Van Hardenbroek et al. 2018).

A new diatom community state
Changes in diatom assemblages associated with the recent wet conditions were observed in both wetlands. Water-level fluctuations could alter diatom communities by changing both physical (habitat zonation) and chemical (salinity) conditions in prairie ecosystems (Wigdahl et al. 2014;Wigdahl-Perry et al. 2016). Benthic diatom species (Am. libyca, Ps. brevistriata, Co. placentula, and Ni. amphibia) were abundant until ca. 1966 and ca. 1994 in wetland P1 and P8, respectively. Am. libyca and Ni. amphibia are benthic diatom species that indicate a lower trophic level and clear-water state with high coverage of macrophytes in shallow prairie lakes within the Northern Great Plains (Hobbs et al. 2014). After ca. 1966, the percentage of planktonic species, that is, Cy. meneghiniana and S. hanzschii, increased in wetland P1, suggesting loss of littoral habitats of macrophytes (Hayashi 2011) or increase input of nutrient (especially TP) (Hobbs et al. 2014). The loss of macrophytes might be the result of an increase in water level as there was a wet period since 1966 (Fig. 2c). Alternatively, increase input of nutrient could lead to high abundance of Cy. meneghiniana and S. hanzschii that are planktonic species that prefer high total phosphorus in turbid prairie lakes (Hobbs et al. 2014). Furthermore, Chl a was the main driver of Cy. meneghiniana in both wetlands on our seasonal contemporary survey (K. Hu et al. unpubl.). The increase of planktonic taxa in both wetlands after the 1993 drought was likely a result of increased water inputs from precipitation (McKenna et al. 2017). The unexpected decrease in the planktonic species Cy. meneghiniana after ca. 1993 may be due to increasing water salinity (LaBaugh et al. 2016;Levy et al. 2018) that benefited the more salt tolerant species Co. placentula in wetland P1 (Figs. 2e, 4a; Supporting Information Table S2) (Fritz et al. 1991;Laird et al. 1996). Research in wetland P1 has shown that salts precipitated along the wetland edge and in deeper sediments during dry periods were resuspended into water when the water level increase and wetland area expanded (LaBaugh et al. 2016;Levy et al. 2018). The increasing abundance of planktonic species Au. ambigua and Cy. meneghiniana was compatible with the stable higher water level in wetland P8, until this stable high-water level favored an increased growth of macrophytes, resulting in a decrease in planktonic taxa after ca. 2008 (Figs. 2d, 4b). Even though the peripheral area could not increase in wetland P8 during the wet period due to the natural outlet, cattails could and did extend into the open water areas as floating mats (Winter 2003). In general, the recent shift to sustained higher precipitation resulted in large shifts in diatom community structure in both study wetlands.
Prior to the most recent ecoclimatic shift, the diatom community response at our study sites during earlier climate fluctuations (e.g., 1930s extreme drought), are considerably more muted, consistent with ostracod and another diatom-based research in the region (Winter and Rosenberry 1998). Sediment records from Devils Lake, North Dakota, indicated that the 1988-1992 drought may have been the second worst of the 20 th century, but that droughts of equal or greater magnitude were common during the past 500 yr (Winter and Rosenberry 1998). By contrast, the recent novel wet period is possibly unprecedented during the past 500 yr (Winter and Rosenberry 1998). Long-term patterns of natural variation in climate show that the Northern Great Plains is characterized by a high frequency and intensity of drought events (Fritz et al. 1991;Laird et al. 1996). Previous diatom-based studies have demonstrated that there were numerous shifts between fresh and saline conditions over the past 2300 yr in the PPR (Fritz et al. 1991;Laird et al. 1996). However, diatom communities in our sediment records were relatively stable until the recent wet period. The lack of an observable diatom assemblage shift in the 1930s "dust bowl" drought in our study could be diatom adapted to the extreme climate events in this region.
A potential new trophic state related to increased methane production Methanogenesis is an important pathway for the breakdown of organic matter under anoxic conditions (Rudd and Hamilton 1978;Belle et al. 2015;Grey 2016). Methane formed in aquatic systems can be oxidized by MOB, especially in well-oxygenated, shallow, polymictic systems (Bastviken et al. 2002). MOB can then be assimilated by metazoan organisms and as a complementary food source for higher trophic levels in food webs (Rudd and Hamilton 1978;Van Hardenbroek et al. 2013). Due to the high trophic fractionation effect of methanogens, biogenic methane usually has very depleted D (δD ranges from À400‰ to À250‰ for the acetate fermentation and from À250‰ to À150‰ for carbon dioxide reduction) (Whiticar 1999) and depleted 13 C values (δ 13 C ranges from À70‰ to À50‰) (Summons et al. 1994). Like chironomid, Daphnia, a planktonic filterer, can consume MOB from the water column (Taipale et al. 2007;Devlin et al. 2015). Moreover, studies of chironomid head-capsule δD, δ 13 C, and ancient DNA in sedimentary cores have shown that chironomids can use MOB as an important food source (20-54% of their diet) in European lakes (Belle et al. 2015). Hence, δD in chironomid head capsules and δ 13 C in Daphnia ephippia can be used to quantify the importance of methanogenesis as a carbon source for aquatic invertebrates (Belle et al. 2015;Schilder et al. 2015aSchilder et al. , 2017. Both chironomid head-capsule δD and Daphnia ephippia δ 13 C suggest a possible shift in food sources to methanotropic bacteria in the new ecosystem state in wetland P8. The increased wetland water levels, high organic matter loading, and increased temperatures in the recent novel period have favored methanogenesis and led to high methane fluxes in these wetlands, some of the highest reported for any wetlands worldwide (Bansal et al. 2016;Dalcin Martins et al. 2017). In wetlands P8, the decrease of chironomid head capsule δD values after 1993 (Fig. 3b) is concurrent with the low Daphnia ephippia δ 13 C values (Fig. 3c). This may suggest there was increase of methane derived carbon as food for invertebrate taxa in this nonsulfate-dominated water wetland P8 during the new state. The lack of signals in sediment of methane derived carbon as food source in wetland P1 could because the higher sulfated levels that suppress methanogenesis (Dalcin Martins et al. 2017) and lower macrophytes coverage (Bansal et al. 2020).
Alternatively, the relatively low δ 13 C values in Daphnia ephippia could be the result of increased consumption of phytoplankton (Schilder et al. 2015a,b). However, in our sedimentary records, δ 15 N was relatively constant during the period when there was a rapid decrease of δ 13 C (Fig. 3c, d).
Reported δ 13 C values in Daphnia ephippia that correlated from methane-derived carbon could range from À25.9‰ to 41.9‰ (Schilder et al. 2015b. In addition, the relative stable trend after ca. 1993 of ephippia δ 15 N values could exclude the possibility of Daphnia consumption of phytoplankton instead of MOB organic matter as their main food source in the new state (Leng and Marshall 2004;Grey 2016). The decrease of ephippia δ 15 N after ca. 2000 is consistent with the use of MOB as a food source and potential ammonium oxidation (Grey 2016). Thus, our observed changes in chironomid headcapsule δD, together with changes in δ 13 C and δ 15 N in Daphnia ephippia could suggest that there is more methane is being produced and consumed under the new, wetter conditions in these prairie-pothole wetland ecosystems. Taxon-specific diet preference of chironomids and Daphnia could have an impact on the δD in chironomids and δ 13 C in Daphnia ephippia, as there was no significant correlation between δD in chironomid and δ 13 C in ephippia in both wetlands during the wet period. Further studies examining taxonspecific diet preference and methane derived organic matter in this study area could help us gain insights into this process (Heiri et al. 2012;Van Hardenbroek et al. 2018). In all, changes in methanogenesis and methanotrophy in the new wetter period may have significant implications for net greenhouse-gas fluxes.

Implications for wetland conservation
Differences in basin morphometry (i.e., closed vs. open) likely contributed to some of the differences we observed in timing and magnitude of climate-change signals preserved in the sediments of our two study wetlands. For example, the lower organic matter in wetland P1 compared to wetland P8 could be the result of higher desiccation and oxidation of sediment due to longer exposure time to air during the 1988-1992 drought. Diatom-based salinity transfer functions are commonly used to infer the frequency and intensity of extreme droughts, especially for closed-basin lakes in which changes in precipitation inputs lead to the concentration or dilution of dissolved salts (Fritz et al. 1991;Laird et al. 1996). For example, in the Northern Great Plains, changes in salt concentrations resulting from climate changes can be inferred from diatom assemblages (Fritz et al. 1991;Laird et al. 1996). However, studies have shown that basin morphometry influences the degree and accuracy of diatom-based salinity transfer functions (Wilson et al. 1996;Wigdahl et al. 2014). Similarly, the ratio of benthic vs. planktonic diatom species has been used to infer microhabitat changes resulting from climate changes (Wilson et al. 1996;Wigdahl et al. 2014). In small freshwater basins, changes in the ratio of benthic vs. planktonic diatom species may be a better proxy for indicating drought (Wigdahl et al. 2014). In our two systems, the diatom community in the open-basin system had quicker and higher magnitude responses to the recent wet period.
In our analyses of invertebrate, isotope-based, methane proxies, we found that methane dynamics in the open-basin wetland, P8 containing lower sulfate levels, was more sensitive to the recent shift to wet conditions. Basin morphometry can affect water chemistry changes, which in turn influence wetland vegetation, thereby having an important indirect effect on methane production (Bansal et al. 2020). Wetland P1 is known for its high sulfate concentrations derived from weathered glacier tills with this closed-basin wetland displaying the highest sulfate concentrations at our study sites, especially with the addition of salts from its expanded periphery area and deep sediments during the wet period (LaBaugh et al. 1996(LaBaugh et al. , 2016Levy et al. 2018). The high sulfate levels in wetland P1 could restrain methanogeneic processes, thereby decreasing methane concentrations (Dalcin Martins et al. 2017). In contrast, freshwater-preferring cattail (cattail rings and cattail floating mat) and macrophytes canopy can increase methane concentrations by providing organic matter and anoxia environment for methane production (Agasild et al. 2014;Bansal et al. 2020). Thus, open-basin wetlands such as P8 could be useful systems for identifying long-term methane production and the relative contribution of the two pathways of methanogenesis in the prairie-pothole wetland ecosystems of the Northern Great Plains.
Our multiple bio-indicators from the sediment record responded consistently to recent observed ecological shifts identified by Winter and Rosenberry (1998) and McKenna et al. (2017). Different aquatic organisms may show heterogeneous responses to environmental changes due to different ecological sensitivities to and interactions with different environmental variables (Hu et al. 2019;Klamt et al. 2021). Primary producers in sediment (e.g., diatoms and algae) are more sensitive to nutrient change than to changes in hydrology (Hu et al. 2019). In our study, multiple paleolimnological indicators showed a synchronous, unprecedented change in hydrology, freshwater microhabitat, and carbon cycling in response to recent changes in climate. Overall, the shifts in multiple biological indicators in both wetlands provide further evidence that these prairie-pothole wetlands have shifted into a new ecosystem state, a state that may also have altered their ecosystem function and value to society in terms of the ecosystem services they provide. Improved understanding of how ecosystem may be altered by these novel ecosystem shifts will be essential for long-term ecological management in this region and for developing a better understanding of changes in methane fluxes related to global climate change.
Wetlands in this region are important habitat for biodiversity, as well as potential sources of greenhouse gases (Winter 2003;Bansal et al. 2016). The novel changes in wetland hydrology and water chemistry can influence freshwater biological communities, carbon cycling, and food webs (McLean et al. 2016). Changes in diatom communities, a vital part of aquatic primary production, can affect waterfowl populations and higher trophic levels by bottom-up effects (Hobbs et al. 2014;McLean et al. 2016). Our study suggests that recent climate shifts have resulted in an increased uptake of methane-derived carbon to the planktonic food web in our study sites, which may indicate changes in the methane dynamics of these wetlands due to altered ecosystem functioning. Further research may be required to understand how changing climate might alter methane dynamics, as wetlands are the largest natural source of methane (Dalcin Martins et al. 2017). Thus, this novel ecohydrological shift may require a refinement of previous knowledge related to wetland functioning in the PPR.

Data Availability Statement
The datasets generated during this study will be made through the Dryad data repository at https://doi.org/10.5061/ dryad.8cz8w9gv0. Data generated by National Oceanic and Atmospheric Administration and used in this study are available https://www.noaa.gov. Data generated by USGS and used in this study are available in four data releases (Mushet et al. , 2017a.