Influence of thermal regime, oxygen conditions and land use on source and pathways of carbon in lake pelagic food webs

ABSTRACT Intensification of anthropogenic activities in many lake catchments during the twentieth century led to increased autochthonous organic matter sedimentation and degradation of hypolimnetic oxygen conditions due to the intensification of heterotrophic processes. These processes can be amplified by the effect of climate warming on thermal stratification in lakes. This study aimed to assess how metabolic disruptions affect carbon sources and pathways in lake pelagic food webs, focusing on methanogenic carbon. The studied lakes showed strong seasonal variations of carbon source availability and transfers to pelagic food webs, characterized by increased methanogenic carbon transfers to Daphnia populations in winter. The magnitude of these winter transfers seems to largely depend on the amount of methane stored in the hypolimnion during the stratification period, and thus on the amount of methane released with autumnal turnover. Methane production, storage and transfer mechanisms partly depend on thermal stratification intensity, but also on external factors such as land use. This study provides new insights into the impacts of global changes on the sources and pathways of carbon in pelagic food webs through their influence on lake metabolism and thermal regimes. These functional changes may lead to greater production and release of greenhouse gases into the atmosphere.


Introduction
Lakes are important ecosystems for the transport, emission and storage of considerable quantities of carbon. Although they cover only 0.9% of the earth's surface (Downing et al. 2006), they exert a disproportionate effect on carbon cycle (Tranvik et al. 2009). However, the carbon balance of lakes is still under-documented (Tranvik et al. 2009) due to the complexity of the carbon cycle in lakes. Indeed, lakes play a dual role as ecosystems that, on the one hand, fix inorganic carbon (atmospheric CO 2 ) through photosynthesis and, on the other hand, mineralize organic carbon from autochthonous and allochthonous sources (production of CO 2 and CH 4 ). Understanding the trophic functioning of lakes and in particular the mechanisms of carbon production and transfer in food webs is therefore a major challenge in the context of global change.
One of the main carbon pathways in lakes is the fixation by primary producers of dissolved carbon which can be of atmospheric origin (CO 2atm ) or released by heterotrophic activities (CO 2resp ). Fixation of methanogenic carbon (C-CH 4 ) by methane oxidising bacteria (MOBs) is another carbon pathway linked to a different form of heterotrophic activity, the methanogenesis, performed by methanogenic archaea that degrade organic matter under anoxic conditions (Rudd and Hamilton 1978).
Global changes are likely to cause disruptions in the trophic functioning of lakes with profound cascading consequences on the associated ecosystem goods and services. On one hand, it is now well established that global warming will have serious impact on the thermal cycle of lakes (Butcher et al. 2015;Woolway and Merchant 2019). An increase in the strength and duration of summer stratification, and less frequent mixing for mid-altitude lakes, will have deep consequences on hypolimnetic oxygen conditions (Golosov et al. 2012;Butcher et al. 2015;Pilla and Williamson 2021). Production, storage and transfers process of C-CH 4 to pelagic food webs are partly controlled by lake thermal regimes (Tranvik et al. 2009;Duc et al. 2010;Yvon-Durocher et al. 2014). In stratified lakes, a large part of methane produced in sediment can indeed accumulate in hypolimnion during summer stratification period, and be released in large quantity into water column in oxic conditions during autumnal turnover (Utsumi et al. 1998). CH 4 can then be oxidised by MOBs and support higher trophic level (Rudd and Taylor 1980;Utsumi et al. 1998). However, knowledge remains limited on the influence of stratification and turnover mechanisms on the intensity of gas production and storage process in deep layers of lakes, and subsequently on the extent of CH 4 transfers in pelagic food webs.
On the other hand, the accelerated eutrophication of waters linked to intensification of anthropogenic activities on the watershed is one of the main forcing factors that may influence the carbon cycle and the trophic functioning of lakes. Over the last decades, the intensification of agropastoral practices and associated disturbances (e.g., increase in organic matter and nutrients inputs in lakes) have led to a strong increase in autochthonous organic sedimentation rates (Anderson et al. 2014), and to the development of unprecedented anoxic conditions in hypolimnion of many lakes due to increasing heterotrophic activities (Frossard et al. 2014;Jane et al. 2021). Numerous studies assessing past trophic functioning of lakes highlighted that these trophic state changes, marked by increasing heterotrophic activities, are associated to an increase in CO 2resp and CH 4 transfers to consumers in lakes food webs (Frossard et al. 2014;Belle et al., 2016aBelle et al., ,2016bSchilder et al. 2017). Indeed, intensity of CH 4 production and oxidation is strongly dependent on oxygen conditions (Rudd and Hamilton 1978), which could be strongly affected by eutrophication. Moreover, several experiments have highlighted that the methanogenic activities can be strongly promoted by the increase in autochthonous organic matter inputs (Duc et al. 2010;West et al. 2012;Praetzel et al. 2020), the phytoplankton-derived organic matter being rich in nitrogen and then easily degradable for methanogenic Archaea. Belle et al. (2016a) suggested that agropastoral practice intensity on watershed may partly drive increased methane transfers to benthic consumers in 14 French lakes. Nonetheless, few studies attempted to assess the implication of watershed coverage, or sediment and thermal stratification characteristics on the production and seasonal transfer processes of these heterotrophic derived carbon sources in lake pelagic food webs.
Stable isotope analysis (SIA) of carbon is a tool that has been widely used over the last decades for studying carbon flows in aquatic ecosystems, especially in lakes (Grey et al. 2001;Fry 2006;Taipale et al. 2007;Frossard et al. 2014). Isotopic carbon signature (δ 13 C) of organisms closely reflects the isotopic signature of their diet (DeNiro and Epstein 1978) and are therefore commonly used to track and highlight the importance of heterotrophic process in lakes (Del Giorgio and France 1996). Indeed, CO 2resp usually exhibits δ 13 C values that are consistently lower than CO 2atm as a result of photosynthetic and metabolic fractionation process occurring during carbon recycling (Degens 1969;Keough et al. 1996;Raven 1996;Fry 2006). Biogenic methane is moreover highly depleted in 13 C (Jedrysek 2005), and isotopic fractionation by MOBs during CH 4 oxidation leads to a further 13 C depletion (Templeton et al. 2006). 13 C-depleted carbon transfers to consumers are thus easily recognizable and result in δ 13 C values highly distinct from CO 2atm transfers. This approach involving SIA to track these carbon transfers has been used successfully in pelagic (e.g., Daphnia, Lennon et al. 2006;Wooller et al. 2012;Rinta et al. 2016b;Morlock et al. 2017) andbenthic (e.g., chironomids, Frossard et al. 2014;Belle et al., 2016b) consumers.
Daphnia is a ubiquitous consumer whose isotopic signature can be monitored to assess the origin and transfer of carbon in pelagic food webs. Daphnia diet in lake is mainly composed of phytoplankton and detritus (Perga et al. 2008;Lampert and Kinne 2011;Hiltunen et al. 2019;Grosbois et al. 2020). Its δ 13 C values are therefore often considered to be dependent on phytoplankton δ 13 C (Matthews and Mazumder 2003;Lennon et al. 2006;van Hardenbroek et al. 2014), and reflects the fixation of the different forms of available dissolved inorganic carbon for photosynthesis (e.g., CO 2resp and/or CO 2atm ). However, some studies highlighted that a significant part of the carbon consumed by Daphnia, and zooplankton in general, may originate from methane (CH 4 ) via MOBs consumption (Bastviken et al. 2003;Kankaala et al. 2006b, Lennon et al. 2006Wooller et al. 2012;Rinta et al. 2016b;Morlock et al. 2017). The transfer of C-CH 4 in pelagic food webs seems nevertheless to be seasonal and depends on lakes thermal regime (Taipale et al. , 2008(Taipale et al. , 2009. The aim of this study was to assess the influence of thermal regime, oxygen conditions, land use and sediment characteristics on source and pathways of carbon transfer in lake pelagic food web with a particular attention paid on methane. For this purpose, we implemented a physico-chemical and biological survey on 11 lakes located in the French Jura, characterized by different land use, hypolimnetic hypoxia and thermal regime. Our objectives were to assess (1) the seasonal dynamics of dissolved gas in the water column ([O 2 ], [CO 2 ] and [CH 4 ]) from direct measurements, (2) the seasonal changes in the carbon sources transferred in pelagic food web using carbon isotopic signature of zooplankton (Daphnia), and (3) the main controlling factors of CH 4 storage and transfer in pelagic food webs.

Site description
The 11 studied lakes are located in Jura Mountains (eastern France) (Figure 1). Due to the calcareous nature of the bedrock in the area, the lake water is highly mineralised with high carbonate concentrations and high conductivity. The 11 lakes differ in elevation, geomorphological characteristics (water depth, lake area) and watershed surface and land use ( Table 1). The catchment area of these lakes is typically mainly composed of forest (mid-mountain beech-fir forest) and agricultural areas (mainly fertilized grassland for milk production) in variable proportions. All are medium to small size lakes with area ranging from 5 to 232 ha and maximum depth ranging from 7 to 30.5 m. The lakes studied are located in a temperate climate context, with annual air temperatures ranging from 6 to 11°C depending on the lake. Therefore, the period of ice cover is limited in time for these systems, ranging from a few days to a few weeks.

Sampling strategy
A continuous monitoring of oxygen concentrations and temperatures in the water column during a whole annual cycle was implemented in the 11 lakes by the deployment of a sensor mooring line at the point of maximal depth. Six to 15 temperature loggers (TidbiT MX400, HOBO, Bourne, USA) were installed depending on the depth of the lake. These thermistors were equally distributed on the sensor line from the lake surface to the lake bottom. Additionally, two or three O 2 sensors (Minidot, PME, Vista, USA) were positioned in the hypolimnion of the 11 lakes. Oxygen and temperature loggers performed automatic measurements every 4 hours. The deepest sensor is installed at 1 meter from the bottom. The other oxygen sensors were equally vertically distributed in the hypolimnion of the lakes. In addition, three field campaigns were carried out from September 2020 to March 2021, one at the end of summer stratification period (September 2020), and two after the autumnal turnover, in early (between the end of November 2020 and the beginning of January 2021 depending on lakes) and late winter (in March 2021). During each field campaign, vertical profiles of temperature and dissolved oxygen concentrations ([O 2 ]) were performed along the whole water column using a multiparameter probe (EXO1, YSI, Yellow Springs, USA) at the point of maximum depth. Membrane covered optical sensors (Mini CO 2 /CH 4, Pro-Oceanus, Bridgewater, Canada) were also used to record CO 2 and CH 4 partial pressure at 5 to 6 points along the water column of each lake. Deepest point (Zmax) measures were taken at 30 cm above sediments. Vertical profiles of dissolved methane ([CH 4 ]) and carbon dioxide ([CO 2 ]) concentrations were determined by using CO 2 and CH 4 solubility and taking into account water temperature and air pressure, according to Weiss (1974) and Wiesenburg and Guinasso (1979).
The lake transparency was measured for each field campaign using a Secchi disk, and the depth of the euphotic zone was determined as 2.5 x the Secchi depth. Two integrated zooplankton samples were made with a 200-µm mesh size plankton net: one in the euphotic zone, and one in the whole water column of the 11 lakes. Zooplankton samples were frozen after each field campaign and stored at −20°C.
Surface sediment samples (the first cm) were collected during the early winter campaign at the deepest point using a 90-mm-diameter gravity corer (UWITECH, Mondsee, Austria).

Laboratory analyses δ 13 C analysis of zooplankton and surface sediment
In laboratory, the zooplankton samples were thawed, and the Daphnia sp. were hand sorted using fine forceps under a binocular microscope without species differentiation. For each sample, a minimum of 20 Daphnia were picked up to obtain the minimum dry weight of 0.07 mg required for isotope analysis. All the samples were previously exposed to 3.7% HCl solution for carbonate removal, rinsed 3 times with demineralized water, and put into ultra-clean tin capsules. The samples were then dried at 60°C in an oven for 2 days. Surface sediment samples were also exposed to 3.7% HCl solution for carbonate removal, then dried at 60°C before being put into ultra-clean tin capsules for carbon isotope analysis.
The carbon stable isotope analyses were performed using an isotope ratio mass spectrometer interfaced with an elemental analyser (EA-IRMS) at INRA Nancy (Champenoux, France). The isotope ratios were expressed in the delta notation with Vienna Pee Dee Belemnite as the standard: δ 13 C (‰) = (Rsample/ Rstandard −1) x 1000, where R = 13 C/ 12 C. The replication of samples measured on an internal standard gave analytical errors of ±0.3‰ for δ 13 C measurements.

Nitrogen, organic carbon and pigment analyses in surface sediment samples
The surface sediment samples of the 11 lakes were dried and acidified with a solution of HCl (3.7%) to remove carbonate. Total organic carbon content and total nitrogen content were measured using a VARIO TOC CUBE analyser (Elementar) and a VARIO MAX CNS, respectively. The results are expressed as mass percentage of organic carbon (C org , %) and C org /N ratio.
Photosynthetic pigments were extracted overnight from fresh sediments with a solution of acetone and water (9:1). Supernatant was then removed and used to quantify the total carotenoids (TC) via spectrophotometry following the formula provided by Züllig (1989) and the recommendations of Guilizzoni et al. (2011). Carotenoid concentrations were expressed in micrograms per gram of sediment dry weight (µg/ g dw). TC concentrations can be used as a proxy of autotrophic organic matter sedimentation and may reflect primary production intensity (Guilizzoni et al. 1983;Adams et al. 1978).

Data analyses
The estimation of the watershed surface area used for agropastoral practices relative to lake surface area (relative agricultural area, ha/ha) was derived from Corine Land Cover data. Stratification characteristics were derived from the continuous temperature and oxygen concentration monitoring performed on the 11 lakes during a whole annual cycle (Supplementary material). The hypolimnetic hypoxia duration was determined by the number of days with [O 2 ] <2 mg/L at the relative depth of 80% of the maximum depth. For each lake, the strength of the thermal stratification was assessed by the mean temperature differences between surface and bottom layers during stratification period (Δ°C stratification period ). Summer stratification period was defined as the period when the temperature difference between surface and bottom layer exceed 2°C.
Due to the limited number of observations for each variable (n = 11 lakes), as well as heteroscedasticity and non-normality of the residuals, non-parametric Spearman's rank-order correlation test was used to assess the relationships between gas concentration profiles, carbon isotopic signature of Daphnia and lake properties. Spearman's correlation coefficient (ρ) as well as p-value were calculated. Differences in δ 13 C or gas concentrations between different sampling periods and sampling zones (photic zone vs whole water column) in the 11 studied lakes were determined using Wilcoxon signed-rank test.
All statistical analyses and figure creations were performed using R 3.5.1 statistical software (R Core Team, 2019). The significance level was set at α = 0.05 for all statistical tests.

Summer campaign
The studied lakes were strongly thermally stratified during the late summer campaign (Supplementary material), except for the two Etival lakes which present a very weak thermal stratification at this period ( Figure 2; Supplementary material). All stratified lakes had clinograde oxygen concentration profiles, with high [O 2 ] in the epilimnion ranging from 8.1 to 11.4 mg/L, and low [O 2 ] in the hypolimnion ranging from 0 to 0.87 mg/L. The Etival lakes had a less pronounced hypoxia in deep water than the other studied lakes with [O 2 ] around 4 mg/L. The CO 2 and CH 4 concentration profiles of the 9 stratified lakes were also quite similar between lakes, with increasing concentrations with depth ( Figure 2). In the epilimnion of stratified lakes, [CO 2 ] aq ranged from 13.4 to 31.6 µmol/L and [CH 4 ] aq from 3.5 to 8.7 µmol/L, except for the Lake Barterand which showed particularly high epilimnetic [CO 2 ] aq at 46.8 µmol/L and [CH 4 ] aq at 26.2 µmol/L. In the hypolimnion, [CH 4 ] aq ranged from 6.2 to 49.2 µmol/L whereas [CO 2 ] aq were comprised between 122.7 and 454.9 µmol/L depending on the lakes. The two shallow Etival lakes featured a relatively homogenous CO 2 and CH 4 concentrations profiles. In Grand and Petit Etival, [CO 2 ] aq varied from 20.4 to 21.1 µmol/L and from 16.8 to 41.12 µmol/L respectively, and [CH 4 ] aq ranged from 4.7 to 6.5 µmol/L and from 3 to 3.4 µmol/L respectively. When considering the whole dataset (i.e. including all lakes and depths), Spearman's rank correlation test indicates a significant positive relationship between [ (Figure 3).

Winter campaigns
In early and late winter, the gas concentration profiles were orthograde with relatively low variability along depth for the 11 lakes. [CH 4 ] aq ranged from 2.84 to 8.70 µmol/L in early winter and from 3.14 to 9.44 µmol/L in late winter, while [CO 2 ] aq ranged from 28.5 to 105.7 µmol/L in early winter and from 21.8 to 133.3 µmol/L in late winter for the 11 lakes ( Figure 3 (Figure 3). O 2 concentration profiles measured in winter were also orthograde, indicating a complete re-oxygenation of the water columns in the 11 lakes. [O 2 ] in the deep water were significantly higher during the early and late winter campaigns (from 3 to 12.4 mg/L) than during the summer period according to the Wilcoxon paired sample test ([O 2 ] early winter: p = 9.7e −4 ; [O 2 ] late winter: p = 9.7e −4 ).

δ 13 C analysis
Two Daphnia samples were collected for each studied lake and sampling campaign, one in the euphotic zone and one in the whole water column. According to Wilcoxon rank test for paired samples, there was no significant difference in Daphnia carbon isotopic signatures between the euphotic zone and the whole water column for the three sampling campaigns (stratification period: p = 0.83, early winter: p = 0.11, late winter: p = 0.27) (Figure 4). Therefore, δ 13 C Daphnia values are further discussed in the following sections are the average of euphotic and total column δ 13 C Daphnia . δ 13 C Daphnia showed a seasonal variability between summer and winter with a varying magnitude depending on the lake. The summer δ 13 C Daphnia ranged from −40.9 to −33.9 ‰ for the 11 studied lakes (Table 2). During the winter sampling campaigns, δ 13 C Daphnia range from −54.9 to −37 ‰ and were thus lower than in summer for all lakes except for Petit and Grand Etival (Table 2; Figure 5). According to a Wilcoxon rank test for paired samples, early and late winter δ 13 C Daphnia were significantly lower than δ 13 C Daphnia during the stratification period (early winter: p = 0.004, late winter: p = 0.009, Figure 4). Furthermore, δ 13 C Daphnia during late winter were significantly higher than δ 13 C Daphnia during early winter (p = 0.006) (Figure 4).
There was a positive correlation of δ 13 C Daphnia between summer and winter (spearman rank correlation test, rho = 0.7, p = 0.04), when excluding the two Etival lakes characterized by quite similar δ 13 C Daphnia between the sampling seasons ( Figure 5). Nevertheless, the differences in δ 13 C Daphnia between summer and early winter (Seasonal Δδ 13 C Daphnia ) or between summer and mean winter (Mean seasonal Δδ 13 C Daphnia ) were variable among the studied lakes (Table 2). For the nine stratified lakes, the seasonal Δδ 13 C Daphnia ranged from 3.23 to 18.58 ‰, while mean seasonal Δδ 13 C Daphnia ranged from 2.23 to 10.85 ‰ (Table 2). In comparison, the Etival lakes showed smaller differences in δ 13 C Daphnia between summer and winter (Seasonal Δδ 13 C Daphnia = −0.1 and −1.3; Mean seasonal Δδ 13 C Daphnia = −0.5 and −1.6).

Effects of lake and watershed properties on dissolved gas concentrations
Three types of explanatory variables were considered in order to assess the influence of lake and watershed properties on dissolved gas concentrations in the water column, especially gas storage in the hypolimnion during summer stratification: (1) sediment organic matter features, (2) thermal stratification and oxygen conditions and (3) land use on the watershed (Table 3). The results of Spearman rank correlation test performed on these datasets are detailed in supplementary material. Total carotenoids (TC) was strongly correlated to gas concentrations in the deep water ( Figure 6) Figure 7). There were no significant relationships between dissolved gas concentrations in deep water and hypoxia duration or the proxy used for the thermal stratification intensity.

Relationship between dissolved gas concentrations and δ 13 C Daphnia
During the summer stratification, no significant relationship could be evidenced between mean δ 13 C Daphnia and [CH 4 ] aq or [CO 2 ] aq in epilimnion or in deep water. Nevertheless, there was a significant negative relationship between epilimnetic [CO 2 ] aq and the δ 13 C of Daphnia sampled only in euphotic zone (p = 0.02, rho = −0.69) (Figure 8). The two Etival lakes, which were only weakly thermally stratified during summer period (Figure 2), did not follow this relationship. Indeed, in these lakes, the highly negative summer δ 13 C Daphnia corresponded to relatively low epilimnetic [CO 2 ] aq (Figure 8).
For both winter campaigns, no relationship was evidenced between δ 13 C Daphnia and gas concentrations in the water columns in the 11 lakes. Furthermore, there was not any significant relationship between the amount of CH 4 or CO 2 stored in deep water during summer period and the δ 13 C of Daphnia of the two winter campaigns. However, the decrease in the δ 13 C Daphnia between summer and winter campaigns (seasonal Δδ 13 C Daphnia ) seemed proportional to the amount of gas stored in the hypolimnion during summer stratification. Indeed, according to Spearman correlation rank test, summer hypolimnetic [CH 4 ] aq was positively correlated with seasonal Δδ 13 C Daphnia (p = 0.03, rho = 0.66) and mean seasonal Δδ 13 C Daphnia (p = 0.008, rho = 0.77) (Figure 9a). Summer hypolimnetic [CO 2 ] aq was also Table 3. Sediment, stratification and watershed characteristics of the 11 studied lakes. C org is expressed in %, δ 13 C OM is expressed in ‰ and TC (total carotenoids) is expressed in µg/g of dry sediment. Hypoxia duration = duration of [O2]<2 mg/l at 0.8*maximal depth in days, and Δ °C stratification period = mean temperature difference in °C between the surface and bottom layers during the summer stratification period. Relative agricultural area and relative forest area represent the watershed area used for forest and agricultural activities relative to lake area (in ha), according to CORINE Land Cover data. positively correlated with seasonal Δδ 13 C Daphnia (p = 0.008, rho = 0.77) and mean seasonal Δδ 13 C Daphnia (p = 0.002, rho = 0.83) (Figure 9b).

Relationships between δ 13 C Daphnia , sediment, watershed and stratification characteristics
The mean Δ°C stratification period was significantly positively correlated with seasonal and mean seasonal Δδ 13 C Daphnia (p = 0.003 rho = 0.81, p = 0.005 rho = 0.8, respectively, Figure 10a). The lakes with the strongest thermal stratification in summer thus showed the highest decrease in δ 13 C Daphnia between summer and winter period. According to the Spearman rank test, there was no significant relationship between mean Δ°C stratification period and summer δ 13 C Daphnia , early winter δ 13 C Daphnia , or late winter δ 13 C Daphnia .
Although no significant relationship could be evidenced between the hypoxia duration in the 11 lakes and the summer or winter δ 13 C Daphnia , there was a significant relationship between the hypoxia duration and seasonal Δδ 13 C Daphnia (p = 0.007, rho = 0.78) and mean seasonal Δδ 13 C Daphnia (p = 0.01, rho = 0.70) according to Spearman correlation tests (Figure 10b). A significant positive relationship was observed between pigment concentrations (TC) and seasonal Δδ 13 C Daphnia (spearman rank test p = 0.01, rho = 0.74) or mean seasonal Δδ 13 C Daphnia (p = 0.001, rho = 0.86) ( Figure 11). No significant relationship was found between % C org , C org /N or δ 13 C OM and seasonal Δδ 13 C Daphnia or δ 13 C Daphnia for all sampling periods.
Finally, the relative agricultural area was negatively correlated to winter δ 13 C Daphnia (relative agricultural area ~ early winter δ 13 C Daphnia : p = 0.03, rho = −0.64; ~ late winter δ 13 C Daphnia : p = 0.02, rho = −0.7; ~ mean winter δ 13 C Daphnia : p = 0.02, rho = −0.69). The relative agricultural area was also significantly positively correlated with the seasonal Δδ 13 C Daphnia (p = 0.04 rho = 0.62) and the mean seasonal Δδ 13 C Daphnia (p = 0.02 rho = 0.69). Moreover, relative agricultural area was significantly positively correlated with TC (p = 0.01, rho = 0.74). Lakes with the largest area of agricultural parcels relative to their surface area and the highest pigment concentrations in sediments showed the lowest winter δ 13 C Daphnia and the largest seasonal Δδ 13 C Daphnia. No  relationship was evidenced between the relative agricultural area and the summer δ 13 C Daphnia according to a Spearman correlation rank test.

Seasonal dynamics of dissolved gas concentrations derived from heterotrophic processes
Similar summer clinograde profiles of gas concentrations were observed for the stratified lakes of our study. They were all characterised by an increase in [CH 4 ] aq and [CO 2 ] aq in the hypolimnion which was comparable to that reported in many studies (Rudd et al. 1976;Bastviken et al. 2003;Kortelainen et al. 2006;Miettinen et al. 2015;Rinta et al. 2015;Laas et al. 2016;Thalasso et al. 2020). In this water layer, gas concentrations ranged from 6.2 to 49.2 µmol/L for CH 4 and from 122.7 to 454.9 µmol/L for CO 2 , depending on the lake. These values, comparable to those measured in several other lakes (Bastviken et al. 2003;Kortelainen et al. 2006;Rinta et al. 2015;Thalasso et al. 2020), are highly supersaturated according to the dissolved concentration at air-water equilibrium. Indeed, air water equilibrium concentration range between 0.002 to 0.004 µmol/L for CH 4  and 13 to 17 µmol/L for CO 2 depending on water temperature (according to solubility equations by Weiss (1974) and Wiesenburg and Guinasso (1979) and by considering atmospheric concentrations of 407 ppm of CO 2 and 1850 ppb of CH 4 (www.eea.europa.eu)). The strong summer hypolimnetic anoxia and the negative relationship between O 2 and CO 2 concentrations observed during the stratification period indicated sustained heterotrophic activities (Rantakari and Kortelainen 2005;Laas et al. 2016). The high concentration of CH 4 found in the hypolimnion of our lakes at the same time can be explain by the degradation of organic matter carried out by methanogenic Archea under such anoxic conditions (Rudd and Hamilton 1975;Rudd et al. 1976;Utsumi et al. 1998).
In the 9 stratified lakes, although lower than in hypolimnion, epilimnetic [CO 2 ] aq were close or higher than air-water equilibrium. These results are in accordance with previous studies that evidenced sursaturation of CO 2 in the surface layer of stratified lakes (Laas et al. 2016;Pighini et al. 2018), reflecting high heterotrophic activities in the epilimnion during summer stratification. However, our results about epilimnetic CO 2 concentrations should be considered with caution, as they result from a balance between respiration and CO 2 fixation by primary producers, and are consequently possibly subject to large temporal variability ). Alternatively, several previous studies suggested that, in calcareous lakes analogues to our study sites, CO 2 concentration may be rather linked to the production of CO 2 by calcite precipitation in alkalinitrophic context (Marcé et al. 2015;Khan et al. 2020). CO 2 supersaturation in our study may then be explained by these processes, the Jura lakes being known to have a significant calcic precipitation by littoral macrophytes (Magny 1992).
These epilimnetic CO 2 concentrations could be sustained by some inputs from respiration in the sedimentary compartment, as shown by Kortelainen et al. (2006). Epilimnetic [CH 4 ] aq were always highly supersaturated for all the studied lakes, especially for Lake Barterand which had considerable surface [CH 4 ] aq in summer. Similar epilimnetic CH 4 supersaturation has been evidenced in many lakes during thermal stratification (Rudd and Hamilton 1975;Rudd et al. 1976;Pighini et al. 2018). This may be the result of vertical diffusive gas transport from bottom sediment to the surface water. Nevertheless, this process seems limited in thermally stratified lakes (Bastviken et al. 2008). Other mechanisms may also explain the presence of CH 4 in surface layers. Some studies suggest indeed that methane-rich air bubbles rising from the lake bottom may contribute to the supersaturation of surface waters (McGinnis et al. 2006). The methane supersaturation in oxic conditions may also be the result of CH 4 production by phototroph organisms together with oxygen tolerant methanogens (Bogard et al. 2014;Tang et al. 2014;Wang et al. 2017). Finally, the horizontal transfers of CH 4 produced in the littoral zones of the lakes could explain these epilimnetic concentrations .
In winter, CO 2 and CH 4 concentrations decreased significantly in the hypolimnion compared to summer measurements, and concentrations profiles became orthograde with low variability along water columns. A large part of methane produced in sediment and accumulated in the hypolimnion during summer stratification was released in the whole water column with autumnal turnover (Rudd and Hamilton 1975;Utsumi et al. 1998). During the turnover and throughout winter, favourable oxygen conditions promote oxidation by MOB (methane oxidizing bacteria) of a large part of the released CH 4 (Rudd and Hamilton 1975;Rudd and Taylor 1980;Utsumi et al. 1998;Kankaala et al. 2007).
In contrast to the stratified lakes of our study, the two weakly stratified Etival lakes showed homogeneous and similar gas concentration profiles in summer and in winter. Both lakes of Etival are small polymictic lakes. Their water columns were thus permanently oxygenated and no gas was stored in deep water layer. CH 4 produced in sedimentary compartment of these lakes was then directly available for emission or oxidation by MOBs and the CO 2 produced can be directly used for photosynthesis (Laas et al. 2016) or released to the atmosphere.

Seasonal dynamics of carbon transfers to pelagic food web
During summer stratification, the δ 13 C Daphnia values ranged from −33.8‰ to −40‰ in the stratified lakes. Daphnia diet is usually mainly composed of phytoplankton and terrestrial detritus during summer (Perga et al. 2008;Lampert and Kinne 2011). Nevertheless, in our study, allochthonous detritus did not appear to contribute significantly to Daphnia diet. Indeed, a study about the δ 13 C of the vegetation in the buffer zones of 4 Jura mountains lakes (50 m all around the lake shore) reported values from −24.2‰ (for Phragmitia) to −30,3‰ (for grassland) (Borderelle 2006). The δ 13 C of Daphnia during summer stratification are however in the range of the carbon isotopic signature usually found for phytoplankton (−25‰ to −40‰, Peterson and Fry 1987;France 1995;Vuorio et al. 2006;Masclaux et al. 2013;Wang et al. 2013;de Kluijver et al. 2014).
In the 9 stratified lakes of our study, we evidenced a negative correlation between δ 13 C of Daphnia from the euphotic zone and the epilimnetic CO 2 concentrations. Some previous studies showed analogous relationships between phytoplankton δ 13 C (and by extension zooplankton δ 13 C) and [CO 2 ] aq in lakes (Mohamed and Taylor 2009;Smyntek et al. 2012;de Kluijver et al. 2014). The phytoplankton δ 13 C is determined by a combination of (i) the degree of isotopic fractionation performed during photosynthesis and (ii) the δ 13 C value of the dissolved inorganic carbon used (de Kluijver et al. 2014). The isotope fractionation by phytoplankton increases with increasing dissolved CO 2 concentration. Indeed, the increase in dissolved inorganic carbon availability leads to an increase in 13 C discrimination, and therefore a 13 C depletion of phytoplankton (Peterson and Fry 1987;Laws et al. 1995;Smyntek et al. 2012;de Kluijver et al. 2014). In our study, high CO 2 concentrations in the epilimnion may be due to intense respiration processes (de Kluijver et al. 2014). Respiration-derived CO 2 exhibits 13 C-depleted signatures compared to other dissolved carbon sources available for photosynthesis (Fry 2006;Parker et al. 2010). As δ 13 C Daphnia were in the lower range of phytoplankton δ 13 C according to literature (Peterson and Fry 1987;France 1995;Vuorio et al. 2006;Masclaux et al. 2013;Wang et al. 2013) and according to the relationship between epilimnetic [CO 2 ] aq and δ 13 C Daphnia , the δ 13 C Daphnia could reflect the intensity of respired carbon transfers to phytoplankton and then to primary consumers. Furthermore, other DIC sources can also sustained the 13 C depletion of phytoplankton, such as the CO 2 derived from the oxidation of CH 4 (Wachniew and Różański 1997). Finally, even if the Daphnia diet seems to be mainly based on phytoplankton, other sources may contribute to the observed 13 C-depletion of their signatures, such as MOBs (Templeton et al. 2006), considering the availability of CH 4 in epilimnion during summer stratification.
In winter, the δ 13 C Daphnia values were significantly lower than in summer in the stratified lakes of our study, with values ranging from −54.9 to −37 ‰. Moreover, these winter δ 13 C Daphnia were substantially lower than the δ 13 C of allochthonous detritus and phytoplankton usually reported in the literature for most of the lakes (−25‰ to −40‰, Peterson and Fry 1987;O'Leary 1988;France 1995;Grey et al. 2001;Vuorio et al. 2006;Masclaux et al. 2013;Wang et al. 2013;de Kluijver et al. 2014). The transfer of methanogenic carbon (C-CH 4 ) in food web via MOBs consumption most likely explains these low carbon isotopic signatures of pelagic consumers. Indeed, C-CH 4 is highly 13 C-depleted, with δ 13 C values ranging between −40 and −80 ‰ in European calcareous hypoxic lakes (Rinta et al. 2015). Moreover, the δ 13 C values of MOBs are further reduced due to isotopic fractionation during methanotrophy, and range between −90‰ and −60‰, making them highly distinct from the carbon isotopic signature of phytoplankton and allochthonous detritus (Jedrysek 2005;Templeton et al. 2006). Similar highly 13 C-depleted signatures of pelagic consumers have already been reported in several stratified lakes and indicated the transfers o C-CH 4 to pelagic food webs in winter (Harrod and Grey 2006;Taipale et al. 2008;Morlock et al. 2017). In such lakes, CH 4 accumulated in the hypolimnion during the summer stratification. During the autumnal turnover, the methane accumulated in the hypolimnion is released into the entire water column, which is simultaneously reoxygenated (Utsumi et al. 1998). When CH 4 become available in oxic conditions, it can be oxidized by MOBs (Rudd and Hamilton 1978;Bastviken et al. 2002;Kankaala et al. 2006a). The release of methane thus results in a high production of MOBs after the autumnal turn-over (Utsumi et al. 1998;Kankaala et al. 2007;Schubert et al. 2012), which then become an abundant carbon source for Daphnia (Taipale et al. 2008(Taipale et al. , 2009. For the studied lakes, this hypothesis is further supported by the significant negative relationship observed between bottom CH 4 concentration measured in summer and the δ 13 C Daphnia decrease between the summer and winter campaigns. This relationship suggests that the intensity of C-CH 4 transfer in pelagic food webs after the autumnal turnover is proportional to the amount of methane stored in the hypolimnion during summer stratification. This hypothesis agrees with literature that indicates that a large part of the methane is oxidised by the MOBs after autumnal mixing regardless of the magnitude of the summer storage in the hypolimnion (Kankaala et al. 2006a. Zooplankton δ 13 C can also decrease in winter due to the increased storage of lipid in their tissues (Syväranta and Rautio 2010). Indeed, lipids are known to be depleted in 13 C, leading to lower δ 13 C values for tissues with a higher lipid content (DeNiro and Epstein 1977).
However, winter lipid storage by Daphnia appears to be very limited compared to other zooplanktonic taxa, especially copepods (Grosbois et al. 2017). Furthermore, these lipid storage mechanisms can lead to maximum variations in cladoceran signatures of only 3 to 4.5 ‰ (Smyntek et al. 2007;Syväranta and Rautio 2010). The seasonal variations of δ 13 C Daphnia observed in our study are larger than these values for the majority of the stratified lakes, thus suggesting that these variations are mainly due to changes in the carbon sources consumed by the daphnids rather than winter storage of lipid.
Both lakes of Etival, which were weakly stratified during summer, had the most negative summer δ 13 C Daphnia of the 11 studied lakes with δ 13 C values slightly below −40‰. These signatures are lower than those reported for phytoplankton (Peterson and Fry 1987;France 1995;Vuorio et al. 2006;Masclaux et al. 2013;Wang et al. 2013), and could be explained by the use of a substantial amount of C-CH 4 by Daphnia during summer. Indeed, in unstratified or polymictic lakes, CH 4 is available all year round under oxic conditions and can thus be transferred to food webs at all seasons (Laas et al. 2016;Vachon et al. 2019). Due to the weak thermal stratification, and the absence of methane storage in deep water layer in summer, no significant difference in δ 13 C Daphnia was then observed between summer and winter for these two lakes. Moreover, the lowest values were observed in summer rather than in winter for these two lakes (even if these differences are not significant). Despite winter lipid storage, the δ 13 C Daphnia probably still increased in winter due to less methanogenic activities under lower water temperature.

Influence of thermal stratification and hypoxia duration
Unexpectedly, our results did not reveal a clear impact of thermal stratification and hypoxia duration on gas concentrations in the hypolimnion during summer stratification. Indeed, lakes with strong thermal stratification and prolonged hypolimnetic hypoxia are usually reported to store large quantity of CH 4 in summer due to the high production rates of CH 4 and CO 2 and the limitation of vertical gas exchanges between the hypolimnion and the epilimnion (Juutinen et al. 2009;Rinta et al. 2016a;Vachon et al. 2019). These unexpected results may be partly linked to our study design which involved only point-in-time physico-chemical measurements. The gas concentration measurements may correspond to different moments of the thermal cycle, which is specific to each studied lake. These differences may explain the lack of significant relationship between the gas concentrations in deep water and the thermal stratification strength or the hypolimnetic anoxia duration. Nevertheless, these factors were significantly correlated with the seasonal decrease in δ 13 C of pelagic consumers (Daphnia). Indeed, δ 13 C signature of organisms are probably more time integrative (Tieszen et al. 1983) and are therefore less affected by the one-off nature of the sampling. Our results suggest therefore that the thermal stratification strength (Δ°C stratification period ), as well as hypoxia duration, at least partly explained the amplitude of δ 13 C Daphnia decrease between summer and winter. These relationships suggest that lakes with stronger stratification and prolonged hypolimnetic hypoxia are likely to store a greater quantity of CH 4 in summer and are characterized by an enhanced transfer of C-CH 4 to the pelagic food web in winter. For the lakes with weaker stratification, a greater proportion of the methane produced in the hypolimnion may be released in summer to pelagic consumers, resulting in a smaller δ 13 C Daphnia decrease between summer and winter.
Finally, our study shows that the storage of the methanogenic carbon and its transfer dynamic in pelagic food webs are highly dependent on the physical conditions prevailing in lakes. This finding implies that the methane cycle in particular, and the carbon cycle more generally, may be impacted by global warming. Indeed, the duration and strength of lake stratification will increase with global warming (Butcher et al. 2015;Winslow et al. 2015;Pilla and Williamson 2021). This will be accompanied by a decrease in hypolimnetic oxygen and an increase in anoxia/hypoxia duration (Fang and Stefan 2009;Foley et al. 2012) which could lead to an increase in CH 4 production during the summer stratification (Rudd and Hamilton 1978). As evidenced by our results, a part of this CH 4 could finally be transferred to the pelagic food web after the autumnal turnover. In addition, many temperate lakes are expected to show a shift of their thermal regime under climate change, from dimictic to either monomictic, polymictic and even permanently stratified in some cases (Woolway and Merchant 2019). These shifts would then induce strong changes in the methanogenic carbon seasonal transfer in pelagic food webs.

Influence of land-use
In the 11 lakes involved in the present study, the relative area of the watershed used for agriculture is positively correlated with the amount of CO 2 and CH 4 stored in the hypolimnion during the summer stratification. The influence of agricultural land-use on the CH 4 fluxes to the atmosphere was already suggested by Huttunen et al. (2003). Moreover, Kortelainen et al. (2006) also evidenced relationships between agricultural catchment area and increase in concentrations in CO 2 derived from heterotrophic activities in lake water columns, but to our knowledge, our study is the first to report a significant relationship between agricultural land on watershed and hypolimnetic CH 4 concentrations. Furthermore, we observed a positive relationship between agricultural catchment area and TC (proxy of primary production), as well as between TC and hypolimnetic CH 4 concentrations. These results are consistent with other studies showing that methane production is positively linked to primary productivity variables such as nutrients or chlorophyll a concentration (Kortelainen et al. 2000;Beaulieu et al. 2019). Indeed, agriculture is already identified as a major disturbance factor for the ecological status and functioning of lakes (Carpenter et al. 1998;Knoll et al. 2003), leading to their accelerated eutrophication. The increase in primary production induced by water eutrophication and thus in sedimentation rates of easily degradable organic matter induces oxygen depletion in the hypolimnion (Foley et al. 2012) and promote methanogenesis (Beaulieu et al. 2019). In addition, several studies report that CH 4 fluxes to the atmosphere are greater in lakes where sedimentary organic matter is mainly autochthonous (Huttunen et al. 2003;West et al. 2012). Moreover, a strong organic sedimentation rate combined to low C/N ratio of lake sediments are particularly suitable for methanogenesis (Duc et al. 2010;Praetzel et al. 2020). Unexpectedly, in our dataset, we did not observe any relationship between C org or C/N ratio in sediment organic matter and hypolimnetic gas concentrations in this study. A possible explanation is that we only measured the proportion of organic carbon in sediments and not considered the rate of organic sedimentation, which would be a more direct measurement of the amount of organic matter being sedimented. Furthermore, other factors such as temperature, O 2 concentration in the sediments, or competition with other bacteria for substrate acquisition can influence the methanogenic activity in sediments (Borrel et al. 2011).
Finally, in this study, relative area of catchment used for agriculture was clearly correlated with the winter δ 13 C Daphnia and seasonal Δδ 13 C Daphnia . TC in sediments was also clearly correlated with seasonal Δδ 13 C Daphnia . These results suggest that agricultural activities in the watershed and primary production of a lake enhance C-CH 4 transfers during winter in the pelagic compartment. Indeed, lakes with large agricultural catchments relative to lake size and intense primary production tend to accumulate more methane in the hypolimnion during the summer period according to our data. These lakes then release more methane during autumnal turnover, resulting in sustained transfer of C-CH 4 to the pelagic food web. Belle et al., (2016a) also suggested that agropastoral practices intensity on catchment may partly drive methane transfers to primary consumers, but to our knowledge, this study is the first to show a direct relationship between catchment occupation, primary production, and transfer of 13 C-depleted carbon sources to lakes pelagic food webs.

Conclusion
The nine stratified lakes of this study, in contrast to the two weakly stratified lakes, showed marked seasonal dynamics of dissolved gas concentrations. These seasonal dynamics were characterized by an accumulation of CO 2 and CH 4 coupled with a decrease in oxygenation of the hypolimnion during stratification, and a reoxygenation of the deep layers associated with a decrease in CO 2 and CH 4 concentrations after autumnal turnover. Daphnia δ 13 C showed strong seasonal variations in the lakes studied, except for the two weakly stratified lakes during summer, marked by a significant decrease of δ 13 C values between summer and winter. These seasonal dynamics suggests the transfer of a non-negligible amount of C-CH 4 into the pelagic food webs after autumnal turnover. According to our results, the intensity of these winters C-CH 4 transfers largely depends on the amount of CH 4 stored in the hypolimnion during the stratification period. These C-CH 4 storage and transfer mechanisms seem partly dependent on the intensity of thermal stratification and the duration of hypoxia periods, but also on external forcing factors such as the land use on watersheds that influence origin and quantity of sedimented organic matter promoting methanogenesis. The ongoing global changes, characterized by global warming and accelerated eutrophication of waters linked to intensification of anthropogenic activities on the watershed, may therefore have a severe impact on the origin and pathways of carbon to pelagic food webs through their influence on thermal regimes and trophic functioning of lakes. More broadly, although some studies suggest that an increase in lake primary production could trap a significant amount of carbon in the sediments (Anderson et al. 2014), our results are consistent with other studies showing on the contrary, that increase in primary production could be at the origin of a greater production and release of greenhouse gases into the atmosphere (Beaulieu et al. 2019). Further studies need to quantify the relative importance of CO 2 and CH 4 production and emission relative to their incorporation in food webs to improve our understanding of carbon cycle in lakes in the context of global changes.