Freshwater ponds create highly dynamic Arctic tundra landscapes

Freshwater ponds create highly dynamic Arctic tundra landscapes | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Freshwater ponds create highly dynamic Arctic tundra landscapes Jakob Assmann, Cengiz Akandil, Elena Plekhanova, Alizée Le Moigne, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5581925/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Greenhouse gas emissions from Arctic tundra ponds and permafrost thaw provide important positive feedbacks to global warming. However, a high landscape heterogeneity and small size of ponds make it challenging to assess trends in surface water extent and associated carbon and energy fluxes, especially in the understudied Eastern Siberian tundra. Here, we show that surface water extent in these landscapes can be highly dynamic, shaped by pond-scale processes that cannot be detected in satellite products. Using a time-series of aerial imagery at 12 cm resolution spanning eight years (2014–2021), we classified surface water at three sites in Kytalyk National Park and traced the 465 ponds larger than 1 m 2 . The surface water extent varied between 102%-124% relative to the time-series mean, without significant trends in contrast to previous reports. Individual pond area fluctuated by 52% on average, and two thirds of ponds were present for less than six years. One-quarter of ponds showed evidence for thermokarst or vegetation colonisation as drivers of change, based on our high-resolution surface elevation models. These findings highlight a highly dynamic nature of tundra ponds and stress the need for improved change detection to better model carbon and energy fluxes in this biome. Main Text Earth and environmental sciences/Hydrology Earth and environmental sciences/Ecology/Climate-change ecology Earth and environmental sciences/Ecology/Ecosystem ecology Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Greenhouse gas emissions from tundra ponds and permafrost thaw (thermokarst) provide important feedbacks to global warming (Wik et al. 2016 , Olefeldt 2016). Ponds are a common feature of many Arctic landscapes, where they provide critical habitat for breeding birds (Corbeil-Robitaille et al. 2024 ) and form an important source of methane emissions (Wik et al. 2016 ). Methane emissions from ponds are mainly driven by microbial activity (Del Giorgio and del Cole, 1998) which is thought to increase with warming (Negandhi et al., 2016 ; Bartosiewicz et al., 2019 ). While permafrost thaw itself may release large quantities of carbon (Nitzbon et al. 2020 , Wang et al. 2024 ), it can also be a driver of pond formation, with newly formed ponds having potentially elevated emission rates (Peura et al. 2020 , Prėskienis et al. 2024 ). However, large uncertainties remain in the accounting and modelling of carbon fluxes in Arctic wetlands, especially in relation to small ponds (Negandhi et al., 2013 , Melton et al. 2013 , Muster et al. 2012 , Wik et al. 2016 ) which are difficult to map with data exceeding metre-scale resolution (Muster et al. 2012 ). Further, ponds are an important factor of tundra land surface albedo and energy fluxes during summer (Juszak et al., 2017 , Langer et al., 2011 ). Studying tundra ponds and how they change at fine spatial scales is therefore a key step to better understand the potential feedbacks to global warming. There is currently no agreement on the direction of change in surface water area in response to warming across the Arctic. Some research indicates that water bodies, including small ponds, are decreasing across the Arctic (Webb et al. 2022 , Wik et al. 2016 , Andresen and Lougheed 2015 ). Other studies highlight the large methodological uncertainties in large-scale analyses of Arctic surface water change (Olthof et al. 2023 , Webb et al. 2023 ). Additionally, there is considerable regional variation in trends (Webb and Liljedahl 2023 ), especially for small water bodies (Payette, 2004, Walter et al. 2006 , Wolfe 2011, Magnússon et al. 2021 ). For instance, increases in area and number of ponds has been reported for Northeastern Siberia (Magnússon et al. 2021 ), while declines have been reported in Alaska (Andresen and Lougheed 2015 ). The variation in regional trends for large water bodies has been associated with large-scale patterns in the distribution of continuous vs. discontinuous permafrost (Webb and Liljedahl 2023 ), but too little data are available for small water bodies. A high landscape heterogeneity and water bodies with dynamic extents make it difficult to detect surface water changes in the Arctic tundra (Muster et al. 2012 , Webb and Liljedahl 2023 ), and thus estimate the associated carbon fluxes at landscape scales (Muster et al. 2012 , Hinks et al. 2015 , Treat et al. 2018 ). The geomorphological processes that shape Arctic tundra landscapes can be very complex (Lara et al. 2020 ) resulting in a highly heterogeneous land cover at metre to decametre scales (Muster et al. 2012 , Lara et al. 2015 , Morgenstern et al. 2013 ). Tundra ponds are an important contributor to this metre-scale heterogeneity (Muster et al. 2012 ), with pond development being driven by various fine-scale abiotic and biotic processes (Pienitz et al., 2008 ), including freeze-thaw cycles of ice wedges (Lachenbruch 1962 ), permafrost thaw that creates depressions (Rautio et al. 2011 , Luoto and Seppälä 2003 ) and vegetation colonisation by sedges and mosses (Magnússon et al. 2020 ). Some of these processes may occur on sub-decadal timescales (Magnússon et al. 2020 ) and could, combined with fluctuations in the main drivers the water budget of the tundra (precipitation, evaporation and lateral-flow) (Helbig et al. 2013 ), potentially cause rapid changes in the surface water area of the ponds. High-resolution (below 4–5 m) and high frequency spatial data are therefore needed to map tundra ponds at the landscape scale (Muster et al. 2012 ), but these are often not available over long time frames (Webb and Liljedahl 2023 ). Here, we assess the change in surface water extent at the landscape-scale and for individual ponds in the tundra of Northeastern Siberia. We also test for evidence of thermokarst and vegetation colonisation as drivers of pond area change. To this end, we collected time-series of high-resolution (12 cm) drone imagery and digital surface elevation models (DSMs) at three study sites in the Kytalyk National Park for the time period 2014 to 2021 (Fig. 1 ). Two study sites are in a drained thaw lake basin with high and medium surface water occurrence (Fig. 1 a) and one site is on a Yedoma ridge with low surface water occurrence functioning as a control. Our study thus covers two of the major landscape types in the region (Nitzbon et al. 2020 , Schirrmeister et al. 2012 ). Next, we classified surface water in each image using a simple threshold in the blue chromatic coordinate (BCC) informed by annotated training data, and identified all individual ponds that exceeded 1 m 2 and were present for at least 3 years (465 in total; 357, 108, 0 respectively for the “high”, “medium” and “low” surface water occurrence sites). We then assessed how surface water extent changed at the landscape- and pond-level throughout each time-series. Finally, we tested whether thermokarst and vegetation colonisation contributed to individual pond change, by estimating the drop or gain in surface elevation in the drone-derived DSMs associated with the change. Our study thus allows us to gain novel insights into the rapid nature of changes in surface water and their ecological drivers in the understudied tundra landscapes of Northeastern Siberia. Results Surface water extent in Northeastern Siberian tundra landscapes is highly dynamic The proportion of surface water detected was highly dynamic in the landscapes where ponds occurred regularly (“medium” and “high”, Fig. 1 a). In the landscape with “low” pond occurrence, we only detected surface water in 2017, with no ponds occurring otherwise. The year 2017 was a strong outlier year with exceptionally high surface water cover across all study sites (Fig. 1 b). The time-series for the “medium” and “high” sites showed high surface water variability (CV: 124% and 102%) even when the outlier year was removed (CV: 42% and 25% respectively). None of the time-series showed a significant trend over time (with and without 2017, Table S6, S7, S8, S9, S10, S11). We also did not detect any correlations with climate variables (Supplementary Methods, Table S12), but the time-series of surface water extent for the “high” site indicates some correspondence with mean summer air temperature (June-August) and autumn precipitation in the preceding calendar year (September-November) (Fig. S1 , Table S12). The outlier year 2017 corresponded to a cold summer and high snowfall at the beginning of the preceding cold season (Tei et al. 2018), but annual temperature and precipitation values did not appear unusual in the context of the previous decades (Fig. S2). Most ponds were short-lived and many strongly fluctuated in size The majority of ponds were intermittently present in our time-series and the average variability in individual pond area was high (Fig. 2 ). Across all study sites and ponds, approximately two-thirds of ponds (65%) were present for less than six years and only one-third of ponds (35%) was “continuously” present, i.e., surface water was detected for six or seven years (allowing for one year of failed detection). The proportion of ponds “continuously” present was similar between the “medium” (38%) and “high” (34%) sites (Fig. S3a). The large majority of ponds experienced strong fluctuations in surface area (Fig. 2 b). Excluding the outlier year 2017, 92% of ponds had a standardised variability (coefficient of variation, CV) larger than 10% of their mean surface area. The mean CV of pond area change was 52% across all time-series, with comparable values for the “medium” (52%) and “high” (52%) sites (Fig. S3b). We observed only 5 out of 465 (1.1%) ponds that were “stable” in the sense that they persisted for six or more years and had a CV of less than 10%, all five were found at the “medium” site. Thermokarst and vegetation colonisation as drivers of pond change We detected evidence for thermokarst (Fig. 3 ) and vegetation colonisation (Fig. 4 ) as drivers of surface area change in 24% of the ponds at the “medium” and “high” sites. Thermokarst may lead to pond expansion or movement (e.g., Fig. 3 a). We observed expansion or movement of 68 ponds (15% of all ponds) across both sites, defined by a mean drop in surface elevation that exceeded 0.1 m (Fig. 3 b) averaged across the pond area gained between 2014 and 2024. This value serves as our threshold for detecting subsidence due to thermokarst (Online Methods). The proportion of these ponds with evidence for thermokarst was higher at the “medium” site (36 ponds = 33%) than at the “high” site (32 ponds = 9%) (Fig. S4). Vegetation colonisation, e.g., incursion of the pond area by sedges or mosses, can result in pond shrinking or even complete infilling (e.g., Fig. 4 a). We observed vegetation colonisation in 83 ponds (18%) across both sites, defined by a gain in surface elevation that exceeded 0.1 m (Fig. 4 b) averaged across the pond area lost between 2014 and 2021. This value serves as our threshold for detecting surface elevation gain indicating vegetation colonisation. The proportion of ponds with evidence for vegetation colonisation was higher at the “medium” site (40 ponds = 37%) than at the “high” site (43 ponds = 12%) (Fig. S5). Discussion Our findings demonstrate that tundra thermokarst landscapes and ponds can have highly dynamic extents - even over short time-periods such as three or five years. Very few ponds (five) in the studied landscape were stable in area and position. In contrast, the majority of studied ponds were short-lived and showed strong fluctuations in size. We found evidence for thermokarst and vegetation colonisation as drivers of change in approximately one quarter of the studied ponds, but many ponds changed without either driver being detected. Together, these findings have considerable implications for how we monitor change, estimate carbon and energy fluxes, and understand drivers of change in surface water extent in tundra thermokarst landscapes. High temporal surface water variability complicates detection of long-term trends High inter-annual variability of surface water at scales of metres to decametres restricts our ability to derive estimates of long-term trends, such as wetting or drying, using remote sensing in tundra systems with pond presence. Multiple studies have already highlighted the insufficiencies of moderate resolution (10–250 m) satellite imagery for surface water (and trend) detection in such tundra systems (Muster 2012, Webb and Liljedahl 2023 , Mullen et al. 2023). While centimetre-resolution drone data (as shown by us) and, to some degree, commercial high-resolution satellite imagery (Magnusson et al. 2020, Magnusson et al. 2021, Mullen et al. 2023) can provide data to detect surface area changes in small water bodies at the relevant spatial scales, our findings highlight the need to also sample these systems over long time-frames and at a high temporal frequencies (at least annual, likely better: intra-annual, see also Mullen et al. 2023). Even if an underlying long-term trend was present, time-series like ours that span almost a decade are not statistically powerful enough to detect a trend, given the high variability observed. We therefore suggest caution when interpreting trends based on time-series with moderate to low spatial (10–250 m) and temporal resolution (i.e., larger than annual) in those tundra systems where small ponds are present and encourage the continued development of improved monitoring methods (Webb and Liljedahl 2023 , and Mullen et al. 2023). Pond dynamics driven by more than just thermokarst and vegetation colonisation Many ponds in our study showed high surface water variability without evidence for either thermokarst or vegetation colonisation. Pond-dominated tundra landscapes have been described as “fill-and-spill” systems (Helbig et al. 2013 ), where surface water dynamics are driven by snow melt (a function of winter precipitation) during spring and evaporation during summer, as well as surface and subsurface lateral flow (Woo and Guan 2006 , Wolfe et al. 2011 , Helbig et al. 2013 ). These processes are likely also key drivers of surface water dynamics observed in our study, as illustrated by the outlier year 2017: The year 2017 was marked by the inundation of landscapes across the region, that was driven by a high amount of snowmelt lasting into mid-summer, following a larger than unusual precipitation in the preceding October (Tei et al. 2020 ). Quantifying and modelling the relative contributions of the different drivers of landscape-level pond change (for example, using drone-derived surface models) will be critical for improving our predictions of future landscape dynamics and the associated carbon fluxes. Surface elevation changes integrate vegetation change and permafrost processes Our findings confirm that thermokarst and vegetation colonisation are processes contributing to pond change in the study region (Magnússon et al. 2020 ), but our method cannot resolve the complex causes of these processes themselves. At the “medium” and “high” study sites the ice-content of the first 60 cm of soil ranges from 50–70% (Wang et al. 2019 ). If such soils are thawed, the release of water will induce subsidence (Li et al. 2017 ) causing thermal erosion. By measuring the drop in surface elevation we detect this subsidence but we also integrate the loss of the vegetation canopy. Indeed, thermokarst is often associated with loss of vegetation, which can be both cause and effect of thermal erosion (Kanevsky et al. 2017, Li et al. 2017 , Magnússon et al, 2020 ). Similarly a gain in surface elevation integrates vegetation growth in previously water covered areas and the formation of ice rich permafrost, which also can be cause and effect of each other (Liljedahl 2016). Deriving digital elevation and canopy height models using drone-based LiDAR or dense ground control points could enable the separation of these processes, but in the absence of these data our study is limited to measuring the integrated effects. Studying fine-scale pond dynamics will better our understanding of ecosystem functioning The strong variation in the surface area of individual ponds observed in our study suggests a need to better understand the associated consequences for the ecosystem and carbon and energy cycle. Many ponds in our study fluctuated notably in size and many had a short life-span; temporary flooding (Woo and Guan 2006 ) and drying out (Smol and Douglas 2007 ) were common across the time-series, as was the establishment and breakage of surface water connections between ponds (Wrona et al. 2016 ). How these dynamic processes affect the habitat availability for breeding birds (Corbeil-Robitaille et al. 2024 ) or the assembly of microbial communities (Le Moigne et al. 2020 , Gurney et al. 2022 ) and their impact on the functioning of the ponds in the carbon cycle or other landscape-scale processes (Blackburn-Desbiens et al. 2023 ) are not well understood. Furthermore, processes like thermokarst and vegetation colonisation have different implications for the carbon cycle. Thermokarst will likely release old carbon from the thawed soil (Wauthy et al. 2018 ), while vegetation incursion by sedges will act as a conduit for the release of methane from the pond water (Knoblauch et al. 2015 ). Combining time-series of high-resolution imagery with in-situ field studies of macro fauna or microbial dynamics could therefore greatly advance our understanding of the effects of pond dynamics on landscape-scale ecosystem function. Not all tundra landscapes experience directional trends or stability Overall, our study highlights the possibility that not all tundra systems are experiencing clear directional trends or stable states. Instead, our data shows that tundra landscapes with a high amount of pond presence can be very dynamic in space and time (Helbig et al. 2013 ). Those dynamics are likely to increase further as extreme weather events become more common while the tundra continues to warm (Wolfe et al. 2011 , Tei et al. 2020 ). Considering the dynamic nature of these systems is critical when assessing long-term trends in surface cover of tundra landscapes, including surface water change (e.g., Webb et al 2022 , Olthof et al. 2023 ), and their associated impacts on ecosystem function. The dynamic ponds further contribute to the complexity of greening and browning (Myers-Smith et al. 2020 , Beamish et al. 2020 ) and albedo trends (Plekhanova et al., 2022 ) in medium- to low-resolution satellite data. Only by accounting for the dynamic nature of some tundra landscapes will we be able to successfully estimate these trends and model the carbon and energy budget in the Arctic. Online methods Study sites The three study sites are located in Kytalyk National Park in Northeastern Siberia, Russia in proximity to the Chokurdakh Research Station (70.83°N, 147.49°E, Interact: https://eu-interact.org/field-sites/chokurdakh-scientific-tundra-station/). The climate in the region is cold, with an average annual temperature of -13.4 °C. Summers are mild with an average July temperature 10.3 °C. The mean annual precipitation is low (196 mm) with the majority falling in autumn and winter as snow, approximately 39% of the precipitation falling over summer (June-August). These climate data are based on the Chokurdakh weather station, WMO 21946, 28 km southeast of the Chokurdakh Research Station and averages refer to the period between 1981 and 2010 (Nauta et al. 2015). The landscapes in the region are underlain by Ice rich permafrost with an average active layer thickness of 42 cm and the soils are composed of silt with a shallow upper layer of peat (Grysko et al. 2021). The most common landscape types in the region include drained thaw lake beds (Alas depressions) and Yedoma ridges (Nitzbon et al. 2020). Ponds and patterned ground (ice-wege polygons) are common in the drained thaw beds (Nitzbon et al. 2020), where the vegetation is composed of dwarf shrubs ( Betula nana ), moss tundra and sedges (Iturrate-Garcia et al. 2016). Ponds are rare on the Yedoma ridges (Nitzbon et al. 2020), where the vegetation primarily consists of tussock sedges ( Eriophorum vaginatum ) (Iturrate-Garcia et al. 2016). Holocene deposits, thermokarst lakes (not drained) and river channels are also present in the region (Nitzbon et al. 2020), but we did not consider these in our study. Within the constraints of the drone data available, we chose the three study sites with an average dimension of 475 m x 435 m to capture the range of pond density occurring in the landscapes of the region. We selected a section of a drained thaw lake bed with a high pond occurrence (site “high”, area: 19.75 ha, centre: 70.8322°N 147.4779°E); a section of a drained thaw lake bed with a medium pond occurrence (site “medium”; area: 21.67 ha; centre: 70.8336°N 147.4951°E); and a section of a Yedoma ridge with no permanent pond occurrence as a control site (site “low”, area: 16.36 ha, centre: 70.8292°N 147.4638°E) (Fig. 1a). For further details see Supplementary Methods - “Determination of site boundaries” and Table S1. Drone data collection We collected Red-Green-Blue (RGB / true colour) drone imagery at a near annual basis from 2014 to 2021. For a detailed description of all flights see Table S2. In brief, we used eBee Classic and eBee X PPK drone systems (AgEagle, formerly SenseFly, Lausanne, Switzerland) equipped with either a SenseFly S110 RGB (AgEagle) or SenseFly S.O.D.A (AgEagle) sensor. Average ground sampling distances varied between 2.00 cm and 4.36 cm based on sensor and flight altitude (Table S2). We carried out all flights during summer between 11 July and 18 August of a given year. Repeat surveys within a year were available from 2019 for the “medium” (3x) and “high” sites (2x). No data was collected in 2015 and the 2016 data at the “low” site was of insufficient quality. In some instances, a merging of multiple flights within a year was required to achieve complete coverage for a site. Illumination conditions varied widely (sunny, overcast, time-of day, etc.) (Table S2). Drone data processing, geolocation and colour correction We used Pix4D Mapper (Pix4D S.A. Prilly, Switzerland) to generate and co-register the RGB raster mosaics and surface models (DSMs) for each drone survey. We generated the RGB mosaics and surface models using the “3D Maps” template in Pix4D and set the output resolution to 12 cm to obtain the mosaics and DSM with a uniform ground sampling distance (GSD). This distance is approximately three-times the maximum GSD in our dataset (Table S2). In four cases, the illumination of the individual raw images varied strongly, resulting in patchy mosaics. We addressed this issue using the colorCorrection function of the CRImage R package (Failmezger et al. 2024) to standardise the colour profiles of the individual images within the four mosaics prior processing (Table S2). High-precision geo-location using dGNSS post processing kinematics (PPK) was only possible for the 2020 and 2021 surveys using the newer eBee X systems. We therefore chose the 2021 mosaics at each site as the reference mosaic and co-registered all preceding mosaics in the time-series using manual tie-points in Pix4D. We verified the geolocation accuracy relative to the 2021 reference using five independent check points. The resulting estimate of the mean geolocation accuracy did not exceed 46 cm (three-times the 12 cm resolution of the mosaics) in all cases (Fig. S6). For further details see Supplementary Methods “Geolocation and accuracy assessment”. Finally, we standardised the colour profiles of all RGB drone mosaics relative to the mosaic from the 2017 survey at the “high” site using the colorCorrection function of the CRImage R package (Failmezger et al. 2024) to aid with the next surface water detection steps. A calibration using empirical lines (Wang and Myint 2015) was not possible as standardised targets were not deployed during the drone surveys, instead we used a surface water detection approach that worked on a similar empirical principle using a mosaic specific threshold based on annotated training data. Surface water detection We used a simple threshold in the Blue Chromatic Coordinate (BCC) determined from supervised training data to classify surface water in each RGB mosaic. We calculated the BCC as the ratio between the digital number (DN) of the blue band and the sum of all bands (BCC = DN blue / (DN red + DN green + DN blue ). We then produced a set of manually annotated reference polygons with two cover classes (“surface water” and “other”) for each drone mosaic (Table S3) and extracted the BCC for all pixels in these polygons. Next, we iterated through all potential BCC threshold values from 0 to 1 at 0.01 intervals and determined the classification accuracy for each threshold to separate surface water from other surfaces based on the annotated reference data. We then chose the threshold providing the highest overall classification accuracy for each drone mosaic (Table S4) and generated a corresponding raster file with two surface classes (“surface water” and “other”). For further details on training and classification see Supplementary Methods - Training annotations & accuracy assessment. Finally, we removed all clusters of surface water pixels less than 1 m 2 , which was the size of the smallest annotated pond in our training data. The mean pond size was 17.89 m 2 . Analysis of landscape-scale dynamics in surface water extent To assess the dynamics of surface water extent at the landscape-scale, we calculated the area of surface water for each site and survey-time combination, and then analysed its variability within the three annual time-series. To this end, we determined the coefficient of variation (CV, the standard deviation divided by the mean and multiplied by 100%) for each time-series, once with and once without the outlier year 2017. For those year and site combinations for which repeat surveys were available (2019 and “high” and “medium” sites) we used the mean value of all surveys for the given year when calculating the CV. Based on the few available repeat surveys, the variation in surface water area extent within 2019 relative to the mean of the whole time series (sd 2019 / mean all x 100) was small compared to coefficient of variation across the whole time series (sd all / mean all x 100): 3% (2019) vs 124% (all years) for the “high” site and 15% (2019) vs 102% (all years) for the “low” site. See Table S5 for a detailed listing of all surface water area and proportion values. We also tested for trends in all time-series using OLS and correlations with climate variables (Supplementary Materials - Surface water trend and climate analysis). Analysis of dynamics in individual pond area To assess the dynamics in surface water extent at the pond-scale, we identified all individual ponds in each time-series and then analysed the change in their surface water area. We identified the individual ponds by generating a binary composite of all surface water extent maps for each time-series. In this composite, a cell was set to one (water present) if water was detected in this cell at any point in the time-series. All other cells were set to NA. In this component of the analysis, we excluded the extreme outlier year 2017 as many ponds were connected during this year. We also excluded all repeat surveys within a year and only kept one observation per year. Here we excluded all surveys that were furthest from the median within-year timing of observations across all years (mosaic IDs: “cbh_2019”, “tlb_2019_a” and “tlb_2019_b”). We then obtained the polygon geometries of all clusters of surface water in the composite (hereafter “reference ponds”) and tracked their presence across the time-series. We tracked the reference ponds identified from the composite across the time-series by testing for geometric intersection with all ponds detected in a given year. As surface water at the “low” site was only detected in 2017, we removed this site from all subsequent analysis. Next, we assigned each remaining cluster of surface water an individual pond identifier and determined the number of years the pond was present in the time-series by testing for geometric intersection between polygons of the clusters derived from the composite with the polygons of all clusters of water detected in each year. To exclude temporary “puddles”, we kept only those clusters of surface water that were present at least three times in each time-series. Experimental manipulation of ponds had previously been carried out at the “medium” site; we identified and removed these ponds (total: 8) from all subsequent analyses. Finally, we calculated the pond area in each year as the total area of all clusters of surface water intersecting with the pond geometry derived from the composite (some ponds split), and then determined the CV as a measure of pond area variability, excluding the outlier year 2017. Analysis of drivers of pond-change We used the drone derived digital surface model to identify the presence of thermokarst and vegetation colonisation as drivers of pond surface area change. We restricted our analysis to changes in the relative surface elevation in the surroundings of each pond, as we were unable to precisely geolocate the three-dimensional point clouds in the absence of high-precision geolocation of image tags and ground control points. To this end, we cropped the drone DSMs for each year with a buffer area of 10 m around each pond and masked all areas that were detected as water in any point of the time-series (including 2017). We then standardised the DSMs relative to the minimum surface elevation within this area. For the detection of thermokarst as a driver of pond change, we calculated the mean drop in surface elevation in the DSMs between 2021 and 2014 within the pond area gained (Fig. S7a). For the detection of vegetation colonisation as a driver of pond change, we calculated the mean gain in surface elevation in the DSMs between 2021 and 2014 for the pond area lost (Fig. S7b). We calculated the average drop / gain in surface elevation for the pond area gained / lost rather than the total absolute change in volume as the ponds varied strongly in size. In other words, we standardised the volumetric differences by area gained / lost to obtain a comparable measure of change in surface volume across ponds independent of size. It is important to highlight two key contributors to uncertainty in the above-described method. First, standardising the DSMs by the minimum elevation within the buffered surroundings of each pond assumes that both the minimum elevation as well as the measurement error are constant across the time-series space. The assumption will likely not hold true in all cases and may introduce both random and systematic errors. Second, for the calculations of the volume differences we assumed that any area covered by water has a relative surface height of 0 m. This assumption may lead to a systematic under- or overestimation of the volume lost in the thermokarst and vegetation colonisation detection analyses respectively. Considering all of the above, we estimated that we are only able to reliably detect volume changes in the surface model that exceed 0.1 m 3 per m 2 , which is equivalent to a drop (or gain) in the surface elevation of 10 cm. Hence, we only assigned the detection of thermokarst or vegetation colonisation to a pond where the threshold was exceeded in the time-series. We empirically validated this threshold by manually inspecting the RGB / DSM time-series of all ponds. Software We conducted all data processing and statistical analysis in R ver 4.4.0 (R Core Team 2024). We used the sf package ver 1.0-16 (Pebesma and Bivand 2023, Pebesma 2018) for all geometrical operations; the terra package ver 1.7-78 (Hijmans 2024) for raster handling and calculations; the tidyverse ver 2.0.0.0 (Whickam et al. 2019) and pbapply ver 1.7-2 (Solymos and Zawdazki 2023) packages for general data wrangling and parallel processing; the ggplot2 ver 3.5.1 (Wickham 2016), cowplot ver 1.1.3 (Wilke 2024) and tidyterra ver 0.61 (Hernangómez 2023) packages for generating the maps and figures, and the gt package ver 0.11.0 (Iannone et al. 2024) for generating tables. References - Online methods Failmezger, H., Yuan, Y., Rueda, O., Markowetz, F., 2024. CRImage: CRImage a package to classify cells and calculate tumour cellularity. Grysko, R., Plekhanova, E., Oehri, J., Karsanaev, S.V., Maximov, T.C., Schaepman-Strub, G., 2021. Design of the tundra rainfall experiment (TRainEx) to simulate future summer precipitation scenarios. MethodsX 8, 101331. https://doi.org/10.1016/j.mex.2021.101331 Hernangómez, D., 2023. Using the tidyverse with terra objects: the tidyterra package. Journal of Open Source Software 8, 5751. https://doi.org/10.21105/joss.05751 Hijmans, R.J., 2024. terra: Spatial Data Analysis. https://CRAN.R-project.org/package=terra Iannone, R., Cheng, J., Schloerke, B., Hughes, E., Lauer, A., Seo, J., Brevoort, K., Roy, O., 2024. gt: Easily Create Presentation-Ready Display Tables. https://CRAN.R-project.org/package=gt Iturrate-Garcia, M., O’Brien, M.J., Khitun, O., Abiven, S., Niklaus, P.A., Schaepman-Strub, G., 2016. Interactive effects between plant functional types and soil factors on tundra species diversity and community composition. Ecology and Evolution 6, 8126–8137. https://doi.org/10.1002/ece3.2548 Nauta, A.L., Heijmans, M.M.P.D., Blok, D., Limpens, J., Elberling, B., Gallagher, A., Li, B., Petrov, R.E., Maximov, T.C., van Huissteden, J., Berendse, F., 2015. Permafrost collapse after shrub removal shifts tundra ecosystem to a methane source. Nature Clim Change 5, 67–70. https://doi.org/10.1038/nclimate2446 Nitzbon, J., Westermann, S., Langer, M., Martin, L.C.P., Strauss, J., Laboor, S., Boike, J., 2020. Fast response of cold ice-rich permafrost in northeast Siberia to a warming climate. Nat Commun 11, 2201. https://doi.org/10.1038/s41467-020-15725-8 R Core Team, 2024. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Pebesma, E., 2018. Simple Features for R: Standardized Support for Spatial Vector Data. The R Journal 10, 439–446. https://doi.org/10.32614/RJ-2018-009 Pebesma, E., Bivand, R., 2023. Spatial Data Science: With Applications in R. Chapman and Hall/CRC, New York. https://doi.org/10.1201/9780429459016 Solymos, P., Zawadzki, Z., 2023. pbapply: Adding Progress Bar to “*apply” Functions. https://CRAN.R-project.org/package=pbapply Wang, C., Myint, S.W., 2015. A Simplified Empirical Line Method of Radiometric Calibration for Small Unmanned Aircraft Systems-Based Remote Sensing. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 8, 1876–1885. https://doi.org/10.1109/JSTARS.2015.2422716 Wickham, H., Averick, M., Bryan, J., Chang, W., McGowan, L.D., François, R., Grolemund, G., Hayes, A., Henry, L., Hester, J., Kuhn, M., Pedersen, T.L., Miller, E., Bache, S.M., Müller, K., Ooms, J., Robinson, D., Seidel, D.P., Spinu, V., Takahashi, K., Vaughan, D., Wilke, C., Woo, K., Yutani, H., 2019. Welcome to the tidyverse. Journal of Open Source Software 4, 1686. https://doi.org/10.21105/joss.01686 Wickham, H., 2016. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. https://ggplot2.tidyverse.org Wilke, C.O., 2024. cowplot: Streamlined Plot Theme and Plot Annotations for “ggplot2.” https://CRAN.R-project.org/package=cowplot Declarations Acknowledgements We would like to thank all drone-pilots and field team members who contributed to the data collection during the field seasons 2014 – 2021, including Inge Grünberg (nee Juszak), Maitane Iturrate-Garcia and Vitalii Zemlianskii. This study was supported by the Swiss National Science Foundation (grant no. 178753) and the University Research Priority Program on Global Change and Biodiversity of the University of Zurich. Code and Data availability All code, tabular and geometric data required to regenerate the analyses are available via the following GitHub repository: https://github.com/jakobjassmann/pond_project The raster data (RGB drone mosaics and DSMs) are also required and are available on Zenodo: https://doi.org/10.5281/zenodo.13992090 The drone RGB rasters, surface water classifications and individual pond time-series have been visualised in a Web-GIS hosted on AWS: https://pondproject.s3.eu-central-1.amazonaws.com/index.html References Andresen, C.G., Lougheed, V.L., 2015. Disappearing Arctic tundra ponds: Fine-scale analysis of surface hydrology in drained thaw lake basins over a 65 year period (1948–2013). Journal of Geophysical Research: Biogeosciences 120, 466–479. https://doi.org/10.1002/2014JG002778 Bartosiewicz, M., Przytulska, A., Lapierre, J.-F., Laurion, I., Lehmann, M.F., Maranger, R., 2019. Hot tops, cold bottoms: Synergistic climate warming and shielding effects increase carbon burial in lakes. Limnology and Oceanography Letters 4, 132–144. https://doi.org/10.1002/lol2.10117 Beamish, A., Raynolds, M.K., Epstein, H., Frost, G.V., Macander, M.J., Bergstedt, H., Bartsch, A., Kruse, S., Miles, V., Tanis, C.M., Heim, B., Fuchs, M., Chabrillat, S., Shevtsova, I., Verdonen, M., Wagner, J., 2020. Recent trends and remaining challenges for optical remote sensing of Arctic tundra vegetation: A review and outlook. Remote Sensing of Environment 246, 111872. https://doi.org/10.1016/j.rse.2020.111872 Blackburn-Desbiens, P., Grosbois, G., Power, M., Culp, J., Rautio, M., 2023. Integrating hydrological connectivity and zooplankton composition in Arctic ponds and lakes. Freshwater Biology 68, 2131–2150. https://doi.org/10.1111/fwb.14181 Corbeil-Robitaille, M.-Z., Duchesne, É., Fortier, D., Kinnard, C., Bêty, J., 2024. Linking geomorphological processes and wildlife microhabitat selection: nesting birds select refuges generated by permafrost degradation in the Arctic. Biogeosciences 21, 3401–3423. https://doi.org/10.5194/bg-21-3401-2024 Del Giorgio, P.A., Cole, J.J., 1998. BACTERIAL GROWTH EFFICIENCY IN NATURAL AQUATIC SYSTEMS. Annual Review of Ecology, Evolution, and Systematics 29, 503–541. https://doi.org/10.1146/annurev.ecolsys.29.1.503 Gurney, K.E.B., Koch, J.C., Schmutz, J.A., Schmidt, J.H., Wipfli, M.S., 2022. In hot water? Patterns of macroinvertebrate abundance in Arctic thaw ponds and relationships with environmental variables. Freshwater Biology n/a. https://doi.org/10.1111/fwb.13978 Helbig, M., Boike, J., Langer, M., Schreiber, P., Runkle, B.R.K., Kutzbach, L., 2013. Spatial and seasonal variability of polygonal tundra water balance: Lena River Delta, northern Siberia (Russia). Hydrogeol J 21, 133–147. https://doi.org/10.1007/s10040-012-0933-4 Hinks, A.E., Cole, E.F., Daniels, K.J., Wilkin, T.A., Nakagawa, S., Sheldon, B.C., 2015. Scale-Dependent Phenological Synchrony between Songbirds and Their Caterpillar Food Source. The American Naturalist 186, 84–97. https://doi.org/10.1086/681572 Juszak, I., Iturrate-Garcia, M., Gastellu-Etchegorry, J.-P., Schaepman, M.E., Maximov, T.C., Schaepman-Strub, G., 2017. Drivers of shortwave radiation fluxes in Arctic tundra across scales. Remote Sensing of Environment 193, 86–102. https://doi.org/10.1016/j.rse.2017.02.017 Kanevskiy, M., Shur, Y., Jorgenson, T., Brown, D.R.N., Moskalenko, N., Brown, J., Walker, D.A., Raynolds, M.K., Buchhorn, M., 2017. Degradation and stabilization of ice wedges: Implications for assessing risk of thermokarst in northern Alaska. Geomorphology 297, 20–42. https://doi.org/10.1016/j.geomorph.2017.09.001 Knoblauch, C., Spott, O., Evgrafova, S., Kutzbach, L., Pfeiffer, E.-M., 2015. Regulation of methane production, oxidation, and emission by vascular plants and bryophytes in ponds of the northeast Siberian polygonal tundra. Journal of Geophysical Research: Biogeosciences 120, 2525–2541. https://doi.org/10.1002/2015JG003053 Lachenbruch, A.H., 1962. Mechanics of thermal contraction cracks and ice-wedge polygons in permafrost. Geological Society of America. Special Papers 70, 69. Langer, M., Westermann, S., Muster, S., Piel, K., Boike, J., 2011. The surface energy balance of a polygonal tundra site in northern Siberia – Part 1: Spring to fall. The Cryosphere 5, 151–171. https://doi.org/10.5194/tc-5-151-2011 Lara, M.J., McGuire, A.D., Euskirchen, E.S., Tweedie, C.E., Hinkel, K.M., Skurikhin, A.N., Romanovsky, V.E., Grosse, G., Bolton, W.R., Genet, H., 2015. Polygonal tundra geomorphological change in response to warming alters future CO2 and CH4 flux on the Barrow Peninsula. Global Change Biology 21, 1634–1651. https://doi.org/10.1111/gcb.12757 Lara, M.J., McGuire, A.D., Euskirchen, E.S., Genet, H., Yi, S., Rutter, R., Iversen, C., Sloan, V., Wullschleger, S.D., 2020. Local-scale Arctic tundra heterogeneity affects regional-scale carbon dynamics. Nat Commun 11, 4925. https://doi.org/10.1038/s41467-020-18768-z Le Moigne, A., Bartosiewicz, M., Schaepman-Strub, G., Abiven, S., Pernthaler, J., 2020. The biogeochemical variability of Arctic thermokarst ponds is reflected by stochastic and niche-driven microbial community assembly processes. Environmental Microbiology 22, 4847–4862. https://doi.org/10.1111/1462-2920.15260 Li, B., Heijmans, M.M.P.D., Blok, D., Wang, P., Karsanaev, S.V., Maximov, T.C., van Huissteden, J., Berendse, F., 2017. Thaw pond development and initial vegetation succession in experimental plots at a Siberian lowland tundra site. Plant Soil 420, 147–162. https://doi.org/10.1007/s11104-017-3369-8 Liljedahl, A.K., Boike, J., Daanen, R.P., Fedorov, A.N., Frost, G.V., Grosse, G., Hinzman, L.D., Iijma, Y., Jorgenson, J.C., Matveyeva, N., Necsoiu, M., Raynolds, M.K., Romanovsky, V.E., Schulla, J., Tape, K.D., Walker, D.A., Wilson, C.J., Yabuki, H., Zona, D., 2016. Pan-Arctic ice-wedge degradation in warming permafrost and its influence on tundra hydrology. Nature Geosci 9, 312–318. https://doi.org/10.1038/ngeo2674 Luoto, M., Seppälä, M., 2003. Thermokarst ponds as indicators of the former distribution of palsas in Finnish Lapland. Permafrost and Periglacial Processes 14, 19–27. https://doi.org/10.1002/ppp.441 Magnússon, R.Í., Limpens, J., Huissteden, J., Kleijn, D., Maximov, T.C., Rotbarth, R., Sass-Klaassen, U., Heijmans, M.M.P.D., 2020. Rapid Vegetation Succession and Coupled Permafrost Dynamics in Arctic Thaw Ponds in the Siberian Lowland Tundra. J. Geophys. Res. Biogeosci. 125. https://doi.org/10.1029/2019JG005618 Magnússon, R.Í., Limpens, J., Kleijn, D., van Huissteden, K., Maximov, T.C., Lobry, S., Heijmans, M.M.P.D., 2021. Shrub decline and expansion of wetland vegetation revealed by very high resolution land cover change detection in the Siberian lowland tundra. Science of The Total Environment 782, 146877. https://doi.org/10.1016/j.scitotenv.2021.146877 Melton, J.R., Wania, R., Hodson, E.L., Poulter, B., Ringeval, B., Spahni, R., Bohn, T., Avis, C.A., Beerling, D.J., Chen, G., Eliseev, A.V., Denisov, S.N., Hopcroft, P.O., Lettenmaier, D.P., Riley, W.J., Singarayer, J.S., Subin, Z.M., Tian, H., Zürcher, S., Brovkin, V., van Bodegom, P.M., Kleinen, T., Yu, Z.C., Kaplan, J.O., 2013. Present state of global wetland extent and wetland methane modelling: conclusions from a model inter-comparison project (WETCHIMP). Biogeosciences 10, 753–788. https://doi.org/10.5194/bg-10-753-2013 Morgenstern, A., Ulrich, M., Günther, F., Roessler, S., Fedorova, I.V., Rudaya, N.A., Wetterich, S., Boike, J., Schirrmeister, L., 2013. Evolution of thermokarst in East Siberian ice-rich permafrost: A case study. Geomorphology 201, 363–379. https://doi.org/10.1016/j.geomorph.2013.07.011 Mullen, A.L., Watts, J.D., Rogers, B.M., Carroll, M.L., Elder, C.D., Noomah, J., Williams, Z., Caraballo-Vega, J.A., Bredder, A., Rickenbaugh, E., Levenson, E., Cooley, S.W., Hung, J.K.Y., Fiske, G., Potter, S., Yang, Y., Miller, C.E., Natali, S.M., Douglas, T.A., Kyzivat, E.D., 2023. Using High-Resolution Satellite Imagery and Deep Learning to Track Dynamic Seasonality in Small Water Bodies. Geophysical Research Letters 50, e2022GL102327. https://doi.org/10.1029/2022GL102327 Muster, S., Langer, M., Heim, B., Westermann, S., Boike, J., 2012. Subpixel heterogeneity of ice-wedge polygonal tundra: a multi-scale analysis of land cover and evapotranspiration in the Lena River Delta, Siberia. Tellus B: Chemical and Physical Meteorology 64, 17301. https://doi.org/10.3402/tellusb.v64i0.17301 Myers-Smith, I.H., Kerby, J.T., Phoenix, G.K., Bjerke, J.W., Epstein, H.E., Assmann, J.J., John, C., Andreu-Hayles, L., Angers-Blondin, S., Beck, P.S.A., Berner, L.T., Bhatt, U.S., Bjorkman, A.D., Blok, D., Bryn, A., Christiansen, C.T., Cornelissen, J.H.C., Cunliffe, A.M., Elmendorf, S.C., Forbes, B.C., Goetz, S.J., Hollister, R.D., de Jong, R., Loranty, M.M., Macias-Fauria, M., Maseyk, K., Normand, S., Olofsson, J., Parker, T.C., Parmentier, F.-J.W., Post, E., Schaepman-Strub, G., Stordal, F., Sullivan, P.F., Thomas, H.J.D., Tømmervik, H., Treharne, R., Tweedie, C.E., Walker, D.A., Wilmking, M., Wipf, S., 2020. Complexity revealed in the greening of the Arctic. Nat. Clim. Chang. 10, 106–117. https://doi.org/10.1038/s41558-019-0688-1 Negandhi, K., Laurion, I., Whiticar, M.J., Galand, P.E., Xu, X., Lovejoy, C., 2013. Small Thaw Ponds: An Unaccounted Source of Methane in the Canadian High Arctic. PLOS ONE 8, e78204. https://doi.org/10.1371/journal.pone.0078204 Negandhi, K., Laurion, I., Lovejoy, C., 2016. Temperature effects on net greenhouse gas production and bacterial communities in arctic thaw ponds. FEMS Microbiology Ecology 92, fiw117. https://doi.org/10.1093/femsec/fiw117 Nitzbon, J., Westermann, S., Langer, M., Martin, L.C.P., Strauss, J., Laboor, S., Boike, J., 2020. Fast response of cold ice-rich permafrost in northeast Siberia to a warming climate. Nat Commun 11, 2201. https://doi.org/10.1038/s41467-020-15725-8 Olefeldt, D., Goswami, S., Grosse, G., Hayes, D., Hugelius, G., Kuhry, P., McGuire, A.D., Romanovsky, V.E., Sannel, A.B.K., Schuur, E. a. G., Turetsky, M.R., 2016. Circumpolar distribution and carbon storage of thermokarst landscapes. Nat Commun 7, 13043. https://doi.org/10.1038/ncomms13043 Olthof, I., Fraser, R.H., van der Sluijs, J., Travers-Smith, H., 2023. Detecting long-term Arctic surface water changes. Nat. Clim. Chang. 13, 1191–1193. https://doi.org/10.1038/s41558-023-01836-9 Payette, S., Delwaide, A., Caccianiga, M., Beauchemin, M., 2004. Accelerated thawing of subarctic peatland permafrost over the last 50 years. Geophysical Research Letters 31. https://doi.org/10.1029/2004GL020358 Peura, S., Wauthy, M., Simone, D., Eiler, A., Einarsdóttir, K., Rautio, M., Bertilsson, S., 2020. Ontogenic succession of thermokarst thaw ponds is linked to dissolved organic matter quality and microbial degradation potential. Limnology and Oceanography 65, S248–S263. https://doi.org/10.1002/lno.11349 Pienitz, R., Doran, P.T., Lamoureux, S.F., 2008. Origin and geomorphology of lakes in the polar regions., in: Polar Lakes and Rivers: Limnology of Arctic and Antarctic Aquatic Ecosystems. Oxford University Press, Oxford, England, pp. 25–41. Planet Labs PBC, 2022. Planet Application Program Interface: In Space for Life on Earth. https://api.planet.com Plekhanova, E., Kim, J.-S., Oehri, J., Erb, A., Schaaf, C., Schaepman-Strub, G., 2022. Mid-summer snow-free albedo across the Arctic tundra was mostly stable or increased over the past two decades. Environ. Res. Lett. 17, 124026. https://doi.org/10.1088/1748-9326/aca5a1 Prėskienis, V., Fortier, D., Douglas, P.M.J., Rautio, M., Laurion, I., 2024. Permafrost degradation and soil erosion as drivers of greenhouse gas emissions from tundra ponds. Environ. Res. Lett. 19, 014072. https://doi.org/10.1088/1748-9326/ad1433 Rautio, M., Dufresne, F., Laurion, I., Bonilla, S., Vincent, W.F., Christoffersen, K.S., 2011. Shallow freshwater ecosystems of the circumpolar Arctic. Écoscience 18, 204–222. https://doi.org/10.2980/18-3-3463 Schirrmeister, L., Pestryakova, L., Wetterich, S., Tumskoy, V., 2012. Joint Russian-German polygon project: East Siberia 2011–2014 ; the expedition Kytalyk 2011 [WWW Document]. Berichte zur Polar- und Meeresforschung = Reports on polar and marine research. https://doi.org/10.2312/BzPM_0653_2012 Smol, J.P., Douglas, M.S.V., 2007. Crossing the final ecological threshold in high Arctic ponds. Proceedings of the National Academy of Sciences 104, 12395–12397. https://doi.org/10.1073/pnas.0702777104 Tei, S., Morozumi, T., Nagai, S., Takano, S., Sugimoto, A., Shingubara, R., Fan, R., Fedorov, A., Gavrilyeva, T., Tananaev, N., Maximov, T., 2020. An extreme flood caused by a heavy snowfall over the Indigirka River basin in Northeastern Siberia. Hydrological Processes 34, 522–537. https://doi.org/10.1002/hyp.13601 Treat, C.C., Marushchak, M.E., Voigt, C., Zhang, Y., Tan, Z., Zhuang, Q., Virtanen, T.A., Räsänen, A., Biasi, C., Hugelius, G., Kaverin, D., Miller, P.A., Stendel, M., Romanovsky, V., Rivkin, F., Martikainen, P.J., Shurpali, N.J., 2018. Tundra landscape heterogeneity, not interannual variability, controls the decadal regional carbon balance in the Western Russian Arctic. Global Change Biology 24, 5188–5204. https://doi.org/10.1111/gcb.14421 Walter, K.M., Zimov, S.A., Chanton, J.P., Verbyla, D., Chapin, F.S., 2006. Methane bubbling from Siberian thaw lakes as a positive feedback to climate warming. Nature 443, 71–75. https://doi.org/10.1038/nature05040 Wang, P., de Jager, J., Nauta, A., van Huissteden, J., Trofim, M.C., Limpens, J., 2019. Exploring near-surface ground ice distribution in patterned-ground tundra: correlations with topography, soil and vegetation. Plant Soil 444, 251–265. https://doi.org/10.1007/s11104-019-04276-7 Wang, G., Peng, Y., Chen, L., Abbott, B.W., Ciais, P., Kang, L., Liu, Y., Li, Q., Peñuelas, J., Qin, S., Smith, P., Song, Y., Strauss, J., Wang, J., Wei, B., Yu, J., Zhang, D., Yang, Y., 2024. Enhanced response of soil respiration to experimental warming upon thermokarst formation. Nat. Geosci. 1–7. https://doi.org/10.1038/s41561-024-01440-2 Wauthy, M., Rautio, M., Christoffersen, K.S., Forsström, L., Laurion, I., Mariash, H.L., Peura, S., Vincent, W.F., 2018. Increasing dominance of terrigenous organic matter in circumpolar freshwaters due to permafrost thaw. Limnology and Oceanography Letters 3, 186–198. https://doi.org/10.1002/lol2.10063 Webb, E.E., Liljedahl, A.K., Cordeiro, J.A., Loranty, M.M., Witharana, C., Lichstein, J.W., 2022. Permafrost thaw drives surface water decline across lake-rich regions of the Arctic. Nat. Clim. Chang. 1–6. https://doi.org/10.1038/s41558-022-01455-w Webb, E.E., Liljedahl, A.K., 2023. Diminishing lake area across the northern permafrost zone. Nat. Geosci. 16, 202–209. https://doi.org/10.1038/s41561-023-01128-z Webb, E.E., Liljedahl, A.K., Loranty, M.M., Witharana, C., Lichstein, J.W., 2023. Reply to: Detecting long-term Arctic surface water changes. Nat. Clim. Chang. 13, 1194–1196. https://doi.org/10.1038/s41558-023-01837-8 Wik, M., Varner, R.K., Anthony, K.W., MacIntyre, S., Bastviken, D., 2016. Climate-sensitive northern lakes and ponds are critical components of methane release. Nature Geosci 9, 99–105. https://doi.org/10.1038/ngeo2578 Wolfe, B.B., Light, E.M., Macrae, M.L., Hall, R.I., Eichel, K., Jasechko, S., White, J., Fishback, L., Edwards, T.W.D., 2011. Divergent hydrological responses to 20th century climate change in shallow tundra ponds, western Hudson Bay Lowlands. Geophysical Research Letters 38. https://doi.org/10.1029/2011GL049766 Woo, M.-K., Guan, X.J., 2006. Hydrological connectivity and seasonal storage change of tundra ponds in a polar oasis environment, Canadian High Arctic. Permafrost and Periglacial Processes 17, 309–323. https://doi.org/10.1002/ppp.565 Wrona, F.J., Johansson, M., Culp, J.M., Jenkins, A., Mård, J., Myers-Smith, I.H., Prowse, T.D., Vincent, W.F., Wookey, P.A., 2016. Transitions in Arctic ecosystems: Ecological implications of a changing hydrological regime. Journal of Geophysical Research: Biogeosciences 121, 650–674. https://doi.org/10.1002/2015JG003133 Additional Declarations There is NO Competing Interest. Supplementary Files AssmannetalTundraPondsSupplementaryMaterials.pdf Supplementary Materials Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5581925","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":387191784,"identity":"8d736115-2f52-4e6b-996d-e85a8f8b2897","order_by":0,"name":"Jakob Assmann","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA30lEQVRIiWNgGAWjYBACxgY2BmYEt4KBwYCBgY0ULWeI0AKSR2hhbCNCC3P7scTHBQx29vLTDj97XDjvsL05A/OzB3gd1pN22HgGQ3Lihttp5sYztx1mtmxgMzfA75f0NmkeBuYEA+kEM2nebYfZDA7wsEng1dL/HKSl3l5+dvo3ad45h3kIa5mRdgyo5TBjw+0coC0NhyWI0PIs2XiGwXGgX3LKpHmOpRsYHGYzw6vFsD/N8HFBRTXIYdukeWqs7Q2ONz/Dr6UBRKKEEDNWhQggT0B+FIyCUTAKRgEDAwAc5D9L0mdcOgAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0002-3492-8419","institution":"University of Zurich","correspondingAuthor":true,"prefix":"","firstName":"Jakob","middleName":"","lastName":"Assmann","suffix":""},{"id":387191785,"identity":"e6983492-be4e-4986-b877-968df0a12da5","order_by":1,"name":"Cengiz Akandil","email":"","orcid":"","institution":"University of Zurich","correspondingAuthor":false,"prefix":"","firstName":"Cengiz","middleName":"","lastName":"Akandil","suffix":""},{"id":387191786,"identity":"bb7133c6-0d6a-42e1-8fa7-d98bf546f6bb","order_by":2,"name":"Elena Plekhanova","email":"","orcid":"","institution":"Swiss Federal Research Institute WSL","correspondingAuthor":false,"prefix":"","firstName":"Elena","middleName":"","lastName":"Plekhanova","suffix":""},{"id":387191787,"identity":"c28e0513-a525-4caa-9741-8ef4a5851228","order_by":3,"name":"Alizée Le Moigne","email":"","orcid":"","institution":"Institut National de la Recherche Scientifique","correspondingAuthor":false,"prefix":"","firstName":"Alizée","middleName":"Le","lastName":"Moigne","suffix":""},{"id":387191788,"identity":"9943c183-c063-4cfb-b4d7-f798ee43b43b","order_by":4,"name":"Sergey Karsanaev","email":"","orcid":"","institution":"Institute for Biological Problems of the Cryolithozone Siberian Branch Russian Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Sergey","middleName":"","lastName":"Karsanaev","suffix":""},{"id":387191789,"identity":"58223e69-c009-4532-a4f7-18756f5dead6","order_by":5,"name":"Trofim Maximov","email":"","orcid":"","institution":"Institute for Biological Problems of Cryolithozone SB RAS","correspondingAuthor":false,"prefix":"","firstName":"Trofim","middleName":"","lastName":"Maximov","suffix":""},{"id":387191790,"identity":"cb216c55-b66b-4b75-acb9-97c911bf129b","order_by":6,"name":"Gabriela Schaepman-Strub","email":"","orcid":"https://orcid.org/0000-0002-4069-1884","institution":"University of Zurich","correspondingAuthor":false,"prefix":"","firstName":"Gabriela","middleName":"","lastName":"Schaepman-Strub","suffix":""}],"badges":[],"createdAt":"2024-12-04 18:35:41","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5581925/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5581925/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":74983520,"identity":"032b6ed2-8fd6-4a04-a4c2-207708e08624","added_by":"auto","created_at":"2025-01-29 05:29:48","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":5975634,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSurface water extent in Northeastern Siberian landscapes can be highly dynamic.\u003c/strong\u003e (a) Map showing the locations of the three Arctic tundra landscapes with high, medium and (very) low pond occurrences in Northeastern Siberia. (b) Time-series of surface water proportion for each site based on classifications of true colour drone-imagery indicating high variability. Dashed lines in the graph indicate gaps where no drone data was available. Coefficients of variation (CV) for each time-series are given in insert on the top-right, values in parentheses represent the CV with outlier year 2017 removed. (c) Surface water maps for each study site, showing the surface water (light blue) on top of the desaturated true-colour drone imagery. No data was collected in 2015 across all sites, and the data from 2016 for the “low” site was of insufficient quality for accurate geolocation. Satellite Imagery in (a): © Planet Labs 2022 (Planet Labs PBC 2022).\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-5581925/v1/e4af6638bee38ed8d859363d.png"},{"id":74982006,"identity":"b549a983-3838-4f7c-848a-2baeb6a4e8b9","added_by":"auto","created_at":"2025-01-29 05:05:48","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":88931,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePonds were short-lived and fluctuated strongly in size. \u003c/strong\u003eThe majority of ponds occurred for less than six years in the time-series and individual ponds showed a high variability in their surface area across the time-series. (a) Distribution of the number of years each pond was detected in the time-series. The dark blue line indicates the proportional split between those ponds present less than six years and those present “continuously” for six or more years (allowing for one year of failed detection). (b) Distribution of the coefficient of variation (CV) of the pond’s area across each time-series. The dark blue line indicates the mean CV value for all ponds. CV values were calculated excluding the outlier year 2017. Please note: the panels only show statistics for the pond-time series where the pond occurred in more than three years (our threshold of detection). No ponds fulfilled that condition at the “low” site.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-5581925/v1/996854b0245101a6c5a03763.png"},{"id":74982010,"identity":"e958dc33-eac4-4e34-9763-29908a5895e3","added_by":"auto","created_at":"2025-01-29 05:05:48","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1052808,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePonds were short-lived and fluctuated strongly in size. \u003c/strong\u003eThe majority of ponds occurred for less than six years in the time-series and individual ponds showed a high variability in their surface area across the time-series. (a) Distribution of the number of years each pond was detected in the time-series. The dark blue line indicates the proportional split between those ponds present less than six years and those present “continuously” for six or more years (allowing for one year of failed detection). (b) Distribution of the coefficient of variation (CV) of the pond’s area across each time-series. The dark blue line indicates the mean CV value for all ponds. CV values were calculated excluding the outlier year 2017. Please note: the panels only show statistics for the pond-time series where the pond occurred in more than three years (our threshold of detection). No ponds fulfilled that condition at the “low” site.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-5581925/v1/38a366bd91076535d20470a5.png"},{"id":74982015,"identity":"1e562ddc-ce8f-43fa-b77e-997510a7dbef","added_by":"auto","created_at":"2025-01-29 05:05:48","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1117275,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVegetation colonisation contributes to shrinking or complete infilling of ponds. \u003c/strong\u003eExample for a pond (“cbh_049”) that showed clear signs of vegetation colonisation in the pond area between 2014 and 2021, observed as a gain in elevation in the drone-derived digital surface model (DSM). Upper row: true-colour (RGB) images with the outline of the pond (white / orange) in each year based on our classifications. Lower row: relative surface elevation derived from the DSM standardised to the minimum of the absolute elevation of the area shown; surface water extent in the given year (blue) and pond outline at the start of the time-series (white). (b) Distribution of the mean gain in surface elevation for the land area gained (2014 vs 2021) for all pond time-series. (c) Surface height profiles of the transect in (a) for the years 2014 and 2021, showing the gain in surface elevation due to the vegetation colonisation of pond “cbh_049” between the two years. The mean gain in surface elevation for the land area gained in pond “cbh_049” was 0.15 m.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-5581925/v1/f1d4e9c71f9ec8ea927f668b.png"},{"id":81489748,"identity":"114ec950-fa4a-46f3-8569-c32151b53424","added_by":"auto","created_at":"2025-04-27 20:13:23","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":11061935,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5581925/v1/c2e0580e-d98a-417a-bac7-5373611a0afa.pdf"},{"id":74982007,"identity":"d404824d-67b0-4069-ba1a-6ad4aef75fef","added_by":"auto","created_at":"2025-01-29 05:05:48","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":1000816,"visible":true,"origin":"","legend":"Supplementary Materials","description":"","filename":"AssmannetalTundraPondsSupplementaryMaterials.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5581925/v1/4d1583df89bd4f9f00f22cde.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Freshwater ponds create highly dynamic Arctic tundra landscapes","fulltext":[{"header":"Introduction","content":"\u003cp\u003eGreenhouse gas emissions from tundra ponds and permafrost thaw (thermokarst) provide important feedbacks to global warming (Wik et al. \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2016\u003c/span\u003e, Olefeldt 2016). Ponds are a common feature of many Arctic landscapes, where they provide critical habitat for breeding birds (Corbeil-Robitaille et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) and form an important source of methane emissions (Wik et al. \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Methane emissions from ponds are mainly driven by microbial activity (Del Giorgio and del Cole, 1998) which is thought to increase with warming (Negandhi et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Bartosiewicz et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). While permafrost thaw itself may release large quantities of carbon (Nitzbon et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2020\u003c/span\u003e, Wang et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), it can also be a driver of pond formation, with newly formed ponds having potentially elevated emission rates (Peura et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2020\u003c/span\u003e, Prėskienis et al. \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). However, large uncertainties remain in the accounting and modelling of carbon fluxes in Arctic wetlands, especially in relation to small ponds (Negandhi et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2013\u003c/span\u003e, Melton et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2013\u003c/span\u003e, Muster et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2012\u003c/span\u003e, Wik et al. \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) which are difficult to map with data exceeding metre-scale resolution (Muster et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Further, ponds are an important factor of tundra land surface albedo and energy fluxes during summer (Juszak et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2017\u003c/span\u003e, Langer et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Studying tundra ponds and how they change at fine spatial scales is therefore a key step to better understand the potential feedbacks to global warming.\u003c/p\u003e \u003cp\u003eThere is currently no agreement on the direction of change in surface water area in response to warming across the Arctic. Some research indicates that water bodies, including small ponds, are decreasing across the Arctic (Webb et al. \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2022\u003c/span\u003e, Wik et al. \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2016\u003c/span\u003e, Andresen and Lougheed \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Other studies highlight the large methodological uncertainties in large-scale analyses of Arctic surface water change (Olthof et al. \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2023\u003c/span\u003e, Webb et al. \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Additionally, there is considerable regional variation in trends (Webb and Liljedahl \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), especially for small water bodies (Payette, 2004, Walter et al. \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2006\u003c/span\u003e, Wolfe 2011, Magn\u0026uacute;sson et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). For instance, increases in area and number of ponds has been reported for Northeastern Siberia (Magn\u0026uacute;sson et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), while declines have been reported in Alaska (Andresen and Lougheed \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). The variation in regional trends for large water bodies has been associated with large-scale patterns in the distribution of continuous vs. discontinuous permafrost (Webb and Liljedahl \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), but too little data are available for small water bodies.\u003c/p\u003e \u003cp\u003eA high landscape heterogeneity and water bodies with dynamic extents make it difficult to detect surface water changes in the Arctic tundra (Muster et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2012\u003c/span\u003e, Webb and Liljedahl \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), and thus estimate the associated carbon fluxes at landscape scales (Muster et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2012\u003c/span\u003e, Hinks et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2015\u003c/span\u003e, Treat et al. \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). The geomorphological processes that shape Arctic tundra landscapes can be very complex (Lara et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) resulting in a highly heterogeneous land cover at metre to decametre scales (Muster et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2012\u003c/span\u003e, Lara et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2015\u003c/span\u003e, Morgenstern et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Tundra ponds are an important contributor to this metre-scale heterogeneity (Muster et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2012\u003c/span\u003e), with pond development being driven by various fine-scale abiotic and biotic processes (Pienitz et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2008\u003c/span\u003e), including freeze-thaw cycles of ice wedges (Lachenbruch \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e1962\u003c/span\u003e), permafrost thaw that creates depressions (Rautio et al. \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2011\u003c/span\u003e, Luoto and Sepp\u0026auml;l\u0026auml; \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2003\u003c/span\u003e) and vegetation colonisation by sedges and mosses (Magn\u0026uacute;sson et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Some of these processes may occur on sub-decadal timescales (Magn\u0026uacute;sson et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) and could, combined with fluctuations in the main drivers the water budget of the tundra (precipitation, evaporation and lateral-flow) (Helbig et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2013\u003c/span\u003e), potentially cause rapid changes in the surface water area of the ponds. High-resolution (below 4\u0026ndash;5 m) and high frequency spatial data are therefore needed to map tundra ponds at the landscape scale (Muster et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2012\u003c/span\u003e), but these are often not available over long time frames (Webb and Liljedahl \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eHere, we assess the change in surface water extent at the landscape-scale and for individual ponds in the tundra of Northeastern Siberia. We also test for evidence of thermokarst and vegetation colonisation as drivers of pond area change. To this end, we collected time-series of high-resolution (12 cm) drone imagery and digital surface elevation models (DSMs) at three study sites in the Kytalyk National Park for the time period 2014 to 2021 (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Two study sites are in a drained thaw lake basin with high and medium surface water occurrence (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea) and one site is on a Yedoma ridge with low surface water occurrence functioning as a control. Our study thus covers two of the major landscape types in the region (Nitzbon et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2020\u003c/span\u003e, Schirrmeister et al. \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Next, we classified surface water in each image using a simple threshold in the blue chromatic coordinate (BCC) informed by annotated training data, and identified all individual ponds that exceeded 1 m\u003csup\u003e2\u003c/sup\u003e and were present for at least 3 years (465 in total; 357, 108, 0 respectively for the \u0026ldquo;high\u0026rdquo;, \u0026ldquo;medium\u0026rdquo; and \u0026ldquo;low\u0026rdquo; surface water occurrence sites). We then assessed how surface water extent changed at the landscape- and pond-level throughout each time-series. Finally, we tested whether thermokarst and vegetation colonisation contributed to individual pond change, by estimating the drop or gain in surface elevation in the drone-derived DSMs associated with the change. Our study thus allows us to gain novel insights into the rapid nature of changes in surface water and their ecological drivers in the understudied tundra landscapes of Northeastern Siberia.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eSurface water extent in Northeastern Siberian tundra landscapes is highly dynamic\u003c/h2\u003e \u003cp\u003eThe proportion of surface water detected was highly dynamic in the landscapes where ponds occurred regularly (\u0026ldquo;medium\u0026rdquo; and \u0026ldquo;high\u0026rdquo;, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). In the landscape with \u0026ldquo;low\u0026rdquo; pond occurrence, we only detected surface water in 2017, with no ponds occurring otherwise. The year 2017 was a strong outlier year with exceptionally high surface water cover across all study sites (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb). The time-series for the \u0026ldquo;medium\u0026rdquo; and \u0026ldquo;high\u0026rdquo; sites showed high surface water variability (CV: 124% and 102%) even when the outlier year was removed (CV: 42% and 25% respectively). None of the time-series showed a significant trend over time (with and without 2017, Table S6, S7, S8, S9, S10, S11). We also did not detect any correlations with climate variables (Supplementary Methods, Table S12), but the time-series of surface water extent for the \u0026ldquo;high\u0026rdquo; site indicates some correspondence with mean summer air temperature (June-August) and autumn precipitation in the preceding calendar year (September-November) (Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Table S12). The outlier year 2017 corresponded to a cold summer and high snowfall at the beginning of the preceding cold season (Tei et al. 2018), but annual temperature and precipitation values did not appear unusual in the context of the previous decades (Fig. S2).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eMost ponds were short-lived and many strongly fluctuated in size\u003c/h3\u003e\n\u003cp\u003eThe majority of ponds were intermittently present in our time-series and the average variability in individual pond area was high (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Across all study sites and ponds, approximately two-thirds of ponds (65%) were present for less than six years and only one-third of ponds (35%) was \u0026ldquo;continuously\u0026rdquo; present, i.e., surface water was detected for six or seven years (allowing for one year of failed detection). The proportion of ponds \u0026ldquo;continuously\u0026rdquo; present was similar between the \u0026ldquo;medium\u0026rdquo; (38%) and \u0026ldquo;high\u0026rdquo; (34%) sites (Fig. S3a). The large majority of ponds experienced strong fluctuations in surface area (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). Excluding the outlier year 2017, 92% of ponds had a standardised variability (coefficient of variation, CV) larger than 10% of their mean surface area. The mean CV of pond area change was 52% across all time-series, with comparable values for the \u0026ldquo;medium\u0026rdquo; (52%) and \u0026ldquo;high\u0026rdquo; (52%) sites (Fig. S3b). We observed only 5 out of 465 (1.1%) ponds that were \u0026ldquo;stable\u0026rdquo; in the sense that they persisted for six or more years and had a CV of less than 10%, all five were found at the \u0026ldquo;medium\u0026rdquo; site.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eThermokarst and vegetation colonisation as drivers of pond change\u003c/h3\u003e\n\u003cp\u003eWe detected evidence for thermokarst (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) and vegetation colonisation (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) as drivers of surface area change in 24% of the ponds at the \u0026ldquo;medium\u0026rdquo; and \u0026ldquo;high\u0026rdquo; sites. Thermokarst may lead to pond expansion or movement (e.g., Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea). We observed expansion or movement of 68 ponds (15% of all ponds) across both sites, defined by a mean drop in surface elevation that exceeded 0.1 m (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb) averaged across the pond area gained between 2014 and 2024. This value serves as our threshold for detecting subsidence due to thermokarst (Online Methods). The proportion of these ponds with evidence for thermokarst was higher at the \u0026ldquo;medium\u0026rdquo; site (36 ponds\u0026thinsp;=\u0026thinsp;33%) than at the \u0026ldquo;high\u0026rdquo; site (32 ponds\u0026thinsp;=\u0026thinsp;9%) (Fig. S4). Vegetation colonisation, e.g., incursion of the pond area by sedges or mosses, can result in pond shrinking or even complete infilling (e.g., Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea). We observed vegetation colonisation in 83 ponds (18%) across both sites, defined by a gain in surface elevation that exceeded 0.1 m (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb) averaged across the pond area lost between 2014 and 2021. This value serves as our threshold for detecting surface elevation gain indicating vegetation colonisation. The proportion of ponds with evidence for vegetation colonisation was higher at the \u0026ldquo;medium\u0026rdquo; site (40 ponds\u0026thinsp;=\u0026thinsp;37%) than at the \u0026ldquo;high\u0026rdquo; site (43 ponds\u0026thinsp;=\u0026thinsp;12%) (Fig. S5).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur findings demonstrate that tundra thermokarst landscapes and ponds can have highly dynamic extents - even over short time-periods such as three or five years. Very few ponds (five) in the studied landscape were stable in area and position. In contrast, the majority of studied ponds were short-lived and showed strong fluctuations in size. We found evidence for thermokarst and vegetation colonisation as drivers of change in approximately one quarter of the studied ponds, but many ponds changed without either driver being detected. Together, these findings have considerable implications for how we monitor change, estimate carbon and energy fluxes, and understand drivers of change in surface water extent in tundra thermokarst landscapes.\u003c/p\u003e\n\u003ch3\u003eHigh temporal surface water variability complicates detection of long-term trends\u003c/h3\u003e\n\u003cp\u003eHigh inter-annual variability of surface water at scales of metres to decametres restricts our ability to derive estimates of long-term trends, such as wetting or drying, using remote sensing in tundra systems with pond presence. Multiple studies have already highlighted the insufficiencies of moderate resolution (10\u0026ndash;250 m) satellite imagery for surface water (and trend) detection in such tundra systems (Muster 2012, Webb and Liljedahl \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2023\u003c/span\u003e, Mullen et al. 2023). While centimetre-resolution drone data (as shown by us) and, to some degree, commercial high-resolution satellite imagery (Magnusson et al. 2020, Magnusson et al. 2021, Mullen et al. 2023) can provide data to detect surface area changes in small water bodies at the relevant spatial scales, our findings highlight the need to also sample these systems over long time-frames and at a high temporal frequencies (at least annual, likely better: intra-annual, see also Mullen et al. 2023). Even if an underlying long-term trend was present, time-series like ours that span almost a decade are not statistically powerful enough to detect a trend, given the high variability observed. We therefore suggest caution when interpreting trends based on time-series with moderate to low spatial (10\u0026ndash;250 m) and temporal resolution (i.e., larger than annual) in those tundra systems where small ponds are present and encourage the continued development of improved monitoring methods (Webb and Liljedahl \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2023\u003c/span\u003e, and Mullen et al. 2023).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003ePond dynamics driven by more than just thermokarst and vegetation colonisation\u003c/h2\u003e \u003cp\u003eMany ponds in our study showed high surface water variability without evidence for either thermokarst or vegetation colonisation. Pond-dominated tundra landscapes have been described as \u0026ldquo;fill-and-spill\u0026rdquo; systems (Helbig et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2013\u003c/span\u003e), where surface water dynamics are driven by snow melt (a function of winter precipitation) during spring and evaporation during summer, as well as surface and subsurface lateral flow (Woo and Guan \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2006\u003c/span\u003e, Wolfe et al. \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2011\u003c/span\u003e, Helbig et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). These processes are likely also key drivers of surface water dynamics observed in our study, as illustrated by the outlier year 2017: The year 2017 was marked by the inundation of landscapes across the region, that was driven by a high amount of snowmelt lasting into mid-summer, following a larger than unusual precipitation in the preceding October (Tei et al. \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Quantifying and modelling the relative contributions of the different drivers of landscape-level pond change (for example, using drone-derived surface models) will be critical for improving our predictions of future landscape dynamics and the associated carbon fluxes.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eSurface elevation changes integrate vegetation change and permafrost processes\u003c/h3\u003e\n\u003cp\u003eOur findings confirm that thermokarst and vegetation colonisation are processes contributing to pond change in the study region (Magn\u0026uacute;sson et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), but our method cannot resolve the complex causes of these processes themselves. At the \u0026ldquo;medium\u0026rdquo; and \u0026ldquo;high\u0026rdquo; study sites the ice-content of the first 60 cm of soil ranges from 50\u0026ndash;70% (Wang et al. \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). If such soils are thawed, the release of water will induce subsidence (Li et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) causing thermal erosion. By measuring the drop in surface elevation we detect this subsidence but we also integrate the loss of the vegetation canopy. Indeed, thermokarst is often associated with loss of vegetation, which can be both cause and effect of thermal erosion (Kanevsky et al. 2017, Li et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2017\u003c/span\u003e, Magn\u0026uacute;sson et al, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Similarly a gain in surface elevation integrates vegetation growth in previously water covered areas and the formation of ice rich permafrost, which also can be cause and effect of each other (Liljedahl 2016). Deriving digital elevation and canopy height models using drone-based LiDAR or dense ground control points could enable the separation of these processes, but in the absence of these data our study is limited to measuring the integrated effects.\u003c/p\u003e\n\u003ch3\u003eStudying fine-scale pond dynamics will better our understanding of ecosystem functioning\u003c/h3\u003e\n\u003cp\u003eThe strong variation in the surface area of individual ponds observed in our study suggests a need to better understand the associated consequences for the ecosystem and carbon and energy cycle. Many ponds in our study fluctuated notably in size and many had a short life-span; temporary flooding (Woo and Guan \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2006\u003c/span\u003e) and drying out (Smol and Douglas \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2007\u003c/span\u003e) were common across the time-series, as was the establishment and breakage of surface water connections between ponds (Wrona et al. \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). How these dynamic processes affect the habitat availability for breeding birds (Corbeil-Robitaille et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) or the assembly of microbial communities (Le Moigne et al. \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2020\u003c/span\u003e, Gurney et al. \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and their impact on the functioning of the ponds in the carbon cycle or other landscape-scale processes (Blackburn-Desbiens et al. \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) are not well understood. Furthermore, processes like thermokarst and vegetation colonisation have different implications for the carbon cycle. Thermokarst will likely release old carbon from the thawed soil (Wauthy et al. \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), while vegetation incursion by sedges will act as a conduit for the release of methane from the pond water (Knoblauch et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Combining time-series of high-resolution imagery with in-situ field studies of macro fauna or microbial dynamics could therefore greatly advance our understanding of the effects of pond dynamics on landscape-scale ecosystem function.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eNot all tundra landscapes experience directional trends or stability\u003c/h2\u003e \u003cp\u003eOverall, our study highlights the possibility that not all tundra systems are experiencing clear directional trends or stable states. Instead, our data shows that tundra landscapes with a high amount of pond presence can be very dynamic in space and time (Helbig et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Those dynamics are likely to increase further as extreme weather events become more common while the tundra continues to warm (Wolfe et al. \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2011\u003c/span\u003e, Tei et al. \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Considering the dynamic nature of these systems is critical when assessing long-term trends in surface cover of tundra landscapes, including surface water change (e.g., Webb et al \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2022\u003c/span\u003e, Olthof et al. \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), and their associated impacts on ecosystem function. The dynamic ponds further contribute to the complexity of greening and browning (Myers-Smith et al. \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2020\u003c/span\u003e, Beamish et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) and albedo trends (Plekhanova et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) in medium- to low-resolution satellite data. Only by accounting for the dynamic nature of some tundra landscapes will we be able to successfully estimate these trends and model the carbon and energy budget in the Arctic.\u003c/p\u003e \u003c/div\u003e"},{"header":"Online methods","content":"\u003ch2\u003eStudy sites\u003c/h2\u003e\n\u003cp\u003eThe three study sites are located in Kytalyk National Park in Northeastern Siberia, Russia in proximity to the Chokurdakh Research Station (70.83\u0026deg;N, 147.49\u0026deg;E, Interact: https://eu-interact.org/field-sites/chokurdakh-scientific-tundra-station/).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe climate in the region is cold, with an average annual temperature of -13.4 \u0026deg;C. Summers are mild with an average July temperature 10.3 \u0026deg;C. The mean annual precipitation is low (196 mm) with the majority falling in autumn and winter as snow, approximately 39% of the precipitation falling over summer (June-August). These climate data are based on the Chokurdakh weather station, WMO 21946, 28 km southeast of the Chokurdakh Research Station and averages refer to the period between 1981 and 2010 (Nauta et al. 2015).\u003c/p\u003e\n\u003cp\u003eThe landscapes in the region are underlain by Ice rich permafrost with an average active layer thickness of 42 cm and the soils are composed of silt with a shallow upper layer of peat (Grysko et al. 2021). The most common landscape types in the region include drained thaw lake beds (Alas depressions) and Yedoma ridges (Nitzbon et al. 2020). Ponds and patterned ground (ice-wege polygons) are common in the drained thaw beds (Nitzbon et al. 2020), where the vegetation is composed of dwarf shrubs (\u003cem\u003eBetula nana\u003c/em\u003e), moss tundra and sedges (Iturrate-Garcia et al. 2016). Ponds are rare on the Yedoma ridges (Nitzbon et al. 2020), where the vegetation primarily consists of tussock sedges (\u003cem\u003eEriophorum vaginatum\u003c/em\u003e) (Iturrate-Garcia et al. 2016). Holocene deposits, thermokarst lakes (not drained) and river channels are also present in the region (Nitzbon et al. 2020), but we did not consider these in our study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWithin the constraints of the drone data available, we chose the three study sites with an average dimension of 475 m x 435 m to capture the range of pond density occurring in the landscapes of the region. We selected a section of a drained thaw lake bed with a high pond occurrence (site \u0026ldquo;high\u0026rdquo;, area: 19.75 ha, centre: 70.8322\u0026deg;N 147.4779\u0026deg;E); a section of a drained thaw lake bed with a medium pond occurrence (site \u0026ldquo;medium\u0026rdquo;; area: 21.67 ha; centre: 70.8336\u0026deg;N 147.4951\u0026deg;E); and a section of a Yedoma ridge with no permanent pond occurrence as a control site (site \u0026ldquo;low\u0026rdquo;, area: 16.36 ha, centre: 70.8292\u0026deg;N 147.4638\u0026deg;E) (Fig. 1a). For further details see Supplementary Methods - \u0026ldquo;Determination of site boundaries\u0026rdquo; and Table S1.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eDrone data collection\u003c/h2\u003e\n\u003cp\u003eWe collected Red-Green-Blue (RGB / true colour) drone imagery at a near annual basis from 2014 to 2021. For a detailed description of all flights see Table S2. In brief, we used eBee Classic and eBee X PPK drone systems (AgEagle, formerly SenseFly, Lausanne, Switzerland) equipped with either a SenseFly S110 RGB (AgEagle) or SenseFly S.O.D.A \u0026nbsp;(AgEagle) sensor. Average ground sampling distances varied between 2.00 cm and 4.36 cm based on sensor and flight altitude (Table S2). We carried out all flights during summer between 11 July and 18 August of a given year. Repeat surveys within a year were available from 2019 for the \u0026ldquo;medium\u0026rdquo; \u0026nbsp; (3x) and \u0026ldquo;high\u0026rdquo; sites (2x). No data was collected in 2015 and the 2016 data at the \u0026ldquo;low\u0026rdquo; site was of insufficient quality. In some instances, a merging of multiple flights within a year was required to achieve complete coverage for a site. Illumination conditions varied widely (sunny, overcast, time-of day, etc.) (Table S2).\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eDrone data processing, geolocation and colour correction\u003c/h2\u003e\n\u003cp\u003eWe used Pix4D Mapper (Pix4D S.A. Prilly, Switzerland) to generate and co-register the RGB raster mosaics and surface models (DSMs) for each drone survey.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe generated the RGB mosaics and surface models using the \u0026ldquo;3D Maps\u0026rdquo; template in Pix4D and set the output resolution to 12 cm to obtain the mosaics and DSM with a uniform ground sampling distance (GSD). This distance is approximately three-times the maximum GSD in our dataset (Table S2). In four cases, the illumination of the individual raw images varied strongly, resulting in patchy mosaics. We addressed this issue using the \u003cem\u003ecolorCorrection\u0026nbsp;\u003c/em\u003efunction of the \u003cem\u003eCRImage\u0026nbsp;\u003c/em\u003eR package (Failmezger et al. 2024) to standardise the colour profiles of the individual images within the four mosaics prior processing (Table S2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHigh-precision geo-location using dGNSS post processing kinematics (PPK) was only possible for the 2020 and 2021 surveys using the newer eBee X systems. We therefore chose the 2021 mosaics at each site as the reference mosaic and co-registered all preceding mosaics in the time-series using manual tie-points in Pix4D. We verified the geolocation accuracy relative to the 2021 reference using five independent check points. The resulting estimate of the mean geolocation accuracy did not exceed 46 cm (three-times the 12 cm resolution of the mosaics) in all cases (Fig. S6). For further details see Supplementary Methods \u0026ldquo;Geolocation and accuracy assessment\u0026rdquo;.\u003c/p\u003e\n\u003cp\u003eFinally, we standardised the colour profiles of all RGB drone mosaics relative to the mosaic from the 2017 survey at the \u0026ldquo;high\u0026rdquo; site using the \u003cem\u003ecolorCorrection\u0026nbsp;\u003c/em\u003efunction of the \u003cem\u003eCRImage\u0026nbsp;\u003c/em\u003eR package (Failmezger et al. 2024) to aid with the next surface water detection steps. A calibration using empirical lines (Wang and Myint 2015) was not possible as standardised targets were not deployed during the drone surveys, instead we used a surface water detection approach that worked on a similar empirical principle using a mosaic specific threshold based on annotated training data. \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eSurface water detection\u003c/h2\u003e\n\u003cp\u003eWe used a simple threshold in the Blue Chromatic Coordinate (BCC) determined from supervised training data to classify surface water in each RGB mosaic. We calculated the BCC as the ratio between the digital number (DN) of the blue band and the sum of all bands (BCC = DN\u003csub\u003eblue\u003c/sub\u003e / (DN\u003csub\u003ered\u003c/sub\u003e + DN\u003csub\u003egreen\u003c/sub\u003e + DN\u003csub\u003eblue\u003c/sub\u003e). We then produced a set of manually annotated reference polygons with two cover classes (\u0026ldquo;surface water\u0026rdquo; and \u0026ldquo;other\u0026rdquo;) for each drone mosaic (Table S3) and extracted the BCC for all pixels in these polygons. Next, we iterated through all potential BCC threshold values from 0 to 1 at 0.01 intervals and determined the classification accuracy for each threshold to separate surface water from other surfaces based on the annotated reference data. We then chose the threshold providing the highest overall classification accuracy for each drone mosaic (Table S4) and generated a corresponding raster file with two surface classes (\u0026ldquo;surface water\u0026rdquo; and \u0026ldquo;other\u0026rdquo;). For further details on training and classification see Supplementary Methods - Training annotations \u0026amp; accuracy assessment. Finally, we removed all clusters of surface water pixels less than 1 m\u003csup\u003e2\u003c/sup\u003e, which was the size of the smallest annotated pond in our training data. The mean pond size was 17.89 m\u003csup\u003e2\u003c/sup\u003e.\u003c/p\u003e\n\u003ch2\u003eAnalysis of landscape-scale dynamics in surface water extent\u003c/h2\u003e\n\u003cp\u003eTo assess the dynamics of surface water extent at the landscape-scale, we calculated the area of surface water for each site and survey-time combination, and then analysed its variability within the three annual time-series. To this end, we determined the coefficient of variation (CV, the standard deviation divided by the mean and multiplied by 100%) for each time-series, once with and once without the outlier year 2017. For those year and site combinations for which repeat surveys were available (2019 and \u0026ldquo;high\u0026rdquo; and \u0026ldquo;medium\u0026rdquo; sites) we used the mean value of all surveys for the given year when calculating the CV. Based on the few available repeat surveys, the variation in surface water area extent within 2019 relative to the mean of the whole time series (sd\u003csub\u003e2019\u003c/sub\u003e / mean\u003csub\u003eall\u003c/sub\u003e x 100) was small compared to coefficient of variation across the whole time series (sd\u003csub\u003eall\u003c/sub\u003e / mean\u003csub\u003eall\u003c/sub\u003e x 100): 3% (2019) vs 124% (all years) for the \u0026ldquo;high\u0026rdquo; site and 15% (2019) vs 102% (all years) for the \u0026ldquo;low\u0026rdquo; site. See Table S5 for a detailed listing of all surface water area and proportion values. We also tested for trends in all time-series using OLS and correlations with climate variables (Supplementary Materials - Surface water trend and climate analysis).\u003c/p\u003e\n\u003ch2\u003eAnalysis of dynamics in individual pond area\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eTo assess the dynamics in surface water extent at the pond-scale, we identified all individual ponds in each time-series and then analysed the change in their surface water area.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe identified the individual ponds by generating a binary composite of all surface water extent maps for each time-series. In this composite, a cell was set to one (water present) if water was detected in this cell at any point in the time-series. All other cells were set to NA. In this component of the analysis, we excluded the extreme outlier year 2017 as many ponds were connected during this year. We also excluded all repeat surveys within a year and only kept one observation per year. Here we excluded all surveys that were furthest from the median within-year timing of observations across all years (mosaic IDs: \u0026ldquo;cbh_2019\u0026rdquo;, \u0026ldquo;tlb_2019_a\u0026rdquo; and \u0026ldquo;tlb_2019_b\u0026rdquo;). We then obtained the polygon geometries of all clusters of surface water in the composite (hereafter \u0026ldquo;reference ponds\u0026rdquo;) and tracked their presence across the time-series.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe tracked the reference ponds identified from the composite across the time-series by testing for geometric intersection with all ponds detected in a given year. As surface water at the \u0026ldquo;low\u0026rdquo; site was only detected in 2017, we removed this site from all subsequent analysis. Next, we assigned each remaining cluster of surface water an individual pond identifier and determined the number of years the pond was present in the time-series by testing for geometric intersection between polygons of the clusters derived from the composite with the polygons of all clusters of water detected in each year. To exclude temporary \u0026ldquo;puddles\u0026rdquo;, we kept only those clusters of surface water that were present at least three times in each time-series. Experimental manipulation of ponds had previously been carried out at the \u0026ldquo;medium\u0026rdquo; site; we identified and removed these ponds (total: 8) from all subsequent analyses. Finally, we calculated the pond area in each year as the total area of all clusters of surface water intersecting with the pond geometry derived from the composite (some ponds split), and then determined the CV as a measure of pond area variability, excluding the outlier year 2017. \u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eAnalysis of drivers of pond-change\u003c/h2\u003e\n\u003cp\u003eWe used the drone derived digital surface model to identify the presence of thermokarst and vegetation colonisation as drivers of pond surface area change. We restricted our analysis to changes in the relative surface elevation in the surroundings of each pond, as we were unable to precisely geolocate the three-dimensional point clouds in the absence of high-precision geolocation of image tags and ground control points. To this end, we cropped the drone DSMs for each year with a buffer area of 10 m around each pond and masked all areas that were detected as water in any point of the time-series (including 2017). We then standardised the DSMs relative to the minimum surface elevation within this area.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor the detection of thermokarst as a driver of pond change, we calculated the mean drop in surface elevation in the DSMs between 2021 and 2014 within the pond area gained (Fig. S7a). For the detection of vegetation colonisation as a driver of pond change, we calculated the mean gain in surface elevation in the DSMs between 2021 and 2014 for the pond area lost (Fig. S7b). We calculated the average drop / gain in surface elevation for the pond area gained / lost rather than the total absolute change in volume as the ponds varied strongly in size. In other words, we standardised the volumetric differences by area gained / lost to obtain a comparable measure of change in surface volume across ponds independent of size.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIt is important to highlight two key contributors to uncertainty in the above-described method. First, standardising the DSMs by the minimum elevation within the buffered surroundings of each pond assumes that both the minimum elevation as well as the measurement error are constant across the time-series space. The assumption will likely not hold true in all cases and may introduce both random and systematic errors. Second, for the calculations of the volume differences we assumed that any area covered by water has a relative surface height of 0 m. This assumption may lead to a systematic under- or overestimation of the volume lost in the thermokarst and vegetation colonisation detection analyses respectively. Considering all of the above, we estimated that we are only able to reliably detect volume changes in the surface model that exceed 0.1 m\u003csup\u003e3\u003c/sup\u003e per m\u003csup\u003e2\u003c/sup\u003e, which is equivalent to a drop (or gain) in the surface elevation of 10 cm. Hence, we only assigned the detection of thermokarst or vegetation colonisation to a pond where the threshold was exceeded in the time-series. We empirically validated this threshold by manually inspecting the RGB / DSM time-series of all ponds.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eSoftware\u003c/h2\u003e\n\u003cp\u003eWe conducted all data processing and statistical analysis in R ver 4.4.0 (R Core Team 2024). We used the sf package ver 1.0-16 (Pebesma and Bivand 2023, Pebesma 2018) for all geometrical operations; the terra package ver 1.7-78 (Hijmans 2024) for raster handling and calculations; the tidyverse ver 2.0.0.0 (Whickam et al. 2019) and pbapply ver 1.7-2 (Solymos and Zawdazki 2023) packages for general data wrangling and parallel processing; the ggplot2 ver 3.5.1 (Wickham 2016), cowplot ver 1.1.3 (Wilke 2024) and tidyterra ver 0.61 (Hernang\u0026oacute;mez 2023) packages for generating the maps and figures, and the gt package ver 0.11.0 (Iannone et al. 2024) for generating tables.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eReferences - Online methods\u003c/h2\u003e\n\u003cp\u003eFailmezger, H., Yuan, Y., Rueda, O., Markowetz, F., 2024. CRImage: CRImage a package to classify cells and calculate tumour cellularity.\u003c/p\u003e\n\u003cp\u003eGrysko, R., Plekhanova, E., Oehri, J., Karsanaev, S.V., Maximov, T.C., Schaepman-Strub, G., 2021. Design of the tundra rainfall experiment (TRainEx) to simulate future summer precipitation scenarios. MethodsX 8, 101331.\u003ca href=\"https://doi.org/10.1016/j.mex.2021.101331\"\u003e\u0026nbsp;\u003c/a\u003ehttps://doi.org/10.1016/j.mex.2021.101331\u003c/p\u003e\n\u003cp\u003eHernang\u0026oacute;mez, D., 2023. Using the tidyverse with terra objects: the tidyterra package. Journal of Open Source Software 8, 5751.\u003ca href=\"https://doi.org/10.21105/joss.05751\"\u003e\u0026nbsp;\u003c/a\u003ehttps://doi.org/10.21105/joss.05751\u003c/p\u003e\n\u003cp\u003eHijmans, R.J., 2024. terra: Spatial Data Analysis. https://CRAN.R-project.org/package=terra\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIannone, R., Cheng, J., Schloerke, B., Hughes, E., Lauer, A., Seo, J., Brevoort, K., Roy, O., 2024. gt: Easily Create Presentation-Ready Display Tables. https://CRAN.R-project.org/package=gt\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIturrate-Garcia, M., O\u0026rsquo;Brien, M.J., Khitun, O., Abiven, S., Niklaus, P.A., Schaepman-Strub, G., 2016. Interactive effects between plant functional types and soil factors on tundra species diversity and community composition. Ecology and Evolution 6, 8126\u0026ndash;8137.\u003ca href=\"https://doi.org/10.1002/ece3.2548\"\u003e\u0026nbsp;\u003c/a\u003ehttps://doi.org/10.1002/ece3.2548\u003c/p\u003e\n\u003cp\u003eNauta, A.L., Heijmans, M.M.P.D., Blok, D., Limpens, J., Elberling, B., Gallagher, A., Li, B., Petrov, R.E., Maximov, T.C., van Huissteden, J., Berendse, F., 2015. Permafrost collapse after shrub removal shifts tundra ecosystem to a methane source. Nature Clim Change 5, 67\u0026ndash;70.\u003ca href=\"https://doi.org/10.1038/nclimate2446\"\u003e\u0026nbsp;\u003c/a\u003ehttps://doi.org/10.1038/nclimate2446\u003c/p\u003e\n\u003cp\u003eNitzbon, J., Westermann, S., Langer, M., Martin, L.C.P., Strauss, J., Laboor, S., Boike, J., 2020. Fast response of cold ice-rich permafrost in northeast Siberia to a warming climate. Nat Commun 11, 2201.\u003ca href=\"https://doi.org/10.1038/s41467-020-15725-8\"\u003e\u0026nbsp;\u003c/a\u003ehttps://doi.org/10.1038/s41467-020-15725-8\u003c/p\u003e\n\u003cp\u003eR Core Team, 2024. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.\u003c/p\u003e\n\u003cp\u003ePebesma, E., 2018. Simple Features for R: Standardized Support for Spatial Vector Data. The R Journal 10, 439\u0026ndash;446.\u003ca href=\"https://doi.org/10.32614/RJ-2018-009\"\u003e\u0026nbsp;\u003c/a\u003ehttps://doi.org/10.32614/RJ-2018-009\u003c/p\u003e\n\u003cp\u003ePebesma, E., Bivand, R., 2023. Spatial Data Science: With Applications in R. Chapman and Hall/CRC, New York.\u003ca href=\"https://doi.org/10.1201/9780429459016\"\u003e\u0026nbsp;\u003c/a\u003ehttps://doi.org/10.1201/9780429459016\u003c/p\u003e\n\u003cp\u003eSolymos, P., Zawadzki, Z., 2023. pbapply: Adding Progress Bar to \u0026ldquo;*apply\u0026rdquo; Functions. https://CRAN.R-project.org/package=pbapply\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWang, C., Myint, S.W., 2015. A Simplified Empirical Line Method of Radiometric Calibration for Small Unmanned Aircraft Systems-Based Remote Sensing. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 8, 1876\u0026ndash;1885.\u003ca href=\"https://doi.org/10.1109/JSTARS.2015.2422716\"\u003e\u0026nbsp;\u003c/a\u003ehttps://doi.org/10.1109/JSTARS.2015.2422716\u003c/p\u003e\n\u003cp\u003eWickham, H., Averick, M., Bryan, J., Chang, W., McGowan, L.D., Fran\u0026ccedil;ois, R., Grolemund, G., Hayes, A., Henry, L., Hester, J., Kuhn, M., Pedersen, T.L., Miller, E., Bache, S.M., M\u0026uuml;ller, K., Ooms, J., Robinson, D., Seidel, D.P., Spinu, V., Takahashi, K., Vaughan, D., Wilke, C., Woo, K., Yutani, H., 2019. Welcome to the tidyverse. Journal of Open Source Software 4, 1686.\u003ca href=\"https://doi.org/10.21105/joss.01686\"\u003e\u0026nbsp;\u003c/a\u003ehttps://doi.org/10.21105/joss.01686\u003c/p\u003e\n\u003cp\u003eWickham, H., 2016. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. https://ggplot2.tidyverse.org\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWilke, C.O., 2024. cowplot: Streamlined Plot Theme and Plot Annotations for \u0026ldquo;ggplot2.\u0026rdquo; https://CRAN.R-project.org/package=cowplot\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch3\u003eAcknowledgements\u003c/h3\u003e\n\u003cp\u003eWe would like to thank all drone-pilots and field team members who contributed to the data collection during the field seasons 2014 \u0026ndash; 2021, including Inge Gr\u0026uuml;nberg (nee Juszak), Maitane Iturrate-Garcia and Vitalii Zemlianskii. This study was supported by the Swiss National Science Foundation (grant no. 178753) and the University Research Priority Program on Global Change and Biodiversity of the University of Zurich.\u003c/p\u003e\n\u003ch2\u003eCode and Data availability\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eAll code, tabular and geometric data required to regenerate the analyses are available via the following GitHub repository: https://github.com/jakobjassmann/pond_project \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe raster data (RGB drone mosaics and DSMs) are also required and are available on Zenodo: https://doi.org/10.5281/zenodo.13992090\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe drone RGB rasters, surface water classifications and individual pond time-series have been visualised in a Web-GIS hosted on AWS: https://pondproject.s3.eu-central-1.amazonaws.com/index.html \u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAndresen, C.G., Lougheed, V.L., 2015. Disappearing Arctic tundra ponds: Fine-scale analysis of surface hydrology in drained thaw lake basins over a 65 year period (1948\u0026ndash;2013). Journal of Geophysical Research: Biogeosciences 120, 466\u0026ndash;479. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/2014JG002778\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eBartosiewicz, M., Przytulska, A., Lapierre, J.-F., Laurion, I., Lehmann, M.F., Maranger, R., 2019. Hot tops, cold bottoms: Synergistic climate warming and shielding effects increase carbon burial in lakes. Limnology and Oceanography Letters 4, 132\u0026ndash;144. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/lol2.10117\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eBeamish, A., Raynolds, M.K., Epstein, H., Frost, G.V., Macander, M.J., Bergstedt, H., Bartsch, A., Kruse, S., Miles, V., Tanis, C.M., Heim, B., Fuchs, M., Chabrillat, S., Shevtsova, I., Verdonen, M., Wagner, J., 2020. Recent trends and remaining challenges for optical remote sensing of Arctic tundra vegetation: A review and outlook. Remote Sensing of Environment 246, 111872. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.rse.2020.111872\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eBlackburn-Desbiens, P., Grosbois, G., Power, M., Culp, J., Rautio, M., 2023. Integrating hydrological connectivity and zooplankton composition in Arctic ponds and lakes. Freshwater Biology 68, 2131\u0026ndash;2150. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/fwb.14181\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eCorbeil-Robitaille, M.-Z., Duchesne, \u0026Eacute;., Fortier, D., Kinnard, C., B\u0026ecirc;ty, J., 2024. Linking geomorphological processes and wildlife microhabitat selection: nesting birds select refuges generated by permafrost degradation in the Arctic. Biogeosciences 21, 3401\u0026ndash;3423. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5194/bg-21-3401-2024\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eDel Giorgio, P.A., Cole, J.J., 1998. BACTERIAL GROWTH EFFICIENCY IN NATURAL AQUATIC SYSTEMS. Annual Review of Ecology, Evolution, and Systematics 29, 503\u0026ndash;541. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1146/annurev.ecolsys.29.1.503\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eGurney, K.E.B., Koch, J.C., Schmutz, J.A., Schmidt, J.H., Wipfli, M.S., 2022. In hot water? Patterns of macroinvertebrate abundance in Arctic thaw ponds and relationships with environmental variables. Freshwater Biology n/a. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/fwb.13978\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eHelbig, M., Boike, J., Langer, M., Schreiber, P., Runkle, B.R.K., Kutzbach, L., 2013. Spatial and seasonal variability of polygonal tundra water balance: Lena River Delta, northern Siberia (Russia). Hydrogeol J 21, 133\u0026ndash;147. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10040-012-0933-4\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eHinks, A.E., Cole, E.F., Daniels, K.J., Wilkin, T.A., Nakagawa, S., Sheldon, B.C., 2015. Scale-Dependent Phenological Synchrony between Songbirds and Their Caterpillar Food Source. The American Naturalist 186, 84\u0026ndash;97. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1086/681572\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eJuszak, I., Iturrate-Garcia, M., Gastellu-Etchegorry, J.-P., Schaepman, M.E., Maximov, T.C., Schaepman-Strub, G., 2017. Drivers of shortwave radiation fluxes in Arctic tundra across scales. Remote Sensing of Environment 193, 86\u0026ndash;102. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.rse.2017.02.017\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eKanevskiy, M., Shur, Y., Jorgenson, T., Brown, D.R.N., Moskalenko, N., Brown, J., Walker, D.A., Raynolds, M.K., Buchhorn, M., 2017. Degradation and stabilization of ice wedges: Implications for assessing risk of thermokarst in northern Alaska. Geomorphology 297, 20\u0026ndash;42. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.geomorph.2017.09.001\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eKnoblauch, C., Spott, O., Evgrafova, S., Kutzbach, L., Pfeiffer, E.-M., 2015. Regulation of methane production, oxidation, and emission by vascular plants and bryophytes in ponds of the northeast Siberian polygonal tundra. Journal of Geophysical Research: Biogeosciences 120, 2525\u0026ndash;2541. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/2015JG003053\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eLachenbruch, A.H., 1962. Mechanics of thermal contraction cracks and ice-wedge polygons in permafrost. Geological Society of America. Special Papers 70, 69.\u003c/li\u003e\n\u003cli\u003eLanger, M., Westermann, S., Muster, S., Piel, K., Boike, J., 2011. The surface energy balance of a polygonal tundra site in northern Siberia \u0026ndash; Part 1: Spring to fall. The Cryosphere 5, 151\u0026ndash;171. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5194/tc-5-151-2011\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eLara, M.J., McGuire, A.D., Euskirchen, E.S., Tweedie, C.E., Hinkel, K.M., Skurikhin, A.N., Romanovsky, V.E., Grosse, G., Bolton, W.R., Genet, H., 2015. Polygonal tundra geomorphological change in response to warming alters future CO2 and CH4 flux on the Barrow Peninsula. Global Change Biology 21, 1634\u0026ndash;1651. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/gcb.12757\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eLara, M.J., McGuire, A.D., Euskirchen, E.S., Genet, H., Yi, S., Rutter, R., Iversen, C., Sloan, V., Wullschleger, S.D., 2020. Local-scale Arctic tundra heterogeneity affects regional-scale carbon dynamics. Nat Commun 11, 4925. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41467-020-18768-z\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eLe Moigne, A., Bartosiewicz, M., Schaepman-Strub, G., Abiven, S., Pernthaler, J., 2020. The biogeochemical variability of Arctic thermokarst ponds is reflected by stochastic and niche-driven microbial community assembly processes. Environmental Microbiology 22, 4847\u0026ndash;4862. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/1462-2920.15260\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eLi, B., Heijmans, M.M.P.D., Blok, D., Wang, P., Karsanaev, S.V., Maximov, T.C., van Huissteden, J., Berendse, F., 2017. Thaw pond development and initial vegetation succession in experimental plots at a Siberian lowland tundra site. Plant Soil 420, 147\u0026ndash;162. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11104-017-3369-8\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eLiljedahl, A.K., Boike, J., Daanen, R.P., Fedorov, A.N., Frost, G.V., Grosse, G., Hinzman, L.D., Iijma, Y., Jorgenson, J.C., Matveyeva, N., Necsoiu, M., Raynolds, M.K., Romanovsky, V.E., Schulla, J., Tape, K.D., Walker, D.A., Wilson, C.J., Yabuki, H., Zona, D., 2016. Pan-Arctic ice-wedge degradation in warming permafrost and its influence on tundra hydrology. Nature Geosci 9, 312\u0026ndash;318. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/ngeo2674\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eLuoto, M., Sepp\u0026auml;l\u0026auml;, M., 2003. Thermokarst ponds as indicators of the former distribution of palsas in Finnish Lapland. Permafrost and Periglacial Processes 14, 19\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/ppp.441\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eMagn\u0026uacute;sson, R.\u0026Iacute;., Limpens, J., Huissteden, J., Kleijn, D., Maximov, T.C., Rotbarth, R., Sass-Klaassen, U., Heijmans, M.M.P.D., 2020. Rapid Vegetation Succession and Coupled Permafrost Dynamics in Arctic Thaw Ponds in the Siberian Lowland Tundra. J. Geophys. Res. Biogeosci. 125. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1029/2019JG005618\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eMagn\u0026uacute;sson, R.\u0026Iacute;., Limpens, J., Kleijn, D., van Huissteden, K., Maximov, T.C., Lobry, S., Heijmans, M.M.P.D., 2021. Shrub decline and expansion of wetland vegetation revealed by very high resolution land cover change detection in the Siberian lowland tundra. Science of The Total Environment 782, 146877. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.scitotenv.2021.146877\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eMelton, J.R., Wania, R., Hodson, E.L., Poulter, B., Ringeval, B., Spahni, R., Bohn, T., Avis, C.A., Beerling, D.J., Chen, G., Eliseev, A.V., Denisov, S.N., Hopcroft, P.O., Lettenmaier, D.P., Riley, W.J., Singarayer, J.S., Subin, Z.M., Tian, H., Z\u0026uuml;rcher, S., Brovkin, V., van Bodegom, P.M., Kleinen, T., Yu, Z.C., Kaplan, J.O., 2013. Present state of global wetland extent and wetland methane modelling: conclusions from a model inter-comparison project (WETCHIMP). Biogeosciences 10, 753\u0026ndash;788. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5194/bg-10-753-2013\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eMorgenstern, A., Ulrich, M., G\u0026uuml;nther, F., Roessler, S., Fedorova, I.V., Rudaya, N.A., Wetterich, S., Boike, J., Schirrmeister, L., 2013. Evolution of thermokarst in East Siberian ice-rich permafrost: A case study. Geomorphology 201, 363\u0026ndash;379. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.geomorph.2013.07.011\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eMullen, A.L., Watts, J.D., Rogers, B.M., Carroll, M.L., Elder, C.D., Noomah, J., Williams, Z., Caraballo-Vega, J.A., Bredder, A., Rickenbaugh, E., Levenson, E., Cooley, S.W., Hung, J.K.Y., Fiske, G., Potter, S., Yang, Y., Miller, C.E., Natali, S.M., Douglas, T.A., Kyzivat, E.D., 2023. Using High-Resolution Satellite Imagery and Deep Learning to Track Dynamic Seasonality in Small Water Bodies. Geophysical Research Letters 50, e2022GL102327. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1029/2022GL102327\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eMuster, S., Langer, M., Heim, B., Westermann, S., Boike, J., 2012. Subpixel heterogeneity of ice-wedge polygonal tundra: a multi-scale analysis of land cover and evapotranspiration in the Lena River Delta, Siberia. Tellus B: Chemical and Physical Meteorology 64, 17301. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3402/tellusb.v64i0.17301\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eMyers-Smith, I.H., Kerby, J.T., Phoenix, G.K., Bjerke, J.W., Epstein, H.E., Assmann, J.J., John, C., Andreu-Hayles, L., Angers-Blondin, S., Beck, P.S.A., Berner, L.T., Bhatt, U.S., Bjorkman, A.D., Blok, D., Bryn, A., Christiansen, C.T., Cornelissen, J.H.C., Cunliffe, A.M., Elmendorf, S.C., Forbes, B.C., Goetz, S.J., Hollister, R.D., de Jong, R., Loranty, M.M., Macias-Fauria, M., Maseyk, K., Normand, S., Olofsson, J., Parker, T.C., Parmentier, F.-J.W., Post, E., Schaepman-Strub, G., Stordal, F., Sullivan, P.F., Thomas, H.J.D., T\u0026oslash;mmervik, H., Treharne, R., Tweedie, C.E., Walker, D.A., Wilmking, M., Wipf, S., 2020. Complexity revealed in the greening of the Arctic. Nat. Clim. Chang. 10, 106\u0026ndash;117. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41558-019-0688-1\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eNegandhi, K., Laurion, I., Whiticar, M.J., Galand, P.E., Xu, X., Lovejoy, C., 2013. Small Thaw Ponds: An Unaccounted Source of Methane in the Canadian High Arctic. PLOS ONE 8, e78204. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0078204\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eNegandhi, K., Laurion, I., Lovejoy, C., 2016. Temperature effects on net greenhouse gas production and bacterial communities in arctic thaw ponds. FEMS Microbiology Ecology 92, fiw117. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/femsec/fiw117\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eNitzbon, J., Westermann, S., Langer, M., Martin, L.C.P., Strauss, J., Laboor, S., Boike, J., 2020. Fast response of cold ice-rich permafrost in northeast Siberia to a warming climate. Nat Commun 11, 2201. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41467-020-15725-8\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eOlefeldt, D., Goswami, S., Grosse, G., Hayes, D., Hugelius, G., Kuhry, P., McGuire, A.D., Romanovsky, V.E., Sannel, A.B.K., Schuur, E. a. G., Turetsky, M.R., 2016. Circumpolar distribution and carbon storage of thermokarst landscapes. Nat Commun 7, 13043. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/ncomms13043\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eOlthof, I., Fraser, R.H., van der Sluijs, J., Travers-Smith, H., 2023. Detecting long-term Arctic surface water changes. Nat. Clim. Chang. 13, 1191\u0026ndash;1193. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41558-023-01836-9\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003ePayette, S., Delwaide, A., Caccianiga, M., Beauchemin, M., 2004. Accelerated thawing of subarctic peatland permafrost over the last 50 years. Geophysical Research Letters 31. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1029/2004GL020358\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003ePeura, S., Wauthy, M., Simone, D., Eiler, A., Einarsd\u0026oacute;ttir, K., Rautio, M., Bertilsson, S., 2020. Ontogenic succession of thermokarst thaw ponds is linked to dissolved organic matter quality and microbial degradation potential. Limnology and Oceanography 65, S248\u0026ndash;S263. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/lno.11349\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003ePienitz, R., Doran, P.T., Lamoureux, S.F., 2008. Origin and geomorphology of lakes in the polar regions., in: Polar Lakes and Rivers: Limnology of Arctic and Antarctic Aquatic Ecosystems. Oxford University Press, Oxford, England, pp. 25\u0026ndash;41.\u003c/li\u003e\n\u003cli\u003ePlanet Labs PBC, 2022. Planet Application Program Interface: In Space for Life on Earth. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://api.planet.com\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003ePlekhanova, E., Kim, J.-S., Oehri, J., Erb, A., Schaaf, C., Schaepman-Strub, G., 2022. Mid-summer snow-free albedo across the Arctic tundra was mostly stable or increased over the past two decades. Environ. Res. Lett. 17, 124026. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1088/1748-9326/aca5a1\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003ePrėskienis, V., Fortier, D., Douglas, P.M.J., Rautio, M., Laurion, I., 2024. Permafrost degradation and soil erosion as drivers of greenhouse gas emissions from tundra ponds. Environ. Res. Lett. 19, 014072. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1088/1748-9326/ad1433\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eRautio, M., Dufresne, F., Laurion, I., Bonilla, S., Vincent, W.F., Christoffersen, K.S., 2011. Shallow freshwater ecosystems of the circumpolar Arctic. \u0026Eacute;coscience 18, 204\u0026ndash;222. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2980/18-3-3463\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eSchirrmeister, L., Pestryakova, L., Wetterich, S., Tumskoy, V., 2012. Joint Russian-German polygon project: East Siberia 2011\u0026ndash;2014 ; the expedition Kytalyk 2011 [WWW Document]. Berichte zur Polar- und Meeresforschung\u0026thinsp;=\u0026thinsp;Reports on polar and marine research. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2312/BzPM_0653_2012\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eSmol, J.P., Douglas, M.S.V., 2007. Crossing the final ecological threshold in high Arctic ponds. Proceedings of the National Academy of Sciences 104, 12395\u0026ndash;12397. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1073/pnas.0702777104\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eTei, S., Morozumi, T., Nagai, S., Takano, S., Sugimoto, A., Shingubara, R., Fan, R., Fedorov, A., Gavrilyeva, T., Tananaev, N., Maximov, T., 2020. An extreme flood caused by a heavy snowfall over the Indigirka River basin in Northeastern Siberia. Hydrological Processes 34, 522\u0026ndash;537. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/hyp.13601\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eTreat, C.C., Marushchak, M.E., Voigt, C., Zhang, Y., Tan, Z., Zhuang, Q., Virtanen, T.A., R\u0026auml;s\u0026auml;nen, A., Biasi, C., Hugelius, G., Kaverin, D., Miller, P.A., Stendel, M., Romanovsky, V., Rivkin, F., Martikainen, P.J., Shurpali, N.J., 2018. Tundra landscape heterogeneity, not interannual variability, controls the decadal regional carbon balance in the Western Russian Arctic. Global Change Biology 24, 5188\u0026ndash;5204. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/gcb.14421\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWalter, K.M., Zimov, S.A., Chanton, J.P., Verbyla, D., Chapin, F.S., 2006. Methane bubbling from Siberian thaw lakes as a positive feedback to climate warming. Nature 443, 71\u0026ndash;75. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/nature05040\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWang, P., de Jager, J., Nauta, A., van Huissteden, J., Trofim, M.C., Limpens, J., 2019. Exploring near-surface ground ice distribution in patterned-ground tundra: correlations with topography, soil and vegetation. Plant Soil 444, 251\u0026ndash;265. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11104-019-04276-7\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWang, G., Peng, Y., Chen, L., Abbott, B.W., Ciais, P., Kang, L., Liu, Y., Li, Q., Pe\u0026ntilde;uelas, J., Qin, S., Smith, P., Song, Y., Strauss, J., Wang, J., Wei, B., Yu, J., Zhang, D., Yang, Y., 2024. Enhanced response of soil respiration to experimental warming upon thermokarst formation. Nat. Geosci. 1\u0026ndash;7. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41561-024-01440-2\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWauthy, M., Rautio, M., Christoffersen, K.S., Forsstr\u0026ouml;m, L., Laurion, I., Mariash, H.L., Peura, S., Vincent, W.F., 2018. Increasing dominance of terrigenous organic matter in circumpolar freshwaters due to permafrost thaw. Limnology and Oceanography Letters 3, 186\u0026ndash;198. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/lol2.10063\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWebb, E.E., Liljedahl, A.K., Cordeiro, J.A., Loranty, M.M., Witharana, C., Lichstein, J.W., 2022. Permafrost thaw drives surface water decline across lake-rich regions of the Arctic. Nat. Clim. Chang. 1\u0026ndash;6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41558-022-01455-w\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWebb, E.E., Liljedahl, A.K., 2023. Diminishing lake area across the northern permafrost zone. Nat. Geosci. 16, 202\u0026ndash;209. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41561-023-01128-z\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWebb, E.E., Liljedahl, A.K., Loranty, M.M., Witharana, C., Lichstein, J.W., 2023. Reply to: Detecting long-term Arctic surface water changes. Nat. Clim. Chang. 13, 1194\u0026ndash;1196. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41558-023-01837-8\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWik, M., Varner, R.K., Anthony, K.W., MacIntyre, S., Bastviken, D., 2016. Climate-sensitive northern lakes and ponds are critical components of methane release. Nature Geosci 9, 99\u0026ndash;105. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/ngeo2578\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWolfe, B.B., Light, E.M., Macrae, M.L., Hall, R.I., Eichel, K., Jasechko, S., White, J., Fishback, L., Edwards, T.W.D., 2011. Divergent hydrological responses to 20th century climate change in shallow tundra ponds, western Hudson Bay Lowlands. Geophysical Research Letters 38. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1029/2011GL049766\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWoo, M.-K., Guan, X.J., 2006. Hydrological connectivity and seasonal storage change of tundra ponds in a polar oasis environment, Canadian High Arctic. Permafrost and Periglacial Processes 17, 309\u0026ndash;323. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/ppp.565\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003cli\u003eWrona, F.J., Johansson, M., Culp, J.M., Jenkins, A., M\u0026aring;rd, J., Myers-Smith, I.H., Prowse, T.D., Vincent, W.F., Wookey, P.A., 2016. Transitions in Arctic ecosystems: Ecological implications of a changing hydrological regime. Journal of Geophysical Research: Biogeosciences 121, 650\u0026ndash;674. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/2015JG003133\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5581925/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5581925/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eGreenhouse gas emissions from Arctic tundra ponds and permafrost thaw provide important positive feedbacks to global warming. However, a high landscape heterogeneity and small size of ponds make it challenging to assess trends in surface water extent and associated carbon and energy fluxes, especially in the understudied Eastern Siberian tundra. Here, we show that surface water extent in these landscapes can be highly dynamic, shaped by pond-scale processes that cannot be detected in satellite products. Using a time-series of aerial imagery at 12 cm resolution spanning eight years (2014\u0026ndash;2021), we classified surface water at three sites in Kytalyk National Park and traced the 465 ponds larger than 1 m\u003csup\u003e2\u003c/sup\u003e. The surface water extent varied between 102%-124% relative to the time-series mean, without significant trends in contrast to previous reports. Individual pond area fluctuated by 52% on average, and two thirds of ponds were present for less than six years. One-quarter of ponds showed evidence for thermokarst or vegetation colonisation as drivers of change, based on our high-resolution surface elevation models. These findings highlight a highly dynamic nature of tundra ponds and stress the need for improved change detection to better model carbon and energy fluxes in this biome.\u003c/p\u003e \u003cp\u003eMain Text\u003c/p\u003e","manuscriptTitle":"Freshwater ponds create highly dynamic Arctic tundra landscapes","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-01-29 05:05:43","doi":"10.21203/rs.3.rs-5581925/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"68a5d81c-ea71-4b7d-9c21-66f3d6f6552e","owner":[],"postedDate":"January 29th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":41264318,"name":"Earth and environmental sciences/Hydrology"},{"id":41264319,"name":"Earth and environmental sciences/Ecology/Climate-change ecology"},{"id":41264320,"name":"Earth and environmental sciences/Ecology/Ecosystem ecology"}],"tags":[],"updatedAt":"2025-04-27T20:05:11+00:00","versionOfRecord":[],"versionCreatedAt":"2025-01-29 05:05:43","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5581925","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5581925","identity":"rs-5581925","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}