Controls on vertical root distribution dynamics in a temperate grassland across daily and seasonal scales

Controls on vertical root distribution dynamics in a temperate grassland across daily and seasonal scales | 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 Research Article Controls on vertical root distribution dynamics in a temperate grassland across daily and seasonal scales Samuele Ceolin, Stanislaus Josef Schymanski, Julian Klaus This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7847447/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Background and aims “Hydromatching” is a phenomenon consisting of the daily-scale promotion of root growth in a newly wetted soil layer and/or a decline in root growth in drier layers. This phenomenon was previously observed on individual plants under controlled conditions. The aim of this study was to determine if hydromatching occurs also under natural settings at a grassland community scale. Our goal was also to assess what environmental parameter was driving root growth in the grassland community and to analyze seasonal shifts in root distribution. Methods We installed twelve minirhizotron tubes in a natural grassland. We imaged the tubes from May 2022 until August 2023 and we carried out image analyses to extrapolate root length and growth rates. We determined the major environmental driver of root growth through regression analyses and monitored the root growth dynamics after major rain events and during the spring-summer transition. Results Soil moisture was the strongest predictor of root growth. Following rain events, root growth shifted from deeper layers to shallow layers within 1-3 days, indicating the occurrence of hydromatching. During the spring-summer transition, we observed significant promotion of growth in deeper soil layers and decline in root length in shallower layers. Conclusions Root distribution responded to seasonality with drastic shifts at the whole root system level and to precipitation with smaller but significant rapid shifts. Both types of response could adhere to an optimization strategy, consisting of the promotion of root growth in resourceful areas while discarding roots where resources are less accessible. Root dynamics Root allocation Hydromatching Root seasonal shifts Water uptake Root growth predictors Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Root water uptake is one of the key aspects studied in the context of agricultural productivity. Management practices and selection of plant varieties that encourage water capture are some of the tools that can enable us to improve agricultural productivity to meet population growth (Eshel and Beeckman 2013). Therefore, studying the limitations to water acquisition and the mechanisms implemented by plants to overcome such limitations is especially important. Studying root growth dynamics as a response to changing conditions in soil moisture is crucial in the context of climate change and drought stress. Unprecedented summer droughts are becoming more frequent in central Europe (Torbenson et al. 2023). These are expected to increase in frequency, duration, and intensity in the future with potentially severe effects on vegetation (Klaus et al. 2022; Van Loon et al. 2016). In fact, the drought of summer 2022, when this study began, lasted from May until August and was one of the most intense drought in Europe in recent history (Tripathy and Mishra 2023). Plants have evolved strategies and root morphological and physiological plasticity in order to deal with the fluctuations in soil water availability in time and space (Fromm 2019; Hodge 2004). Among such strategies we find “Hydrotropism” and “Hydropatterning”. The former refers to directional root curvature and the latter to promotion of asymmetrical lateral root formation towards wet soil patches (Fromm 2019). At the seasonal scale, several studies documented root systems changing their vertical distribution seasonally as a response to soil moisture changes in different plant communities and under different climates (Hendrick and Pregitzer 1996; Peek et al. 2006; Wan et al. 2002; Wedderburn et al. 2010; Zhang et al. 2019; Zwetsloot and Bauerle 2021). Overall, these studies reported that root length was higher in shallow soil layers during the wet season. During the dry season, root growth increased towards deeper layers where soil moisture is usually higher. Ceolin et al. (2025) investigated experimentally daily time-scale root growth responses to rapid and local variations of soil moisture along the soil profile. The study was carried out under controlled conditions on 5-week old individual maize plants. In their study, a phenomenon coined as “hydromatching” was documented, which consists of enhanced local root proliferation (involving both new root emergence and elongation of pre-existing roots) in a wetted soil layer within 48 hours from a water pulse application. The increase in the wetted layer was accompanied by a decrease in root growth in other non-wetted layers. Ceolin et al. (2025) suspected that this was part of a strategy enabling the plant to explore dynamic soil moisture sources while economizing on root carbon investments. A logical progression of such moisture-manipulation lab studies would be to assess whether the observed rapid dynamics are also visible in a naturally established plant community (e.g., a grassland), where variations of other environmental factors can potentially affect the dynamics. Transferring such findings from lab to field might not be straightforward. In fact, in a study on ryegrass populations, roots did proliferate following re-wetting of a dry topsoil, but started to do so only one month after the re-watering event (Wedderburn et al. 2010). The delay in the response was interpreted as a way to ensure that carbon is invested in layers with sustained resource availability. Different studies observed different root responses to soil moisture deficit in plant communities. Some studies detected increased fine root production under drought (Leuschner et al. 2001; Teskey and Hinckley 1981). This would be consistent with the “functional balance theory” which states that plants actively adjust shoot:root carbon allocation to improve the uptake of the most limiting resource (Thornley 1972). Other studies observed a decrease in root production under drought (Metcalfe et al. 2008; Zwetsloot and Bauerle 2021). Different responses to dry conditions could be explained by different physiological processes and limitations, and by niche separation among species within a community (Leuschner et al. 2001; Zwetsloot and Bauerle 2021). When considering grasslands specifically, mixed species grasslands could be more resistant to drought effects in terms of ecosystem productivity (Craven et al. 2016; Isbell et al. 2015). This is because the two main components of grasslands, grasses and forbs, are known to respond differently to drought and to occupy different niches to avoid competition for water. Grasses tend to grow their root systems in the topsoil, effectively extracting water even under conditions of scarcity. In contrast, forbs tend to avoid growing roots in upper soil strata and overcome water scarcity by tapping into deeper soil layers (Nippert and Knapp 2007). Although it is widely recognized that soil moisture highly influences root growth dynamics, it is not clear whether it is the main driver of root growth in temperate grasslands. For example, one study found that root growth rates were strongly influenced by plant water status in a mesic grassland (Slette et al. 2023). In contrast, Wedderburn et al. (2010) reported that temperature was the main controller of root growth peaks in ryegrass populations. In another temperate grassland, root growth correlated positively with radiation but not with temperature (Edwards et al. 2004). Instead, precipitation was identified as the main determinant of vertical root distribution patterns in Mongolian temperate grasslands (Ma et al. 2008). Phenological factors may also play a role in driving root growth. While the importance of phenological drivers of root growth in temperate grasslands remain understudied, Joslin et al. (2001) documented that phenological factors were stronger than environmental variables in influencing root elongation rates in an oak stand. These scattered findings makes it challenging to identify the environmental variable with the strongest control on root growth in temperate grasslands, mostly because they result from comparisons that considered only two variables at a time. These observations also indicate that root growth dynamics might be dominated by different environmental (and phenological) factors in different situations. Moreover, the apparent importance of each environmental parameter might change based on observation frequency. In fact, low-frequency observations might inaccurately emphasize the importance of phenological factors (operating over long time scales) over rapidly changing environmental parameters such as soil moisture, temperature and solar radiation. In line with this, most studies investigating root growth dynamics as a response to seasonal moisture fluctuations and to drought measured at biweekly, monthly and even yearly intervals (Peek et al. 2006; Slette et al. 2023; Wan et al. 2002; Wedderburn et al. 2010; Zhang et al., 2019; Zwetsloot and Bauerle 2021), failing to capture potential rapid shifts in root allocation driven by sub-daily fluctuations in drying and rewetting conditions. Such studies might have missed short-term dynamics potentially entailing high rates of root productivity and turnover, as it was demonstrated that one-month sampling intervals might lead to 60% underestimation of root growth and mortality (Stewart and Frank 2008). In addition, insights on how soil water fluctuations affect the spatial and temporal organization of root systems could also enhance soil vegetation atmosphere transfer (SVAT) models, enabling more accurate predictions of water fluxes mediated by vegetation under dynamic water availability (Schwinning and Ehleringer 2001). In this study we build on earlier results of Ceolin et al. (2025) and (1) test if the phenomenon of hydromatching (observed on individual young plants under controlled conditions in unexplored soil) occurs also in response to natural rainfall variability in an established grassland community. We then aim to (2) capture vertical shifts in root distribution during the transition from wet to drier soil conditions, e.g. between a wet late spring and a dry early summer. Finally, our aim is to (3) test if soil moisture is an overall stronger driver than air and soil temperature, precipitation, irradiance, air humidity and wind speed for root growth rates in a temperate grassland under natural conditions. Specifically, we intend to answer the following research questions: • Question 1 (occurrence of hydromatching): Do the root systems of the grassland community exhibit hydromatching in response to rapid and localized changes in soil moisture? • Question 2 (seasonality of vertical root profiles): Do the root systems of the grassland community exhibit a vertical shift in root growth distribution at the seasonal scale, and is this correlated with soil moisture dynamics? • Question 3 (hierarchy of drivers of root growth): What are the environmental drivers of root growth in a temperate grassland? Materials and methods Field site and data collection The study was conducted in a managed grassland in Southern Luxembourg (49.5137°N, 6.0284°E, Fig. 1) that is mowed once per year and sporadically subjected to grazing (never during the study period). The weather station at the site is operated by the Luxembourg Institute of Science and Technology (LIST). The soil at the site is clay loam down to 40 cm depth and heavy clay from 40 to 130 cm (Marx and Flammang 2015). The plant species populating the site include both forbs and grasses and are common European temperate grassland species such as Lolium perenne, Trifolium pratense , Trifolium repens , Taraxacum officinale , Jacobaea vulgaris (or erratica ), Carduus crispus and Daucus carota . The climate is oceanic and characterized by mild winters and moderate summers. At the study site the long-term annual mean precipitation from 2004 to 2024 was 736 mm while the long-term mean temperature from 2006 to 2024 was 9.99 °C. Aboveground, the weather station is equipped with instruments measuring hourly air temperature and humidity (Humidity and Temperature Probe HMP155, Vaisala, Finland), precipitation intensity (Tipping Bucket Precipitation Sensor 15189, LAMBRECHT meteo GmbH, Germany), wind speed (A100LK-L Anemometer, Campbell Scientific, UT, USA) and irradiance (CMP3 Pyranometer, Kipp & Zonen, The Netherlands). Belowground, soil moisture and temperature sensors collecting half-hourly data are located at depths of 10, 20, 40 and 60 cm (CS650 30 cm Soil Moisture and Temperature Sensor and SoilVue 10 TDR Soil Moisture and Temperature Profile Sensor, Campbell Scientific, UT, USA). Three months before the start of data collection, we installed 12 acrylic minirhizotron tubes (180 cm long and 5 cm of diameter, CID Bioscience, Inc. Camas, WA, USA) along the fences of the weather station at an angle of 45° pointing outwards under the fences, approximately 50 cm apart and with 20 cm of tube sticking out of the ground (Fig. 1). Installation under the fence allowed us to capture the dynamics of undisturbed plants growing immediately outside the fenced weather station (and thus belonging to the grassland). Another reason for not installing the tubes outside the fence was to avoid tube damage by heavy machinery during annual mowing and to monitor root dynamics of individuals not exposed to mowing (the mower did not reach plants directly adjacent to the fence, as visible in Fig. 1). We believe the fence did not introduce any significant edge effect: the mesh was fine enough to avoid affecting light or wind passage, and although the fence poles could have altered soil structure and created preferential root growth zones, the minirhizotron tubes would have caused similar effects, so the potential bias was deemed negligible. The tubes reached a soil depth of approximately 115 cm. When not in use, polystyrene foam plugs were inserted inside the tubes and PVC caps were used to cover the protruding part of the tubes and avoid light disturbance. Roots growing along the external wall of the tubes were imaged using the CI-602 Narrow Gauge Root Imager (CID Bioscience, Inc. Camas, WA, USA). We imaged to a soil depth of 85 cm from May 2022 to September 2022 and to a soil depth of 115 cm from September 2022 until August 2023, after purchasing additional extension rods allowing us to image deeper. We took images every two weeks but increased the sampling frequency shortly after major precipitation events during the growing season. Major precipitation events were defined as events that increased the matric suction head (MSH) by at least 3 m at 10 cm depth, equivalent to the effect caused by experimental water pulses in Ceolin et al. (2025). To document the pre-existing root distribution before a major precipitation event, we imaged 1 to 4 days prior forecasted moderate rain events (>2 mm h −1 ) and kept measuring every 2 to 4 days after the start of the event for a total of approximately 12 days, resulting in at least three measurements taken closely post-precipitation. Image analysis Root images were corrected for overlaps between different depths and merged to obtain single images of the entire root profiles along each of the 12 minirhizotrons. We subdivided each root profile image in different sections covering different soil depths in order to better capture the root growth occurring in proximity of the soil moisture and temperature sensors located at 10, 20, 40 and 60 cm, which we term as “depths of interest”. Specifically, we considered the portions of the root profile images within 5 cm above and below the depths of 10 and 20 cm, and 10 cm above and below the depths of 40 and 60 cm. Therefore, despite having detected some roots down to the deepest part of the tubes (115 cm depth), we did not consider these roots in the analyses. Instead, we considered the roots growing nearby the 60 cm depth sensor as indicators of deep root activity. The Software “RootPainter” (originated by Smith et al. 2020) was used to detect alive roots in the images in an automated way. The software is a Convolutional Neural Networks (CNN) that, once trained on a suitable dataset, is able to segment roots in soil and to differentiate them from the background. During the training, roots were considered alive if they appeared bright white to the human eye. Subjectivity in root vitality detection based on visual color cues is common in minirhizotron studies (Vamerali et al., 2011). While this subjectivity was present during the training process, the use of the CNN on the full image set likely minimized additional user bias. The software “Rhizovision Explorer” was then used to extrapolate values of alive root length (Seethepalli et al. 2021). The detection of alive roots allowed us to detect root growth and root decline. Corrections through image manipulation have been implemented to reduce image noise. For example, edges of water droplets condensing along the minirhizotron wall were often confused as roots by the Software, so the segmented perimeter of the droplets was manually removed from the images. Also, images taken from 21/07/2023 onward presented problems of noise coming from camera malfunction, leading to the formation of straight bright stripes that were identified as roots by the Software. As in the previous case, a manual removal of the segmented stripes was performed. It is important to note that our data can be used to detect vertical (and relative) distributions of root length and sudden, local shifts in growth rates. However, the absolute values of root growth and root length decline are not reliable. In fact, our machine learning model was not yet accurate enough to consistently detect every root across all images, and a certain degree of noise was unavoidable. In addition, a precise recognition of root decay would require much more rigorous criteria of colour thresholding than the ones considered in this study. Data and statistical analysis We converted soil moisture data (measured as volumetric water content in m 3 m -3 ) to meters of matric suction head (MSH) by using the equations of the water retention model formulated by van Genuchten (1980). This was done to account for the role of textural classes in affecting soil water retention and availability to plants. Van Genuchten parameters for clay loam (down to 40 cm) and clay (from 40 cm downward) were obtained from Carsel and Parrish (1988) (Table 1). Table 1 : Van Genucthen parameters used for the conversion from volumetric water content (m 3 m −3 ) to soil water head (m).θ r (cm 3 cm −3 ) is residual soil water content, θ s (cm 3 cm −3 ) is saturated soil water content and a vG (cm−1), m vG and n vG are empirical parameters of the van Genuchten water retention model. We considered the clay loam parameters for the soil depth 0-40 cm and the clay parameters for the deeper depths. Van Genucthen parameters θr (cm3 cm−3) θs (cm3 cm−3) a vG (cm -3 ) nvG (-) Clay loam 0.095 0.410 0.019 1.310 Clay 0.068 0.380 0.008 1.090 We converted root length (mm) to net growth rates (mm day -1 ) by subtracting the root length of the previous sampling from the next sampling and by dividing that value by the number of days between events. To answer Question 1 (occurrence of hydromatching), we considered the growth rates observed in proximity of the sensors located at 10 and 60 cm and investigated the link between root growth and matric suction head. We also considered the differences between growth rate at 10 cm and 60 cm (with positive values indicating higher growth rate in the topsoil and vice versa). This was done to reveal changes in growth allocation between the top and subsoil, especially in circumstances of high root mortality (e.g.: under dry conditions). In this case, the differences in growth allocation between top and subsoil could still be meaningful while no clear response would be detected when considering the values of net growth rates only. Depths of 10 and 60 cm were considered in this analysis because they represented two distinct soil zones with contrasting moisture dynamics. The 10 cm topsoil experienced rapid and pronounced fluctuations in soil moisture, making it ideal for assessing root responsiveness to rewetting events due to rainfall. In contrast, the 60 cm subsoil remained consistently saturated and represented the “deep root pool”, where water was constantly available but harder for roots to access. Intermediate depths (20 and 40 cm) showed moisture patterns similar to 10 cm but with much weaker fluctuations, so including them would not add meaningful insight to this analysis. In our study, we selected a subset of the monitored events, specifically, three precipitation events that led to a substantial increase in soil water potential by at least 30 kPa (approximately 3 m of MSH) at 10 cm depth and for which we had root images right before and several right after the event. To determine if hydromatching occurred, we first used the Shapiro test to check for normal distribution of the values of net growth rates for each sampling event. We then compared them between subsequent sampling events by either using the Mann-Whitney test (when at least one of the two compared groups did not follow normality) or the Student t-test (when both of the considered groups followed normality) to check for significant differences. For hydromatching to occur, we expected a significant increase in growth rates in the topsoil and/or a significant decrease in the subsoil following major precipitation events. The Shapiro test, the Mann-Whitney test and the Student t-test were carried out using the Python library “scipy” (v1.1.0) and the package “stats” (Virtanen et al. 2020). To answer Question 2 (seasonality of vertical root profiles) we compared the vertical root length distributions along each depth of the 12 tube profiles on two sampling days, one in May (late spring, wetter season) and one in June (early summer, drier season) of both 2022 and 2023. These two time points differ in their seasonal characteristics but are both periods of high root productivity, making them suitable for comparing root responses to seasonally induced variations in soil moisture availability. Transitions involving autumn and winter were not considered, as dormancy-related processes could have confounded the interpretation of root dynamics. To test for significant vertical shifts in root length allocation we lumped together the root lengths at each depth from all 12 minirhizotrons. We compared the root lengths at each depth on the sampling event in May with the root lengths at the same depth in June for both years. We first used the Shapiro test to check for normal distribution of the root length values at each depth and in each sampling event. We then carried out the comparison with the Mann-Whitney test (when at least one of the two compared groups of root length did not follow normality) or with the Student t-test (when both groups followed normality). To answer Question 1 on the hierarchy of drivers of root growth, we used the median root growth rate at each depth, calculated across the 12 minirhizotrons, as the dependent variable in our analysis. Mean values between sampling times of soil moisture (expressed as both volumetric water content and matric suction head), air temperature, soil temperature, irradiance, relative humidity and wind speed were considered as the independent variables. Additionally, we considered the interactions of soil moisture at different depths with air temperature and soil temperature at different depths (e.g., soil moisture at 10 cm X air temperature; soil moisture at 20 cm X soil temperature at 60 cm) as independent variables. We calculated the interaction by multiplying the values of the interacting features, a recognized method in multiple regression analyses (Allison 1977; Li et al. 2020). We disregarded the data from the winter months to exclude the impact of plant dormancy on the results. We then used the dependent and independent variables to perform a Random forest analysis, a Ridge regression analysis and a Partial Least Square (PLS) analysis using the Python library “scikit-learn” (Pedregosa et al. 2011). Random Forest, Ridge regression and PLS have been used for analysis of feature importance and feature selection in multiple studies and fields (Hu et al. 2018; Shahhosseini et al. 2019; Toğaçar et al. 2020). These analyses compute coefficients indicating the importance of the independent variables (in our case the environmental parameters) in explaining the dependent variable (in our case root growth rate). We performed these analyses to determine the strongest predictors of root growth separately for each soil depth of interest (10, 20, 40 and 60 cm). For each soil depth, we performed the analysis with these three methods twice, once considering volumetric water content and once considering matric suction head instead to avoid redundancy and an excessive amount of "independent" variables per analysis. Eventually, we selected the analysis that yielded the highest R-squared for each soil depth of interest and considered the five independent features that ranked as the best predictors of root growth in each selected analysis. We used this top five of most influential independent variables to draw our conclusions on what affected root growth. The Python libraries “NumPy” (v1.15.4) (Harris et al., 2020), “pandas” (v0.21.0) (McKinney, 2010), “Matplotlib” (v3.1.0) (Hunter, 2007), and “seaborn” (v0.9.0) (Waskom, 2021) were used for numerical computation, data manipulation, and visualization. OpenAI was used to assist in code preparation during the initial round of analyses prior to subsequent adjustments. Results During the observation period from 01/05/2022 to 31/08/2023, MSH at 10 cm depth varied between saturation (during most of winter and following intense rain events) and -29 m (in late July 2023) (blue line in Fig. 2b). Root length showed clear fluctuations reflecting seasonality, with peaks at 10 cm in late spring (beginning-middle of May), peaks at 60 cm depth in early summer (end of May-early July) and reduced root length during winter months at both 10 and 60 cm depths (Fig. 2c and d). Mean annual precipitation and mean MSH at 10 cm were much lower in 2022 than in 2023, whereas mean air temperature was very similar (Table 2). Table 2 : Annual and long-term means of total precipitation (mm), air temperature (°C) and matric suction head (MSH, m) at 10 cm depth. Long-term means consider the period 2004-2024 for total precipitation, 2006-2024 for air temperature and 2018-2024 for MSH. Total precipitation (mm) Air temperature mean (°C) MSH mean at 10 cm depth (m) 2022 683.7 11.1 - 0.75 2023 965.3 10.9 - 0.38 Long-term mean 736.0 9.99 - 0.46 Occurrence of hydromatching We evaluated the occurrence of hydromatching by comparing the net growth rates and the top-bottom growth difference between sampling dates on three selected periods, termed as “September 2022”, “June 2023” and “July/August 2023”. These periods were characterized by major precipitation events replenishing the topsoil moisture, along with frequent imaging of the tubes (Fig. 3). Specifically, the September 2022 period included the post-drought rewetting event, which established an end to the long drought period that started in May. We observed the largest fluctuations in MSH over the entire study period at 10 cm depth, where values ranged from saturation (0 m) to a minimum of -30 m in July/August 2023 (Fig. 3). The soil at 60 cm was at saturation for the entirety of the study. Note that the highly clayey texture of the soil at this depth could have affected the readings of the sensors. In September 2022 we observed a root growth promotion in the topsoil 3 to 5 days after the soil wetting. The net growth rates at 10 cm (purple boxes in Fig. 3a) increased significantly between 27/08/2022 and 05/09/2022 (median of 0.05 mm day −1 to 2.33 mm day −1 ), as also evidenced by the progressive root length increment (purple boxes in Fig 3b). A change in image contrast between 27/08/2022 and 02/09/2022 did not allow to compare the growth rates on those days, preventing us to assess whether the response occurred potentially already on 02/09/2022. At 60 cm net growth rates decreased to -0.45 mm day −1 on 07/09/2022, the day when net growth rates at 10 cm were at their highest. The difference in growth rate between the top and subsoil increased between 02/09 and 07/09, indicating root growth allocation at 10 cm depth (at the expense of growth at 60 cm). In June 2023, 1 to 3 days after the onset of rain events the net growth rates at 10 cm increased significantly (from a median of -2.38 mm day −1 on 19/06/2023 to a median of 2.99 mm day −1 on 22/06/2023, purple boxes in Fig. 3c) and so did the top-bottom growth difference (grey line in Fig. 3c) indicating root growth promotion in the topsoil. At 60 cm, net growth rates significantly declined 1 to 3 days after the soil rewetting, reaching negative values and a minimum median of -25.76 mm day −1 on 22/06/2023 (light blue boxes in Fig. 3c). This was the same day when we observed the maximum positive growth rate reached at 10 cm. When soil started drying up, root growth at 60 cm progressively increased and reached 12.44 mm day −1 on 28/06/2023, with the top-bottom growth difference significantly decreasing and reaching negative values on 28/06/2023. This indicates that root promotion progressively switched from the top to the bottom of the profile when dry conditions developed again. The root length variations at 10 and 60 cm depths (Fig. 3d) also reflect these growth dynamics, although less clearly. July/August 2023 was characterized by a long rainy period during which the growth rates at 10 and at 60 cm remained near or below zero (Fig. 3e), with noticeable declines in root length especially at 60 cm (Fig. 3f). Net growth rates at both depths started increasing only towards the end of the period. Vertical root distribution changes between spring and summer MSH progressively decreased at 10 cm depth from May to June in both 2022 and 2023. However, the change in MSH from May to June was much more dramatic in 2023 compared to 2022. At 10 cm depth, MSH dropped from 0 m (saturation) to -15 m in 2023, compared to a smaller decrease from -2.75 to -5.85 in 2022 (Fig. 4:4). Despite the two years having different magnitudes of soil moisture change over the spring-summer transition, we observed a similar and significant shift in root length from the shallower soil depths to the deeper soil depths with the progression of the dry season in both years (Fig. 4:4). The shift consisted of a decline in root length at shallow depths and a promotion of root growth in the deeper layers. In both years, between May and June, root length decreased significantly at 5-30 cm depth and increased significantly at 60-80 cm depth (Fig. 4:4). Importance of water or temperature on root growth dynamics Random Forest, Ridge analysis and Partial Least Square (PLS) regressions performed differently for each depth of interest and differently depending on whether we considered volumetric water content (VWC) or matric suction head (MSH) as an independent variable. The R-squared resulting from the Random Forest, Ridge regression and PLS considering the growth rates at 10, 20, 40 and 60 cm and considering either VWC or MSH are presented in Table 3. Several reported R-squared values are negative. Although seemingly counter-intuitive, a negative value of R-squared computed by regression models can occur when the regression model’s predictions are worse than those of a null model represented by a constant value (Nakagawa and Schielzeth 2013). The growth rates at 10 cm were best predicted by the Ridge regression considering VWC (R2=0.5), the growth rates at 20 cm were best modeled by the Random Forest considering VWC (R2=0.26), the growth rates at 40 cm were best predicted by the Ridge regression considering MSH (R2=0.55) and the growth rates at 60 cm were best predicted by the PLS considering MSH (R2=0.24). The top 5 predictors of root growth varied between soil depths and differed in their directional influence (Fig. 5). We consistently observed soil moisture (as VWC and MSH) ranking highest among the top 5 features at all depths except 20 cm and 60 cm. At 20 cm, one soil moisture feature (VWC at 10 cm) ranked 3rd right after irradiance and air humidity. At 60 cm, all of the top 5 most influential features were interactions between MSH at 10 and 20 cm and temperatures from the above soil layers or air temperature. These results could indicate a larger influence of temperature on root growth at 60 cm, but are most likely shaped by the invariability of soil moisture at this depth. Table 3 : R-squared values deriving from the Random Forest, Ridge Regression and Partial Least Square Regression computed for each depth considering root growth rate as the dependent variable. We reported the results considering soil moisture either as volumetric water content (VWC) or matric suction head (MSH) R2 values Random forest Ridge regression Partial least square regression Dependent variable Considering VWC Considering MSH Considering VWC Considering MSH Considering VWC Considering MSH Growth rate at 10 cm 0.26 0.26 0.50 0.12 -2.22 -0.88 Growth rate at 20 cm 0.27 -0.65 -0.30 -0.43 -1.47 -0.41 Growth rate at 40cm -0.48 -1.04 0 0.55 -0.35 -1.02 Growth rate at 60 cm 0.03 0.03 -0.72 -0.69 -0.06 0.23 Discussion Grassland root systems exhibit hydromatching Hydromatching was previously observed as rapid switches in local root growth patterns occurring within 48 hours following local changes in soil moisture (Ceolin et al. 2025). However, that study was performed on young individual maize plants grown in confined, unexplored soil. The plants never experienced critical water stress, as soil moisture was maintained above 6% at all time. In contrast, the plants in our temperate grassland community experienced prolonged periods of water deficit, ultimately leading to leaf senescence. A declining temperate grassland community is expected to behave differently than healthy, individual maize plants. Therefore, the effects of severe water stress need to be considered when interpreting our results and comparing them with the laboratory study of Ceolin et al. (2025), with the expectation of more altered or dampened root plastic responses to soil moisture variations. Another caveat is that the lab-based study used Magnetic Resonance Imaging (MRI) to visualize entire root systems in three dimensions, whereas the minirhizotron technique used here captures only two-dimensional images along a plexiglass-soil interface. Despite these caveats, we observed evidence of hydromatching within 2-5 days in the September 2022 event and within 1-3 days in the June 2023 event (Fig. 3a and c). Plants rapidly adapted their local root growth rates to exploit the most resourceful soil layers and interrupting root investment into soil layers requiring more carbon investment per amount of water uptake. When soil at 10 cm depth was re-wetted, soil at 60 cm became a more “costly” source of water. This could be due to its heavy clay composition, likely retaining water more tightly, imposing a higher mechanical resistance and creating conditions of low aeration (da Silva and Kay 1997). Top and subsoil likely swapped the role of “most resourceful” soil layer in a matter of days, mostly depending on the moisture conditions of the topsoil. These results and interpretation reveal a highly dynamic nature of the root systems in this grassland and are in line with the ones from Ceolin et al. (2025). It then appears that hydromatching is also observable in natural settings at a grassland community level, and not only under controlled conditions at the individual plant level (as in Ceolin et al., 2025). Hydromatching occurred as a collective response, as we did not discriminate between roots belonging to different species or ecological groups. In particular, it was not possible to distinguish between roots belonging to forbs and grasses. Both groups were present throughout the whole sampling campaign, as individuals of Lolium perenne (perennial grass) and a series of other annual forbs were consistently present at the study site. Therefore, we can assume that during the entire observation period forbs and grasses coexisted, meaning that hydromatching was identifiable despite niche separations and despite the different drought-adaptation strategies between forbs and grasses (Nippert and Knapp 2007). Previous studies on root morphological adjustments to soil moisture variations focused on large time scales that mostly captured seasonal or even annual dynamics of root growth (Peek et al. 2006; Wan et al. 2002; Wedderburn et al. 2010; Zhang et al. 2019; Zwetsloot and Bauerle 2021). However, studies at such large time scales overlook the daily adjustments to quick changes in soil moisture, which were revealed in the present study. To our knowledge, studies documenting local root growth patterns in natural grassland communities at such high frequency are lacking, especially ones comparing pre and post precipitation events. The daily time scale dynamics may also be important in the context of carbon and nutrient cycling, ecosystem productivity and in soil-vegetation-atmosphere transfer models. Our results also allow to identify factors potentially limiting the manifestation and the observation of hydromatching, i.e. soil dryness persistence prior to the rewetting event, water logging, energy limitation, herbivory and nutrient remobilization. In September 2022, high levels of water stress prior to the rain event were caused by the prolonged drought period started in May, which led to apparent signs of leaf senescence. This could have caused a decrease in photosynthetic rates and a reduction of carbon allocatable to roots (Zwetsloot and Bauerle 2021) which, in turn, might have delayed the response to the re-wetting in September 2022 compared to June 2023. Furthermore, a rapid and significant shift in net growth rates shortly after the re-wetting was probably less of a “risk factor” in June 2023 than it could have been in September 2022. In September 2022, the root growth shift towards the topsoil could have been delayed to mitigate the risk of losing newly grown roots due to potentially renewed dry conditions (Wedderburn et al. 2010). Another strong limitation to the occurrence of hydromatching was given by water logging and energy limitation under the form of solar radiation and temperature, as evidenced by the results from July/August 2023. Decreased irradiance and/or lower temperatures during the 14-day rain event likely inhibited root growth (Fig. 3e), suggesting that root investment in the grassland was likely simultaneously guided by shoot water demand and limited by shoot carbon supply. Also, the induced conditions of soil saturation could have hindered root growth, as soil saturation is known to restrict root activity and lead to root decay due to lack of aeration (Drew 1992). Waterlogging conditions and severe levels of anoxia were likely the cause of the strong decline in root length observed at 60 cm from 24/07 to 05/08 (Fig. 3f). Herbivory represented another potential limitation in the detection of hydromatching. This could explain the sudden and significant levels of root length decline observed at 10 cm depth at the end of the September 2022 event (Fig. 3a and b), given the presence of unidentified invertebrates in several images taken at shallow depths in this period. This means that root loss due to predation should be regarded as a potential limitation in minirhizotron studies focusing on root dynamics. Finally, another possible explanation that cannot be ruled out for the fast switch in root growth patterns could be nutrient scavenging in the (normally) nutrient-rich topsoil, triggered by the re-wetting (Armas et al. 2012). Future studies should investigate the effect of nutrient remobilization on root growth rates in grasslands shortly after rain events. Root growth shifts from spring to summer The increase in root length in deeper soil layers during the transition from late spring to early summer (Fig. 4) is consistent with previous studies observing root growth promotion in the subsoil as a dry period progressed (Hendrick and Pregitzer, 1996; Peek et al. 2006; Wan et al. 2002; Weddenburn et al., 2010, Zhang et al. 2019, Zweetslot and Bauerle, 2021). These studies suggested that plants can redistribute roots at different depths according to the long-term seasonality of water availability along the soil profile. In this study, not only root length significantly increased in the deeper soil (60-80 cm) but it also significantly decreased in the upper soil layers (0-30 cm) during the transition to summer. Such behaviour was also observed in a juniper stand study, where a vertical transition in soil water availability was accompanied by a progressive downward shift in fine root growth and soil water use (Peek et al. 2006). Our observation of lower root lengths in May 2023 compared to May 2022 might have been due to near-saturation conditions across all depths between March and May 2023, which are known to hinder root growth (Drew 1992). Another explanation for the overall lower root length in 2023 could be seedbank depletion caused by the intense drought of summer 2022 (Zhou et al. 2022). Different studies demonstrated how droughts can lead to legacy effects impacting root production in the following years (Slette et al. 2023; Zhou et al. 2022). While the accuracy in the measurements of root length decline was limited, the significant decline observed consistently in both years suggests that root loss in the topsoil likely occurred and was not an artefact caused by uncertainty in the image analysis. The significant decline of root length at 20 cm depth in both 2022 and 2023, despite relatively moist soil at that depth (-1.2 m MSH in June 2022 and -2.7 m in June 2023, Fig. 4:4b and d), indicates that at least a portion of the root senescence occurring over a month in the upper layers was not caused by water deprivation, but was likely a programmed mechanism. This could be seen as a mechanism to reduce root maintenance costs and save resources to be used for growing and maintaining roots in the subsoil. The potential existence of such a cost-effective strategy of carbon allocation reflecting soil moisture availability was already discussed by Ceolin et al. (2025), after observing both root growth promotion in wetter soil layers and root growth suppression (and potential root loss) in drier layers. The shifts in vertical root distribution that we detected indicate a top-bottom carbon allocation transition, with a clear separation between shallow and deep parts of the root system. It is arguable that the observed shift in root abundance from topsoil to subsoil reflects a change in dominance from shallow- to deep-rooted species. However, such a profound shift in community composition is unlikely to occur within just over a month. Moreover, the similar root length at 40-50 cm in May and June in both years indicates that deeper roots were already established and remained stable at that depth over time, suggesting they likely belonged to the same individuals. If a dominance shift had occurred, it is improbable that root length at 40-50 cm would remain unchanged. This supports the idea that the observed changes were not due to species turnover or mortality of shallow-rooted plants, but rather to plastic responses within the same individuals shifting root growth from more populated shallow layers to less populated deeper ones. Furthermore, it is unlikely that newly established deep-rooted species extended roots to 80 cm within a month; it is more plausible that pre-existing deep root systems expanded further while surface root growth declined. The similar extents of the shifts from May to June in both 2022 and 2023, despite the different levels of MSH recorded in early summer, suggest that these shifts were phenologically-driven processes. This partially aligns with the ‘’phenological programming’’ theory (Hendrick and Pregitzer, 1996; Joslin et al., 2001), which applies to deciduous trees that evolved in environments where summer droughts are the norm. It proposes that a programmed burst of root growth following foliar expansion in late spring is advantageous, as it normally allows to establish an extensive root system when moisture and temperature conditions are favourable and carbon supply is adequate. However, this theory does not account for potential phenologically programmed vertical shifts in root growth that can occur during the transition from spring to summer, which we observed here. Furthermore, to our knowledge, there have been no previous reports of root decay as a programmed mechanism. Overall, our findings suggest the existence of two types of root morphological adaptations occurring at different time scales in our grassland community. At daily time scales and at local, small spatial scales (within a soil layer of few cm), root growth rates appear to be strongly influenced by rapidly changing environmental factors, such as soil moisture. Such daily-scale local adjustments in growth rates could be guided by both local changes in soil moisture and potentially by changes in soil moisture occurring elsewhere, according to the theory of the “whole-root system coordination” described by Ceolin et al. (2025). At the long-term (seasonal) and at the whole-root system scale, root distribution appears to undergo drastic and significant adjustments, occurring as a programmed, phenologically-dictated processes, independent of rapidly changing environmental variables. Both short-scale, environmentally dictated and long-scale, phenologically dictated adaptations seem to share a common goal, which is meeting the water requirements while flexibly managing carbon budgets. Carbon is spent where a return is expected, while the cost of root maintenance is cut where no resource can be gained. In our case, such adaptations were community responses, although similar adjustments are observable in individual plants as well (Ceolin et al. 2025; Engels et al., 1994). Moisture controls growth rates Random Forest, Ridge regression and PLS ranked soil moisture (as VWC and MSH) as the overall strongest driver of grassland root growth rate at the study site (Fig. 5). This result aligns with a study on temperature grasslands in inner Mongolia, which found that precipitation, rather than temperature, was the main driver of root vertical distribution patterns. In contrast, prior research identified temperature, solar radiation and phenological factors as the stronger drivers of root growth in grasslands and tree stands (Edwards et al. 2004; Joslin, Wolfe, and Hanson 2001; Wedderburn et al. 2010). However, these studies employed a low-frequency (weekly-monthly) fixed sampling schedule, so they were not designed to detect the fast responses to wetting as observed in our study, where soil moisture appeared as a key driver of root growth. This result is consistent with the controlled lab experiment by Ceolin et al. (2025), where soil moisture was manipulated, but the other environmental variables such as temperature, irradiance and air humidity were kept constant. In this present study we found that variations in soil moisture had the strongest influence on root growth even under natural conditions, where other environmental factors co-varied. Soil moisture was particularly important in influencing root growth at 10 cm and 40 cm depth, where the best predictors were VWC at 10 and 20 cm and MSH at 20 and 10 cm, respectively (Fig. 5a and c). Overall, soil moisture ranked higher than temperature at all depths except 60 cm, where soil moisture was constant. This is particularly remarkable considering that our grassland is located in an energy-limited environment (McVicar et al. 2012). Ma et al. (2008) found that root distribution patterns were strongly driven by precipitation in temperate grasslands, but they state that this was likely due to the overall dry climate and water-limited environment of inner Mongolia. Instead, previous studies suggested that radiation flux (Edwards et al. 2004) or temperature (Kaspar and Bland 1992) is likely the main driver of root growth in temperate climates, as well as in northern forest ecosystems where water limitation is rare (Montagnoli et al. 2014). However, Wedderburn et al. (2010), who conducted their study in an energy-limited area, found that, after a drought period, the typically dominant effect of temperature on root development was overtaken by the effect of soil rewetting. The stimulative effect of an environmental parameter on root growth might therefore depend on the history of the stresses experienced by the plant community: under typical temperate conditions without abnormal drought, energy-related factors generally control root growth (e.g., Edwards et al. 2004; Kaspar and Bland 1992) but, after severe water limitation, sensitivity of root dynamics may shift from energy to water-sensible and become more responsive to moisture variation (e.g., Wedderburn et al. 2010). As our study period included the notable 2022 drought, it is possible that the water stress experienced during that time gave soil moisture a greater influence over growth rates than it would have had in previous years. This change in influence could have persisted during the following growing season of 2023, as droughts are known to have legacy effects on root development lasting years (Slette et al. 2023; Zhou et al. 2022). These mechanisms represent a distinct form of root plasticity related to stress memory, suggesting that soil moisture may become an increasingly dominant driver of root dynamics under climate change in central European temperate grasslands. Such flexibility may be crucial for temperate grassland communities to cope with climatic extremes. Nevertheless, further research is needed to clarify whether what we observed represents a true “priority shift’’ in root growth drivers and how widespread or persistent this phenomenon may be. A limitation of the performed regression analyses is the limited size of the dataset. Although the higher frequency of observation allowed us to better capture the effects of rapidly changing environmental parameters, the impossibility to analyze subsets of data prevented a clear distinction of effects on root growth given by seasonality/phenology and effects dictated by rapid and isolated soil environmental changes. Conclusions The root systems of the grassland community implemented drastic seasonal adjustments (between spring and summer) involving root length decline in shallow soil layers and root growth promotion in deeper layers, offering evidence in response to Question 2 regarding the seasonality of vertical root profiles. At the same time, the grassland’s root systems were able to deploy ‘’hydromatching’’ in two out of our three selected periods, by favouring root growth in the topsoil at the expense of the subsoil, 1–3 days after a rain event. This offers proof in response to Question 1, related to the occurrence of hydromatching. The seasonal shifts involved the entire root systems and were likely driven by phenological factors, whereas hydromatching consisted of more localized adjustments following rapid changes in soil moisture and was therefore dictated by environmental factors. Energy limitation (given by lower solar radiation and temperature) and a saturated soil in July/August 2023, together with the prolonged conditions of water stress prior to the rain event in September 2022, could have been important limitations to hydromatching. Although we cannot fully rule out the potential role of nutrients in influencing root growth dynamics, our results still strongly suggest that hydromatching is a naturally occurring phenomenon observable in established grassland communities. Both observed long-term and short-term growth rate modifications might represent strategies evolved to cope with soil moisture heterogeneity alongside carbon budgeting. Lastly, soil moisture had the strongest control on root growth rates at the investigated temperate grassland, followed by soil temperature. This provides evidence in response to Question 3, related to the hierarchy of environmental drivers of root growth. Our findings hold important implications for an improved understanding of plant carbon allocation strategies, plant population dynamics and biogeochemical processes such as carbon and nutrient cycling. Future studies could focus on quantifying fine root shedding, as it seems to be deeply integrated in the root growth dynamics as a response to soil moisture heterogeneity. It may also be important to investigate to what extent these adjustments facilitate the fulfilment of the transpiration demand. Future studies should also examine the extent to which typically energy-limited grassland communities can implement soil moisture-driven root dynamics following abnormal drought events, a key question in vision of climate change and increasing drought stress in temperate ecosystems. Ultimately, consideration of plant water stress indicators could help explain contradicting findings about root responsiveness to re-hydration. Other factors might also play a role, such as legacy effects of previous droughts, nutrient limitation, and prolonged periods of low temperature and solar radiation. Abbreviations VWC: Volumetric Water Content; MSH: Matric Suction Head Declarations Acknowledgements We thank Jérôme Juilleret for his valuable assistance during the sampling campaign and Jean François Iffly for his crucial support in site selection, minirhizotron installation and for configuring and maintaining the weather station instrumentation. We also thank Abraham George Smith for his significant assistance in operating the RootPainter software. Funding This research has been supported by the Fonds National de la Recherche Luxembourg (grant no. AFR PhD/19/SR/13577787). Competing interests The authors declare no competing interests. Author contributions Concept and experimental design: SC, SJS and JK. Field study preparation and measurements: SC and SJS. Data analysis: SC. Data interpretation: SC, SJS and JK. Paper preparation: SC, SJS and JK. All authors commented on previous versions of the manuscript . All authors read and approved the final manuscript. Supervision: SJS (main supervisor), JK (co-supervisor). Data availability The datasets generated during the current study and code for data analysis are available at https://doi.org/10.5281/zenodo.10836040, https://doi.org/10.5281/zenodo.10528310, https://doi.org/10.5281/zenodo.10528980, https://doi.org/10.5281/zenodo.10530073, https://doi.org/10.5281/zenodo.10551976, https://doi.org/10.5281/zenodo.10679137. Multiple repositories were required to upload all the minirhizotron images due to storage limitations. References Allison, Paul D. 1977. ‘Testing for Interaction in Multiple Regression’. American Journal of Sociology 83(1):144–53. doi:10.1086/226510. Armas, Cristina, John H. Kim, Timothy M. Bleby, and Robert B. Jackson. 2012. ‘The Effect of Hydraulic Lift on Organic Matter Decomposition, Soil Nitrogen Cycling, and Nitrogen Acquisition by a Grass Species’. Oecologia 168(1):11–22. doi:10.1007/s00442-011-2065-2. Carsel, Robert F., and Rudolph S. Parrish. 1988. ‘Developing Joint Probability Distributions of Soil Water Retention Characteristics’. Water Resources Research 24(5):755–69. doi:10.1029/WR024i005p00755. Ceolin, Samuele, Stanislaus J. Schymanski, Dagmar van Dusschoten, Robert Koller, and Julian Klaus. 2025. “Root Growth Dynamics and Allocation as a Response to Rapid and Local Changes in Soil Moisture.” Biogeosciences 22(3):691–703. doi:10.5194/bg-22-691-2025. Conant, Richard T., Keith Paustian, and Edward T. Elliott. 2001. ‘Grassland Management and Conversion into Grassland: Effects on Soil Carbon’. Ecological Applications 11(2):343–55. doi:10.1890/1051-0761(2001)011[0343:GMACIG]2.0.CO;2. Craven, Dylan, Forest Isbell, Pete Manning, John Connolly, Helge Bruelheide, Anne Ebeling, Christiane Roscher, Jasper van Ruijven, Alexandra Weigelt, Brian Wilsey, Carl Beierkuhnlein, Enrica de Luca, John N. Griffin, Yann Hautier, Andy Hector, Anke Jentsch, Jürgen Kreyling, Vojtech Lanta, Michel Loreau, Sebastian T. Meyer, Akira S. Mori, Shahid Naeem, Cecilia Palmborg, H. Wayne Polley, Peter B. Reich, Bernhard Schmid, Alrun Siebenkäs, Eric Seabloom, Madhav P. Thakur, David Tilman, Anja Vogel, and Nico Eisenhauer. 2016. ‘Plant Diversity Effects on Grassland Productivity Are Robust to Both Nutrient Enrichment and Drought’. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 371(1694):20150277. doi:10.1098/rstb.2015.0277. Drew, Malcolm C. 1992. ‘SOIL AERATION AND PLANT ROOT METABOLISM’. Soil Science 154(4):259. Edwards, Everard J., David G. Benham, Louise A. Marland, and Alastair H. Fitter. 2004. ‘Root Production Is Determined by Radiation Flux in a Temperate Grassland Community’. Global Change Biology 10(2):209–27. doi:10.1111/j.1365-2486.2004.00729.x. Engels, Christof, Martin Mollenkopf, and Horst Marschner. 1994. ‘Effect of Drying and Rewetting the Topsoil on Root Growth of Maize and Rape in Different Soil Depths’. Zeitschrift Für Pflanzenernährung Und Bodenkunde 157(2):139–44. doi:10.1002/jpln.19941570213. Eshel, Amram, and Tom Beeckman, eds. 2013. Plant Roots: The Hidden Half, Fourth Edition . 4 edition. Boca Raton, FL: CRC Press. Fay, Philip A., Suzanne M. Prober, W. Stanley Harpole, Johannes M. H. Knops, Jonathan D. Bakker, Elizabeth T. Borer, Eric M. Lind, Andrew S. MacDougall, Eric W. Seabloom, Peter D. Wragg, Peter B. Adler, Dana M. Blumenthal, Yvonne M. Buckley, Chengjin Chu, Elsa E. Cleland, Scott L. Collins, Kendi F. Davies, Guozhen Du, Xiaohui Feng, Jennifer Firn, Daniel S. Gruner, Nicole Hagenah, Yann Hautier, Robert W. Heckman, Virginia L. Jin, Kevin P. Kirkman, Julia Klein, Laura M. Ladwig, Qi Li, Rebecca L. McCulley, Brett A. Melbourne, Charles E. Mitchell, Joslin L. Moore, John W. Morgan, Anita C. Risch, Martin Schütz, Carly J. Stevens, David A. Wedin, and Louie H. Yang. 2015. ‘Grassland Productivity Limited by Multiple Nutrients’. Nature Plants 1(7):1–5. doi:10.1038/nplants.2015.80. Fromm, Hillel. 2019. ‘Root Plasticity in the Pursuit of Water’. Plants 8(7):236. doi:10.3390/plants8070236. van Genuchten, M. Th. 1980. ‘A Closed-Form Equation for Predicting the Hydraulic Conductivity of Unsaturated Soils’. Soil Science Society of America Journal 44(5):892–98. doi:https://doi.org/10.2136/sssaj1980.03615995004400050002x. Gill, Richard A., Ingrid C. Burke, William K. Lauenroth, and Daniel G. Milchunas. 2002. ‘Longevity and Turnover of Roots in the Shortgrass Steppe: Influence of Diameter and Depth’. Plant Ecology 159(2):241–51. doi:10.1023/A:1015529507670. Hayes, D. C., and T. R. Seastedt. 1987. ‘Root Dynamics of Tallgrass Prairie in Wet and Dry Years’. Canadian Journal of Botany 65(4):787–91. doi:10.1139/b87-105. Hendrick, Ronald L., and Kurt S. Pregitzer. 1996. ‘Temporal and Depth-Related Patterns of Fine Root Dynamics in Northern Hardwood Forests’. The Journal of Ecology 84(2):167. doi:10.2307/2261352. Hodge, Angela. 2004. ‘The Plastic Plant: Root Responses to Heterogeneous Supplies of Nutrients’. New Phytologist 162(1):9–24. doi:https://doi.org/10.1111/j.1469-8137.2004.01015.x. Hu, Yutong, Xiaorong Wei, Mingde Hao, Wei Fu, Jing Zhao, and Zhe Wang. 2018. ‘Partial Least Squares Regression for Determining Factors Controlling Winter Wheat Yield’. Agronomy Journal 110(1):281–92. doi:10.2134/agronj2017.02.0108. Isbell, Forest, Dylan Craven, John Connolly, Michel Loreau, Bernhard Schmid, Carl Beierkuhnlein, T. Martijn Bezemer, Catherine Bonin, Helge Bruelheide, Enrica de Luca, Anne Ebeling, John N. Griffin, Qinfeng Guo, Yann Hautier, Andy Hector, Anke Jentsch, Jürgen Kreyling, Vojtěch Lanta, Pete Manning, and Sebastian T. Meyer. 2015. ‘Biodiversity Increases the Resistance of Ecosystem Productivity to Climate Extremes’. Nature 526(7574):574–77. doi:10.1038/nature15374. Joslin, J. D., M. H. Wolfe, and P. J. Hanson. 2001. ‘Factors Controlling the Timing of Root Elongation Intensity in a Mature Upland Oak Stand’. Plant and Soil 228(2):201–12. Kaspar, T. C., and W. L. Bland. 1992. ‘SOIL TEMPERATURE AND ROOT GROWTH’: Soil Science 154(4):290–99. doi:10.1097/00010694-199210000-00005. Klaus, Julian, Wendy A. Monk, Lu Zhang, and David M. Hannah. 2022. ‘Ecohydrological Interactions during Drought’. Ecohydrology 15(5):e2456. doi:10.1002/eco.2456. Leuschner, Ch, K. Backes, D. Hertel, F. Schipka, U. Schmitt, O. Terborg, and M. Runge. 2001. ‘Drought Responses at Leaf, Stem and Fine Root Levels of Competitive Fagus Sylvatica L. and Quercus Petraea (Matt.) Liebl. Trees in Dry and Wet Years’. Forest Ecology and Management 149(1):33–46. doi:10.1016/S0378-1127(00)00543-0. Li, Xinhai, Baidu Li, Guiming Wang, Xiangjiang Zhan, and Marcel Holyoak. 2020. ‘Deeply Digging the Interaction Effect in Multiple Linear Regressions Using a Fractional-Power Interaction Term’. MethodsX 7:101067. doi:10.1016/j.mex.2020.101067. Ma, WenHong, YuanHe Yang, JinSheng He, Hui Zeng, and JingYun Fang. 2008. ‘Above- and Belowground Biomass in Relation to Environmental Factors in Temperate Grasslands, Inner Mongolia’. Science in China Series C: Life Sciences 51(3):263–70. doi:10.1007/s11427-008-0029-5. Marx, Simone, and Frank Flammang. 2015. ‘La cartographie des sols au Grand-Duché de Luxembourg’. McVicar, Tim R., Michael L. Roderick, Randall J. Donohue, Ling Tao Li, Thomas G. Van Niel, Axel Thomas, Jürgen Grieser, Deepak Jhajharia, Youcef Himri, Natalie M. Mahowald, Anna V. Mescherskaya, Andries C. Kruger, Shafiqur Rehman, and Yagob Dinpashoh. 2012. ‘Global Review and Synthesis of Trends in Observed Terrestrial Near-Surface Wind Speeds: Implications for Evaporation’. Journal of Hydrology 416–417:182–205. doi:10.1016/j.jhydrol.2011.10.024. Metcalfe, Daniel B., Patrick Meir, Luiz Eduardo O. C. Aragão, Antonio C. L. da Costa, Alan P. Braga, Paulo H. L. Gonçalves, Joao de Athaydes Silva Junior, Samuel S. de Almeida, Lorna A. Dawson, Yadvinder Malhi, and Mathew Williams. 2008. ‘The Effects of Water Availability on Root Growth and Morphology in an Amazon Rainforest’. Plant and Soil 311(1):189–99. doi:10.1007/s11104-008-9670-9. Montagnoli, A., A. Di Iorio, M. Terzaghi, D. Trupiano, G. S. Scippa, and D. Chiatante. 2014. ‘Influence of Soil Temperature and Water Content on Fine-Root Seasonal Growth of European Beech Natural Forest in Southern Alps, Italy’. European Journal of Forest Research 133(5):957–68. doi:10.1007/s10342-014-0814-6. Nakagawa, Shinichi, and Holger Schielzeth. 2013. ‘A General and Simple Method for Obtaining R 2 from Generalized Linear Mixed‐effects Models’ edited by R. B. O’Hara. Methods in Ecology and Evolution 4(2):133–42. doi:10.1111/j.2041-210x.2012.00261.x. Nippert, Jesse B., and Alan K. Knapp. 2007. ‘Soil Water Partitioning Contributes to Species Coexistence in Tallgrass Prairie’. Oikos 116(6):1017–29. doi:10.1111/j.0030-1299.2007.15630.x. Pedregosa, Fabian, Gael Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, Jake Vanderplas, Alexandre Passos, and David Cournapeau. 2011. ‘Scikit-Learn: Machine Learning in Python’. MACHINE LEARNING IN PYTHON . Peek, M. S., A. J. Leffler, L. Hipps, S. Ivans, R. J. Ryel, and M. M. Caldwell. 2006. ‘Root Turnover and Relocation in the Soil Profile in Response to Seasonal Soil Water Variation in a Natural Stand of Utah Juniper (Juniperus Osteosperma)’. Tree Physiology 26(11):1469–76. doi:10.1093/treephys/26.11.1469. Saelim, Sakanan, Sayan Sdoodee, and Rawee Chiarawipa. 2019. ‘Monitoring Seasonal Fine Root Dynamics of Hevea Brasiliensis Clone RRIM 600 in Southern Thailand Using Minirhizotron Technique’. 8. Schwinning, Susanne, and James R. Ehleringer. 2001. ‘Water Use Trade-Offs and Optimal Adaptations to Pulse-Driven Arid Ecosystems’. Journal of Ecology 89(3):464–80. Seethepalli, Anand, Kundan Dhakal, Marcus Griffiths, Haichao Guo, Gregoire T. Freschet, and Larry M. York. 2021. RhizoVision Explorer: Open-Source Software for Root Image Analysis and Measurement Standardization . preprint . Plant Biology. doi:10.1101/2021.04.11.439359. Shahhosseini, Mohsen, Rafael A. Martinez-Feria, Guiping Hu, and Sotirios V. Archontoulis. 2019. ‘Maize Yield and Nitrate Loss Prediction with Machine Learning Algorithms’. Environmental Research Letters 14(12):124026. doi:10.1088/1748-9326/ab5268. da Silva, Alvaro Pires, and B. D. Kay. 1997. ‘Estimating the Least Limiting Water Range of Soils from Properties and Management’. Soil Science Society of America Journal 61(3):877–83. doi:10.2136/sssaj1997.03615995006100030023x. Slette, Ingrid J., David L. Hoover, Melinda D. Smith, and Alan K. Knapp. 2023. ‘Repeated Extreme Droughts Decrease Root Production, but Not the Potential for Post-Drought Recovery of Root Production, in a Mesic Grassland’. Oikos 2023(1):e08899. doi:10.1111/oik.08899. Smith, Abraham George, Eusun Han, Jens Petersen, Niels Alvin Faircloth Olsen, Christian Giese, Miriam Athmann, Dorte Bodin Dresbøll, and Kristian Thorup-Kristensen. 2020. RootPainter: Deep Learning Segmentation of Biological Images with Corrective Annotation . preprint . Plant Biology. doi:10.1101/2020.04.16.044461. Stewart, Anna M., and Douglas A. Frank. 2008. ‘Short Sampling Intervals Reveal Very Rapid Root Turnover in a Temperate Grassland’. Oecologia 157(3):453–58. doi:10.1007/s00442-008-1088-9. Teskey, Robert O., and Thomas M. Hinckley. 1981. ‘Influence of Temperature and Water Potential on Root Growth of White Oak’. Physiologia Plantarum 52(3):363–69. doi:10.1111/j.1399-3054.1981.tb06055.x. Thornley, J. H. M. 1972. ‘A Balanced Quantitative Model for Root: Shoot Ratios in Vegetative Plants’. Annals of Botany 36(145):431–41. Toğaçar, Mesut, Burhan Ergen, and Zafer Cömert. 2020. ‘Application of Breast Cancer Diagnosis Based on a Combination of Convolutional Neural Networks, Ridge Regression and Linear Discriminant Analysis Using Invasive Breast Cancer Images Processed with Autoencoders’. Medical Hypotheses 135:109503. doi:10.1016/j.mehy.2019.109503. Torbenson, Max C. A., Ulf Büntgen, Jan Esper, Otmar Urban, Jan Balek, Frederick Reinig, Paul J. Krusic, Edurne Martinez Del Castillo, Rudolf Brázdil, Daniela Semerádová, Petr Štěpánek, Natálie Pernicová, Tomáš Kolář, Michal Rybníček, Eva Koňasová, Juliana Arbelaez, and Miroslav TRNKAc. 2023. ‘Central European Agroclimate over the Past 2000 Years’. Journal of Climate 36(13):4429–41. doi:10.1175/JCLI-D-22-0831.1. Tripathy, Kumar Puran, and Ashok Kumar Mishra. 2023. ‘How Unusual Is the 2022 European Compound Drought and Heatwave Event?’ Geophysical Research Letters 50(15):e2023GL105453. doi:10.1029/2023GL105453. Vamerali, Teofilo, Marianna Bandiera, and Giuliano Mosca. 2011. ‘Minirhizotrons in Modern Root Studies’. Pp. 341–61 in Measuring Roots . Van Loon, Anne F., Tom Gleeson, Julian Clark, Albert I. J. M. Van Dijk, Kerstin Stahl, Jamie Hannaford, Giuliano Di Baldassarre, Adriaan J. Teuling, Lena M. Tallaksen, Remko Uijlenhoet, David M. Hannah, Justin Sheffield, Mark Svoboda, Boud Verbeiren, Thorsten Wagener, Sally Rangecroft, Niko Wanders, and Henny A. J. Van Lanen. 2016. ‘Drought in the Anthropocene’. Nature Geoscience 9(2):89–91. doi:10.1038/ngeo2646. Virtanen, Pauli, Ralf Gommers, Travis E. Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, Stéfan J. van der Walt, Matthew Brett, Joshua Wilson, K. Jarrod Millman, Nikolay Mayorov, Andrew R. J. Nelson, Eric Jones, Robert Kern, Eric Larson, C. J. Carey, İlhan Polat, Yu Feng, Eric W. Moore, Jake VanderPlas, Denis Laxalde, Josef Perktold, Robert Cimrman, Ian Henriksen, E. A. Quintero, Charles R. Harris, Anne M. Archibald, Antônio H. Ribeiro, Fabian Pedregosa, and Paul van Mulbregt. 2020. ‘SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python’. Nature Methods 17(3):261–72. doi:10.1038/s41592-019-0686-2. Wan, Changgui, Ibrahim Yilmaz, and Ronald E. Sosebee. 2002. ‘Seasonal Soil–Water Availability Influences Snakeweed Root Dynamics’. Journal of Arid Environments 51(2):255–64. doi:10.1006/jare.2001.0942. Wedderburn, ME, JR Crush, WJ Pengelly, and JL Walcroft. 2010. ‘Root Growth Patterns of Perennial Ryegrasses under Well-Watered and Drought Conditions’. New Zealand Journal of Agricultural Research 53(4):377–88. doi:10.1080/00288233.2010.514927. Zhang, Bingwei, Marc W. Cadotte, Shiping Chen, Xingru Tan, Cuihai You, Tingting Ren, Minling Chen, Shanshan Wang, Weijing Li, Chengjin Chu, Lin Jiang, Yongfei Bai, Jianhui Huang, and Xingguo Han. 2019. ‘Plants Alter Their Vertical Root Distribution Rather than Biomass Allocation in Response to Changing Precipitation’. Ecology 100(11):e02828. doi:https://doi.org/10.1002/ecy.2828. Zhou, Huailin, Lulu Hou, Xiaomin Lv, Guang Yang, Yuhui Wang, and Xu Wang. 2022. ‘Compensatory Growth as a Response to Post-Drought in Grassland’. Frontiers in Plant Science 13. https://www.frontiersin.org/articles/10.3389/fpls.2022.1004553. Zwetsloot, Marie J., and Taryn L. Bauerle. 2021. ‘Repetitive Seasonal Drought Causes Substantial Species-Specific Shifts in Fine-Root Longevity and Spatio-Temporal Production Patterns in Mature Temperate Forest Trees’. New Phytologist 231(3):974–86. doi:10.1111/nph.17432. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 14 Jan, 2026 Reviewers invited by journal 13 Jan, 2026 Editor assigned by journal 13 Jan, 2026 First submitted to journal 12 Jan, 2026 Editorial decision: Major revisions 20 Nov, 2025 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-7847447","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":573992071,"identity":"3961f6fa-72e3-4fca-9768-1a63856b7828","order_by":0,"name":"Samuele Ceolin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6klEQVRIie2QMWvCQBTH/yJkepj1imK+wlsL4mdJCGQy4CTuQrIUXPs1SiF0fBJwMegacDG7gYwKhfau7dLl0rHD/Ya794bfvf87wOH4tyy/zoGEmE08U156Ff6+tJKQhyEQ/lkBSkKfEuSHRHU8g59vRJq3E438kyDKLO9XafHwzAlUtQ8lqs7kqRh2BWkxJi7BasESZUYZ2pVg2xbjd/4AB1ejHMnzS7uCWk8Bi55CRhH9Y33B6vb18YljUlWid8liswtLeLQFS1/q23o+9fNy19yz+TTY7ppLt7IE+4F+ddIvOBwOh8PGJ1atTT13i2mDAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-4969-1397","institution":"Luxembourg Institute of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"Samuele","middleName":"","lastName":"Ceolin","suffix":""},{"id":573992072,"identity":"07824fa4-9dbe-4fbb-b651-1f011db69ab8","order_by":1,"name":"Stanislaus Josef Schymanski","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Stanislaus","middleName":"Josef","lastName":"Schymanski","suffix":""},{"id":573992073,"identity":"80a90885-1fbb-4fd2-97c6-e49b1c310278","order_by":2,"name":"Julian Klaus","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Julian","middleName":"","lastName":"Klaus","suffix":""}],"badges":[],"createdAt":"2025-10-13 09:54:31","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7847447/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7847447/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102747803,"identity":"17653074-cb30-4a8f-a004-cf6745f5f4bf","added_by":"auto","created_at":"2026-02-16 09:05:24","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":470761,"visible":true,"origin":"","legend":"\u003cp\u003eThe study site located within a natural grassland in southern Luxembourg. The caps covering the minirhizotron tubes are visible along the fence.\u003c/p\u003e","description":"","filename":"1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7847447/v1/c1702f91b372f698093a36b0.jpeg"},{"id":102593149,"identity":"8526966e-dbd7-499f-9131-fa71337a8b9b","added_by":"auto","created_at":"2026-02-13 11:46:38","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":159013,"visible":true,"origin":"","legend":"\u003cp\u003eEnvironmental variables and root length observed during the entire study period. (a) precipitation (mm), (b) air temperature (°C, orange line) and MSH at 10 cm depth (m, blue line), (c) root length (mm) measured at 10 cm depth and (d) root length measured at 60 cm depth. Points represent the median value of root length, while whiskers extend to the observed maximum and minimum values. Minirhizotrons were imaged 64 times during the study period. Sampling frequency differs between growing and dormant season.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7847447/v1/466776196f45b020f1a1fb67.png"},{"id":102593145,"identity":"6896407c-d450-4fca-bbe9-2d8bb392fe74","added_by":"auto","created_at":"2026-02-13 11:46:38","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":344804,"visible":true,"origin":"","legend":"\u003cp\u003eGrowth rates and root lengths observed during the three selected precipitation events of September 2022 (a-b), June 2023 (c-d) and July/August 2023 (e-f). In the plots on the left, net growth rates at 10 cm (mm day\u003csup\u003e−1\u003c/sup\u003e) are represented by the purple boxes, at 60 cm by the light blue boxes and the top-bottom growth rate difference between 10 and 60 cm is shown by the grey lines. Black horizontal lines inside the boxes indicate the median value, the squares indicate the mean value, the edges of the boxes mark the interquartile range (lower edge marks the 25\u003csup\u003eth\u003c/sup\u003e percentile and the upper edge marks the 75\u003csup\u003eth\u003c/sup\u003e percentile) and the whiskers show the range of data within 1.5 times the interquartile range. Purple and light blue points outside the whiskers’ range represent outliers. Stars above the boxes indicate a significant increase or decline (p value \u0026lt;0.05) on that date compared to the previous sampling date for growth rates at 10 cm (purple star), growth rates at 60 cm (light blue star). Indigo lines in a), c) and e) depict the matric suction head (MSH, m) and the blue bars represent precipitation (mm). In the plots on the right (b, d, f), the left panel shows root lengths (mm) observed at 10 cm depth (purple boxes) while the right panel display root lengths observed at 60 cm depth (light blue boxes) on five days selected from each precipitation event. Outliers are not shown.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7847447/v1/82714871d8359b6f2a1c39db.png"},{"id":102593150,"identity":"19282081-0db0-4bc6-8b54-0485e8bca711","added_by":"auto","created_at":"2026-02-13 11:46:39","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":749752,"visible":true,"origin":"","legend":"\u003cp\u003eSeasonal shift in vertical root distribution. a) and b) show the root lengths (mm) along the vertical profile on 05/05/2022 and on 16/06/2022, respectively. c) and d) show the root lengths on 12/05/2023 and on 19/06/2023, respectively. Blue bars indicate root length medians obtained from the 12 tubes at all the measured depths. Whiskers represent the 25th and 75th percentiles for each bar. Numbers report the value of root length median and stars indicate a significant change (p-value \u0026lt;0.05) in root length between May and June. Orange dots represent MSH measurements at 10, 20, 40 and 60 cm.\u003c/p\u003e","description":"","filename":"4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7847447/v1/7fccee605c62414ccb0cb513.jpeg"},{"id":102747728,"identity":"a976c86f-b04e-4845-a2d3-55650042a2c1","added_by":"auto","created_at":"2026-02-16 09:05:18","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":722862,"visible":true,"origin":"","legend":"\u003cp\u003eFeature importance and coefficients of contribution from the best-performing regression analyses to determine the most influential features on root growth. Coefficients were normalized within the range 0-1 for each selected analysis. a) Normalized coefficients of contribution of the independent variables from the Ridge regression model of root growth at 10 cm depth. b) Normalized feature importance of the independent variables from the Random Forest model of root growth at 20 cm depth. c) Normalized coefficients of contribution from the Ridge regression model at 40 cm depth. d) Normalized coefficients of contribution from the Partial Least Square (PLS) regression at 60 cm depth. Note that matric suction head (MSH) values were considered positive during the analysis, meaning that a negative relationship between root growth and MSH indicate a positive relationship between root growth and soil moisture. The directionality of the influence of the parameter on root growth (positive or negative) is shown by the ‘+’ or ‘-’ signs above the bars.\u003c/p\u003e","description":"","filename":"5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7847447/v1/152cd17c3720d15ee2c1a66b.jpeg"},{"id":102750851,"identity":"93438dce-d83a-43ae-866b-3efeac4973cf","added_by":"auto","created_at":"2026-02-16 09:22:30","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2993597,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7847447/v1/3709a6e5-3d6d-447d-97b6-26f13301344e.pdf"}],"financialInterests":"","formattedTitle":"Controls on vertical root distribution dynamics in a temperate grassland across daily and seasonal scales","fulltext":[{"header":"Introduction","content":"\u003cp\u003eRoot water uptake is one of the key aspects studied in the context of agricultural productivity. Management practices and selection of plant varieties that encourage water capture are some of the tools that can enable us to improve agricultural productivity to meet population growth (Eshel and Beeckman 2013). Therefore, studying the limitations to water acquisition and the mechanisms implemented by plants to overcome such limitations is especially important. Studying root growth dynamics as a response to changing conditions in soil moisture is crucial in the context of climate change and drought stress. Unprecedented summer droughts are becoming more frequent in central Europe (Torbenson et al. 2023). These are expected to increase in frequency, duration, and intensity in the future with potentially severe effects on vegetation (Klaus et al. 2022; Van Loon et al. 2016). In fact, the drought of summer 2022, when this study began, lasted from May until August and was one of the most intense drought in Europe in recent history (Tripathy and Mishra 2023).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePlants have evolved strategies and root morphological and physiological plasticity in order to deal with the fluctuations in soil water availability in time and space (Fromm 2019; Hodge 2004). Among such strategies we find \u0026ldquo;Hydrotropism\u0026rdquo; and \u0026ldquo;Hydropatterning\u0026rdquo;. The former refers to directional root curvature and the latter to promotion of asymmetrical lateral root formation towards wet soil patches (Fromm 2019). At the seasonal scale, several studies documented root systems changing their vertical distribution seasonally as a response to soil moisture changes in different plant communities and under different climates (Hendrick and Pregitzer 1996; Peek et al. 2006; Wan et al. 2002; Wedderburn et al. 2010; Zhang et al. 2019; Zwetsloot and Bauerle 2021). Overall, these studies reported that root length was higher in shallow soil layers during the wet season. During the dry season, root growth increased towards deeper layers where soil moisture is usually higher. Ceolin et al. (2025) investigated experimentally daily time-scale root growth responses to rapid and local variations of soil moisture along the soil profile. The study was carried out under controlled conditions on 5-week old individual maize plants. In their study, a phenomenon coined as \u0026ldquo;hydromatching\u0026rdquo; was documented, which consists of enhanced local root proliferation (involving both new root emergence and\u0026nbsp;elongation of pre-existing roots) in a wetted soil layer within 48 hours from a water pulse application. The increase in the wetted layer was accompanied by a decrease in root growth in other non-wetted layers.\u0026nbsp;Ceolin et al. (2025)\u0026nbsp;suspected that this was part of a strategy enabling the plant to explore dynamic soil moisture sources while economizing on root carbon investments. A logical progression of such moisture-manipulation lab studies would be to assess whether the observed rapid dynamics are also visible in a naturally established plant community (e.g., a grassland), where variations of other environmental factors can potentially affect the dynamics. Transferring such findings from lab to field might not be straightforward. In fact, in a study on ryegrass populations, roots did proliferate following re-wetting of a dry topsoil, but started to do so only one month after the re-watering event\u0026nbsp;(Wedderburn et al. 2010). The delay in the response was interpreted as a way to ensure that carbon is invested in layers with sustained resource availability.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDifferent studies observed different root responses to soil moisture deficit in plant communities. Some studies detected increased fine root production under drought (Leuschner et al. 2001; Teskey and Hinckley 1981). This would be consistent with the \u0026ldquo;functional balance theory\u0026rdquo; which states that plants actively adjust shoot:root carbon allocation to improve the uptake of the most limiting resource (Thornley 1972). Other studies observed a decrease in root production under drought (Metcalfe et al. 2008; Zwetsloot and Bauerle 2021). Different responses to dry conditions could be explained by different physiological processes and limitations, and by niche separation among species within a community (Leuschner et al. 2001; Zwetsloot and Bauerle 2021). When considering grasslands specifically, mixed species grasslands could be more resistant to drought effects in terms of ecosystem productivity (Craven et al. 2016; Isbell et al. 2015). This is because the two main components of grasslands, grasses and forbs, are known to respond differently to drought and to occupy different niches to avoid competition for water. Grasses tend to grow their root systems in the topsoil, effectively extracting water even under conditions of scarcity. In contrast, forbs tend to avoid growing roots in upper soil strata and overcome water scarcity by tapping into deeper soil layers (Nippert and Knapp 2007).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAlthough it is widely recognized that soil moisture highly influences root growth dynamics, it is not clear whether it is the main driver of root growth in temperate grasslands. For example, one study found that root growth rates were strongly influenced by plant water status in a mesic grassland (Slette et al. 2023). In contrast, Wedderburn et al. (2010) reported that temperature was the main controller of root growth peaks in ryegrass populations. In another temperate grassland, root growth correlated positively with radiation but not with temperature (Edwards et al. 2004). Instead, precipitation was identified as the main determinant of vertical root distribution patterns in Mongolian temperate grasslands (Ma et al. 2008). Phenological factors may also play a role in driving root growth. While the importance of phenological drivers of root growth in temperate grasslands remain understudied, Joslin et al. (2001) documented that phenological factors were stronger than environmental variables in influencing root elongation rates in an oak stand.\u0026nbsp;These scattered findings makes it challenging to identify the environmental variable with the strongest control on root growth in temperate grasslands, mostly because they result from comparisons that considered only two variables at a time. These observations also indicate that root growth dynamics might be dominated by different environmental (and phenological) factors in different situations. Moreover, the apparent importance of each environmental parameter might change based on observation frequency. In fact, low-frequency observations might inaccurately emphasize the importance of phenological factors (operating over long time scales) over rapidly changing environmental parameters such as soil moisture, temperature and solar radiation. In line with this, most studies investigating root growth dynamics as a response to seasonal moisture fluctuations and to drought measured at biweekly, monthly and even yearly intervals (Peek et al. 2006; Slette et al. 2023; Wan et al. 2002; Wedderburn et al. 2010; Zhang et al., 2019; Zwetsloot and Bauerle 2021), failing to capture potential rapid shifts in root allocation driven by sub-daily fluctuations in drying and rewetting conditions. Such studies might have missed short-term dynamics potentially entailing high rates of root productivity and turnover, as it was demonstrated that one-month sampling intervals might lead to 60% underestimation of root growth and mortality\u0026nbsp;(Stewart and Frank 2008).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn addition, insights on how soil water fluctuations affect the spatial and temporal organization of root systems could also enhance soil vegetation atmosphere transfer (SVAT) models, enabling more accurate predictions of water fluxes mediated by vegetation under dynamic water availability (Schwinning and Ehleringer 2001).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn this study we build on earlier results of Ceolin et al. (2025) and (1) test if the phenomenon of hydromatching (observed on individual young plants under controlled conditions in unexplored soil) occurs also in response to natural rainfall variability in an established grassland community. We then aim to (2) capture vertical shifts in root distribution during the transition from wet to drier soil conditions, e.g. between a wet late spring and a dry early summer. Finally, our aim is to (3) test if soil moisture is an overall stronger driver than air and soil temperature, precipitation, irradiance, air humidity and wind speed for root growth rates in a temperate grassland under natural conditions. Specifically, we intend to answer the following research questions:\u003c/p\u003e\n\u003cp\u003e\u0026bull; Question 1 (occurrence of hydromatching): Do the root systems of the grassland community exhibit hydromatching in response to rapid and localized changes in soil moisture?\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026bull; Question 2 (seasonality of vertical root profiles): Do the root systems of the grassland community exhibit a vertical shift in root growth distribution at the seasonal scale, and is this correlated with soil moisture dynamics?\u003c/p\u003e\n\u003cp\u003e\u0026bull; Question 3 (hierarchy of drivers of root growth): What are the environmental drivers of root growth in a temperate grassland?\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cp\u003eField site and data collection\u003c/p\u003e\n\u003cp\u003eThe study was conducted in a managed grassland in Southern Luxembourg (49.5137\u0026deg;N, 6.0284\u0026deg;E, Fig. 1) that is mowed once per year and sporadically subjected to grazing (never during the study period).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe weather station at the site is operated by the Luxembourg Institute of Science and Technology (LIST). The soil at the site is clay loam down to 40 cm depth and heavy clay from 40 to 130 cm (Marx and Flammang 2015). The plant species populating the site include both forbs and grasses and are common European temperate grassland species such as \u003cem\u003eLolium perenne, Trifolium pratense\u003c/em\u003e,\u003cem\u003e\u0026nbsp;Trifolium repens\u003c/em\u003e,\u003cem\u003e\u0026nbsp;Taraxacum officinale\u003c/em\u003e,\u003cem\u003e\u0026nbsp;Jacobaea vulgaris\u003c/em\u003e (or \u003cem\u003eerratica\u003c/em\u003e), \u003cem\u003eCarduus crispus\u003c/em\u003e and \u003cem\u003eDaucus carota\u003c/em\u003e. The climate is oceanic and characterized by mild winters and moderate summers. At the study site the long-term annual mean precipitation from 2004 to 2024 was 736 mm while the long-term mean temperature from 2006 to 2024 was 9.99 \u0026deg;C. Aboveground, the weather station is equipped with instruments measuring hourly air temperature and humidity (Humidity and Temperature Probe HMP155, Vaisala, Finland), precipitation intensity (Tipping Bucket Precipitation Sensor 15189, \u0026nbsp;LAMBRECHT meteo GmbH, Germany), wind speed (A100LK-L Anemometer, Campbell Scientific, UT, USA) and irradiance (CMP3 Pyranometer, Kipp \u0026amp; Zonen, The Netherlands). Belowground, soil moisture and temperature sensors collecting half-hourly data are located at depths of 10, 20, 40 and 60 cm (CS650 30 cm Soil Moisture and Temperature Sensor and SoilVue 10 TDR Soil Moisture and Temperature Profile Sensor, Campbell Scientific, UT, USA). Three months before the start of data collection, we installed 12 acrylic minirhizotron tubes (180 cm long and 5 cm of diameter, CID Bioscience, Inc. Camas, WA, USA) along the fences of the weather station at an angle of 45\u0026deg; pointing outwards under the fences, approximately 50 cm apart and with 20 cm of tube sticking out of the ground (Fig. 1). Installation under the fence allowed us to capture the dynamics of undisturbed plants growing immediately outside the fenced weather station (and thus belonging to the grassland). Another reason for not installing the tubes outside the fence was to avoid tube damage by heavy machinery during annual mowing and to monitor root dynamics of individuals not exposed to mowing (the mower did not reach plants directly adjacent to the fence, as visible in Fig. 1). We believe the fence did not introduce any significant edge effect: the mesh was fine enough to avoid affecting light or wind passage, and although the fence poles could have altered soil structure and created preferential root growth zones, the minirhizotron tubes would have caused similar effects, so the potential bias was deemed negligible. The tubes reached a soil depth of approximately 115 cm. When not in use, polystyrene foam plugs were inserted inside the tubes and PVC caps were used to cover the protruding part of the tubes and avoid light disturbance.\u003c/p\u003e\n\u003cp\u003eRoots growing along the external wall of the tubes were imaged using the CI-602 Narrow Gauge Root Imager (CID Bioscience, Inc. Camas, WA, USA). We imaged to a soil depth of 85 cm from May 2022 to September 2022 and to a soil depth of 115 cm from September 2022 until August 2023, after purchasing additional extension rods allowing us to image deeper. We took images every two weeks but increased the sampling frequency shortly after major precipitation events during the growing season. Major precipitation events were defined as events that increased the matric suction head (MSH) by at least 3 m at 10 cm depth, equivalent to the effect caused by experimental water pulses in Ceolin et al. (2025). To document the pre-existing root distribution before a major precipitation event, we imaged 1 to 4 days prior forecasted moderate rain events (\u0026gt;2 mm h\u003csup\u003e\u0026minus;1\u003c/sup\u003e) and kept measuring every 2 to 4 days after the start of the event for a total of approximately 12 days, resulting in at least three measurements taken closely post-precipitation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eImage analysis\u003c/p\u003e\n\u003cp\u003eRoot images were corrected for overlaps between different depths and merged to obtain single images of the entire root profiles along each of the 12 minirhizotrons. We subdivided each root profile image in different sections covering different soil depths in order to better capture the root growth occurring in proximity of the soil moisture and temperature sensors located at 10, 20, 40 and 60 cm, which we term as \u0026ldquo;depths of interest\u0026rdquo;. Specifically, we considered the portions of the root profile images within 5 cm above and below the depths of 10 and 20 cm, and 10 cm above and below the depths of 40 and 60 cm. Therefore, despite having detected some roots down to the deepest part of the tubes (115 cm depth), we did not consider these roots in the analyses. Instead, we considered the roots growing nearby the 60 cm depth sensor as indicators of deep root activity. The Software \u0026ldquo;RootPainter\u0026rdquo; (originated by Smith et al. 2020) was used to detect alive roots in the images in an automated way. The software is a Convolutional Neural Networks (CNN) that, once trained on a suitable dataset, is able to segment roots in soil and to differentiate them from the background. During the training, roots were considered alive if they appeared bright white to the human eye. Subjectivity in root vitality detection based on visual color cues is common in minirhizotron studies (Vamerali et al., 2011). While this subjectivity was present during the training process, the use of the CNN on the full image set likely minimized additional user bias. The software \u0026ldquo;Rhizovision Explorer\u0026rdquo; was then used to extrapolate values of alive root length (Seethepalli et al. 2021). The detection of alive roots allowed us to detect root growth and root decline. Corrections through image manipulation have been implemented to reduce image noise. For example, edges of water droplets condensing along the minirhizotron wall were often confused as roots by the Software, so the segmented perimeter of the droplets was manually removed from the images. Also, images taken from 21/07/2023 onward presented problems of noise coming from camera malfunction, leading to the formation of straight bright stripes that were identified as roots by the Software. As in the previous case, a manual removal of the segmented stripes was performed.\u003c/p\u003e\n\u003cp\u003eIt is important to note that our data can be used to detect vertical (and relative) distributions of root length and sudden, local shifts in growth rates. However, the absolute values of root growth and root length decline are not reliable. In fact, our machine learning model was not yet accurate enough to consistently detect every root across all images, and a certain degree of noise was unavoidable. In addition, a precise recognition of root decay would require much more rigorous criteria of colour thresholding than the ones considered in this study.\u003c/p\u003e\n\u003cp\u003eData and statistical analysis\u003c/p\u003e\n\u003cp\u003eWe converted soil moisture data (measured as volumetric water content in m\u003csup\u003e3\u003c/sup\u003e m\u003csup\u003e-3\u003c/sup\u003e) to meters of matric suction head (MSH) by using the equations of the water retention model formulated by van Genuchten (1980). This was done to account for the role of textural classes in affecting soil water retention and availability to plants. Van Genuchten parameters for clay loam (down to 40 cm) and clay (from 40 cm downward) were obtained from Carsel and Parrish (1988) (Table 1).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e1\u003c/strong\u003e: Van Genucthen parameters used for the conversion from volumetric water content (m\u003csup\u003e3\u003c/sup\u003e m\u003csup\u003e\u0026minus;3\u003c/sup\u003e) to soil water head (m).\u0026theta;\u003csub\u003er\u003c/sub\u003e (cm\u003csup\u003e3\u003c/sup\u003e cm\u003csup\u003e\u0026minus;3\u003c/sup\u003e) is residual soil water content,\u0026nbsp;\u0026theta;\u003csub\u003es\u003c/sub\u003e (cm\u003csup\u003e3\u003c/sup\u003e cm\u003csup\u003e\u0026minus;3\u003c/sup\u003e) is saturated soil water content and\u0026nbsp;a\u003csub\u003evG\u003c/sub\u003e (cm\u0026minus;1), m\u003csub\u003evG\u003c/sub\u003e and n\u003csub\u003evG\u003c/sub\u003e are empirical parameters of the van Genuchten water retention model. We considered the clay loam parameters for the soil depth 0-40 cm and the clay parameters for the deeper depths.\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"518\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 23.7911%;\"\u003e\n \u003cp\u003eVan Genucthen\u003c/p\u003e\n \u003cp\u003eparameters\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.7911%;\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026theta;r\u003c/em\u003e (cm3 cm\u0026minus;3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.7911%;\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026theta;s\u003c/em\u003e (cm3 cm\u0026minus;3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.8607%;\"\u003e\n \u003cp\u003e\u003cem\u003ea\u003csub\u003evG\u003c/sub\u003e\u003c/em\u003e (cm\u003csup\u003e-3\u003c/sup\u003e)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12.766%;\"\u003e\n \u003cp\u003e\u003cem\u003envG\u003c/em\u003e (-)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 23.7911%;\"\u003e\n \u003cp\u003eClay loam\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.7911%;\"\u003e\n \u003cp\u003e0.095\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.7911%;\"\u003e\n \u003cp\u003e0.410\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.8607%;\"\u003e\n \u003cp\u003e0.019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12.766%;\"\u003e\n \u003cp\u003e1.310\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 23.7911%;\"\u003e\n \u003cp\u003eClay\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.7911%;\"\u003e\n \u003cp\u003e0.068\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.7911%;\"\u003e\n \u003cp\u003e0.380\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15.8607%;\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 12.766%;\"\u003e\n \u003cp\u003e1.090\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eWe converted root length (mm) to net growth rates (mm day\u003csup\u003e-1\u003c/sup\u003e) by subtracting the root length of the previous sampling from the next sampling and by dividing that value by the number of days between events.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo answer Question 1 (occurrence of hydromatching), we considered the growth rates observed in proximity of the sensors located at 10 and 60 cm and investigated the link between root growth and matric suction head. We also considered the differences between growth rate at 10 cm and 60 cm (with positive values indicating higher growth rate in the topsoil and vice versa). This was done to reveal changes in growth allocation between the top and subsoil, especially in circumstances of high root mortality (e.g.: under dry conditions). In this case, the differences in growth allocation between top and subsoil could still be meaningful while no clear response would be detected when considering the values of net growth rates only. Depths of 10 and 60 cm were considered in this analysis because they represented two distinct soil zones with contrasting moisture dynamics. The 10 cm topsoil experienced rapid and pronounced fluctuations in soil moisture, making it ideal for assessing root responsiveness to rewetting events due to rainfall. In contrast, the 60 cm subsoil remained consistently saturated and represented the \u0026ldquo;deep root pool\u0026rdquo;, where water was constantly available but harder for roots to access. Intermediate depths (20 and 40 cm) showed moisture patterns similar to 10 cm but with much weaker fluctuations, so including them would not add meaningful insight to this analysis. In our study, we selected a subset of the monitored events, specifically, three precipitation events that led to a substantial increase in soil water potential by at least 30 kPa (approximately 3 m of MSH) at 10 cm depth and for which we had root images right before and several right after the event. To determine if hydromatching occurred, we first used the Shapiro test to check for normal distribution of the values of net growth rates for each sampling event. We then compared them between subsequent sampling events by either using the Mann-Whitney test (when at least one of the two compared groups did not follow normality) or the Student t-test (when both of the considered groups followed normality) to check for significant differences. For hydromatching to occur, we expected a significant increase in growth rates in the topsoil and/or a significant decrease in the subsoil following major precipitation events. The Shapiro test, the Mann-Whitney test and the Student t-test were carried out using the Python library \u0026ldquo;scipy\u0026rdquo; (v1.1.0) and the package \u0026ldquo;stats\u0026rdquo; (Virtanen et al. 2020).\u003c/p\u003e\n\u003cp\u003eTo answer Question 2 (seasonality of vertical root profiles) we compared the vertical root length distributions along each depth of the 12 tube profiles on two sampling days, one in May (late spring, wetter season) and one in June (early summer, drier season) of both 2022 and 2023. These two time points differ in their seasonal characteristics but are both periods of high root productivity, making them suitable for comparing root responses to seasonally induced variations in soil moisture availability. Transitions involving autumn and winter were not considered, as dormancy-related processes could have confounded the interpretation of root dynamics. To test for significant vertical shifts in root length allocation we lumped together the root lengths at each depth from all 12 minirhizotrons. We compared the root lengths at each depth on the sampling event in May with the root lengths at the same depth in June for both years. We first used the Shapiro test to check for normal distribution of the root length values at each depth and in each sampling event. We then carried out the comparison with the Mann-Whitney test (when at least one of the two compared groups of root length did not follow normality) or with the Student t-test (when both groups followed normality).\u003c/p\u003e\n\u003cp\u003eTo answer Question 1 on the hierarchy of drivers of root growth, we used the median root growth rate at each depth, calculated across the 12 minirhizotrons, as the dependent variable in our analysis. Mean values between sampling times of soil moisture (expressed as both volumetric water content and matric suction head), air temperature, soil temperature, irradiance, relative humidity and wind speed were considered as the independent variables. Additionally, we considered the interactions of soil moisture at different depths with air temperature and soil temperature at different depths (e.g., soil moisture at 10 cm X air temperature; soil moisture at 20 cm X soil temperature at 60 cm) as independent variables. We calculated the interaction by multiplying the values of the interacting features, a recognized method in multiple regression analyses (Allison 1977; Li et al. 2020). We disregarded the data from the winter months to exclude the impact of plant dormancy on the results. We then used the dependent and independent variables to perform a Random forest analysis, a Ridge regression analysis and a Partial Least Square (PLS) analysis using the Python library \u0026ldquo;scikit-learn\u0026rdquo; (Pedregosa et al. 2011). Random Forest, Ridge regression and PLS have been used for analysis of feature importance and feature selection in multiple studies and fields (Hu et al. 2018; Shahhosseini et al. 2019; Toğa\u0026ccedil;ar et al. 2020). These analyses compute coefficients indicating the importance of the independent variables (in our case the environmental parameters) in explaining the dependent variable (in our case root growth rate). We performed these analyses to determine the strongest predictors of root growth separately for each soil depth of interest (10, 20, 40 and 60 cm). For each soil depth, we performed the analysis with these three methods twice, once considering volumetric water content and once considering matric suction head instead to avoid redundancy and an excessive amount of \u0026quot;independent\u0026quot; variables per analysis. Eventually, we selected the analysis that yielded the highest R-squared for each soil depth of interest and considered the five independent features that ranked as the best predictors of root growth in each selected analysis. We used this top five of most influential independent variables to draw our conclusions on what affected root growth.\u003c/p\u003e\n\u003cp\u003eThe Python libraries \u0026ldquo;NumPy\u0026rdquo; (v1.15.4) (Harris et al., 2020), \u0026ldquo;pandas\u0026rdquo; (v0.21.0) (McKinney, 2010), \u0026ldquo;Matplotlib\u0026rdquo; (v3.1.0) (Hunter, 2007), and \u0026ldquo;seaborn\u0026rdquo; (v0.9.0) (Waskom, 2021) were used for numerical computation, data manipulation, and visualization. OpenAI was used to assist in code preparation during the initial round of analyses prior to subsequent adjustments.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eDuring the observation period from 01/05/2022 to 31/08/2023, MSH at 10 cm depth varied between saturation (during most of winter and following intense rain events) and -29 m (in late July 2023) (blue line in Fig. 2b). Root length showed clear fluctuations reflecting seasonality, with peaks at 10 cm in late spring (beginning-middle of May), peaks at 60 cm depth in early summer (end of May-early July) and reduced root length during winter months at both 10 and 60 cm depths (Fig. 2c and d).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMean annual precipitation and mean MSH at 10 cm were much lower in 2022 than in 2023, whereas mean air temperature was very similar (Table 2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e2\u003c/strong\u003e: Annual and long-term means of total precipitation (mm), air temperature (\u0026deg;C) and matric suction head (MSH, m) at 10 cm depth. Long-term means consider the period 2004-2024 for total precipitation, 2006-2024 for air temperature and 2018-2024 for MSH.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"601\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003eTotal precipitation (mm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003eAir temperature mean\u0026nbsp;(\u0026deg;C)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003eMSH mean at 10 cm depth (m)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e683.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e11.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e- 0.75\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e965.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e10.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e- 0.38\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003eLong-term mean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e736.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e9.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 25%;\"\u003e\n \u003cp\u003e- 0.46\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eOccurrence of hydromatching\u003c/p\u003e\n\u003cp\u003eWe evaluated the occurrence of hydromatching by comparing the net growth rates and the top-bottom growth difference between sampling dates on three selected periods, termed as \u0026ldquo;September 2022\u0026rdquo;, \u0026ldquo;June 2023\u0026rdquo; and \u0026ldquo;July/August 2023\u0026rdquo;. These periods were characterized by major precipitation events replenishing the topsoil moisture, along with frequent imaging of the tubes (Fig. 3). Specifically, the September 2022 period included the post-drought rewetting event, which established an end to the long drought period that started in May. We observed the largest fluctuations in MSH over the entire study period at 10 cm depth, where values ranged from saturation (0 m) to a minimum of -30 m in July/August 2023 (Fig. 3). The soil at 60 cm was at saturation for the entirety of the study. Note that the highly clayey texture of the soil at this depth could have affected the readings of the sensors.\u003c/p\u003e\n\u003cp\u003eIn September 2022 we observed a root growth promotion in the topsoil 3 to 5 days after the soil wetting. The net growth rates at 10 cm (purple boxes in Fig. 3a) increased significantly between 27/08/2022 and 05/09/2022 (median of 0.05 mm day\u003csup\u003e\u0026minus;1\u003c/sup\u003e to 2.33 mm day\u003csup\u003e\u0026minus;1\u003c/sup\u003e), as also evidenced by the progressive root length increment (purple boxes in Fig 3b). A change in image contrast between 27/08/2022 and 02/09/2022 did not allow to compare the growth rates on those days, preventing us to assess whether the response occurred potentially already on 02/09/2022. At 60 cm net growth rates decreased to -0.45 mm day\u003csup\u003e\u0026minus;1\u003c/sup\u003e on 07/09/2022, the day when net growth rates at 10 cm were at their highest. The difference in growth rate between the top and subsoil increased between 02/09 and 07/09, indicating root growth allocation at 10 cm depth (at the expense of growth at 60 cm).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn June 2023, 1 to 3 days after the onset of rain events the net growth rates at 10 cm increased significantly (from a median of -2.38 mm day\u003csup\u003e\u0026minus;1\u0026nbsp;\u003c/sup\u003eon 19/06/2023 to a median of 2.99 mm day\u003csup\u003e\u0026minus;1\u003c/sup\u003e on 22/06/2023, purple boxes in Fig. 3c) and so did the top-bottom growth difference (grey line in Fig. 3c) indicating root growth promotion in the topsoil. At 60 cm, net growth rates significantly declined 1 to 3 days after the soil rewetting, reaching negative values and a minimum median of -25.76 mm day\u003csup\u003e\u0026minus;1\u003c/sup\u003e on 22/06/2023 (light blue boxes in Fig. 3c). This was the same day when we observed the maximum positive growth rate reached at 10 cm. When soil started drying up, root growth at 60 cm progressively increased and reached 12.44 mm day\u003csup\u003e\u0026minus;1\u003c/sup\u003e on 28/06/2023, with the top-bottom growth difference significantly decreasing and reaching negative values on 28/06/2023. This indicates that root promotion progressively switched from the top to the bottom of the profile when dry conditions developed again. The root length variations at 10 and 60 cm depths (Fig. 3d) also reflect these growth dynamics, although less clearly.\u003c/p\u003e\n\u003cp\u003eJuly/August 2023 was characterized by a long rainy period during which the growth rates at 10 and at 60 cm remained near or below zero (Fig. 3e), with noticeable declines in root length especially at 60 cm (Fig. 3f). Net growth rates at both depths started increasing only towards the end of the period.\u003c/p\u003e\n\u003cp\u003eVertical root distribution changes between spring and summer\u003c/p\u003e\n\u003cp\u003eMSH progressively decreased at 10 cm depth from May to June in both 2022 and 2023. However, the change in MSH from May to June was much more dramatic in 2023 compared to 2022. At 10 cm depth, MSH dropped from 0 m (saturation) to -15 m in 2023, compared to a smaller decrease from -2.75 to -5.85 in 2022 (Fig. 4:4). Despite the two years having different magnitudes of soil moisture change over the spring-summer transition, we observed a similar and significant shift in root length from the shallower soil depths to the deeper soil depths with the progression of the dry season in both years (Fig. 4:4). The shift consisted of a decline in root length at shallow depths and a promotion of root growth in the deeper layers. In both years, between May and June, root length decreased significantly at 5-30 cm depth and increased significantly at 60-80 cm depth (Fig. 4:4).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eImportance of water or temperature on root growth dynamics\u003c/p\u003e\n\u003cp\u003eRandom Forest, Ridge analysis and Partial Least Square (PLS) regressions performed differently for each depth of interest and differently depending on whether we considered volumetric water content (VWC) or matric suction head (MSH) as an independent variable. The R-squared resulting from the Random Forest, Ridge regression and PLS considering the growth rates at 10, 20, 40 and 60 cm and considering either VWC or MSH are presented in Table 3. Several reported R-squared values are negative. Although seemingly counter-intuitive, a negative value of R-squared computed by regression models can occur when the regression model\u0026rsquo;s predictions are worse than those of a null model represented by a constant value (Nakagawa and Schielzeth 2013). The growth rates at 10 cm were best predicted by the Ridge regression considering VWC (R2=0.5), the growth rates at 20 cm were best modeled by the Random Forest considering VWC (R2=0.26), the growth rates at 40 cm were best predicted by the Ridge regression considering MSH (R2=0.55) and the growth rates at 60 cm were best predicted by the PLS considering MSH (R2=0.24). The top 5 predictors of root growth varied between soil depths and differed in their directional influence (Fig. 5). We consistently observed soil moisture (as VWC and MSH) ranking highest among the top 5 features at all depths except 20 cm and 60 cm. At 20 cm, one soil moisture feature (VWC at 10 cm) ranked 3rd right after irradiance and air humidity. At 60 cm, all of the top 5 most influential features were interactions between MSH at 10 and 20 cm and temperatures from the above soil layers or air temperature. These results could indicate a larger influence of temperature on root growth at 60 cm, but are most likely shaped by the invariability of soil moisture at this depth.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e3\u003c/strong\u003e:\u0026nbsp;R-squared values deriving from the Random Forest, Ridge Regression and Partial Least Square Regression computed for each depth considering root growth rate as the dependent variable. We reported the results considering soil moisture either as volumetric water content (VWC) or matric suction head (MSH)\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"601\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003eR2 values\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 183px;\"\u003e\n \u003cp\u003eRandom forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 172px;\"\u003e\n \u003cp\u003eRidge regression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 172px;\"\u003e\n \u003cp\u003ePartial least square regression\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003eDependent variable\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eConsidering VWC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003eConsidering MSH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eConsidering VWC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eConsidering MSH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eConsidering VWC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eConsidering MSH\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003eGrowth rate at 10 cm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e0.26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e-2.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e-0.88\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003eGrowth rate at 20 cm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e-0.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e-1.47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e-0.41\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003eGrowth rate at 40cm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e-0.48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e-1.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e0.55\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e-1.02\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003eGrowth rate at 60 cm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.72\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e0.23\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Discussion","content":"\u003cp\u003eGrassland root systems exhibit hydromatching\u003c/p\u003e\n\u003cp\u003eHydromatching was previously observed as rapid switches in local root growth patterns occurring within 48 hours following local changes in soil moisture (Ceolin et al. 2025). However, that study was performed on young individual maize plants grown in confined, unexplored soil. The plants never experienced critical water stress, as soil moisture was maintained above 6% at all time. In contrast, the plants in our temperate grassland community experienced prolonged periods of water deficit, ultimately leading to leaf senescence. A declining temperate grassland community is expected to behave differently than healthy, individual maize plants. Therefore, the effects of severe water stress need to be considered when interpreting our results and comparing them with the laboratory study of Ceolin et al. (2025), with the expectation of more altered or dampened root plastic responses to soil moisture variations. Another caveat is that the lab-based study used Magnetic Resonance Imaging (MRI) to visualize entire root systems in three dimensions, whereas the minirhizotron technique used here captures only two-dimensional images along a plexiglass-soil interface. Despite these caveats, we observed evidence of hydromatching within 2-5 days in the September 2022 event and within 1-3 days in the June 2023 event (Fig. 3a and c). Plants rapidly adapted their local root growth rates to exploit the most resourceful soil layers and interrupting root investment into soil layers requiring more carbon investment per amount of water uptake. When soil at 10 cm depth was re-wetted, soil at 60 cm became a more \u0026ldquo;costly\u0026rdquo; source of water. This could be due to its heavy clay composition, likely retaining water more tightly, imposing a higher mechanical resistance and creating conditions of low aeration (da Silva and Kay 1997). Top and subsoil likely swapped the role of \u0026ldquo;most resourceful\u0026rdquo; soil layer in a matter of days, mostly depending on the moisture conditions of the topsoil. These results and interpretation reveal a highly dynamic nature of the root systems in this grassland and are in line with the ones from Ceolin et al. (2025). It then appears that hydromatching is also observable in natural settings at a grassland community level, and not only under controlled conditions at the individual plant level (as in Ceolin et al., 2025). Hydromatching occurred as a collective response, as we did not discriminate between roots belonging to different species or ecological groups. In particular, it was not possible to distinguish between roots belonging to forbs and grasses. Both groups were present throughout the whole sampling campaign, as individuals of \u003cem\u003eLolium perenne\u003c/em\u003e (perennial grass) and a series of other annual forbs were consistently present at the study site. Therefore, we can assume that during the entire observation period forbs and grasses coexisted, meaning that hydromatching was identifiable despite niche separations and despite the different drought-adaptation strategies between forbs and grasses (Nippert and Knapp 2007).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePrevious studies on root morphological adjustments to soil moisture variations focused on large time scales that mostly captured seasonal or even annual dynamics of root growth (Peek et al. 2006; Wan et al. 2002; Wedderburn et al. 2010; Zhang et al. 2019; Zwetsloot and Bauerle 2021). However, studies at such large time scales overlook the daily adjustments to quick changes in soil moisture, which were revealed in the present study. To our knowledge, studies documenting local root growth patterns in natural grassland communities at such high frequency are lacking, especially ones comparing pre and post precipitation events. The daily time scale dynamics may also be important in the context of carbon and nutrient cycling, ecosystem productivity and in soil-vegetation-atmosphere transfer models.\u003c/p\u003e\n\u003cp\u003eOur results also allow to identify factors potentially limiting the manifestation and the observation of hydromatching, i.e. soil dryness persistence prior to the rewetting event, water logging, energy limitation, herbivory and nutrient remobilization.\u003c/p\u003e\n\u003cp\u003eIn September 2022, high levels of water stress prior to the rain event were caused by the prolonged drought period started in May, which led to apparent signs of leaf senescence. This could have caused a decrease in photosynthetic rates and a reduction of carbon allocatable to roots (Zwetsloot and Bauerle 2021) which, in turn, might have delayed the response to the re-wetting in September 2022 compared to June 2023.\u003c/p\u003e\n\u003cp\u003eFurthermore, a rapid and significant shift in net growth rates shortly after the re-wetting was probably less of a \u0026ldquo;risk factor\u0026rdquo; in June 2023 than it could have been in September 2022. In September 2022, the root growth shift towards the topsoil could have been delayed to mitigate the risk of losing newly grown roots due to potentially renewed dry conditions (Wedderburn et al. 2010).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAnother strong limitation to the occurrence of hydromatching was given by water logging and energy limitation under the form of solar radiation and temperature, as evidenced by the results from July/August 2023. Decreased irradiance and/or lower temperatures during the 14-day rain event likely inhibited root growth (Fig. 3e), suggesting that root investment in the grassland was likely simultaneously guided by shoot water demand and limited by shoot carbon supply. Also, the induced conditions of soil saturation could have hindered root growth, as soil saturation is known to restrict root activity and lead to root decay due to lack of aeration (Drew 1992). Waterlogging conditions and severe levels of anoxia were likely the cause of the strong decline in root length observed at 60 cm from 24/07 to 05/08 (Fig. 3f).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHerbivory represented another potential limitation in the detection of hydromatching. This could explain the sudden and significant levels of root length decline observed at 10 cm depth at the end of the September 2022 event (Fig. 3a and b), given the presence of unidentified invertebrates in several images taken at shallow depths in this period. This means that root loss due to predation should be regarded as a potential limitation in minirhizotron studies focusing on root dynamics.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFinally, another possible explanation that cannot be ruled out for the fast switch in root growth patterns could be nutrient scavenging in the (normally) nutrient-rich topsoil, triggered by the re-wetting (Armas et al. 2012). Future studies should investigate the effect of nutrient remobilization on root growth rates in grasslands shortly after rain events.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eRoot growth shifts from spring to summer\u003c/p\u003e\n\u003cp\u003eThe increase in root length in deeper soil layers during the transition from late spring to early summer (Fig. 4) is consistent with previous studies observing root growth promotion in the subsoil as a dry period progressed (Hendrick and Pregitzer, 1996; Peek et al. 2006; Wan et al. 2002; Weddenburn et al., 2010, Zhang et al. 2019, Zweetslot and Bauerle, 2021). These studies suggested that plants can redistribute roots at different depths according to the long-term seasonality of water availability along the soil profile. In this study, not only root length significantly increased in the deeper soil (60-80 cm) but it also significantly decreased in the upper soil layers (0-30 cm) during the transition to summer. Such behaviour was also observed in a juniper stand study, where a vertical transition in soil water availability was accompanied by a progressive downward shift in fine root growth and soil water use (Peek et al. 2006). Our observation of lower root lengths in May 2023 compared to May 2022 might have been due to near-saturation conditions across all depths between March and May 2023, \u0026nbsp;which are known to hinder root growth (Drew 1992). Another explanation for the overall lower root length in 2023 could be seedbank depletion caused by the intense drought of summer 2022 (Zhou et al. 2022). Different studies demonstrated how droughts can lead to legacy effects impacting root production in the following years (Slette et al. 2023; Zhou et al. 2022).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWhile the accuracy in the measurements of root length decline was limited, the significant decline observed consistently in both years suggests that root loss in the topsoil likely occurred and was not an artefact caused by uncertainty in the image analysis. The significant decline of root length at 20 cm depth in both 2022 and 2023, despite relatively moist soil at that depth (-1.2 m MSH in June 2022 and -2.7 m in June 2023, Fig. 4:4b and d), indicates that at least a portion of the root senescence occurring over a month in the upper layers was not caused by water deprivation, but was likely a programmed mechanism. This could be seen as a mechanism to reduce root maintenance costs and save resources to be used for growing and maintaining roots in the subsoil. The potential existence of such a cost-effective strategy of carbon allocation reflecting soil moisture availability was already discussed by Ceolin et al. (2025), after observing both root growth promotion in wetter soil layers and root growth suppression (and potential root loss) in drier layers.\u003c/p\u003e\n\u003cp\u003eThe shifts in vertical root distribution that we detected indicate a top-bottom carbon allocation transition, with a clear separation between shallow and deep parts of the root system. It is arguable that the observed shift in root abundance from topsoil to subsoil reflects a change in dominance from shallow- to deep-rooted species. However, such a profound shift in community composition is unlikely to occur within just over a month. Moreover, the similar root length at 40-50 cm in May and June in both years indicates that deeper roots were already established and remained stable at that depth over time, suggesting they likely belonged to the same individuals. If a dominance shift had occurred, it is improbable that root length at 40-50 cm would remain unchanged. This supports the idea that the observed changes were not due to species turnover or mortality of shallow-rooted plants, but rather to plastic responses within the same individuals shifting root growth from more populated shallow layers to less populated deeper ones. Furthermore, it is unlikely that newly established deep-rooted species extended roots to 80 cm within a month; it is more plausible that pre-existing deep root systems expanded further while surface root growth declined.\u003c/p\u003e\n\u003cp\u003eThe similar extents of the shifts from May to June in both 2022 and 2023, despite the different levels of MSH recorded in early summer, suggest that these shifts were phenologically-driven processes. This partially aligns with the \u0026lsquo;\u0026rsquo;phenological programming\u0026rsquo;\u0026rsquo; theory (Hendrick and Pregitzer, 1996; Joslin et al., 2001), which applies to deciduous trees that evolved in environments where summer droughts are the norm. It proposes that a programmed burst of root growth following foliar expansion in late spring is advantageous, as it normally allows to establish an extensive root system when moisture and temperature conditions are favourable and carbon supply is adequate. However, this theory does not account for potential phenologically programmed vertical shifts in root growth that can occur during the transition from spring to summer, which we observed here. Furthermore, to our knowledge, there have been no previous reports of root decay as a programmed mechanism. Overall, our findings suggest the existence of two types of root morphological adaptations occurring at different time scales in our grassland community. At daily time scales and at local, small spatial scales (within a soil layer of few cm), root growth rates appear to be strongly influenced by rapidly changing environmental factors, such as soil moisture. Such daily-scale local adjustments in growth rates could be guided by both local changes in soil moisture and potentially by changes in soil moisture occurring elsewhere, according to the theory of the \u0026ldquo;whole-root system coordination\u0026rdquo; described by Ceolin et al. (2025). At the long-term (seasonal) and at the whole-root system scale, root distribution appears to undergo \u0026nbsp;drastic and significant adjustments, occurring as a programmed, phenologically-dictated processes, independent of rapidly changing environmental variables. Both short-scale, environmentally dictated and long-scale, phenologically dictated adaptations seem to share a common goal, which is meeting the water requirements while flexibly managing carbon budgets. Carbon is spent where a return is expected, while the cost of root maintenance is cut where no resource can be gained. In our case, such adaptations were community responses, although similar adjustments are observable in individual plants as well (Ceolin et al. 2025; Engels et al., 1994).\u003c/p\u003e\n\u003cp\u003eMoisture controls growth rates\u003c/p\u003e\n\u003cp\u003eRandom Forest, Ridge regression and PLS ranked soil moisture (as VWC and MSH) as the overall strongest driver of grassland root growth rate at the study site (Fig. 5). This result aligns with a study on temperature grasslands in inner Mongolia, which found that precipitation, rather than temperature, was the main driver of root vertical distribution patterns. In contrast, prior research identified temperature, solar radiation and phenological factors as the stronger drivers of root growth in grasslands and tree stands (Edwards et al. 2004; Joslin, Wolfe, and Hanson 2001; Wedderburn et al. 2010). However, these studies employed a low-frequency (weekly-monthly) fixed sampling schedule, so they were not designed to detect the fast responses to wetting as observed in our study, where soil moisture appeared as a key driver of root growth. This result is consistent with the controlled lab experiment by Ceolin et al. (2025), where soil moisture was manipulated, but the other environmental variables such as temperature, irradiance and air humidity were kept constant. In this present study we found that variations in soil moisture had the strongest influence on root growth even under natural conditions, where other environmental factors co-varied. Soil moisture was particularly important in influencing root growth at 10 cm and 40 cm depth, where the best predictors were VWC at 10 and 20 cm and MSH at 20 and 10 cm, respectively (Fig. 5a and c). Overall, soil moisture ranked higher than temperature at all depths except 60 cm, where soil moisture was constant. This is particularly remarkable considering that our grassland is located in an energy-limited environment (McVicar et al. 2012). Ma et al. (2008) found that root distribution patterns were strongly driven by precipitation in temperate grasslands, but they state that this was likely due to the overall dry climate and water-limited environment of inner Mongolia. Instead, previous studies suggested that radiation flux (Edwards et al. 2004) or temperature (Kaspar and Bland 1992) is likely the main driver of root growth in temperate climates, as well as in northern forest ecosystems where water limitation is rare (Montagnoli et al. 2014). However, Wedderburn et al. (2010), who conducted their study in an energy-limited area, found that, after a drought period, the typically dominant effect of temperature on root development was overtaken by the effect of soil rewetting. The stimulative effect of an environmental parameter on root growth might therefore depend on the history of the stresses experienced by the plant community: under typical temperate conditions without abnormal drought, energy-related factors generally control root growth (e.g., Edwards et al. 2004; Kaspar and Bland 1992) but, after severe water limitation, sensitivity of root dynamics may shift from energy to water-sensible and become more responsive to moisture variation (e.g., Wedderburn et al. 2010). As our study period included the notable 2022 drought, it is possible that the water stress experienced during that time gave soil moisture a greater influence over growth rates than it would have had in previous years. This change in influence could have persisted during the following growing season of 2023, as droughts are known to have legacy effects on root development lasting years (Slette et al. 2023; Zhou et al. 2022). These mechanisms represent a distinct form of root plasticity related to stress memory, suggesting that soil moisture may become an increasingly dominant driver of root dynamics under climate change in central European temperate grasslands. Such flexibility may be crucial for temperate grassland communities to cope with climatic extremes. Nevertheless, further research is needed to clarify whether what we observed represents a true \u0026ldquo;priority shift\u0026rsquo;\u0026rsquo; in root growth drivers and how widespread or persistent this phenomenon may be.\u003c/p\u003e\n\u003cp\u003eA limitation of the performed regression analyses is the limited size of the dataset. Although the higher frequency of observation allowed us to better capture the effects of rapidly changing environmental parameters, the impossibility to analyze subsets of data prevented a clear distinction of effects on root growth given by seasonality/phenology and effects dictated by rapid and isolated soil environmental changes.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThe root systems of the grassland community implemented drastic seasonal adjustments (between spring and summer) involving root length decline in shallow soil layers and root growth promotion in deeper layers, offering evidence in response to Question 2 regarding the seasonality of vertical root profiles. At the same time, the grassland\u0026rsquo;s root systems were able to deploy \u0026lsquo;\u0026rsquo;hydromatching\u0026rsquo;\u0026rsquo; in two out of our three selected periods, by favouring root growth in the topsoil at the expense of the subsoil, 1\u0026ndash;3 days after a rain event. This offers proof in response to Question 1, related to the occurrence of hydromatching. The seasonal shifts involved the entire root systems and were likely driven by phenological factors, whereas hydromatching consisted of more localized adjustments following rapid changes in soil moisture and was therefore dictated by environmental factors. Energy limitation (given by lower solar radiation and temperature) and a saturated soil in July/August 2023, together with the prolonged conditions of water stress prior to the rain event in September 2022, could have been important limitations to hydromatching. Although we cannot fully rule out the potential role of nutrients in influencing root growth dynamics, our results still strongly suggest that hydromatching is a naturally occurring phenomenon observable in established grassland communities. Both observed long-term and short-term growth rate modifications might represent strategies evolved to cope with soil moisture heterogeneity alongside carbon budgeting. Lastly, soil moisture had the strongest control on root growth rates at the investigated temperate grassland, followed by soil temperature. This provides evidence in response to Question 3, related to the hierarchy of environmental drivers of root growth.\u003c/p\u003e \u003cp\u003eOur findings hold important implications for an improved understanding of plant carbon allocation strategies, plant population dynamics and biogeochemical processes such as carbon and nutrient cycling. Future studies could focus on quantifying fine root shedding, as it seems to be deeply integrated in the root growth dynamics as a response to soil moisture heterogeneity. It may also be important to investigate to what extent these adjustments facilitate the fulfilment of the transpiration demand. Future studies should also examine the extent to which typically energy-limited grassland communities can implement soil moisture-driven root dynamics following abnormal drought events, a key question in vision of climate change and increasing drought stress in temperate ecosystems. Ultimately, consideration of plant water stress indicators could help explain contradicting findings about root responsiveness to re-hydration. Other factors might also play a role, such as legacy effects of previous droughts, nutrient limitation, and prolonged periods of low temperature and solar radiation.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eVWC: Volumetric Water Content; MSH: Matric Suction Head\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eAcknowledgements\u003c/p\u003e\n\u003cp\u003eWe thank J\u0026eacute;r\u0026ocirc;me Juilleret for his valuable assistance during the sampling campaign and Jean Fran\u0026ccedil;ois Iffly for his crucial support in site selection, minirhizotron installation and for configuring and maintaining the weather station instrumentation. We also thank Abraham George Smith for his significant assistance in operating the RootPainter software.\u003c/p\u003e\n\u003cp\u003eFunding\u003c/p\u003e\n\u003cp\u003eThis research has been supported by the Fonds National de la Recherche Luxembourg (grant no. AFR PhD/19/SR/13577787).\u003c/p\u003e\n\u003cp\u003eCompeting interests\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003eAuthor contributions\u003c/p\u003e\n\u003cp\u003eConcept and experimental design: SC, SJS and JK. Field study preparation and measurements: SC and SJS. Data analysis: SC. Data interpretation: SC, SJS and JK. Paper preparation: SC, SJS and JK. All authors commented on previous versions of the manuscript\u003cem\u003e.\u0026nbsp;\u003c/em\u003eAll authors read and approved the final manuscript. Supervision: SJS (main supervisor), JK (co-supervisor).\u003c/p\u003e\n\u003cp\u003eData availability\u003c/p\u003e\n\u003cp\u003eThe datasets generated during the current study and code for data analysis are available at https://doi.org/10.5281/zenodo.10836040, https://doi.org/10.5281/zenodo.10528310, https://doi.org/10.5281/zenodo.10528980, https://doi.org/10.5281/zenodo.10530073, https://doi.org/10.5281/zenodo.10551976, https://doi.org/10.5281/zenodo.10679137. Multiple repositories were required to upload all the minirhizotron images due to storage limitations.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAllison, Paul D. 1977. \u0026lsquo;Testing for Interaction in Multiple Regression\u0026rsquo;. \u003cem\u003eAmerican Journal of Sociology\u003c/em\u003e 83(1):144\u0026ndash;53. doi:10.1086/226510.\u003c/li\u003e\n\u003cli\u003eArmas, Cristina, John H. Kim, Timothy M. Bleby, and Robert B. Jackson. 2012. \u0026lsquo;The Effect of Hydraulic Lift on Organic Matter Decomposition, Soil Nitrogen Cycling, and Nitrogen Acquisition by a Grass Species\u0026rsquo;. \u003cem\u003eOecologia\u003c/em\u003e 168(1):11\u0026ndash;22. doi:10.1007/s00442-011-2065-2.\u003c/li\u003e\n\u003cli\u003eCarsel, Robert F., and Rudolph S. Parrish. 1988. \u0026lsquo;Developing Joint Probability Distributions of Soil Water Retention Characteristics\u0026rsquo;. \u003cem\u003eWater Resources Research\u003c/em\u003e 24(5):755\u0026ndash;69. doi:10.1029/WR024i005p00755.\u003c/li\u003e\n\u003cli\u003eCeolin, Samuele, Stanislaus J. Schymanski, Dagmar van Dusschoten, Robert Koller, and Julian Klaus. 2025. \u0026ldquo;Root Growth Dynamics and Allocation as a Response to Rapid and Local Changes in Soil Moisture.\u0026rdquo; \u003cem\u003eBiogeosciences\u003c/em\u003e 22(3):691\u0026ndash;703. doi:10.5194/bg-22-691-2025.\u003c/li\u003e\n\u003cli\u003eConant, Richard T., Keith Paustian, and Edward T. Elliott. 2001. \u0026lsquo;Grassland Management and Conversion into Grassland: Effects on Soil Carbon\u0026rsquo;. \u003cem\u003eEcological Applications\u003c/em\u003e 11(2):343\u0026ndash;55. doi:10.1890/1051-0761(2001)011[0343:GMACIG]2.0.CO;2.\u003c/li\u003e\n\u003cli\u003eCraven, Dylan, Forest Isbell, Pete Manning, John Connolly, Helge Bruelheide, Anne Ebeling, Christiane Roscher, Jasper van Ruijven, Alexandra Weigelt, Brian Wilsey, Carl Beierkuhnlein, Enrica de Luca, John N. Griffin, Yann Hautier, Andy Hector, Anke Jentsch, J\u0026uuml;rgen Kreyling, Vojtech Lanta, Michel Loreau, Sebastian T. Meyer, Akira S. Mori, Shahid Naeem, Cecilia Palmborg, H. Wayne Polley, Peter B. Reich, Bernhard Schmid, Alrun Siebenk\u0026auml;s, Eric Seabloom, Madhav P. Thakur, David Tilman, Anja Vogel, and Nico Eisenhauer. 2016. \u0026lsquo;Plant Diversity Effects on Grassland Productivity Are Robust to Both Nutrient Enrichment and Drought\u0026rsquo;. \u003cem\u003ePhilosophical Transactions of the Royal Society of London. Series B, Biological Sciences\u003c/em\u003e 371(1694):20150277. doi:10.1098/rstb.2015.0277.\u003c/li\u003e\n\u003cli\u003eDrew, Malcolm C. 1992. \u0026lsquo;SOIL AERATION AND PLANT ROOT METABOLISM\u0026rsquo;. \u003cem\u003eSoil Science\u003c/em\u003e 154(4):259.\u003c/li\u003e\n\u003cli\u003eEdwards, Everard J., David G. Benham, Louise A. Marland, and Alastair H. Fitter. 2004. \u0026lsquo;Root Production Is Determined by Radiation Flux in a Temperate Grassland Community\u0026rsquo;. \u003cem\u003eGlobal Change Biology\u003c/em\u003e 10(2):209\u0026ndash;27. doi:10.1111/j.1365-2486.2004.00729.x.\u003c/li\u003e\n\u003cli\u003eEngels, Christof, Martin Mollenkopf, and Horst Marschner. 1994. \u0026lsquo;Effect of Drying and Rewetting the Topsoil on Root Growth of Maize and Rape in Different Soil Depths\u0026rsquo;. \u003cem\u003eZeitschrift F\u0026uuml;r Pflanzenern\u0026auml;hrung Und Bodenkunde\u003c/em\u003e 157(2):139\u0026ndash;44. doi:10.1002/jpln.19941570213.\u003c/li\u003e\n\u003cli\u003eEshel, Amram, and Tom Beeckman, eds. 2013. \u003cem\u003ePlant Roots: The Hidden Half, Fourth Edition\u003c/em\u003e. 4 edition. Boca Raton, FL: CRC Press.\u003c/li\u003e\n\u003cli\u003eFay, Philip A., Suzanne M. Prober, W. Stanley Harpole, Johannes M. H. Knops, Jonathan D. Bakker, Elizabeth T. Borer, Eric M. Lind, Andrew S. MacDougall, Eric W. Seabloom, Peter D. Wragg, Peter B. Adler, Dana M. Blumenthal, Yvonne M. Buckley, Chengjin Chu, Elsa E. Cleland, Scott L. Collins, Kendi F. Davies, Guozhen Du, Xiaohui Feng, Jennifer Firn, Daniel S. Gruner, Nicole Hagenah, Yann Hautier, Robert W. Heckman, Virginia L. Jin, Kevin P. Kirkman, Julia Klein, Laura M. Ladwig, Qi Li, Rebecca L. McCulley, Brett A. Melbourne, Charles E. Mitchell, Joslin L. Moore, John W. Morgan, Anita C. Risch, Martin Sch\u0026uuml;tz, Carly J. Stevens, David A. Wedin, and Louie H. Yang. 2015. \u0026lsquo;Grassland Productivity Limited by Multiple Nutrients\u0026rsquo;. \u003cem\u003eNature Plants\u003c/em\u003e 1(7):1\u0026ndash;5. doi:10.1038/nplants.2015.80.\u003c/li\u003e\n\u003cli\u003eFromm, Hillel. 2019. \u0026lsquo;Root Plasticity in the Pursuit of Water\u0026rsquo;. \u003cem\u003ePlants\u003c/em\u003e 8(7):236. doi:10.3390/plants8070236.\u003c/li\u003e\n\u003cli\u003evan Genuchten, M. Th. 1980. \u0026lsquo;A Closed-Form Equation for Predicting the Hydraulic Conductivity of Unsaturated Soils\u0026rsquo;. \u003cem\u003eSoil Science Society of America Journal\u003c/em\u003e 44(5):892\u0026ndash;98. doi:https://doi.org/10.2136/sssaj1980.03615995004400050002x.\u003c/li\u003e\n\u003cli\u003eGill, Richard A., Ingrid C. Burke, William K. Lauenroth, and Daniel G. Milchunas. 2002. \u0026lsquo;Longevity and Turnover of Roots in the Shortgrass Steppe: Influence of Diameter and Depth\u0026rsquo;. \u003cem\u003ePlant Ecology\u003c/em\u003e 159(2):241\u0026ndash;51. doi:10.1023/A:1015529507670.\u003c/li\u003e\n\u003cli\u003eHayes, D. C., and T. R. Seastedt. 1987. \u0026lsquo;Root Dynamics of Tallgrass Prairie in Wet and Dry Years\u0026rsquo;. \u003cem\u003eCanadian Journal of Botany\u003c/em\u003e 65(4):787\u0026ndash;91. doi:10.1139/b87-105.\u003c/li\u003e\n\u003cli\u003eHendrick, Ronald L., and Kurt S. Pregitzer. 1996. \u0026lsquo;Temporal and Depth-Related Patterns of Fine Root Dynamics in Northern Hardwood Forests\u0026rsquo;. \u003cem\u003eThe Journal of Ecology\u003c/em\u003e 84(2):167. doi:10.2307/2261352.\u003c/li\u003e\n\u003cli\u003eHodge, Angela. 2004. \u0026lsquo;The Plastic Plant: Root Responses to Heterogeneous Supplies of Nutrients\u0026rsquo;. \u003cem\u003eNew Phytologist\u003c/em\u003e 162(1):9\u0026ndash;24. doi:https://doi.org/10.1111/j.1469-8137.2004.01015.x.\u003c/li\u003e\n\u003cli\u003eHu, Yutong, Xiaorong Wei, Mingde Hao, Wei Fu, Jing Zhao, and Zhe Wang. 2018. \u0026lsquo;Partial Least Squares Regression for Determining Factors Controlling Winter Wheat Yield\u0026rsquo;. \u003cem\u003eAgronomy Journal\u003c/em\u003e 110(1):281\u0026ndash;92. doi:10.2134/agronj2017.02.0108.\u003c/li\u003e\n\u003cli\u003eIsbell, Forest, Dylan Craven, John Connolly, Michel Loreau, Bernhard Schmid, Carl Beierkuhnlein, T. Martijn Bezemer, Catherine Bonin, Helge Bruelheide, Enrica de Luca, Anne Ebeling, John N. Griffin, Qinfeng Guo, Yann Hautier, Andy Hector, Anke Jentsch, J\u0026uuml;rgen Kreyling, Vojtěch Lanta, Pete Manning, and Sebastian T. Meyer. 2015. \u0026lsquo;Biodiversity Increases the Resistance of Ecosystem Productivity to Climate Extremes\u0026rsquo;. \u003cem\u003eNature\u003c/em\u003e 526(7574):574\u0026ndash;77. doi:10.1038/nature15374.\u003c/li\u003e\n\u003cli\u003eJoslin, J. D., M. H. Wolfe, and P. J. Hanson. 2001. \u0026lsquo;Factors Controlling the Timing of Root Elongation Intensity in a Mature Upland Oak Stand\u0026rsquo;. \u003cem\u003ePlant and Soil\u003c/em\u003e 228(2):201\u0026ndash;12.\u003c/li\u003e\n\u003cli\u003eKaspar, T. C., and W. L. Bland. 1992. \u0026lsquo;SOIL TEMPERATURE AND ROOT GROWTH\u0026rsquo;: \u003cem\u003eSoil Science\u003c/em\u003e 154(4):290\u0026ndash;99. doi:10.1097/00010694-199210000-00005.\u003c/li\u003e\n\u003cli\u003eKlaus, Julian, Wendy A. Monk, Lu Zhang, and David M. Hannah. 2022. \u0026lsquo;Ecohydrological Interactions during Drought\u0026rsquo;. \u003cem\u003eEcohydrology\u003c/em\u003e 15(5):e2456. doi:10.1002/eco.2456.\u003c/li\u003e\n\u003cli\u003eLeuschner, Ch, K. Backes, D. Hertel, F. Schipka, U. Schmitt, O. Terborg, and M. Runge. 2001. \u0026lsquo;Drought Responses at Leaf, Stem and Fine Root Levels of Competitive Fagus Sylvatica L. and Quercus Petraea (Matt.) Liebl. Trees in Dry and Wet Years\u0026rsquo;. \u003cem\u003eForest Ecology and Management\u003c/em\u003e 149(1):33\u0026ndash;46. doi:10.1016/S0378-1127(00)00543-0.\u003c/li\u003e\n\u003cli\u003eLi, Xinhai, Baidu Li, Guiming Wang, Xiangjiang Zhan, and Marcel Holyoak. 2020. \u0026lsquo;Deeply Digging the Interaction Effect in Multiple Linear Regressions Using a Fractional-Power Interaction Term\u0026rsquo;. \u003cem\u003eMethodsX\u003c/em\u003e 7:101067. doi:10.1016/j.mex.2020.101067.\u003c/li\u003e\n\u003cli\u003eMa, WenHong, YuanHe Yang, JinSheng He, Hui Zeng, and JingYun Fang. 2008. \u0026lsquo;Above- and Belowground Biomass in Relation to Environmental Factors in Temperate Grasslands, Inner Mongolia\u0026rsquo;. \u003cem\u003eScience in China Series C: Life Sciences\u003c/em\u003e 51(3):263\u0026ndash;70. doi:10.1007/s11427-008-0029-5.\u003c/li\u003e\n\u003cli\u003eMarx, Simone, and Frank Flammang. 2015. \u0026lsquo;La cartographie des sols au Grand-Duch\u0026eacute; de Luxembourg\u0026rsquo;.\u003c/li\u003e\n\u003cli\u003eMcVicar, Tim R., Michael L. Roderick, Randall J. Donohue, Ling Tao Li, Thomas G. Van Niel, Axel Thomas, J\u0026uuml;rgen Grieser, Deepak Jhajharia, Youcef Himri, Natalie M. Mahowald, Anna V. Mescherskaya, Andries C. Kruger, Shafiqur Rehman, and Yagob Dinpashoh. 2012. \u0026lsquo;Global Review and Synthesis of Trends in Observed Terrestrial Near-Surface Wind Speeds: Implications for Evaporation\u0026rsquo;. \u003cem\u003eJournal of Hydrology\u003c/em\u003e 416\u0026ndash;417:182\u0026ndash;205. doi:10.1016/j.jhydrol.2011.10.024.\u003c/li\u003e\n\u003cli\u003eMetcalfe, Daniel B., Patrick Meir, Luiz Eduardo O. C. Arag\u0026atilde;o, Antonio C. L. da Costa, Alan P. Braga, Paulo H. L. Gon\u0026ccedil;alves, Joao de Athaydes Silva Junior, Samuel S. de Almeida, Lorna A. Dawson, Yadvinder Malhi, and Mathew Williams. 2008. \u0026lsquo;The Effects of Water Availability on Root Growth and Morphology in an Amazon Rainforest\u0026rsquo;. \u003cem\u003ePlant and Soil\u003c/em\u003e 311(1):189\u0026ndash;99. doi:10.1007/s11104-008-9670-9.\u003c/li\u003e\n\u003cli\u003eMontagnoli, A., A. Di Iorio, M. Terzaghi, D. Trupiano, G. S. Scippa, and D. Chiatante. 2014. \u0026lsquo;Influence of Soil Temperature and Water Content on Fine-Root Seasonal Growth of European Beech Natural Forest in Southern Alps, Italy\u0026rsquo;. \u003cem\u003eEuropean Journal of Forest Research\u003c/em\u003e 133(5):957\u0026ndash;68. doi:10.1007/s10342-014-0814-6.\u003c/li\u003e\n\u003cli\u003eNakagawa, Shinichi, and Holger Schielzeth. 2013. \u0026lsquo;A General and Simple Method for Obtaining \u003cem\u003eR\u003c/em\u003e \u003csup\u003e2\u003c/sup\u003e from Generalized Linear Mixed‐effects Models\u0026rsquo; edited by R. B. O\u0026rsquo;Hara. \u003cem\u003eMethods in Ecology and Evolution\u003c/em\u003e 4(2):133\u0026ndash;42. doi:10.1111/j.2041-210x.2012.00261.x.\u003c/li\u003e\n\u003cli\u003eNippert, Jesse B., and Alan K. Knapp. 2007. \u0026lsquo;Soil Water Partitioning Contributes to Species Coexistence in Tallgrass Prairie\u0026rsquo;. \u003cem\u003eOikos\u003c/em\u003e 116(6):1017\u0026ndash;29. doi:10.1111/j.0030-1299.2007.15630.x.\u003c/li\u003e\n\u003cli\u003ePedregosa, Fabian, Gael Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, Jake Vanderplas, Alexandre Passos, and David Cournapeau. 2011. \u0026lsquo;Scikit-Learn: Machine Learning in Python\u0026rsquo;. \u003cem\u003eMACHINE LEARNING IN PYTHON\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003ePeek, M. S., A. J. Leffler, L. Hipps, S. Ivans, R. J. Ryel, and M. M. Caldwell. 2006. \u0026lsquo;Root Turnover and Relocation in the Soil Profile in Response to Seasonal Soil Water Variation in a Natural Stand of Utah Juniper (Juniperus Osteosperma)\u0026rsquo;. \u003cem\u003eTree Physiology\u003c/em\u003e 26(11):1469\u0026ndash;76. doi:10.1093/treephys/26.11.1469.\u003c/li\u003e\n\u003cli\u003eSaelim, Sakanan, Sayan Sdoodee, and Rawee Chiarawipa. 2019. \u0026lsquo;Monitoring Seasonal Fine Root Dynamics of Hevea Brasiliensis Clone RRIM 600 in Southern Thailand Using Minirhizotron Technique\u0026rsquo;. 8.\u003c/li\u003e\n\u003cli\u003eSchwinning, Susanne, and James R. Ehleringer. 2001. \u0026lsquo;Water Use Trade-Offs and Optimal Adaptations to Pulse-Driven Arid Ecosystems\u0026rsquo;. \u003cem\u003eJournal of Ecology\u003c/em\u003e 89(3):464\u0026ndash;80.\u003c/li\u003e\n\u003cli\u003eSeethepalli, Anand, Kundan Dhakal, Marcus Griffiths, Haichao Guo, Gregoire T. Freschet, and Larry M. York. 2021. \u003cem\u003eRhizoVision Explorer: Open-Source Software for Root Image Analysis and Measurement Standardization\u003c/em\u003e. \u003cem\u003epreprint\u003c/em\u003e. Plant Biology. doi:10.1101/2021.04.11.439359.\u003c/li\u003e\n\u003cli\u003eShahhosseini, Mohsen, Rafael A. Martinez-Feria, Guiping Hu, and Sotirios V. Archontoulis. 2019. \u0026lsquo;Maize Yield and Nitrate Loss Prediction with Machine Learning Algorithms\u0026rsquo;. \u003cem\u003eEnvironmental Research Letters\u003c/em\u003e 14(12):124026. doi:10.1088/1748-9326/ab5268.\u003c/li\u003e\n\u003cli\u003eda Silva, Alvaro Pires, and B. D. Kay. 1997. \u0026lsquo;Estimating the Least Limiting Water Range of Soils from Properties and Management\u0026rsquo;. \u003cem\u003eSoil Science Society of America Journal\u003c/em\u003e 61(3):877\u0026ndash;83. doi:10.2136/sssaj1997.03615995006100030023x.\u003c/li\u003e\n\u003cli\u003eSlette, Ingrid J., David L. Hoover, Melinda D. Smith, and Alan K. Knapp. 2023. \u0026lsquo;Repeated Extreme Droughts Decrease Root Production, but Not the Potential for Post-Drought Recovery of Root Production, in a Mesic Grassland\u0026rsquo;. \u003cem\u003eOikos\u003c/em\u003e 2023(1):e08899. doi:10.1111/oik.08899.\u003c/li\u003e\n\u003cli\u003eSmith, Abraham George, Eusun Han, Jens Petersen, Niels Alvin Faircloth Olsen, Christian Giese, Miriam Athmann, Dorte Bodin Dresb\u0026oslash;ll, and Kristian Thorup-Kristensen. 2020. \u003cem\u003eRootPainter: Deep Learning Segmentation of Biological Images with Corrective Annotation\u003c/em\u003e. \u003cem\u003epreprint\u003c/em\u003e. Plant Biology. doi:10.1101/2020.04.16.044461.\u003c/li\u003e\n\u003cli\u003eStewart, Anna M., and Douglas A. Frank. 2008. \u0026lsquo;Short Sampling Intervals Reveal Very Rapid Root Turnover in a Temperate Grassland\u0026rsquo;. \u003cem\u003eOecologia\u003c/em\u003e 157(3):453\u0026ndash;58. doi:10.1007/s00442-008-1088-9.\u003c/li\u003e\n\u003cli\u003eTeskey, Robert O., and Thomas M. Hinckley. 1981. \u0026lsquo;Influence of Temperature and Water Potential on Root Growth of White Oak\u0026rsquo;. \u003cem\u003ePhysiologia Plantarum\u003c/em\u003e 52(3):363\u0026ndash;69. doi:10.1111/j.1399-3054.1981.tb06055.x.\u003c/li\u003e\n\u003cli\u003eThornley, J. H. M. 1972. \u0026lsquo;A Balanced Quantitative Model for Root: Shoot Ratios in Vegetative Plants\u0026rsquo;. \u003cem\u003eAnnals of Botany\u003c/em\u003e 36(145):431\u0026ndash;41.\u003c/li\u003e\n\u003cli\u003eToğa\u0026ccedil;ar, Mesut, Burhan Ergen, and Zafer C\u0026ouml;mert. 2020. \u0026lsquo;Application of Breast Cancer Diagnosis Based on a Combination of Convolutional Neural Networks, Ridge Regression and Linear Discriminant Analysis Using Invasive Breast Cancer Images Processed with Autoencoders\u0026rsquo;. \u003cem\u003eMedical Hypotheses\u003c/em\u003e 135:109503. doi:10.1016/j.mehy.2019.109503.\u003c/li\u003e\n\u003cli\u003eTorbenson, Max C. A., Ulf B\u0026uuml;ntgen, Jan Esper, Otmar Urban, Jan Balek, Frederick Reinig, Paul J. Krusic, Edurne Martinez Del Castillo, Rudolf Br\u0026aacute;zdil, Daniela Semer\u0026aacute;dov\u0026aacute;, Petr \u0026Scaron;těp\u0026aacute;nek, Nat\u0026aacute;lie Pernicov\u0026aacute;, Tom\u0026aacute;\u0026scaron; Kol\u0026aacute;ř, Michal Rybn\u0026iacute;ček, Eva Koňasov\u0026aacute;, Juliana Arbelaez, and Miroslav TRNKAc. 2023. \u0026lsquo;Central European Agroclimate over the Past 2000 Years\u0026rsquo;. \u003cem\u003eJournal of Climate\u003c/em\u003e 36(13):4429\u0026ndash;41. doi:10.1175/JCLI-D-22-0831.1.\u003c/li\u003e\n\u003cli\u003eTripathy, Kumar Puran, and Ashok Kumar Mishra. 2023. \u0026lsquo;How Unusual Is the 2022 European Compound Drought and Heatwave Event?\u0026rsquo; \u003cem\u003eGeophysical Research Letters\u003c/em\u003e 50(15):e2023GL105453. doi:10.1029/2023GL105453.\u003c/li\u003e\n\u003cli\u003eVamerali, Teofilo, Marianna Bandiera, and Giuliano Mosca. 2011. \u0026lsquo;Minirhizotrons in Modern Root Studies\u0026rsquo;. Pp. 341\u0026ndash;61 in \u003cem\u003eMeasuring Roots\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eVan Loon, Anne F., Tom Gleeson, Julian Clark, Albert I. J. M. Van Dijk, Kerstin Stahl, Jamie Hannaford, Giuliano Di Baldassarre, Adriaan J. Teuling, Lena M. Tallaksen, Remko Uijlenhoet, David M. Hannah, Justin Sheffield, Mark Svoboda, Boud Verbeiren, Thorsten Wagener, Sally Rangecroft, Niko Wanders, and Henny A. J. Van Lanen. 2016. \u0026lsquo;Drought in the Anthropocene\u0026rsquo;. \u003cem\u003eNature Geoscience\u003c/em\u003e 9(2):89\u0026ndash;91. doi:10.1038/ngeo2646.\u003c/li\u003e\n\u003cli\u003eVirtanen, Pauli, Ralf Gommers, Travis E. Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, St\u0026eacute;fan J. van der Walt, Matthew Brett, Joshua Wilson, K. Jarrod Millman, Nikolay Mayorov, Andrew R. J. Nelson, Eric Jones, Robert Kern, Eric Larson, C. J. Carey, İlhan Polat, Yu Feng, Eric W. Moore, Jake VanderPlas, Denis Laxalde, Josef Perktold, Robert Cimrman, Ian Henriksen, E. A. Quintero, Charles R. Harris, Anne M. Archibald, Ant\u0026ocirc;nio H. Ribeiro, Fabian Pedregosa, and Paul van Mulbregt. 2020. \u0026lsquo;SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python\u0026rsquo;. \u003cem\u003eNature Methods\u003c/em\u003e 17(3):261\u0026ndash;72. doi:10.1038/s41592-019-0686-2.\u003c/li\u003e\n\u003cli\u003eWan, Changgui, Ibrahim Yilmaz, and Ronald E. Sosebee. 2002. \u0026lsquo;Seasonal Soil\u0026ndash;Water Availability Influences Snakeweed Root Dynamics\u0026rsquo;. \u003cem\u003eJournal of Arid Environments\u003c/em\u003e 51(2):255\u0026ndash;64. doi:10.1006/jare.2001.0942.\u003c/li\u003e\n\u003cli\u003eWedderburn, ME, JR Crush, WJ Pengelly, and JL Walcroft. 2010. \u0026lsquo;Root Growth Patterns of Perennial Ryegrasses under Well-Watered and Drought Conditions\u0026rsquo;. \u003cem\u003eNew Zealand Journal of Agricultural Research\u003c/em\u003e 53(4):377\u0026ndash;88. doi:10.1080/00288233.2010.514927.\u003c/li\u003e\n\u003cli\u003eZhang, Bingwei, Marc W. Cadotte, Shiping Chen, Xingru Tan, Cuihai You, Tingting Ren, Minling Chen, Shanshan Wang, Weijing Li, Chengjin Chu, Lin Jiang, Yongfei Bai, Jianhui Huang, and Xingguo Han. 2019. \u0026lsquo;Plants Alter Their Vertical Root Distribution Rather than Biomass Allocation in Response to Changing Precipitation\u0026rsquo;. \u003cem\u003eEcology\u003c/em\u003e 100(11):e02828. doi:https://doi.org/10.1002/ecy.2828.\u003c/li\u003e\n\u003cli\u003eZhou, Huailin, Lulu Hou, Xiaomin Lv, Guang Yang, Yuhui Wang, and Xu Wang. 2022. \u0026lsquo;Compensatory Growth as a Response to Post-Drought in Grassland\u0026rsquo;. \u003cem\u003eFrontiers in Plant Science\u003c/em\u003e 13. https://www.frontiersin.org/articles/10.3389/fpls.2022.1004553.\u003c/li\u003e\n\u003cli\u003eZwetsloot, Marie J., and Taryn L. Bauerle. 2021. \u0026lsquo;Repetitive Seasonal Drought Causes Substantial Species-Specific Shifts in Fine-Root Longevity and Spatio-Temporal Production Patterns in Mature Temperate Forest Trees\u0026rsquo;. \u003cem\u003eNew Phytologist\u003c/em\u003e 231(3):974\u0026ndash;86. doi:10.1111/nph.17432.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":true,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"plant-and-soil","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"plso","sideBox":"Learn more about [Plant and Soil](https://www.springer.com/journal/11104)","snPcode":"11104","submissionUrl":"https://submission.nature.com/new-submission/11104/3","title":"Plant and Soil","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Root dynamics, Root allocation, Hydromatching, Root seasonal shifts, Water uptake, Root growth predictors","lastPublishedDoi":"10.21203/rs.3.rs-7847447/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7847447/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground and aims\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e“Hydromatching” is a phenomenon consisting of the daily-scale promotion of root growth in a newly wetted soil layer and/or a decline in root growth in drier layers. This phenomenon was previously observed on individual plants under controlled conditions. The aim of this study was to determine if hydromatching occurs also under natural settings at a grassland community scale. Our goal was also to assess what environmental parameter was driving root growth in the grassland community and to analyze seasonal shifts in root distribution.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe installed twelve minirhizotron tubes in a natural grassland. We imaged the tubes from May 2022 until August 2023 and we carried out image analyses to extrapolate root length and growth rates. We determined the major environmental driver of root growth through regression analyses and monitored the root growth dynamics after major rain events and during the spring-summer transition.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSoil moisture was the strongest predictor of root growth. Following rain events, root growth shifted from deeper layers to shallow layers within 1-3 days, indicating the occurrence of hydromatching. During the spring-summer transition, we observed significant promotion of growth in deeper soil layers and decline in root length in shallower layers.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRoot distribution responded to seasonality with drastic shifts at the whole root system level and to precipitation with smaller but significant rapid shifts. Both types of response could adhere to an optimization strategy, consisting of the promotion of root growth in resourceful areas while discarding roots where resources are less accessible.\u003c/p\u003e","manuscriptTitle":"Controls on vertical root distribution dynamics in a temperate grassland across daily and seasonal scales","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-13 11:46:34","doi":"10.21203/rs.3.rs-7847447/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2026-01-14T09:06:37+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-01-13T14:30:36+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-13T08:21:58+00:00","index":"","fulltext":""},{"type":"submitted","content":"Plant and Soil","date":"2026-01-12T17:12:53+00:00","index":"","fulltext":""},{"type":"decision","content":"Major revisions","date":"2025-11-21T03:02:00+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"plant-and-soil","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"plso","sideBox":"Learn more about [Plant and Soil](https://www.springer.com/journal/11104)","snPcode":"11104","submissionUrl":"https://submission.nature.com/new-submission/11104/3","title":"Plant and Soil","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"8b07ad2e-3854-4eaf-a571-613bee65bb60","owner":[],"postedDate":"February 13th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-06T00:47:34+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-13 11:46:34","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7847447","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7847447","identity":"rs-7847447","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}