香京julia种子在线播放

This study was set up in ten solar parks (SPs) located in Southern France along the Rhône valley (Figure 1). The SPs are distributed along a climate gradient from the Mediterranean Sea in the south to inland sites in the north and east (Table 1). Distance to the sea was between 37 and 192 km, and sites differed in major climate characteristics from Mediterranean in the south to sub-Mediterranean at highest distances from the coastline (Table 1). Continentality increased to the north, mean annual temperature ranged from 11.8°C to 14.9°C and annual precipitation from 710 mm to 903 mm. Other environmental factors such as soil and technical characteristics are similar between SPs. The size of the SPs ranged from 1 to 7 ha, equipped with solar panel either fixed on the ground or trackers. Ground-fixed panels had a height of 1 m at the southern edge and of 2.5 m at the northern edge (inclination towards south) whereas the incliniation of tracker panels changed according to the sun position. The SP were built on former wasteland, crop land or semi-natural land (Table 1). The solar panels were aligned to form rows with a gap of 2.5 m between rows. All solar parks were managed by extensive sheep grazing. Parks were additionally mown when grazing was not intensive enough or irregular.

Within each SP, five blocks of at least 50 m distance were set up in spring 2021. In each block (Supplementary Appendix S1), three plots of 4 m² were delimited for the following treatments: outside solar panels (outside), between two rows of solar panels (between) and under solar panels (under) (N = 10 solar parks X 5 blocks X 3 treatments = 150).

The effect of solar panels on plant community composition and vegetation cover (vertical projection of above-ground plant organs) of all vascular plant species was evaluated in each plot in late spring. To evaluate the effects of solar panels on physiological performance of the vegetation, a chlorophyll index was measured by Multiplex^® 3 (FORCE-A, Orsay, France) with ten records of chlorphyll fluorescence within each plot. The Multiplex ResearchTM allows in situ, non-destructive and real-time measurements of chlorophyll content (Agati et al., 2011).

Soil was sampled in the same plots as the vegetation. In April 2021, three soil samples were randomly collected (10 cm depth) in each plot and mixed to one bulk sample. Bulk samples were sieved (mesh size: 2 mm) prior to analysis. An aliquot of samples was air-dried (1 week, 30 C) for physico-chemical analyses and another aliquot stored at 4°C for microbial analyses. Soil water content was determined after drying samples (24 h, 105°C). Total carbon (TC) and total nitrogen (TN) contents were measured by combustion in the CN FlashEA 1112 (ThermoFisher) (NF ISO 10694, NFISO 13878) elemental analyser.

In April 2021, the soil mesofauna was sampled in stainless steel tubes (5.5 cm diameter, 6 cm depth) using two core samples per plot after vegetation removal. Mites and springtails were extracted during a period of 7 days at 25°C using the Berlese-Tullgren method (Bano and Roy, 2016) and then stored in 70% ethanol. Springtails and mites were counted under a stereomicroscope.

Phospholipid fatty acids (PLFA) are essential components of all living cells, but several PLFA are specific to bacterial and fungal biomass. PLFA were analyzed according to Buyer and Sasser (2012) and Aupic-Samain et al. (2021). After freeze-drying, 1 g of dry soil of each sample was extracted in 4 mL of Bligh–Dyer extractant containing 4 μL of 1,2-dinonadecanoyl-sn-glycero-3-phosphocholine (C19:0; Avanti^® Polar lipids, Inc.). Lipids were separated using solid-phase extraction (SPE) on Phenomenex^® (Strata SI-1 with 50 mg of silica, 55 μm, 70 Å). The resulting fatty acids methyl ester (FAME) were analysed by gas-chromatography/mass-spectrometry (GC-MS) on an Agilent 7890 system equipped with an MSD5977. A network mass detector, an ALS7693 automatic injector and an HP5-MS apolar column (30 m × 0.25 mm × 0.25 μm; JandW Agilent Technologies). Qualitative analysis of FAMEs resulted from retention time comparison of the FAME mixture (range between C4 to C24). We analyzed 85 specific biomarkers out of 98 identified PLFAs. Each biomarker was attributed to a specific taxon such as arbuscular mycorrhiza, other fungi, actinobacteria, gram-negative bacteria, gram-positive bacteria, in accordance with Frostegård et al. (1993). Bacterial biomass was calculated as the sum of actynobacteria, gram-negative and gram-positive bacteria biomass. Total microbial biomass was calculated as the sum of fungal and bacterial biomass. The details of the attribution of each peak to the taxa are presented in Supplementary Appendix S2.

The activity of luorescein diacetate hydrolase (FDAse) was measured according to Green et al. (2006) in order to determine the catabolic potential of microbial communities. FDAse is a complex of enzymes comprising phosphatase, cellulase, and lipase involved in the decomposition of cellulose and carbohydrates (Guénon et al., 2017).

In May 2021, in situ soil CO₂ effluxes (g.CO₂.m^-1.h^-1) were recorded using a portable gas analyser (EGM-4, PP-system) after removal of aboveground vegetation. The soil CO₂ effluxes represent the respiration of plant roots, soil organisms and chemical oxidation of C compounds. The device was connected to a closed soil respiration chamber (SRC-1, PP systems Massachusetts, United States). To prevent leakage of CO₂ when placing the chamber on the soil, a PVC tube (10 cm × 11 cm) was buried 1 cm deep into the soil prior to measurements. In each plot, soil temperature was recorded in situ in a depth of 7 cm using the soil temperature probe (STP-1, PP-system) connected to the portable gas analyzer.

Climatic variables were extracted from WorldClim version 2.1 dataset using the period from 1970 to 2000 (Fick and Hijmans, 2017). The 19 climatic variables were extracted on a grid of 1 km² and represented annual means of temperature and precipitation, seasonality, and extreme values resulting from MODIS images of the same period.

All data were analysed using R software (3.6.1, R core Team, 2020). Relationships between plant community composition, bioclimatic variables, plot position (under vs. outside panels) and one-way interaction of plot position and climatic variables were analysed using Redundancy analysis (RDA) in the R package “vegan” (Oksanen et al., 2007). Prior to RDA, multicollinearity of the 19 WorldClim explanatory variables and of their interactions with panel was tested by calculating the variance inflation factor (VIF) and using a cut value of 3. VIF retained four of them as non-collinear (Precipitation of Wettest Month, Annual Temperature Range, Mean Temperature of Wettest Quarter, Temperature Seasonality). A first RDA model tested the effect of panel and these four climatic variables on plant community composition including panel-climate interactions. Starting with an initial model of explanatory variables and interactions not showing collinearity, a backward selection procedure was applied using the “adespatial” package. Significance of explanatory variables was tested using PERMANOVA. Since most climatic variables showed significant interactions with plot position, separate RDA were additionally run for plots under and outside panels to explore the influence of climatic gradients.

To analyze the effect of solar panels on plant community characteristics, Shannon index, Simpson index and evenness were calculated. For each species, ecological indicator values were obtained using Julve (1998) adapted from Ellenberg et al. (1992). We separated species according to light indicator values in two groups: shade tolerant (1000–5,000 lux) and heliophilous species (>5,000 lux). We tested the habitat preference of each species in order to evaluate whether solar panels change grassland communities to shade-tolerant understory vegetation. The following response variables were tested: diversity indices, total vegetation cover, cover of shade-tolerant and heliophilous plants, chlorophyll index, mite and springtail abundance, biomass of microbes (AM fungi, Total Fungi, total Bacteria, Actynobacteria, Gram −, Gram +), FDase activity, soil respiration, soil temperature and moisture, total carbon and total nitrogen N. Linear mixed-effect models (LMMs) were calculated including position (under vs. between vs. outside panels) as fixed factor and SP as random factor. A Tukey post-hoc test was run to test for differences between the three positions when a significant effect of solar panel was detected. When necessary, data were transformed using the “bestNormalize” package (Peterson, 2021) to meet the assumptions of normality and homoscedasticity of variances.

To evaluate the influence of vegetation cover and main climatic variables on soil food web and soil CO₂ effluxes (a key soil function), a path analysis was performed using R package “lavaan” (Rosseel, 2012). The analysis was run separately for measurements outside, between and under solar panels. Path analysis is a structural equation modelling (SEM) technique used to show causal relationships between several measured variables. To build a full model (Supplementary Appendix S3), we assumed a causal relationship between climatic variables (i.e., annual mean temperature and precipitation) and vegetation cover on total carbon and total nitrogen content, total bacterial and fungal biomass, abundance of springtails and mites, and soil respiration. The full models were simplified by progressively excluding insignificant variables until an appropriate minimum model was obtained. The fit of each model was evaluated by the significance of differences between the predicted and observed covariance matrices (χ²-squared tests, p > 0.05), by the mean square root error of the approximation index (RMSEA <0.1), by comparative fit index (CFI >0.90) and the Tucker–Lewis index (TLI >0.90) (Xia and Yang, 2019).

Solar panels significantly affected plant community composition and the interactions of panels with annual temperature range (P:TAR) and temperature of the wettest month (P:TW) (Table 3; Figure 2A). Solar panels and interactions constrained the first axis of the RDA. The second RDA axis was constrained by temperature seasonality (TS) and precipitation of the wettest month (PW). Plant communities under solar panels showed less variation on the first RDA axis than those outside solar panels. Vegetation under solar panels was characterized by Anisantha sterilis and outside solar panels by Lotus dorycnium.

Outside solar panels (Figure 2B), temperature of the warmest month (TW), temperature seasonality (TS) and precipitation of the wettest month constrained the first axis of the RDA, while annual temperature range (TAR) constrained the second one (Figure 2B). The solar parks 1 and 6 were negatively correlated to precipitation of the wettest month and temperature seasonality. And they were characterized by L. dorycnium. Solar parks 2 to 5 were positively correlated to PW and negatively to TAR. Solar parks 8 to 10 were positively correlated to TW and TS and were characterized by Cynodon dactylon and Bromus rubens.

Under solar panels, TS and PW constrained the first axis, while TAR and TW constrained the second axis (Figure 2C). Solar panels 1–5 characterized by A. sterilis, grouped together at low precipitation of the wettest month and low temperature seasonality. The solar parks 7 to 10 characterized by L. dorycnium also grouped together but at the left part of the biplot characterized by higher precipitation of the wettest month and higher temperature seasonality. The solar park 6 was apart from the others occurring in the upper part of the biplot that was related to lower temperatures of the warmest month and characterized by Erigeron sumatrensis/canadensis and Argyrolobium zanonii.

Plant communities of solar parks were dominated by ruderal and grassland species. Herbaceous species cover outside and under solar panels was 70 and 10 times higher, respectively, than forest type species cover (Table 2; Supplementary Appendix S5). Two non-native species (E. sumatrensis/canadensis, Senecio inaequidens) were recorded in 55 plots. Their cover represented on average 2% of total vegetation cover (Table 2; Supplementary Appendix S5).

The species richness (Figure 3A) and total vegetation cover (Figure 3B) were significantly lower under solar panels than outside panels (p < 0.05). The chlorophyll index (Figure 3C; Supplementary Appendix 6) increased under solar panels. The cover of shade-tolerant species (Figure 3D) was twice as high and the cover of heliophilous species (Figure 3E) 2.5 times lower (p < 0.001) under solar panels than outside. The cover of non-native species was four times higher (p < 0.001) under than outside and between solar panels (Figure 3F). Shannon index, Simpson index and evenness were significantly higher outside than under solar panels (Supplementary Appendix S6).

Total mite abundance was higher (p < 0.001) outside than under or between solar panels (0.5 and 1.5 times, respectively, Figure 4A; Supplementary Appendix S6). Springtails were 1.5 times more abundant (p < 0.001) outside than under solar panels (Figure 4B; Supplementary Appendix S6). Soil respiration was 5.5 times lower under than between and outside solar panels (Figure 4L, p < 0.001). Fungi and gram-negative bacteria biomass were significantly reduced by 25% under solar panels compared with outside (Figures 4C, G), Supplementary Appendix S6). Total biomass of microorganisms was 1.25 times higher outside than under solar panels (Supplementary Appendix S6; Figure 4J). FDAse activity was 20% lower under than between and outside solar panels (Supplementary Appendix S6; Figure 4K).

Solar panels further decreased soil temperature, total carbon and total nitrogen contents but did not significantly change soil water content and C:N (ratio (Table 3; Supplementary Appendix S6). Between solar panels, the soil temperature was about 30% higher than under solar panels Table 4.

No significant differences were found between the different modalities for AM fungi, Bacterial biomass, Actinobacteria, Gram positives biomass, Fungi:Bacteria and Microbial biomass (respectively Figures 4D–F, H, I).

Outside panels, the most parsimonious model for predicting soil bacterial biomass, springtail abundance and CO₂ effluxes included precipitation, temperature, plant cover and total C contents as significant explanatory variables. The model was characterized by ten significant relationships. Plant cover, the abundance of soil organisms and CO₂ effluxes were mainly controlled by mean annual temperature and precipitation. Annual precipitation had an indirect effect on CO₂ effluxes via the influence on bacterial biomass (−), vegetation cover (+), and soil organic carbon concentrations (+) (Figure 5A). An increase in vegetation cover and then total C content led to an increase in springtail abundance.

Between solar panels, the most parsimonious model for predicting soil bacterial biomass, abundance of mesofauna organisms and CO₂ effluxes included precipitation, plant cover and C and N contents as significant explicative variables (Figure 5B). The model is characterized by ten significant relationships. Plant cover, abundance of soil organisms and CO₂ effluxes were mainly controlled by mean annual precipitation. Collembola and mites were affected by organic carbon (+) and nitrogen content (−) and springtails additionally by vegetation cover (+). CO₂ effluxes increased with fungi biomass increase and bacterial biomass decrease.

Under panels, the most parsimonious model for predicting soil bacterial biomass and abundance of mesofauna organisms included precipitation, temperature, plant cover and C and N contents as significant explicative variables (Figure 5C). The model is characterized by seven significant relations. Annual mean temperature and precipitation controlled microbial biomass (−) only. Vegetation cover directly controlled (−) mite abundance and indirectly (+) springtail abundance by an increase in organic carbon. Although bacterial biomass decreased with increasing precipitation, CO₂ effluxes under panels were explained by climatic, chemical, and biological variables. Similarly to the SEM outside and between panel, no significant interaction between bacterial biomass and mesofauna abundance was detected.

In the studied solar parks, plant communities outside panels were distributed along a climatic gradient. Through grading and clearing of vegetation, both soil quality and plant community composition change during the construction of solar park (Armstrong et al., 2016; Lambert et al., 2021). Plant establishment and spatial distribution of communities after a disturbance are driven by local climatic and soil physico-chemical parameters (Zarzycki and Bedla, 2017; Boonman et al., 2021). Chauvier et al. (2021) showed that artificial land cover (i.e., urban and agricultural area) clearly affects plant distribution along a climatic gradient in lower parts of the European Alps. Our results showed that climatic variables such as precipitation of the wettest month and temperature annual range still drive plant community composition of solar parks suggesting that solar parks do not totally disintegrate the relationship between climate and vegetation.

Outside the panels, vegetation was shaped by a climatic gradient, in particular by variations in temperature and precipitation. More precisely, we found that plant community composition was driven by the mean temperature of wettest quarter, temperature seasonality, annual temperature range and the precipitation of the wettest month. Our results suggest a progressive decrease in the sensitivity of plant species towards lower latitude to an increase in temperature variability and in the range of extreme temperature and precipitation conditions characteristic for the more continental climate of our inland sites (Maestre et al., 2009). The solar panels affected plant community composition and formed two distinct communities shaped by temperature. Under panels, mean temperature of the wettest month and annual temperature range had a smaller effect on plant communities than outside panels. Plant composition of grasslands is often driven by climate-related water availability (Cherwin and Knapp, 2012; Cleland et al., 2013). Armstrong et al. (2016) reported a buffer effect of panels on climatic variations leading to a reduction of ecological niche amplitude and a decrease in plant richness under a temperate-oceanic climate. We also observed a decrease in plant species richness under solar panels. In semi-natural dry grasslands, plants are often more affected by the lack of precipitation during the wettest months in autumn and winter (Dostalek and Frantik, 2011; Fischer et al., 2014). Rainfall during these periods is essential for seed germination and provides a soil water stock for the dry summer. Reduced rainfall under the panels may have limited plant establishment (Akinci, 2013; Lambert et al., 2022).

Contrary to Uldrijan et al. (2022) who showed that the cover of heliophilous plants was greater under the panels in the Czech Republic, we found that the light reduction under panels resulted in a shift from heliophilous to shade-tolerant plant species that are more competitive under such conditions (Chen et al., 2004). This reduction of light also resulted in an increase in chlorophyll content, confirming the results of Lambert et al. (2022) who found a higher chlorophyll content in leaves of Brachypodium retusum under panels. Solar panels change the metabolism of the plants that need to allocate more resources to the production of chlorophyll in order to maintain a sufficient photosynthetic activity under panels (Ma et al., 2010). This higher resource allocation in chlorophyll production may have resulted in a reduction of leave, root and flower biomass. Total vegetation cover and species richness were smaller under solar panels which may be a consequence such a reduction in plant performance. These changes in plant performance and plant community composition may affect ecological functioning. Uldrijan et al. (2022) showed that plant species under solar panels had fewer interactions with other taxa than species than outside panels. Such changes in the number of interactions may negatively affect ecosystem services provided by natural grasslands such as carbon sequestration (Uldrijan et al., 2022).

During daytime, we measured a clear reduction of temperature by 4°C under solar panels confirming results of Moscatelli et al. (2022) and Lambert et al. (2021) under similar Mediterranean climate. However, continuous measurements showed a homogenizing effect of solar panels on temperature being lower during daytime but higher at night (Armstrong et al., 2016). Solar panels intercept rainfalls but reduce evapotranspiration (Lambert et al., 2021) resulting in non-significant panel effects on soil moisture. Accordingly, the area between panels (inter-rows) still affected by shading of panels but not protected from rainfall showed the highest soil moisture. Effects of panels on soil temperature and moisture may further depend on seasonal variations (Armstrong et al., 2016). Lower C and N contents were observed under panels resulting from a lower plant biomass production and subsequently a lower organic matter content.

We found that FDAse activity, soil respiration and microbial biomass were reduced under solar panels. Climate and organic matter content are the most important drivers of soil microbial biomass and activities (Craine et al., 2012). Fluctuations in climatic parameters such as warming or drought may have important consequences for microbial activities (Brinkman and Sombroek, 1996). Microbial biomass and activities are indicators of soil functioning because they are involved in the decomposition of soil organic matter and provisioning of nutrients controlling plant growth (Abrahão et al., 2022). The lower FDase activity may be related to the lower microbial biomass under panels, especially of gram-negative bacteria. Gram-negative bacteria are particularly sensitive to drought and lower temperatures. Bacteria interact with the soil mesofauna to recycle nutrients and sequester carbon (Basile-Doelsch et al., 2020), a key ecosystem service provided by semi-natural grasslands (Bengtsson et al., 2019). In our study, the lower abundance of mites and springtails under panels may be explained by the reduction of soil temperature but also by a lower plant biomass. Mesofaunal soil communities are particularly sensitive to temperature changes. Aupic-Samain et al. (2018, 2021) showed that abundance and diversity of the mesofauna organisms were lower in systems with lower temperatures. Moreover, the lower content in soil organic carbon under panels limits the growth of such microbial and mesofaunal decomposers.

Path analysis revealed either positive indirect (via organic carbon content) or negative direct effects of vegetation cover on soil mesofauna. An increase in plant cover increases the soil organic matter content, a food source for springtails that contribute to the microfragmentation of leaf litter (Berg and Laskowski, 2005). The negative direct effects of plant cover may be explained by a lower plant diversity in most productive plant communities. Zhang et al. (2022) reported in a metanalysis that soil fauna is most abundant and diverse in plant mixtures of high diversity through an increase in microhabitat complexity. Solar panels decreased plant diversity in our study and thus microhabitat diversity. In contrast, panels did not affect the positive indirect effect of vegetation cover on the abundance of mesofauna organisms. This result suggests that changes in microclimate under panels changed resource availability for such decomposers (Aupic-Samain et al., 2021).

In this study, we did not observe an effect of plant cover on microbial biomass and related interactions along the climatic along the climate gradient, neither outside nor under panels. However, plant litter is known to supply nutrients to soil microbes, and root exudates and rhizodeposition attract bacteria, fungi and nematodes (Chakraborty et al., 2012) suggesting that microbial communities are more sensitive to changes in climate than in plant cover. Changes in mean annual temperature and precipitation were the main drivers of bacterial biomass. Drenovsky et al. (2010) showed under a Mediterranean climate that land-use type and level of soil disturbance (i.e., tillage and irrigation regimes) are more important factors for the composition and biomass of the microbial community than vegetation.

We found more trophic interactions between panels compared to outside and under panels suggesting a protection effect of panels between rows on the soil food web. In the gap between panels, wind, albedo, soil temperature and solar radiation are generally lower than outside panels (Armstrong et al., 2014). Under panels, the effect of mean annual temperature and precipitation on trophic interactions diminished. The result may be explained by the homogenizing effect of panels on climate reducing maximum temperatures during daytime and summer and increasing minimum temperatures at night and during winter (Armstrong et al., 2016). Changes in precipitation and temperature are likely resulting in changes in the soil biodiversity and can alter soil biological processes and functions with potential consequences for ecosystem services (Pritchard, 2011; Nielsen and Ball, 2015). Soil CO₂ emission is an indicator of soil organic matter decomposition which integrates several biotic and abiotic components. CO₂ effluxes mainly depend on soil heterotrophic organisms (bacteria, fungi and fauna), plant roots (Bond-Lamberty et al., 2004; Moinet et al., 2019), soil temperature and moisture (Moinet et al., 2019). By changing plant communities, soil organisms and microclimate, the panels reduced direct and indirect effects of organisms in lower trophic levels, of mean annual temperature and of precipitation on CO₂ effluxes. Solar panels may thus limit ecosystem services such as carbon sequestration and nutrient cycling.