FIGURE 6 - available via license: Creative Commons Attribution-NonCommercial 4.0 International
Content may be subject to copyright.
The mean annual soil temperature (SBIO1, 1 x 1 km resolution) modelled here is consistently cooler than ERA5L (9 x 9 km) soil temperature in forested areas. (a) Spatial representation of the difference between SBIO1 based on our model and based on ERA5L soil temperature data. Negative values (blue colours) indicate areas where our model predicts cooler soil temperature. Dark grey areas (Greenland and Antarctica) are excluded from our models. Asterisk in Scandinavia indicates the highlighted area in panels d to f (see below). (b) Distribution of the difference between SBIO1 and ERA5L along the macroclimatic gradient (represented by SBIO1 itself) based on a random subsample of 50,000 points from the map in a). Red line from a Generalized Additive Model (GAM) with k = 4. (c-e) High-resolution zoomed panels of an area of high elevational contrast in Norway (from 66.0-66.4°N, 15.0-16.0°E) visualizing SBIO1 (c), ERA5L (d) and their difference (e), to highlight the higher spatial resolution as obtained with SBIO1
Source publication
Research in global change ecology relies heavily on global climatic grids derived from estimates of air temperature in open areas at around 2 m above the ground. These climatic grids do not reflect conditions below vegetation canopies and near the ground surface, where critical ecosystem functions occur and most terrestrial species reside. Here, we...
Contexts in source publication
Context 1
... we trained RF models (with the same hyperparameters as selected during the grid-search procedure) using each of 100 bootstrap iterations. Each of these trained RF models was then used to classify the predictor layer stack, to generate per-pixel 95% confidence intervals and standard deviation for the modelled monthly offsets (Figure 5a, Figure S6a). The mean R² value of the RF models for the monthly mean temperature offset was 0.70 (from 0.64 to 0.78) at 0-5 cm and 0.76 (0.63-0.85) at 5 to 15 cm across all 12 monthly models. ...
Context 2
... bootstrap approach to validate modelled monthly offsets indicated high consistency among the outcomes of 100 bootstrapped models ( Figure 5, Figure S6a), with standard deviations in most months and across most parts of the globe around or below ±1°C. One exception to this was the temperature offset at high latitudes of the Northern ...
Context 3
... highlight that the current availability of in situ soil temperature measurements is significantly lower in the tropics (Table S5), where our model had to extrapolate temperatures beyond the range used to calibrate the model (Figure 5b, Figure S6b). ...
Context 4
... our comparison with a mean annual soil temperature product derived from the coarse-resolution ERA5L topsoil temperature showed that spatial variability, for example, driven by topographic heterogeneity, is much better captured here than in the coarser resolution of the ER A5L-based product (Figure 6c-e). Nevertheless, our predictions at the coarse scale showed to be condensed within a 5°C range of values from the ERA5L-predictions, for more than 95% of pixels globally. ...
Context 5
... our predictions at the coarse scale showed to be condensed within a 5°C range of values from the ERA5L-predictions, for more than 95% of pixels globally. Noteworthy, our predictions resulted in consistently cooler soil temperature predictions than topsoil conditions provided by ERA5L across large areas, such as the boreal and tropical forest biomes (Figure 6a,b). Additionally, our models predicted lower values for SBIO1 than ERA5L in all regions with mean annual soil temperature below 0°C, except for a few locations around Greenland and Svalbard (Figure 6a,b). ...
Context 6
... our predictions resulted in consistently cooler soil temperature predictions than topsoil conditions provided by ERA5L across large areas, such as the boreal and tropical forest biomes (Figure 6a,b). Additionally, our models predicted lower values for SBIO1 than ERA5L in all regions with mean annual soil temperature below 0°C, except for a few locations around Greenland and Svalbard (Figure 6a,b). ...
Similar publications
Via sheltering, decoupling and buffering mechanisms, tree canopies have the capacity to mitigate impacts of multiple global‐change drivers on below‐canopy processes and organisms in forests. As a result, canopies have an important potential as nature‐based solution.
The optimal combinations of forest canopy structural attributes to jointly mitigate...
Citations
... Because scorpions are among the soil fauna, we also used the soil temperature layers. We used the available data from Lembrechts et al. (2022). The resolution of the soil temperature layers was also 1 km 2 , and a list of these 11 soil temperature layers is also reported in Table S1. ...
Iran is a biodiversity hotspot of scorpions with 80 recorded species. Thus, scorpion envenomation is a serious public health problem in the country. Here, we used habitat suitability modeling to assess the spatial distribution of scorpions in Iran. Only 45 species had sufficient georeferenced data. We used bioclimatic variables, soil temperature layers, and 9 modeling algorithms to perform habitat suitability modeling. We employed an ensemble approach to obtain the final models. We calculated the richness map and drew distribution maps for genera with more than 1 species. Also, we assessed the scorpions’ species richness inside and outside of national parks. Finally, we created a risk map of encountering a venomous scorpion. The results showed that the highest scorpion richness is in the southern and southwestern parts of Iran, especially in the coastal areas of the Persian Gulf. We observed 3 biodiversity hotspots for scorpions that are located in the south and southwestern, central parts, and eastern parts of Iran. Except for northern parts, there is a possibility of encountering a venomous scorpion in other parts of Iran. The 3 biodiversity hotspots are also the areas with the highest chance of encountering a venomous species. We found that the hotspots are not protected and are located in areas facing land-use changes. Thus, hotspots have the highest human–scorpion conflicts. Our results provide new insight into the distribution of scorpion species in Iran. Conservation actions that ensure both human safety and species richness are essential and can be achieved by halting further land degradation in scorpion habitats and providing easy-to-understand manuals for local people.
... Open access to microclimate data facilitates easy access for researchers working within the same study system, supports larger-scale collaborative analyses for understanding the role of microclimatic processes in ecology (e.g. Risch et al., 2023) and enables regional or global microclimate mapping, which all in all enhances baseline data availability Lembrechts et al., 2022). Openly sharing such data also contributes to the creation of long-term data sets, essential for assessing microclimate changes over recent decades Finally, sharing data is a major step towards efficient scientific collaboration, but the same argument can be made for any microclimate-related code or script to handle and analyse the data. ...
Most biodiversity dynamics and ecosystem processes on land take place in microclimates that are decoupled from the climate as measured by standardised weather stations in open, unshaded locations. As a result, microclimate monitoring is increasingly being integrated in many studies in ecology and evolution.
Overviews of the protocols and measurement methods related to microclimate are needed, especially for those starting in the field and to achieve more generality and standardisation in microclimate studies.
Here, we present 10 practical guidelines for ground‐based research of terrestrial microclimates, covering methods and best practices from initial conceptualisation of the study to data analyses.
Our guidelines encompass the significance of microclimates; the specifics of what, where, when and how to measure them; the design of microclimate studies; and the optimal approaches for analysing and sharing data for future use and collaborations. The paper is structured as a chronological guide, leading the reader through each step necessary to conduct a comprehensive microclimate study. At the end, we also discuss further research avenues and development in this field.
With these 10 guidelines for microclimate monitoring, we hope to stimulate and advance microclimate research in ecology and evolution, especially under the pressing need to account for buffering or amplifying abilities of contrasting microhabitats in the context of global climate change.
... The type of phytocenosis also affects the dependence of CO 2 emission on soil temperature (Koptsik et al., 2018). In general, the difference between soil and air temperature varies greatly across climate zones, biomes, and seasons of the year (Lembrechts et al., 2022). ...
... Filling this gap is essential for understanding and predicting river corridor biogeochemical functioning, human impacts, and system responses to environmental change. One promising approach is the application of scaling laws, which have been demonstrated to provide generalisable frameworks of the spatial organisation of the ecological and biological structure on Earth (Lembrechts et al., 2022). ...
Understanding the molecular complexity of dissolved organic matter (DOM) is crucial for deciphering ecosystem function and predicting responses to environmental change. Using a continental-scale dataset of molecular formulae generated by Fourier-transform ion cyclotron resonance mass spectrometry through the WHONDRS Project, we reveal fundamental scaling patterns of DOM diversity across watershed characteristics. We show that both local and regional watershed features across latitude reveal universal scaling patterns across compound classes. Our analysis demonstrates that local and regional watershed features—including drainage area, land use, land cover, and water temperature—significantly influence DOM diversity, quantified through DOM richness and chemodiversity. Notably, we document one of the first observed latitudinal gradients in freshwater chemodiversity, suggesting that macroecological factors orchestrate molecular diversity similarly to organismal distributions. These findings offer more than descriptive insights. By identifying universal scaling laws, we propose that simple predictive models can forecast compound responses to climate change, providing critical understanding of watershed functional dynamics under emerging environmental conditions. The research contributes a novel perspective on ecosystem molecular complexity, bridging biogeochemical processes with broader ecological principles and offering a quantitative framework for anticipating ecosystem molecular transformations across scales.
... Climate change in soil largely follow the aboveground trends described above, meaning that soil temperature increases when air temperature increases. There is large spatiotemporal heterogeneity in the global offset between soil and air temperature, often in the order of several degrees annually and up to more than 20 • C during winter months at high latitudes (Lembrechts et al., 2019(Lembrechts et al., , 2022. Such large offset is found in the most continental areas at high northern latitudes where an insulating snowpack causes the large difference between ambient and soil thermal conditions. ...
The Arctic amplification affects the geology, cryosphere, and the total environment of high-latitude maritime influenced lands. This study synthesizes information on recent and future climatic changes within the Nordic boreo-arctic region. The study area includes Greenland, Iceland, and the central and northern parts of Finland, Norway (incl. Svalbard), and Sweden. The climate scenarios used are derived from the CMIP6 ECEarth3 Earth System Model (ESM) data for the period 2015–2100 under the SSP2–4.5 scenario. The synthesis builds upon a comprehensive range of sources, addressing both gradual climatic changes and the frequency of extreme weather events across all seasons. Ongoing and projected changes to the cryosphere, soil, freshwater systems, wind, precipitation, and frequency of hazardous events are comprehensively reviewed and discussed.
... This temperature does not account for soil temperature changes. Soil temperature can differ by 10°C, compared to atmospheric temperature, around the world (Lembrechts et al. 2022), with groundcover vegetation type creating microclimates buffering temperature differences (De Frenne et al. 2019;Huang et al. 2024;Xiao, Ma, and Hu 2019). Under extreme dry conditions, such vegetation reduces high air temperatures and soil moisture loss (Lortie et al. 2022), which is linked to evapotranspiration and rainfall in drylands (Zhou et al. 2021). ...
Introduction
Drylands are a major terrestrial biome, supporting much of the earth's population. Soil microbial communities maintain drylands’ ecosystem functions but are threatened by increasing temperature. Groundcover, such as vegetation or biocrust, drives the patchiness of drylands' soil microbial communities, reflected in fertile islands and rhizosphere soil microbial associations. Groundcover may shelter soil microbial communities from increasingly harsh temperatures under climate change, mitigating effects on microclimate, but few data on the microbial response exists. Understanding the fine‐scale interactions between plants and soil is crucial to improving conservation and management of drylands under climate change.
Materials and Methods
We used open‐top chambers to experimentally increase the temperature on five key groundcover species found in arid Australia, and are commonly present in drylands worldwide; bareground (controls), biocrust, perennial grass, Maireana sp. shrub, Acacia aneura trees, testing soil bacterial diversity and community composition response to the effects of increased temperatures.
Results
We found that groundcover was a stronger driver of soil bacterial composition than increased temperature, but this response varied with groundcover type. Larger groundcover types (Acacia and Maireana) buffered the impact of heat stress on the soil bacterial community. Bacterial diversity and species richness declined with heat stress affecting the bacterial communities associated with perennial grass, Maireana and Acacia. We identified 16 bacterial phyla significantly associated with groundcover types in ambient treatment. But, under heat stress, only three phyla, Verrumicrobiota, Patescibacteria, and Abditibacteriota, had significantly different relative abundance under groundcovers, Acacia and Maireana, compared to bareground controls. The soil bacterial community associated with perennial grass was most affected by increased temperature.
Conclusion
Our findings suggest soil communities may become more homogeneous under climate change, with compositional change, rather than diversity, tracking soil response to heat stress.
... Soil temperature is influenced by various soil properties, such as thermal conductivity and heat capacity, which are affected by factors like bulk density, moisture content, and organic matter (Jury and Horton, 2004). While ML models are often used to predict soil temperature based on historical data, these models can be limited by their reliance on specific observation periods and may not fully capture the underlying causes of temperature variations (Lembrechts et al., 2022). Although ML algorithms can identify nonlinear relationships between soil temperature and air temperature along with other climate variables, they are generally incapable of showing the physical processes involved. ...
Machine learning (ML) applications in soil science have significantly increased over the past two decades, reflecting a growing trend towards data-driven research addressing soil security. This extensive application has mainly focused on enhancing predictions of soil properties, particularly soil organic carbon, and improving the accuracy of digital soil mapping (DSM). Despite these advancements, the application of ML in soil science faces challenges related to data scarcity and the interpretability of ML models. There is a need for a shift towards Soil Science-Informed ML (SoilML) models that use the power of ML but also incorporate soil science knowledge in the training process to make predictions more reliable and generalisable. This paper proposes methodologies for embedding ML models with soil science knowledge to overcome current limitations. Incorporating soil science knowledge into ML models involves using observational priors to enhance training datasets, designing model structures which reflect soil science principles, and supervising model training with soil science-informed loss functions. The informed loss functions include observational constraints, coherency rules such as regularisation to avoid overfitting, and prior or soil-knowledge constraints that incorporate existing information about the parameters or outputs. By way of illustration, we present examples from four fields: digital soil mapping, soil spectroscopy, pedotransfer functions, and dynamic soil property models. We discuss the potential to integrate process-based models for improved prediction, the use of physics-informed neural networks, limitations, and the issue of overparametrisation. These approaches improve the relevance of ML predictions in soil science and enhance the models’ ability to generalise across different scenarios while maintaining soil science principles, transparency and reliability.
... Fifteen environmental predictors were chosen based on established ecological relationships (see Table S1). These predictors were derived from various sources and encompassed climate (Brun et al., 2022), soil characteristics (Lembrechts et al., 2022;Poggio et al., 2021), and human-mediated effects (e.g., livestock density, Gilbert et al., 2018). All environmental layers were averaged to a grid scale of 1 × 1 km grid cells and reprojected to the ESPG:4326 coordinate system using the functions 'resample' and 'project' from the terra R package (Hijmans, 2024). ...
... Gasch et al. (2015) provided the first study on 4D patterns of soil temperature and soil moisture with 10 m resolution on a large agronomy farm, but the model is applicable to that single field site only and it cannot be assumed here that the relationships learned by the model are transferable to other landscapes, including, e.g., different land use types, relief positions, or soil types. If larger and more heterogeneous areas are investigated, current modelling approaches typically work on low resolutions of about 1 km, which is too low to infer regional landscape-scale patterns (e.g., Rötzer et al., 2015;Lembrechts et al., 2022). Landscape units where we expect significant differences in soil temperature or soil moisture, e.g., caused by different land use, are mixed within single pixel, hence preventing a differentiated analysis of the patterns. ...
... Studies have revealed that winter temperatures in northern mid-and high-latitude areas are increasing at a rate exceeding 0.5°C per decade 15 . This increase is nearly 1.8 times faster than the rise in mean annual temperatures, particularly in high-latitude regions 15,16 . Winter warming is expected to heighten the risk of reduced winter crop yields by breaking dormancy 17,18 , advancing phenology 19 , shortening the growing season 20 and photosynthetic activity 21 , and exacerbating the incidence of pests and pathogens 22 . ...
... Winter warming is expected to heighten the risk of reduced winter crop yields by breaking dormancy 17,18 , advancing phenology 19 , shortening the growing season 20 and photosynthetic activity 21 , and exacerbating the incidence of pests and pathogens 22 . Despite limited research on how winter warming affects non-winter crops, it is important to note that winter warming can change soil temperature and moisture 16 , which can affect soil fertility 23 and influence the growth of these crops. ...
... Ignoring these discrepancies could pose significant risks to current food security and potentially destabilise and exacerbate the global food supply chain in forthcoming years. Furthermore, it is imperative to recognise the distinction between the mean annual soil temperature and the mean annual air temperature 16 . Most current studies have relied on air temperature to forecast future food production, introducing potential uncertainties into these predictions. ...
Global warming poses an unprecedented threat to agroecosystems. Although temperature increases are more pronounced during winter than in other seasons, the impact of winter warming on crop biomass carbon has not been elucidated. Here we integrate global observational data with a decade-long field experiment to uncover a significant negative correlation between winter soil temperature and crop biomass carbon. For every degree Celsius increase in winter soil temperature, straw and grain biomass carbon decreased by 6.6 ( ± 1.7) g kg⁻¹ and 10.2 ( ± 2.3) g kg⁻¹, respectively. This decline is primarily attributed to the loss of soil organic matter and micronutrients induced by warming. Ignoring the adverse effects of winter warming on crop biomass carbon could result in an overestimation of total food production by 4% to 19% under future warming scenarios. Our research highlights the critical need to incorporate winter warming into agricultural productivity models for more effective climate adaptation strategies.
