Northern landscapes in transition: Evidence, approach and ways forward using the Krycklan Catchment Study

Improving our ability to detect changes in terrestrial and aquatic systems is a grand challenge in the environmental sciences. In a world experiencing increasingly rapid rates of climate change and ecosystem transformation, our ability to understand and predict how, when, where, and why changes occur is essential for adapting and mitigating human behaviours. In this context, long‐term field research infrastructures have a fundamentally important role to play. For northern boreal landscapes, the Krycklan Catchment Study (KCS) has supported monitoring and research aimed at revealing these changes since it was initiated in 1980. Early studies focused on forest regeneration and microclimatic conditions, nutrient balances and forest hydrology, which included monitoring climate variables, water balance components, and stream water chemistry. The research infrastructure has expanded over the years to encompass a 6790 ha catchment, which currently includes 11 gauged streams, ca. 1000 soil lysimeters, 150 groundwater wells, >500 permanent forest inventory plots, and a 150 m tall tower (a combined ecosystem‐atmosphere station of the ICOS, Integrated Carbon Observation System) for measurements of atmospheric gas concentrations and biosphere‐atmosphere exchanges of carbon, water, and energy. In addition, the KCS has also been the focus of numerous high resolution multi‐spectral LiDAR measurements and large scale experiments. This large collection of equipment and data generation supports a range of disciplinary studies, but more importantly fosters multi‐, trans‐, and interdisciplinary research opportunities. The KCS attracts a broad collection of scientists, including biogeochemists, ecologists, foresters, geologists, hydrologists, limnologists, soil scientists, and social scientists, all of whom bring their knowledge and experience to the site. The combination of long‐term monitoring, shorter‐term research projects, and large‐scale experiments, including manipulations of climate and various forest management practices, has contributed much to our understanding of boreal landscape functioning, while also supporting the development of models and guidelines for research, policy, and management.

of climate and various forest management practices, has contributed much to our understanding of boreal landscape functioning, while also supporting the development of models and guidelines for research, policy, and management.

K E Y W O R D S
boreal region, field research infrastructure, Krycklan Catchment Study, long-term and large scale experiments, long-term monitoring, process-based research 1 | INTRODUCTION Water balance, carbon dynamics, and the ecological integrity of northern regions are expected to change in response to global warming.
Such changes, in combination with a growing human population, increased environmental pressure, and more intensive resource extraction, will amplify the stressors imposed on terrestrial and freshwater resources in the north. Regardless of whether responses to these combined stressors are gradual or abrupt, the outcome is likely to be unexpected, owing to nonlinearities in catchment-scale storage and release of water, carbon, nutrients, and a wide range of additional processes. Already, a cascade of changes to northern landscapes have been observed in response to shorter and milder winters (Spence et al., 2015) and enhanced summer warming (Isles et al., 2016). Understanding and predicting how this trajectory toward a warmer climate will reshape the physical, chemical, and ecological properties of the natural environment at northern latitudes will be a critical challenge for the scientific community in the decades to come.
High quality empirical data of adequate spatial and temporal resolution are central for deciphering patterns, dynamics, and trends in environmental variables. Catchment hydrology, biogeochemical processes, and landscape carbon balance are inherently complex and often scale dependent, making reliable predictions of future conditions difficult . Sound predictions are even more challenging in regions where appropriate empirical data to develop, test, and validate models are sparse. Despite the undeniable value of field observations and experiments, a transition from field-based empirical studies to model-only approaches has been an accelerating trend in environmental science (Burt & McDonnell, 2015). This trend is especially noticeable at northern latitudes, where the number of long-term research sites have declined rapidly during the last decades, leaving only a few locations that generate sufficient empirical data to answer the most pertinent questions about the future of our water resources .
Most changes caused by anthropogenic forcing will be superimposed on natural variability that can mask or delay responses and make changes difficult to detect and mechanistically explain.
Unravelling the processes responsible for various degrees of environmental perturbations in northern ecosystems therefore requires more than just basic monitoring information, standard models, and insights from other regions. The boreal region is dominated by nutrient limited forests and peatlands representing large carbon stores (Loisel et al., 2014) that combined contain at least a third of the Earth's soil carbon pool (Bradshaw & Warkentin, 2015). Despite their global importance and vulnerability to ongoing climate change, boreal catchments have been subject to comparatively little experimental, integrative, and process-oriented research in the past (Song et al., 2019). This knowledge-gap is of concern, yet at the same time represents a tremendous opportunity for cutting edge research on global warming feedbacks to both the atmosphere and water resources.
Most stream hydrological and biogeochemical research is based on individual, well-studied catchments, or alternatively on data from regional monitoring datasets. While a major advantage of small research catchments is the large amount of ancillary data that can provide mechanistic insights, a disadvantage is that the results are often based on limited replication and provide poor geographic representation. Conversely, a limitation of environmental monitoring is that data collection is often not designed to answer process-oriented questions, which can make it difficult to infer causal relationships. To overcome the constraints of both approaches, one way forward is to combine the strengths of these two approaches into a framework that promotes basic research in longterm monitored catchments , especially when they include several catchments of different scales and land-use . The heterogeneity inherent to boreal landscapes provides a unique template in this context because of the large spatial variability in the coverage of forests and peatlands that regulates much of the spatial and temporal complexity of soils, hydrology, and biogeochemistry (Fork et al., 2020;Laudon et al., 2011). The study of stream networks also makes it possible to assess the influence of more seldomly studied headwaters to downstream ecosystems (Bishop et al., 2008).
Understanding the role of thresholds, tipping points, and other nonlinear processes is crucial for assessing the consequences of future environmental change. Detecting these responses requires field studies that go well beyond standard, disciplinary approaches. Instead, what is needed for solving many future challenges are research infrastructures that combine field measurements from a multitude of disciplines in the same catchment. This type of effort goes well beyond what one research project or research group can achieve, and requires a well-coordinated infrastructure that can support and combine the collection of critical long-term monitoring data, provide large sets of ancillary empirical information, and host complementary long-term/ large-scale experiments that are crucial to achieving mechanistic understanding of ecosystem responses to environmental change.
While long-term, process-based research at the landscape scale is clearly needed to address the influence of environmental perturbation on terrestrial and aquatic resources, few research sites exist that capture the spatial and temporal dimensions required. One existing example is the Krycklan Catchment Study (KCS), located in northern Sweden, which has provided a unique opportunity for integrated, process-based research in the boreal region for decades . The over-arching objectives of KCS are to (1) provide a state-of-the-art infrastructure for experimental and hypothesis driven research, (2) maintain a collection of high quality, long-term climatic, biogeochemical, hydrological, and other environmental data, and (3) support the development of models and guidelines for research, policy and management. An important step in this direction is to make the field infrastructure even more visible for potential users at the same time as we make more of the field data openly accessible.

| SITE DESCRIPTION
The Krycklan Catchment Study (KCS, www.slu.se/Krycklan) is located in the heart of the boreal landscape (64 14 0 N, 19 46 0 E), approximately 50 km northwest of the city of Umeå in northern Sweden . The KCS is 6790 ha in area and comprises a mosaic of instrumented and well-studied forests, wetlands, and lakes, all drained and connected by a network of streams and rivers ( Figure 1). Over the past 35 years, the existing field research infrastructure has generated data resulting in approximately 1000 peerreviewed publications, and over 110 PhD theses in diverse areas such as hydrology, biogeochemistry, carbon dynamics, climate change, The Krycklan Catchment Study shown with sub-catchment areas and corresponding tree density (panel a), the stream network, including the Trollberget Experimental Area in the south (panel b), the distribution of quaternary deposits (panel c), and a close up on the central area, which included sub-catchments C2,C4, C5, C6 and C7, the experimental stream segment (between C5 and C6), and approximately 150 groundwater wells (panel d) geography, soil science, ecology, forestry, land-use history, and political science.
Forestry and peatland research began in the area in the 1910s, with the first field station being built in 1923 at the Kulbäcksliden research park (Grip, 2015). Forest regeneration was the primary motivation for establishing the nearby Svartberget research station in the center of the KCS in 1979. The specific questions at that time were primarily related to microclimatic conditions after forest harvesting, nutrient limitation to tree growth, and forest hydrology. Ten years later, the role of acid deposition and natural acidity became the major water related research question at the site (Bishop et al., 1990). The combined focus on climate, forestry, and hydrology resulted in a wide range of high quality field measurements that are now of particular importance for documenting responses to current environmental changes.

| Climate
An overall warming pattern is clear from the long-term air temperature record at Svartberget from 1980 to 2020, but becomes even more visible when extending the time series back to 1891 by using the nearby Stensele site (Figure 2) for extrapolation. Annual air temperature has increased by about 3.0 C since 1891, but the most pronounced increase (2.5 C) has occurred in the last 40 years (Mann-Kendall tests, p < 0.001), with 2020 by far being the warmest year on record. The seasonal air temperature has increased most rapidly dur- Total annual average precipitation equals 623 mm, ranging from 446 (1994) to 918 mm (1982), with no statistical trend over the last 40 years. Of the precipitation, approximately 30% arrives as snow.
The average snow water equivalents (SWE) for the 40 years of record is 180 mm, ranging from 64 (1996) to 321 (1988) mm. The 40-year average duration of winter snow cover is 167 days, but this has been decreasing over time. During the first decade of measurement, the average date of initial snow cover was in early November; since then, this onset has been delayed by $0.5 day year À1 (Laudon & Ottosson Löfvenius, 2016). However, the melting of snow in spring has experienced no significant trend, and on average peaks in late April. The long-term average annual runoff at a forest dominated catchment (C7;  F I G U R E 2 Panel a. Mean annual air temperature at Svartberget from 1890-present with 10-year running average (solid line). Data prior to 1980 are modeled from the scaling relationship with a nearby climate station at Stensele (150 km to northeast). The 24 year overlap  in the air temperature records at these two stations yielded a linear correlation with slope = 1.041, intercept = À0.008, r 2 = 0.93, p < 0.001, and RMSE = 2.786. Panel b shows the seasonal trends in air temperature from the Svartberget station from 1980 to present KCS has experienced relatively low levels of direct human influence. The current population of the area is approximately 1.2 people per km 2 . Historically, the human population was even lower and only hunting, fishing, and reindeer herding occurred before 1750, when the first village settlements were established. Prior to the early 1900's, peatlands constituted a major source of livelihood for farmers.
In the KCS, this meant that up to 22% of the original peatland area was used for hay harvest on mire meadows (Norstedt et  F I G U R E 3 Examples changing hydrological and chemical conditions in the Krycklan Catchment Study over the last 30-60 years. These changes include an increase in evapotranspiration (ET, trend: +0.3% year À1 ) at the C7 catchment measured as the difference between precipitation and discharge (panel a). Long-term increases in forest biomass in the extended Krycklan area relative to estimates in 1957 (trend: +1.0% year À1 ; panel b; Swedish National Forest Inventory, SLU, unpublished data). An increase in mid-winter (March) runoff in the C7 catchment (trend: +0.1 L s À1 year À1 ; panel c). Declines in the timing (date) of lake ice-off in the spring, based on a 55-year record collected 25 km north of the Krycklan (trend: À0.2 days year À1 ; Rune Axelsson, unpublished data; panel d). Declining Ca concentrations in a forest-dominated stream (C2; trend: À0.02 mg L À1 year À1 ), but not for an adjacent mire-domminated stream (C4; panel e). Increasing DOC concentrations in same forested stream (C2; trend: +0.2 mg L À1 year À1 ), but not in the mire-dominated counterpart (C4; no trend, panel f). All trends were calculated using Mann Kendall tests whereas only 3% were drained for more modern agriculture. Presently, only 1% of the original peatland area is still used for agricultural purposes. Beginning around 1900, mires were drained to enhance forest wood production. As a consequence, about 40% of the original peatland area is currently forested (Norstedt et al., 2021). In addition, approximately 162 km of forest drainage ditches can be found within the KCS area , which can be compared with approximately 180 km of natural, permanent streams (Ågren et al., 2015). Prior to the 1940s, selective cutting was the primary method used in forestry. Later, rotation forestry grew to dominate, involving mostly clear-cutting with subsequent planting of conifers (Norstedt & Laudon, 2019).

| Geological and physiographic setting
The KCS is located in the Svecofennian orogenic belt, which traverses large parts of northern Sweden. The bedrock is dominated by 1.92-1.87 Ga old migmatised meta-greywacke or paragneiss, which consists of metamorphosed sediments once deposited outside the Achaean Baltic Shield. The numerous hills in the area with peaks up to 400 m are largely derived from selective weathering of biotiteplagioclase schist in the valleys and more resistant veined gneiss at higher altitudes. Further inland, the meta-sediments are gradually replaced by 1.74-1.82 Ga old granite and granodiorite, which also occur as intrusions in the KCS along with minor intrusions of mafic rocks.
The Quaternary deposits are strongly influenced by the latest glaciation. Drumlins and crag-and-tails are aligned in a SSE direction as the inland ice was moving from NNW. The ice retreated from the area ca. 10 200 a BP (Stroeven et al., 2016), leaving up to 30 m thick till in sheltered areas, but also bare bedrock in more exposed locations. In addition, the large Vindel River Esker passes through the lower parts of the KCS adding large deposits of glaciofluvial material (Figure 1c). The Quaternary deposits are predominately of local origin, displaying a silicate-dominated chemistry with quartz>plagioclase>K feldspar>amphiboles as the main minerals (Lampa et al., 2020). At the termination of the deglaciation, approximately half of the KCS was located below the highest postglacial coastline (situated at ca 257 m above present sea level). This has resulted in locally >60 m deep sand and silt sediments that now cover the lower parts of the KCS, deposited by the Vindel River during the course of the isostatic rebound. In areas with low topographic relief, peat has built up, generally forming oligotrophic minerogenic mires.

| LONG-TERM ENVIRONMENTAL TRENDS
The boreal region encompassing the KCS has experienced some strong environmental trends during the last several decades. Changes in climate, land-use, and long-range transport of air pollutants all have had a role to play in explaining some of these decadal changes.
Despite having a highly developed research infrastructure in place, the co-occurrence, interaction, and synchronicity of several human interventions complicate our efforts to disentangle the causes-andeffects responsible for the observed changes. However, by combining long-time series, large-scale experiments, and modelling we are now beginning to understand the roles climate change, land-use, and atmospheric pollution have played in the past, as well as to predict their relative influences in the future.
Here we highlight some of the major trends in forest biomass growth, lake ice extent, catchment hydrology, and water quality for the KCS (Figure 3). While some of the trends can be directly related to changes in temperature, such as the increasingly earlier lake iceout, other trends can be linked to atmospheric pollution, namely, the decline in stream calcium that is caused by the recovery from acid deposition . Increased forest biomass production, stream water brownification, and increase in ET are likely caused by a combination of interacting factors. Such interactions, and the fact that some catchments respond while adjacent systems do not, call for the need of continued research to disentangle the cause-and-effect mechanisms. In addition, we urgently need to provide predictions for what these large-scale environmental changes will mean for northern environments. This includes understanding the direct and indirect effects on carbon and greenhouse gas (GHG) balances, atmospheric radiative forcing, terrestrial and aquatic biodiversity, water quality, and on the capacity for northern regions to support society. Living up to society's many goals for the region in an environment that is constantly changing, requires maintaining research infrastructures that take a landscape scale perspective and measure the most important processes in the atmosphere, vegetation, soils, bedrock, and water, as well as the interactions between them.

| RESEARCH INFRASTRUCTURE
The ambition of KCS is to take a holistic ecosystem perspective of the boreal landscape to understand, elucidate, and predict the role of internal and external drivers of catchment processes across a range of scales. To do this, we combine state-of-the-art technology to capture various ecosystem processes with traditional research tools and basic environmental monitoring. We focus on processes and dynamics of living and non-living ecosystem compartments, as well as the fluxes of energy, water, carbon, nutrients, metals, and other compounds within and between the atmosphere, lithosphere, cryosphere, and hydrosphere. In the KCS, we do this by combining a large, central, research facilitynamely, the ICOS research towerwith supplemental infrastructures distributed across the entire landscape ( Figure 4). In addition to these facilities, the KCS also offers a number of large scale and/or long-term experimental facilities. Below we outline some of the most central of these facilities and data.

| Surface water program
The hydrology and water chemistry program has been a central feature of the KCS program since the beginning. The 50 ha C7 catchment

| Soil water program
Soil water from three soil profiles (called the S-transect) located 4, 12 and 22 m from the C2 stream have been monitored 5-12 times per year since 1995 for water isotopes and water chemistry (Bishop et al., 2004;Blackburn et al., 2017;Lidman et al., 2017;Nyberg et al., 2001). The S-transect is aligned based on topography, following the lateral flowpath of groundwater. Each profile consists of measurements at six soil depths between 5 and 90 cm using ceramic suction lysimeters (P100). Soil water content using Time-domain reflectometry (TDR) and soil temperatures are measured at the same depths. A similar setup was installed in one of the wetland soils upstream of C4 to monitor soil water chemistry since 1997 using 12 nested wells extending to different depths, ranging from 25 to 350 cm below the ground surface (Lidman et al., 2013;Sponseller et al., 2018). In 2007, the Riparian Observatory of Krycklan, a complementary set of transects to represent a range of topographic situations was also established (Grabs et al., 2012).

| Biosphere-atmosphere exchanges of carbon, water and energy
A combined atmosphere-ecosystem ICOS station is located in the center of KCS (Figures 1 and 4). ICOS is a pan-European research infrastructure with the mission to produce standardized, high-preci- give additional information related to ecosystem properties.
The ICOS infrastructure also serves as a platform for establishing and connecting external research projects. For instance, the installation of two more EC systems at 60 and 85 m height along the ICOS tower provides additional estimates of CO 2 , CH 4 , water, and energy exchanges at the landscape scale (i.e., few km radius), roughly spanning the area of the KCS (Chi et al., 2019). Integration of these landscape EC measurements with aquatic fluxes of carbon species via stream runoff has resulted in a first estimate of the net landscape carbon balance (NLCB) for the KCS . In addition, the combination of EC and sapflow measurements around the ICOS tower has provided an opportunity to partition the forest water cycle components (Kozii et al., 2020). Furthermore, the concentration measurement profile including several levels along the 150 m tall tower has enabled investigations of atmospheric organic pollutants (Bidleman et al., 2017), as well as water isotope and mercury dynamics. The ICOS tower structure also hosts multispectral sensors, and phenology cameras within the SITES-Spectral infrastructure to collect spectral data for estimating ecosystem vegetation properties at various spatial and temporal scales.

| Sapflow measurements
In spring 2016, a network of Granier sap flow sensors (Granier, 1985) was installed in 70 trees (30 P. sylvestris, 30 P. abies and 10 Betula sp.) to continuously measure tree-level transpiration, an integrated measure of whole tree hydraulic stress, in the three tree species that are dominant in Fennoscandian boreal forests. This technique includes a pair of thermocouple sensors that detect changes in the temperature difference (ΔT) from the baseline (ΔT m at zero flow), which in turn reflects the flow rate of water through stems (Granier, 1987). In addition, a field deployed Picarro L2131-i analyzer provides continuous, high temporal resolution isotopic measurements of tree xylem water. Taken together, these measurements provide a unique opportunity to test how changes in environmental conditions affects stand evaporation and tree-level transpiration across a range of temporal scales as well as directly compare the importance of transformations into other water balance components (i.e., streamflow) within a northern boreal headwater catchment. Macroinvertebrate and stream microbial data have been collected repeatedly from a number of streams within the catchment and used in different contexts (e.g., Burrows et al., 2017;Göthe et al., 2013;Jonsson et al., 2017). Survival experiments on fish (Serrano et al., 2008) and invertebrate population studies (Petrin et al., 2007) have been conducted in several of the monitored streams. In addition, the main stem of the Krycklan river network has been used as an unimpacted (by timber floating) reference site in a number of studies of stream hydrogeomorphology (Polvi et al., 2014), riparian plant diversity and composition (Hasselquist et al., 2015), riparian nutrient cycling (Hasselquist et al., 2017), instream ecosystem functioning (Frainer et al., 2018), and biodiversity (Hasselquist et al., 2018). Since 2007, C7 is also a part of the national freshwater monitoring program under which aquatic macroinvertebrates are annually collected to depict long-term biodiversity trends.  (Table 2).

| Lidar data
Artificial drainage of peatlands through ditches have dewatered millions of hectares of northern peatlands for forestry. Recent estimates suggest that up to 1 million km of wetland ditches in Sweden alone have been created (Ågren & Lidberg, 2019), many of which are now not functioning . The future fate of these drainage ditches can be to: (1) clean them to ensure resumed drainage, (2) ecologically restore them to a more natural state, or (3) leave them unmanaged.
At TEA, we have created a side-by-side comparison of these three different management options with the objective of determining their effects on water quality and quantity as well as their role in altering the landatmosphere greenhouse gas exchanges. Specifically, we have the goal of quantifying the impact of peatland forest harvest, ditch cleaning, and filling-in of ditches (ecological restoration) on dissolved organic matter export and quality, greenhouse gas balance, nutrient and sediment export, export and speciation of mercury, as well as water storage. Here we take a catchment-scale approach to monitoring the dynamics and export from our different treatments. In total, six experimental catchments with an average size of 10 ha are being monitored in TEA, where four catchments in the nearby KCS sites serve as controls.
In addition, TEA includes a riparian buffer experiment with the goal to directly compare the functioning of narrow (5 m) to wider (15 m) buffers. In response, we are monitoring nutrient and sediment export, riparian vegetation, and greenhouse gas fluxes in the streams and riparian zone. The unique opportunity of comparing before and after treatment, as well as the differences in responses to buffer widths, will allow for more informed management decisions in the future.

| Soil frost experiment
The KCS supports the longest (>17 years) ongoing soil frost manipulations experiment in the world (Campbell & Laudon, 2019), contributing with knowledge critical for interpreting how changes in winter conditions will affect soils and streams. The ongoing experiment has been used to study the effects of soil frost on DOC (Haei et al., 2010(Haei et al., , 2012, CO 2 emissions (Öquist & Laudon, 2008), decomposition processes (Kreyling et al., 2013)  with three replicates each. The slope and flow of each channel is adjustable ranging from 0 to 1.5 for slope and 0 to 1 L s À1 for flow.
The channels can also be experimentally heated using warming cables.
The bottom substrate in each channel is composed of local, natural gravel and pebbles but can also be manipulated.
In For every water sample ever collected within the KCS research program, duplicate samples have been stored at À18 0 C in acid-washed bottles for chemical analysis, and at +4 C, in glass bottles for water isotopes. In total, the Krycklan archive contains over 25 000 unique samples. Additionally, the KCS database contains more than 15 million datapoints on water chemistry, and even more data from long-term, high-resolution timeseries on physical parameters. The KCS database builds on a concept we call FAIR & Square (Laudon & Taberman, 2016), which is guided by Data FAIR port requirements (https://www. datafairport.org/). While FAIR stands for Findable, Accessible, Interoperable, and Reusable, "Square" symbolizes the importance of acknowledging the original data producer. An important aspect of FAIR & Square is that we try to provide clear, precise and standardized metadata that can answer questions about "when, where, how, and why" samples have been collected, analyzed and quality controlled.
An important principle in the FAIR & Square concept is to acknowledge the effort that has gone into acquiring the data being shared. In short we ask users to: (1) always cite the original source of data used, (2) acknowledge other studies that touch on similar aspects that their work builds upon, (3) not distribute data to third parties in order to allow updates/corrections and to avoid spreading erroneous data, and (4) recognize the original data producer properly in acknowledgments if they provided data, or by offering co-authorship if they contributed with significant work, important ideas, and/or helped with essential interpretation.

| DIRECTIONS FOR THE FUTURE
At a time when our environment is under increasing pressure from global change it is alarming that many leading field research infrastructures are under increasing threat to be down-sized or even closed.
Failure to recognize and prioritize the value of long term research and monitoring compromises the possibilities that science can contribute to long-lasting solutions. Never before has the need been greater to continue the collection of empirical data that, together with past data, can provide baseline conditions before climate change obscures the clues into how ecosystems functioned without this massive human influence. Understanding the fate of surface and groundwater resources, carbon exchange processes, and threats to biomass production are questions of fundamental importance for the future.
Despite this need, the trend in monitoring is going the other way, where empirical studies are declining relative to modelling-based analysis. While modelling will be an important part of environmental science, models must be constrained, tested, and validated by empirical field data to be useful. To cite Sherlock Holmes by Sir Arthur Conan Doyle, "It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts".