Using isotopes to understand landscape‐scale connectivity in a groundwater‐dominated, lowland catchment under drought conditions

The Demnitzer Millcreek catchment (DMC), is a 66 km2 long‐term experimental catchment located 50 km SE of Berlin. Monitoring over the past 30 years has focused on hydrological and biogeochemical changes associated with de‐intensification of farming and riparian restoration in the low‐lying landscape dominated by rain‐fed farming and forestry. However, the hydrological function of the catchment, which is closely linked to nutrient fluxes and highly sensitive to climatic variability, is still poorly understood. In the last 3 years, a prolonged drought period with below‐average rainfall and above‐average temperatures has resulted in marked hydrological change. This caused low soil moisture storage in the growing season, agricultural yield losses, reduced groundwater recharge, and intermittent streamflows in parts of an increasingly disconnected channel network. This paper focuses on a two‐year long isotope study that sought to understand how different parts of the catchment affect ecohydrological partitioning, hydrological connectivity and streamflow generation during drought conditions. The work has shown the critical importance of groundwater storage in sustaining flows, basic in‐stream ecosystem services and the dominant influence of vegetation on groundwater recharge. Recharge was much lower and occurred during a shorter window of time in winter under forests compared to grasslands. Conversely, groundwater recharge was locally enhanced by the restoration of riparian wetlands and storage‐dependent water losses from the stream to the subsurface. The isotopic variability displayed complex emerging spatio‐temporal patterns of stream connectivity and flow duration during droughts that may have implications for in‐stream solute transport and future ecohydrological interactions between landscapes and riverscapes. Given climate projections for drier and warmer summers, reduced and increasingly intermittent streamflows are very likely not just in the study region, but in similar lowland areas across Europe. An integrated land and water management strategy will be essential to sustaining catchment ecosystem services in such catchment systems in future.

stand how different parts of the catchment affect ecohydrological partitioning, hydrological connectivity and streamflow generation during drought conditions. The work has shown the critical importance of groundwater storage in sustaining flows, basic in-stream ecosystem services and the dominant influence of vegetation on groundwater recharge. Recharge was much lower and occurred during a shorter window of time in winter under forests compared to grasslands. Conversely, groundwater recharge was locally enhanced by the restoration of riparian wetlands and storage-dependent water losses from the stream to the subsurface.
The isotopic variability displayed complex emerging spatio-temporal patterns of stream connectivity and flow duration during droughts that may have implications for in-stream solute transport and future ecohydrological interactions between landscapes and riverscapes. Given climate projections for drier and warmer summers, reduced and increasingly intermittent streamflows are very likely not just in the study region, but in similar lowland areas across Europe. An integrated land 1 | INTRODUCTION Global climate change and population growth are increasing pressure on agricultural landscapes, threatening food and water security in many lowland catchments. This underlines the need for experimental observatories in such environments (Turral et al., 2011;Tetzlaff et al., 2017). An example for such pressures was the recent European drought of 2018 (continuing into 2020) where a prolonged period of warm and dry weather severely affected water availability over extensive areas, with a significant number of headwater streams ceasing to flow (Buras et al., 2020;Toreti et al., 2019). Water resources in the extensive, glacially formed, lowland landscape of the North German Plain (NGP) sustain food production (Barkmann et al., 2017;Gutzler et al., 2015) and water supplies to large cities like Berlin. In lowland catchments in the NGP, streams are often dominated by groundwater but it is still unclear how such catchments function hydrologically in both space (e.g., within a catchment) and time (Boulton & Hancock, 2006). Often, lowlands catchments are understudied in favour of landscapes with stronger topographic controls on drainage of surface and subsurface water (Devito et al., 2005). Thus, there is a weak evidence base for understanding how drought affects recharge and streamflows in such lowland areas, including the seasonal cessation of discharge (Germer et al., 2011). This is alarming, given climate change scenarios for the region which predict significantly drier and warmer summers (Mirshel et al., 2020) with a precipitation shift towards winter (Cubasch & Kadow, 2011).
To address this research gap, an ecohydrological study was initiated in the 66 km 2 Demnitzer Millcreek catchment (DMC), in the State of Brandenburg near Berlin (Figure 1) in 2018. This area is part of the NGP. A challenge of working in heavily managed agricultural catchments like DMC is assessing how a long legacy of constantly changing anthropogenic activities affects hydrological function, and how this is affected by non-stationary climatic conditions. In lowland catchments with extensive aquifers, understanding groundwater recharge and groundwater-surface water exchanges is a major research need (Kløve et al., 2014). Residual "blue" water fluxes, sustaining recharge and streamflow, are highly dependent on ecohydrological partitioning and the "green" water fluxes that return moisture back into the atmosphere. It is increasingly recognized how profoundly and subtly catchment hydrology is influenced by land use and vegetation cover . Elucidating the subsequent linkages between ecohydrological partitioning and hydrological connectivity at the catchment scale is crucial to understand groundwater recharge and runoff generation, as well as informing land and water management in lowland catchments.
The impact of droughts over the last decade have also highlighted the importance of intermittent streamsthat is, streams which temporarily cease to flow at some point in time and space (Acuña et al., 2014). Such systems comprise half of the global river network  and their extent and distribution are increasing in many areas as a result of increasing human water withdrawal and climate change (Shumilova et al., 2019). A characteristic feature of the 2018 drought in the DMC (and similar lowland headwater streams) was the prolonged cessation of streamflow until well into the following winter (Smith, Tetzlaff, Gelbrecht, et al., 2020). Thus, the dynamic, seasonal interplay of temporary streamflow and bidirectional streamgroundwater interactions in headwater streams like the DMC highlights the importance of focal areas for groundwater recharge from streams (Zimmer & McGlynn, 2017a). Despite this, we still poorly understand how, why and where streams like the DMC expand, contract and interact with adjacent aquifers (Zimmer & McGlynn, 2017b).
Any assessment of climate impacts on flow performance of groundwater dependent stream networks in anthropogenically influenced lowlands demand thorough understanding of the local hydrological system (van Engelenburg et al., 2018) and the structuring of the landscape (Bertrand et al., 2012) which can be highly place specific (Ward et al., 2020).
Gaining landscape scale understanding of groundwater-surface water interactions and hydrological connectivity requires the use of extensive instrumentation and integrating techniques that go beyond point scale hydrometric measurements. The factors for regional timing and spatial variability of surface water connectivity dynamics and groundwater-surface water interactions are not well constrained in most lowland areas (Lewandowski et al., 2009). In this regard, isotope ratios of oxygen and hydrogen have proven to be effective tools to trace fluxes in the terrestrial water cycle in an integrated way at large scales (Kendall & Mcdonnell, 2012;Penna et al., 2018). Naturally occurring isotopic variation in precipitation can be used as a basis for tracing water in the "critical zone" (the thin dynamic "life zone" of the terrestrial Earth, extending from the top of the vegetation canopy through the soils and down to fresh bedrock and the bottom of groundwater, Grant & Dietrich, 2017). Previous studies have shown how modification of the isotope input signal through dynamic mixing and fractionation provide additional insight into how water that infiltrates into soils, is subsequently evaporated or percolates deeper to recharge groundwater (e.g., . In F I G U R E 1 Maps of elevation, locations of measurements and sampling sites (a), geology (b), soils (c), land use (d), and detailed sampling locations (e-g) of the Demnitzer Millcreek catchment (DMC) addition to such qualitative insights, tracers can provide a more quantitative assessment of the travel time between water entering the system as precipitation ("water age" = 0) and its exit fluxes from various stores (e.g., soil and groundwater compartments) or the entire catchment as streamflow. Moreover, the temporal and spatial dynamics in water ages in various critical zone compartments (Sprenger et al., 2019) help to understand how mixing involves different catchment storage characteristics (Soulsby et al., 2009), as well as illuminating the ecohydrological functioning of the landscape and resilience of the blue and green water fluxes (Kuppel et al., 2020).
The DMC, a tributary of the River Spree, is a long-term research catchment established in 1990 initially to understand agricultural influences on nutrient dynamics (Gelbrecht et al., 2005). Most of the catchment supports rain-fed agricultural systems, which face serious challenges in maintaining food production and other ecosystem services under likely climate change scenarios. Thus, recent years have seen a reorientation of research efforts to better understand the catchment's water balance and hydrological function. For example, the impacts of changes in wetland management and beaver recolonization have been assessed (Smith, Tetzlaff, Gelbrecht, et al., 2020). Investigations of the impact of the drought 2018 on two contrasting soil-vegetation plot-scale units revealed land use dependent differences in the isotopic dynamics and age of soil water at the plot scale suggesting a higher drought resilience of forests . Further, ecohydrological modelling presented differences in age dynamics of soil water and groundwater recharge between contrasting soil-vegetation assemblages (Smith et al., 2020b).
Here, we report an investigation that sought to up-scale insights from these initial plot-scale studies Smith et al., 2020b), through catchment-scale isotope monitoring of precipitation, soil water, groundwater and stream water over a >2-year period. This complemented and leveraged data from an existing, longer term hydrometric infrastructure. The study aimed to use isotopes to better understand the ecohydrological linkages between precipitation and spatial patterns of soil moisture, groundwater levels and streamflow generation at the catchment scale.
The specific objectives of the study were, to: 1. Characterize the spatio-temporal stable isotope dynamics in precipitation, soil water, groundwater and streamflow at the catchment scale.
2. Assess how different catchment characteristics (particularly land use) affect water partitioning, connectivity and the isotopic composition of soil, ground and stream water.
3. Understand how groundwater-surface water interactions and inchannel processes affect the isotopic dynamics of stream water, and the implications for water age estimates.
The wider implications of an improved understanding of vegetation and catchment responses to climate change (Babst et al., 2019) are also examined, as well catchment management for improving the resilience of ecosystem services in similar lowland landscapes is discussed.

| STUDY SITE
The study was conducted at the Demnitzer Millcreek catchment (DMC; 66 km 2 ) in NE Germany, which is a long-term experimental headwater site for hydrological and biogeochemical research at the Leibniz-Institute of Freshwater Ecology and Inland Fisheries (IGB, Berlin, Germany). DMC is representative of the lowland and mixed land use landscape of the NGP (Smith, Tetzlaff, Gelbrecht, et al., 2020). The climate is characterized as mid-continental with annual precipitation (569 mm a À1 ) being exceeded by annual potential evapotranspiration of 650-700 mm a À1 (Smith, Tetzlaff, Gelbrecht, et al., 2020). Precipitation occurs throughout the year with slightly more during summer when infrequent heavy convectional storms dominate. More frequent but lower intensity frontal rain prevails during the winter dormant period. The plain (<2% slope) landscape has an NNE -SSW orientation (Figure 1a). The catchment is characterized by complex hydrogeology formed during the last glaciation about 10-15 k years BP (Gelbrecht et al., 2005). The southern part is in a glacial meltwater valley with glacio-fluvial sediments prevalent extending from Warsaw to Berlin (Figure 1b), where the catchment outlet discharges water into Lake Dehmsee and subsequently into the River Spree. The more elevated northern part of the catchment is dominated by freely draining unconsolidated ground moraines, with glaciofluvial sands and gravels prevalent in the South. Soils are freely draining and have a high fraction of sand (Table 1), though the northerly soils associated with the ground moraine have a higher silt content and retain more water (Figure 1c). Near the stream and in depressions, peat deposits are extensive and remain close to saturation throughout the year. Kettle holes are abundant as depressions in the landscape and are strongly influenced by groundwater (Nitzsche et al., 2017). These small water bodies provide important habitats and ecosystem services (Biggs et al., 2017).
The finer soils in the northern and eastern parts of the catchment are used for agricultural production (Table 1, Figure 1d). Multiple peatlands and fens exist along the stream network that are partly used as meadows. Land use gradually changes to forestry towards the southern region of the catchment (Figure 1d). Large parts of the catchment's forest cover are dominated by stands of Scots pine (Pinus sylvestris) which were intensively managed in the past. More recent management aims to enhance mixed and broadleaved forests (Lasch et al., 2002). The catchment has been historically drained, and fields in wetter areas are widely underlain with tile drainages. There is no irrigation being applied for agriculture. The anthropogenically influenced stream network (Nützmann et al., 2011) is highly droughtsensitive  with flows being intermittent during dry periods (Smith, Tetzlaff, Gelbrecht, et al., 2020). Drainage and the connection of glacial hollows which formerly had no surface outflow expanded the channel network from 20 km (1790) to 88 km (Nützmann et al., 2011). This results in a transformed hydrology, channel morphology, aquatic habitats, and nutrient cycling (Blann et al., 2009). Overall, streamflow generation in the catchment is dominated by groundwater; the catchment has a strong seasonal flow regime, with highest flows generally in winter. Runoff coefficients during storm events are <5% indicating limited contributions from restricted areas of saturated or compacted soils, as well as sealed (urban/road) surfaces (Smith, Tetzlaff, Gelbrecht, et al., 2020).

| DATA AND METHODS
In addition to using long-term hydrometric data, we conducted multiple sampling campaigns to obtain spatially distributed samples of precipitation, throughfall, soil bulk water, groundwater, and stream water for water stable isotope analysis at the catchment scale. The sampling covers the period between January 2018 and April 2020. Due to the evolving sampling infrastructure and logistics, not all time series started simultaneously, and some were conducted for shorter periods (e.g., soil water isotopes). Nevertheless, good spatial coverage of the catchment was achieved. We used daily meteorological data of precipitation, mean air temperature and relative humidity at 2 m height from a nearby climate station (Müncheberg), which is operated by the [%] [%] [%] [%] [%] [%] [%] [%] [%] [%] Marxdorfer St.  Umwelt-Geräte-Technik GmbH, Müncheberg, Germany), which were Demnitz Mill (6) followed by site Fox bridge (7), which is near the terrestrial ecohydrological site of FB ( Figure 1g). The Demnitzer Millcreek then reaches the catchment outlet at Berkenbrück (8). Samples were taken weekly if stream water was present, but data gaps occurred due to the cessation of flow during dry periods. All stream water samples of a given date were taken within a few hours. Some sites (Peat South (3), Demnitz Mill (6), Berkenbrück (8) Dansgaard (1964): We further calculated the line-conditioned excess (short lcexcess) defined by Landwehr and Coplen (2006) which defines the unconformity with the local meteoric water line (LMWL) as: where a is the slope and b the intercept of the local amount weighted  Gupta et al., 2009) within the predefined parameter ranges for the shape factor (α; 0.1 to 2.5) and the scale parameter (β; 2 to 500).
MTT uncertainties are presented as the SD of the fitted MTT. To F I G U R E 2 Timeseries of DWD precipitation, relative humidity and air temperature (a); daily precipitation and δ 18 O at Hasenfelde (b); soil moisture (at 3 depths) and bulk soil water δ 18 O (at 6 depths) and soil storage in the first meter of the forested (c); and grassland site (d); soil storage for all five soil moisture locations (e); groundwater levels and δ 18 O signature (f); discharge and spatial weekly stream water δ 18 O (g) assess the young water fraction (YWF) in stream water, we applied the iteratively re-weighted least squares fitted sine-wave method . The ratio of amplitude values derived from the seasonal cycle of precipitation and stream represent the estimated fraction of water in streams that fell as precipitation within the last $2-3 months (Kirchner, 2016). Uncertainties in the YWF were derived from maximum ranges from variables SE of the YWF fit. We below the long-term mean annual precipitation (569 mm a À1 ; 1990-2018, for details see Smith, Tetzlaff, Gelbrecht, et al., 2020).
Precipitation was generally slightly higher during summer relative to winter.
F I G U R E 3 DMC spatial and seasonal (see Table 1) precipitation patterns of precipitation amounts (a), δ 18 O signature (b), and lc-excess (c) The seasonality in hydroclimate was reflected in the variations in δ 18 O in daily precipitation (Figures 2b and 3b) with heavier δ 18 O signatures in summer precipitation (max = 0.3‰) than winter precipitation (min = À18.3‰). However, large day to day variability obscured a seasonal pattern, and there was an overall SD of 3.4‰.
Spatial differences in precipitation amounts across the entire catchment were limited compared to the temporal dynamics over the seasons (Figure 3a). Spatial differences in the isotopic composition of rainfall were only evident during convective summer events, when differences in precipitation amounts were also higher as a result of more localized storm cells. However, these event-based variations were minor when aggregated over longer (seasonal) time scales (Table 2).  Figure 3 including precipitation seasons, sums, SD and its weighted mean (w.mean) and SD of δ 18 O and lc-excess values ~0.2-0.3‰), which was notably lower than for even the deepest soil horizons.
Measured stream discharge at Demnitz Mill (6) (Figure 2g stream water was frequently absent due to drought conditions causing long gaps in the isotopic dataset.

| Landscape influences on isotope fractionation processes
Along the catchment's N-S gradient there is an increasing percentage of forest along the river corridor ( Figure 1d and Table 1) (7) and Berkenbrück (8)) showed very different characteristics.
At Fox bridge (7), a low number of samples (n = 22), mostly from winter, deviated little from the LMWL and the groundwater signal.
The more complex catchment structure with increasing scale and more marked anthropogenic influences (e.g., drainage and urban areas) at the downstream stream sampling site Peat North (2) resulted in a higher YWF (13%) and reduced MTTs to~3.5 years. Subsequent sites showed increasing YWFs and decreasing MTTs but with lower alpha values (Table 3). The "apparent" YWF increase (YWF = 51%) and MTT decrease (MTT = 0.2 a) at Demnitz Mill (6) is presumably an artefact of fractionation driven stream isotope variability, rather than representing a decrease in water age. Water was always present at Berkenbrück (8), even when disconnected from the surface stream network. The strong groundwater effect resulted in a strong damping of the isotopic precipitation signal, with low YWF (6.5%) even though the MTT (3.0 a) remained lower than at Marxdorfer St.
F I G U R E 6 Dual isotope plots of weekly stream samples and nearest groundwater site coloured by season in order from headwater to catchment outlet (a-h) F I G U R E 7 Precipitation and stream lc-excess of spatial distributed DMC sampling locations from headwater to outlet (a-i) 5 | DISCUSSION

| Insights into hydrological process dynamics from water stable isotopes of the DMC
The overarching aim of this study was to use water isotope signatures of different compartments of the critical zone to provide an improved understanding of the hydrological functioning of the DMC, which is representative of many lowland headwater catchments of the NGP and the North European Plain. For almost three decades, research in the DMC catchment was mostly focused on water quality issues (Dieter et al., 2011;Gelbrecht et al., 1998Gelbrecht et al., , 2005Gücker et al., 2014).
However, given the urgency of climate change, the consequences of riparian restoration and beaver population recovery, and the closely coupled nature of biogeochemical processes and ecohydrological pathways, it became clear that more detailed hydrological understanding was needed at the catchment scale. Though it was not foreseen that the study period included one of the driest spells in the catchment's monitoring history, this provided a preview of future hydroclimatic conditions in the region under climate change (Lüttger et al., 2011). Despite the lower-than-average rainfall and higher-thanaverage temperatures, the seasonal distribution of precipitation was characteristic in terms of less frequent, high intensity summer rainfall events, and more frequent, low intensity winter rain (Smith et al., 2020b). Precipitation in this flat landscape was generally spatially uniform, as often assumed but rarely examined in landscapes without major terrain features (Daly, 2006).
Whilst previous ecohydrological studies in the catchment focused on the plot-scale Smith et al., 2020b), water isotope signatures can also provide a more spatially integrated understanding of hydrological function at the catchment scale (Kendall & Mcdonnell, 2012). Similar to the precipitation amount, the spatial variation in the isotopic composition of rainfall was limited relative to the spatial scale (Bowen & Revenaugh, 2003). Pronounced deviations only occurred at the catchment outlet at Berkenbrück, which reflected the influence of land use (forest) on net precipitation isotopic composition rather than spatial variations in open precipitation. Isotopic transformations in throughfall can be complex and substantial (Allen et al., 2017), however, canopy effects in DMC were limited and restricted to small summer events. This presumably reflects the DMC's high intensity summer precipitation, the canopy structure of the sampling site FA, the sampling frequency and that no stem flow was considered (which was largely negligible in the forest plots). Such limited canopy effects were also recently reported for Scots Pine forests in Scotland .
Bulk soil water isotopes reflected differences in vegetation cover, soil water use and soil characteristics. The free-draining soils under forest (FA) showed a stronger influence of recent precipitation and stronger evaporation effects due to lower net rainfall and limited water storage. Under grassland, the dynamics were similar but less pronounced for bulk soil water. This was also reflected in the soil moisture profiles under both land uses, indicating vegetation and soil dependent water dynamics and age distributions (Smith et al., 2020b).
Observed variability in shallow groundwater isotopes were directly and indirectly influenced by spatial patterns of groundwater recharge at the catchment scale (Lewandowski et al., 2009;Zimmer & McGlynn, 2017a). Previous isotope-based modelling of the dynamics in water ages (Smith et al., 2020b) showed that forest cover results in lower and older recharge to the near-surface groundwater system that feeds streamflow. This is in accordance with other assessments that predict negative trends in groundwater recharge under changing forest and climate characteristics in the region (Natkhin et al., 2012).
Furthermore, low groundwater levels during the dry study period could indicate local groundwater recharge from surface waters, with streams seasonally switching to losing conditions (Nitzsche et al., 2017). This is a common transient phenomena and can have a greater influence than vegetation water use in local groundwater recharge dynamics (Krause et al., 2007), though the process is reversed during wetter conditions when groundwater gradients are likely to be towards the stream (Zimmer & McGlynn, 2017a).
The isotope data also provided evidence that groundwater and riparian storages-flux relationships can be dynamic in space and time in lowland landscapes (Krause et al., 2007) showing limited variability in stream water isotopes during winter, when groundwater levels are high and closely overlap with wells near-by. Summer fractionation is T A B L E 3 Young water fraction and related p-value (significance) of the fitting and ranges from SEs in the parameters mean transit time (MTT) estimations and related fitting efficiency and parameters alpha and beta for the DMC stream sampling sites where peatlands (Sprenger, Tetzlaff, Tunaley, et al., 2017), particularly with beaver dams, are present and provide large open water surfaces (Rosell et al., 2005). Isotopes, thus, have helped to underpin a conceptual model of this groundwater-dominated, lowland catchment; identifying spatial variability in ecohydrological partitioning under differing water storage states that can guide future research efforts, and support more quantitative modelling (Birkel et al., 2011).

| Landscape influences on water partitioning and connectivity
The interplay of interception, rooting depth, transpiration, soils and artificial drainage results in vegetation having a strong influence on spatial patterns of groundwater recharge amounts and timing at the catchment scale (Smith et al., 2020a). The dynamic ecohydrological partitioning of precipitation into groundwater recharge and ET alters seasonal hydraulic gradients in the subsurface under contrasting dominant vegetation types in the DMC (Smith et al., 2020a), with recharge being lower under forest than grassland (cf. Douinot et al., 2019).
Resulting spatial patterns of groundwater discharge in riparian areas are important for sustainable management (Gou et al., 2015), biotic communities (Fritz & Dodds, 2004;Larned et al., 2010) and the transport of organic matter and nutrients (del Campo et al., 2020;Stieglitz et al., 2003).
The effects of groundwater-surface water interactions on catchment runoff were modulated by the high summer ET losses, which resulted in prolonged periods of discontinuous streamflows in summer and below average winter runoff. Nevertheless, the synchrony of the soil re-wetting by autumn rainfall with rising groundwater levels and re-establishment of connectivity in the channel network was consistent with the findings of previous work which emphasized the dominant role of groundwater in streamflow generation in DMC (Nützmann et al., 2011;Smith, Tetzlaff, Gelbrecht, et al., 2020).
Of course, the dynamics of the stream network are related to underlying runoff generation processes (Garbin et al., 2019) as well as the groundwater and riparian storages-flux relationships that can be highly dynamic in lowland landscapes (Krause et al., 2007). (7) site in the wettest conditions. This part of the stream was a "losing reach" with stream water leakage to groundwater throughout the study period indicating another important subsurface storage deficit that was not refilled during rewetting periods. The year-round reemergence of streamflow at Berkenbrück (8) and the low YWF (6.5%) indicate a strong dominance of groundwater during dry periods at the catchment outlet, which is consistent with the distinct hydrochemistry at this site (Smith, Tetzlaff, Gelbrecht, et al., 2020). This sampling location also experienced extensive beaver activity up and downstream of the sampling site. Further influences by the larger groundwater system of the nearby river Spree are likely (Smith, Tetzlaff, Gelbrecht, et al., 2020).
These observed patterns in surface and subsurface hydrological processes illustrate the importance of land use and landscape management (drainage, forestry, restoration, beavers etc.) on spatial dynamics of surface water availability, stream connectivity and modification of the seasonal tracer signal (e.g., YWF) in such lowland landscapes. Catchment-specific patterns of river network disconnection are becoming more widely investigated in contrasting environments and have been related to climate change-induced water balance alterations, very localized site characteristics (e.g., in the steep, forested Pacific Northwest; Ward et al., 2020), and landscape structure (Bertrand et al., 2012). More explicit, an assessment of the role of land use and vegetation age classes (Germer et al., 2011) on groundwater recharge and spatial connectivity patterns could be usefully integrated into further analysis through spatially distributed hydrological modelling (Holman et al., 2017).
The effects of the peat fen on the downstream nutrient dynamics are being assessed in on-going work at DMC and again, isotopes have given invaluable insights into hydrological function (cf. Smith et al., 2020b). Key questions revolve around the way in which runoff derived from groundwater draining the agricultural areas interacts with the organic-rich wetland soils, with slower flows, ponded water and longer residence times, in potentially warmer conditions (Lam et al., 2011). Such changes are likely to affect biogeochemical processes (Stieglitz et al., 2003), riparian vegetation (Pettit & Froend, 2018), as well as local groundwater recharge (Krause et al., 2007), and resulting runoff generation (and associated isotope composition) especially during drought conditions. Reactive tracers would therefore also give additional insight (Li et al., 2020) on instream biogeochemical processes (Dieter et al., 2011). In the lower catchment, the effects of land use, especially forest cover and age classes (Germer et al., 2011), as well as riparian management, also need to be considered to fully understand potential climate change impacts (Holman et al., 2017;Natkhin et al., 2012). Future surface connectivity patterns under climate change will have implications for in-stream biogeochemical processes (Dieter et al., 2011), water ages (Soulsby et al., 2015), and aquatic habitats (Sarremejane et al., 2017).
Catchment scale understanding from isotope studies as provided here can help identify the dominant hotspots for process-based research and climate mitigation.
Most importantly, in our study, vegetation cover emerged as the key land management focus for sustainable water management because of the groundwater recharge implications of forest cover (Natkhin et al., 2012) and the local importance of green water fluxes (Smith, Tetzlaff, Gelbrecht, et al., 2020;Smith et al., 2020b). The use of soil water or groundwater as a transpiration source shows dependency on tree species, age, stand density and distribution (Song et al., 2016), with a tendency for deeper water sources in more arid climates (Evaristo & McDonnell, 2017). Potential impacts of pine plantations (i.e., uniform age distributions in stems) on the regional groundwater levels have been also highlighted by Nützmann et al. (2011) and evidence-based assessment of water footprints of different land use will be essential in studying and managing local subsurface and surface water resources (Neill et al., 2021).

| In-stream processes and water ages
Our study further identified hotspots for in-stream evaporation and the downstream transmission of fractionation signals, especially under low flows. Isotopic fractionation of the groundwater-dominated streamflow occurred in peatlands and sites affected by beaver dams, similar to findings by Sprenger, Tetzlaff and Soulsby (2017) for a peatland in Scotland. In this context, analysis of the YWF showed that caution is needed when applying this method in situations where tracer signals are modified due to evaporative fractionation which is not considered in YWF approach as noted by Kirchner (2016). The MTTs have similar limitations in terms of not considering evaporative fractionation effects (McGuire & McDonnell, 2006), but can also be more generally uncertain as metrics of hydrological function in heterogenous catchments (Kirchner, 2016).
As a result of these fractionation effects, some of the calculated MTTs were unrealistically low and YWFs too high, showing high uncertainties. However, despite these limitations, both methods captured and constrained well some of the differences among the sites ( Figure 9) and the spatial aspects in streamflow generation. The domination of groundwater in discharge was apparent in the high MTT and low YWF at the two upstream sites (Marxdorfer St. (1), Peat North (2)). This is in line with findings from Smith et al. (2020a) and Massmann et al. (2009). Low YWFs ( Figure 9) and high MTTs were further apparent at the catchment outlet Berkenbrück (8)

| CONCLUSION
Water stable isotopes were used to supplement existing data in a longterm research catchment to enhance our understanding of ecohydrological function and catchment-scale connectivity. Our sampling over 2 years coincidentally captured catchment responses to drought conditions. Water isotope signatures enabled us to assess the spatial variations of ecohydrological partitioning between blue and green water fluxes in the DMC that are likely to become more marked under climate change predictions. Isotope dynamics provided some preliminary assessment of stream water ages through estimations of young water fractions and MTTs, which will be more constrained by ongoing monitoring. We also assessed land management and vegetation impacts on blue water fluxes. We found that forested areas are more likely to reduce groundwater recharge under water stress, causing a faster decline of groundwa- We are thankful for trustful collaboration with B. Bösel and technical support by the WLV furthermore access to their well (Wasser und Landschaftspflegeverband Untere Spree).

DATA AVAILABILITY STATEMENT
The data are available from the corresponding author upon reasonable request.
F I G U R E 9 Tentative young water fraction and MTT estimations and conservative uncertainties for the DMC stream sampling sites