Forage Quality Improves but Ecosystem Multifunctionality Declines Under Drought and Frequent Cutting in Dry Grassland Mesocosms

Abstract

Dry grasslands are vast, socioeconomically and ecologically important environments, which are increasingly threatened by multiple stressors. We tested whether plant cover composition could mitigate ecosystem services loss under multiple stressors in dry grassland mesocosms by growing the grass sorghum (Sorghum bicolor) alone (Grass cover) or together with the legume serradella (Ornithopus sativus) (Mixed cover) under frequent cutting and/or increasing water stress. We assessed erosion control, carbon sequestration, forage quantity and quality, and soil fertility, individually and simultaneously (i.e., multifunctionality). Contrary to our hypothesis, the Mixed cover did not improve ecosystem services compared to the Grass cover, except for forage quality, which improved by 30%. In general, the stressors had negative effects: cutting reduced erosion control by 20%, forage quantity by 50%, soil fertility by 40% and multifunctionality by 20%, and severe water stress decreased carbon sequestration by 40%, forage quantity by 30%, soil fertility by 10%, and multifunctionality by 10%. Water stress caused 100% serradella mortality, underscoring this legume’s vulnerability to increasing aridity. Combined stressors yielded the lowest service provision. Forage quality was the only service that improved under stress: cutting improved it by 40% and severe water stress by 60%. Our results suggest that while systems combining grasses and legumes may enhance forage quality, grass-dominated systems appear more resilient to multiple stressors in drylands, largely due to their superior efficiency in accessing and conserving limited water and nutrient resources. Given the ongoing trends of aridification and land-use intensification, future research should explore adaptive management strategies that prioritize resource-efficient plant species, foster belowground resource retention, and optimize grazing regimes to sustain resilience and multifunctionality in dry grasslands.

1. Introduction

Global drylands, including dry sub-humid, semi-arid, arid and hyper-arid areas, cover approximately 41% of the Earth’s land surface [1]. These regions are home to 38% of the global human population and support 20% of global plant biodiversity and 30% of bird biodiversity hotspots [2]. Further, drylands may store up to 45% of global terrestrial carbon (C) stocks [3], highlighting their importance in climate change mitigation. However, between 1982 and 2015, increasing aridity led to the degradation of 12.6% (5.43 million km²) of global drylands, contributing to desertification and affecting the livelihoods of 213 million people [4]. Alarmingly, projections indicate that, globally, drylands may expand by 11–23% by the end of this century [5]. Despite their ecological, socioeconomic, and climatic importance [6], drylands are among the most vulnerable biomes to climate change and ecosystem degradation [7].

Within drylands, grasslands represent the most widespread land cover type [1], supporting roughly 50% of the world’s livestock production [8]. Yet, vegetation in these regions is undergoing substantial change, including the decline or extinction of numerous forage and pasture species due to climate shifts, land-use changes (ranging from land conversion to forestry and crops, to land abandonment) [4,9] and overgrazing [5,10] among other drivers [11]. For example, by comparing eight dry grassland sites of the United States Long-Term Ecological Research Network, Hudson et al. [12] showed that multiple years of drought had larger negative effects than single years on primary production. Furthermore, they showed that droughts, floods, and wildfires altered resource availability and restructured plant communities, with greater impacts on primary production than warming alone. In Australian drylands, overgrazing has been shown to reduce biodiversity by 10%, plant productivity by 40%, and supporting ecosystem services by 20% [13]. And a standardized survey at 98 sites across six continents showed that increasing grazing pressure reduced ecosystem service delivery in warmer and species-poor drylands, while positive effects of grazing were observed in colder and species-rich areas [14]. While it is widely recognized that climate change and intensive land use each influence ecosystem functioning, the nature of their combined effects (whether mitigating, additive, or synergistically detrimental) remains unclear [15]. Despite the importance of grasslands, there has been little progress in finding solutions to halting and reversing grassland degradation, which is compromising sustainable development and the ecosystem services that grasslands provide [11]. Therefore, gaining a better understanding of how increasing aridity and overgrazing interact to impact ecosystem functions and service provision is essential for developing sustainable management strategies to prevent further degradation and desertification in drylands.

With growing evidence that the species influencing ecosystem functioning vary across different functions, local communities [16], and environmental conditions [17], the composition of grassland plant communities may be a key strategy for enhancing dryland resilience to global change. Dry grasslands are often constrained by low forage yield and/or poor forage quality due to suboptimal species composition. As a result, grasslands may be composed of native vegetation or sown with introduced species to improve productivity. For instance, sown biodiverse pastures, composed of mixtures of up to 20 self-reseeding cultivars of highly productive annual legumes and grasses, have been employed to boost pasture yield, combat soil erosion, and prevent land abandonment [18]. Compared to grass-dominated semi-natural pastures, these biodiverse mixtures can deliver higher forage yields of superior nutritional quality, allowing for increased sustainable stocking rates (i.e., more and higher-quality meat production), while also offering multiple environmental co-benefits [19]. However, their long-term persistence and sustainability may be jeopardized by ongoing climate change, particularly by reductions in precipitation [18].

To investigate how plant cover composition influences grassland functioning under multiple stressors (specifically increasing drought and/or overgrazing), we conducted a mesocosm experiment simulating grass-dominated and mixed (grass-legume) dry grasslands. We selected (i) sorghum (Sorghum bicolor (L.) Moench), a C4 grass known for its high photosynthetic efficiency, productivity, and broad tolerance to biotic and abiotic stresses [20], and (ii) serradella (Ornithopus sativus Brot.), a legume increasingly valued in drought-prone regions for its adaptability to nutrient-poor soils [21]. Given that both taxonomic and functional biodiversity can be strategically assembled to support complementary ecological roles and enhance ecosystem functioning [19], we hypothesized that the Mixed cover (i.e., sorghum together with serradella) would outperform the Grass cover (i.e., sorghum alone) in the provision of ecosystem services, particularly forage yield and quality, and soil fertility. According to the stress-gradient hypothesis [22], the frequency of facilitative and competitive interspecific interactions will vary inversely across abiotic stress gradients, with facilitation being more common in conditions of high abiotic stress [22,23] such as those tested in our study. However, despite the relative drought tolerance of both species, water limitation remains a major constraint, often reducing biomass and nutritional quality, especially in monocultures in arid and semi-arid systems [24,25]. Moreover, the documented drought sensitivity of certain sown legumes [18] may limit the advantage of Mixed cover under more extreme water deficits. We evaluated the effects of plant cover composition on a range of ecosystem services, including regulating (erosion control, C sequestration), provisioning (forage yield and quality), and supporting (soil fertility) services. Ecosystem services were assessed individually and simultaneously (i.e., multifunctionality). Multifunctionality corresponds to ‘the simultaneous provision of multiple functions’ or ‘the potential of landscapes to supply multiple benefits to society’ [26]. This approach allows us to observe the trade-offs between ecosystem functions or services, and understand how changes in factors (e.g., biodiversity and land management practices) affect multiple functions simultaneously [27].

2. Materials and Methods

2.1. Experimental Design

We used a fully factorial design incorporating the following three factors:

(i)

Plant cover composition: Mesocosms were either sown with sorghum (Sorghum bicolor (L.) Moench) alone (hereafter Grass), or with both sorghum and serradella (Ornithopus sativus Brot.) (hereafter Mixed).

(ii)

Cutting regime: Mesocosms were either subjected to frequent cutting (Cuts) or left uncut (No cuts). In the Cuts treatment, shoots were clipped 5 cm above the substrate surface every three weeks, starting when seedlings had developed at least two true leaves and were no longer reliant on their cotyledons [28].

(iii)

Water stress: Two irrigation regimes were used to simulate different stress levels—moderate water stress (150 mL per mesocosm, three times per week) and severe water stress (75 mL per mesocosm, three times per week).

This design resulted in eight combinations, each replicated five times, totaling 40 mesocosms. Each mesocosm consisted of a 250 cm² elliptical aluminum container (2.4 L volume; Monouso, Valencia, Spain), filled with 2.25 L of a soil-vermiculite (Projar, Lisboa, Portugal) mixture (1:2 v/v). The soil, collected in November 2022 from a natural pasture in Vaiamonte (Portugal), had a sandy loam texture, pH of 5.1 (measured in a 1:2.5 soil-water suspension), and a cation exchange capacity of 4.8 cmol kg⁻¹. Soil nutrient analysis revealed an organic matter concentration of 0.4% (following ISO standard 10694, using the loss-on-ignition method at 600 °C overnight in a Nabertherm L3/11/C6 muffle furnace, Cornellà de Llobregat, Spain), and the following concentrations of mineral nutrients: 105 ppm of nitrogen and 13 ppm of phosphorus. The soil was sieved through an 8 mm mesh prior to mixing.

Sorghum seeds were obtained from a local seed supplier (Casa das Sementes Augusto A. Dias Lda., Almada, Portugal), and serradella seeds were donated by Fertiprado (Vaiamonte, Portugal). Grass mesocosms were sown with 70 sorghum seeds, while Mixed mesocosms were sown with 35 sorghum seeds and 50 serradella seeds. Sowing densities reflected common agronomic practice, with sorghum sown at lower densities due to its greater height and competitive ability, and legumes sown at higher densities to optimize their functional contribution. In Grass mesocosms, sorghum seeds were evenly distributed across the container surface. In Mixed mesocosms, the two species were sown in spatially distinct halves, simulating a strip intercropping arrangement. To ensure successful germination and seedling establishment, all mesocosms were irrigated daily with 150 mL of tap water during the first three weeks. Both the Cutting and Water stress treatments were initiated three weeks after the beginning of the experiment (see Figure 1 for timing of treatment application) as follows:

• Cutting involved trimming the shoots to ~5 cm above the substrate surface using scissors every three weeks, for a total of three cuts throughout the experiment.
• Water stress was imposed by reducing irrigation to 75 mL per mesocosm, applied three times per week (severe stress), while control mesocosms continued to receive 150 mL but only three times per week (moderate stress).

The experiment was conducted over 10 weeks, from April to June 2023, in a greenhouse at the Faculdade de Ciências da Universidade de Lisboa. The greenhouse provided natural light (approximately 15 h day/9 h night), with photosynthetically active radiation ranging between 600 and 1000 μmol m⁻² s⁻¹, and ambient temperatures fluctuating between 9.5 °C and 44.5 °C (recorded by Govee sensors, Hong Kong; see Figure 1 for daily mean values). Mesocosms were randomized weekly within the greenhouse (Figure S1).

2.2. Harvest and Analysis

At harvest, plants were classified as either living (i.e., healthy or senescing individuals with green leaves) or dead (i.e., completely dry individuals), and counted accordingly. When applicable, sorghum and serradella individuals were separated by species and into shoots and roots, then dried at ambient temperature to a constant mass. Root, shoot (including harvested cuttings, when applicable), and total biomass were subsequently determined. Dried shoot tissues (including harvested cuttings, when applicable) were ground into a fine powder using a ball mill (Retsch MM 4000, Vila Nova de Gaia, Portugal) and analyzed for carbon (C) and nitrogen (N) concentrations using an elemental analyzer (EuroVector, Pavia, Italy) coupled with a mass spectrometer [29].

From each mesocosm, two soil samples were collected and combined into a composite sample. In Mixed mesocosms, one soil sample was taken from the sorghum half and the other from the serradella half. All soil samples were dried at 60 °C to constant mass and analyzed for soil organic matter (SOM) concentration following ISO standard 10694, using the loss on ignition method at 600 °C overnight in a Nabertherm L3/11/C6 muffle furnace.

2.3. Ecosystem Services Provision

Ecosystem services provision was evaluated using two complementary approaches: (i) individual ecosystem services, assessed separately, and (ii) multifunctionality, calculated using the average approach [9,30], which aggregates all standardized ecosystem service indicators. To assess individual ecosystem services, all variables were standardized by dividing each value by the maximum observed value for that variable across all treatments, thereby constraining standardized values between 0 and 1 (refer to Supplementary Material for raw data). When multiple parameters were used to represent a single ecosystem service, the mean of the standardized values was calculated to generate a composite estimate between 0 and 1.

Potential ecosystem services provision was assessed as follows:

• Erosion control (regulating service) was estimated using total plant density (living + dead individuals) per mesocosm. This metric was chosen because vegetative cover, regardless of whether plants are alive or senescing, is known to significantly reduce erosion by intercepting raindrops, reducing runoff, and stabilizing soil structure [31]. We considered both sorghum and serradella densities as equally contributing to erosion control, based on their complementary morphological features, i.e., serradella has a compound leaf structure with alternate phyllotaxy, which increases canopy complexity and effectively dissipates rainfall and wind energy; and sorghum produces dense, deep root systems that enhance soil cohesion and resistance to detachment.
• C sequestration (regulating service), its biotic component, was estimated by combining plant survival rate and total plant biomass. This approach reflects the two key components of C capture in vegetation, i.e., since only living plants can assimilate atmospheric CO₂ through photosynthesis, we used plant survival; and given that approximately 40% of plant biomass is composed of C derived from atmospheric CO₂, we used total plant biomass. Survival rate was calculated as the percentage of living individuals at harvest relative to the total number of plants (Table S3). Total plant biomass was obtained by summing the aboveground and belowground dry biomass per mesocosm at the end of the experiment. The standardized values of survival rate and total biomass were averaged to derive the final C sequestration index for each mesocosm.
• Forage quantity (provisioning service) was estimated using plant shoot biomass.
• Forage quality (provisioning service) was estimated by integrating shoot N concentration and the CN ratio of the plant shoots. In Mixed mesocosms, shoot N concentration and CN ratio were calculated as weighted averages based on each species’ contribution to total aboveground biomass within each mesocosm. Because shoot N concentration and CN ratio provide opposite indications of forage quality (i.e., high N indicates higher quality, whereas a high CN ratio indicates lower quality), we transformed the CN ratio by taking its inverse (1/CN). The final forage quality score for each mesocosm was obtained by averaging the standardized values of shoot N concentration and 1/CN, ensuring both metrics contributed equally to the index.
• Soil fertility (supporting service) was estimated by combining soil organic matter (SOM) increment and root biomass. SOM increment was calculated by subtracting the initial SOM concentration of the substrate (0.13%) from the final SOM concentration measured in each mesocosm. Root biomass, as a key driver of belowground nutrient cycling and SOM formation, was included as a complementary indicator of fertility. Both SOM increment and root biomass values were standardized (scaled from 0 to 1), and the soil fertility score for each mesocosm was computed by averaging these two standardized values.

Average multifunctionality was calculated as the mean of the standardized values of all individual ecosystem services. This approach provides a simple and interpretable metric that integrates multiple ecosystem functions into a single value, facilitating comparisons across treatments and systems [26,32].

2.4. Calculations and Statistics

The change in the potential provision of each ecosystem service, assessed both individually and as multifunctionality, in response to stressors was calculated separately for each plant cover composition. This was achieved by comparing the standardized values under each stressor treatment (cutting, water stress, or both) with those under control conditions (i.e., no cuts + moderate water stress), thereby quantifying the relative impact of stress on service provision within each cover type, as follows:

$$\mathrm{Change} (\%) = {\displaystyle \frac{(Parameter \mathrm{w}\mathrm{i}\mathrm{t}\mathrm{h} 1 or 2 stressors-Average parameter \mathrm{w}\mathrm{i}\mathrm{t}\mathrm{h} 0 stressors)}{Average parameter \mathrm{w}\mathrm{i}\mathrm{t}\mathrm{h} 0 \mathrm{s}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{o}\mathrm{r}\mathrm{s}}} x 100$$

Prior to analysis, data were checked for normality and homogeneity of variances, using the Shapiro–Wilk test and the Levene’s test, respectively. The effects of plant cover composition, cutting and water stress on ecosystem service provision (both individually and simultaneously, i.e., multifunctionality) were assessed using three-way analyses of variance (ANOVA), with plant cover composition, cutting, and water stress as fixed factors. Summary statistics for individual ecosystem services were compared between Grass and Mixed mesocosms using two-sided t-tests (p < 0.05). Multifunctionality was further analyzed using a two-way ANOVA with plant cover composition and treatments (i.e., the four combinations of cutting and water stress) as fixed factors. Differences between treatments and between the number of stressors were evaluated using least significant difference (LSD) post hoc tests (p < 0.05). All statistical analyses were conducted using SPSS Statistics software (version 29.0.0.0, IBM, Inc., Chicago, IL, USA).

3. Results

3.1. Effect of the Three Factors on Ecosystem Services Provision

Potential ecosystem service provision was influenced by at least one of the three tested factors (plant cover composition, cutting, and water stress) and, in some cases, by interactions among them (Figure 2; Table S1). Specifically:

• Erosion control was influenced by cutting, with cut mesocosms exhibiting a 20% reduction in erosion control compared to uncut (no cuts) treatments.
• C sequestration was influenced by plant cover composition, water stress, and the interaction between cover composition and cutting. In Mixed cover mesocosms and those under severe water stress, C sequestration was reduced by 20% and 40% relative to Grass cover and to moderate water stress, respectively.
• Forage quantity was influenced by cutting, water stress, and their interaction, as well as a three-way interaction among plant cover composition, cutting, and water stress. In mesocosms where shoots were cut and those with severe water stress, forage quantity declined by 50% and 30% relative to uncut and those with moderate water stress, respectively.
• Forage quality was influenced by all three factors individually. In Mixed mesocosms, those under cuts and severe water stress, forage quality increased by 30%, 40%, and 60% relative to Grass cover, no cuts, and moderate water stress, respectively.
• Soil fertility was influenced by cutting and water stress. When shoots were cut and when water stress was severe, soil fertility decreased by 40% and 10% compared to no cuts and moderate water stress, respectively.
• Multifunctionality was influenced by cutting and water stress. Under cuts and under severe water stress, multifunctionality was reduced by 20% and 10% compared to no cuts and moderate water stress, respectively.

When significant, plant cover composition had contrasting effects: it enhanced forage quality but reduced C sequestration. In contrast, both cuts and severe water stress generally had negative impacts across most ecosystem services, except for forage quality, which was improved (Figure 2).

3.2. Effect of Multiple Stressors on Ecosystem Services Provision

Ecosystem service provision varied considerably across combinations of plant cover composition, cutting, and water stress treatments (Figure 3), with the exception of the cuts + severe water stress condition, which consistently resulted in the lowest provision of ecosystem services (Figure 3d). Comparative analysis of plant cover compositions revealed that: (i) the Mixed cover only enhanced erosion control in the absence of stressors (no cuts + moderate water stress; Figure 3a), and (ii) the Grass cover improved C sequestration under single-stressor conditions (cuts + moderate water stress and no cuts + severe water stress; Figure 3b,c), and improved forage quantity and soil fertility under cuts + moderate water stress (Figure 3b).

Given that both cutting and water stress negatively influenced individual ecosystem services (Figure 2 and Figure 3), it is not surprising that multifunctionality was lowest under combined stressors (cuts + severe water stress) and highest in the absence of stressors (no cuts + moderate water stress: Figure 4a). Despite having distinct effects on individual services, exposure to a single stressor (either cuts or severe water stress) resulted in similar negative impacts on overall multifunctionality. Finally, plant cover composition had no effect on multifunctionality (Figure 2p and Figure 4a) or on changes in ecosystem service provision, whether assessed individually (Figure S2) or simultaneously (Figure 4b).

4. Discussion

Contrary to our initial hypothesis, the Mixed cover (i.e., sorghum together with serradella) did not mitigate the loss of ecosystem services under multiple stressors. This outcome is likely explained by the greater drought sensitivity of serradella compared to sorghum, which reduced its capacity to cope with scarce water resources. In addition, competition for water and nutrients likely exacerbated the legume’s disadvantage, further limiting its contribution to ecosystem service provision. These findings highlight that resource availability and species interactions are critical determinants of species performance and the resilience of ecosystem services in drylands.

4.1. Effect of Plant Cover Composition

We anticipated significant effects of plant cover composition on ecosystem service provision. Specifically, we expected the Grass cover to perform better in terms of C sequestration, forage quantity, and soil fertility, given sorghum’s known drought tolerance [33,34] and higher biomass production [35,36]. Conversely, we expected the Mixed cover to enhance erosion control, forage quality, and multifunctionality due to its 17% higher sowing density, potential nutritional benefits of legume inclusion [37], and the positive association between species richness, functional diversity, and multifunctionality [7,38,39].

Our results partially confirmed these expectations. Sorghum was indeed more drought-tolerant (Table S3) and produced more biomass than serradella (Table S4), leading to significantly higher C sequestration in Grass mesocosms (Figure 2). Meanwhile, forage quality was higher in the Mixed cover than in the Grass cover (Table S5), likely due to the inclusion of the high-protein legume.

However, plant cover composition did not influence the following services:

• Erosion control. Although Mixed mesocosms were sown at higher density, final plant densities were similar across treatments (Table S2), likely reflecting a saturation point in mesocosm capacity rather than treatment differences.
• Forage quantity. Despite similar final plant densities (Table S2) and species proportions, the expected advantage of the Grass cover in forage quantity was not observed. Surprisingly, sorghum biomass per plant was consistently higher in the Mixed cover, particularly when serradella survival was low (Table S3). Under no cuts + moderate water stress, sorghum shoot biomass increased by 86% (from 188 mg to 350 mg plant⁻¹) in Mixed versus Grass mesocosms. Smaller yet consistent increases were observed across the other treatments: 24% (cuts + moderate water stress), 37% (cuts+ severe water stress), and 56% (no cuts + severe water stress). These results suggest several non-exclusive facilitative or compensatory mechanisms: (i) reduced competition for water, nutrients, and space as serradella individuals died; (ii) decomposition of serradella, particularly its roots, releasing readily available nutrients; (iii) biological nitrogen fixation by serradella improving nutrient availability [40]; and (iv) resource sharing via common mycorrhizal networks between co-occurring species [41], potentially enhancing sorghum growth.
• Soil fertility. Although sorghum produced more root biomass than serradella (Table S4), its root biomass was also higher in the Mixed cover, compensating for the lower density of sorghum plants. Given the similar increments in soil organic matter (Table S6), plant cover composition did not influence soil fertility.
• Multifunctionality. The absence of a clear advantage in any consistent set of services, coupled with trade-offs between them, explains why plant cover composition had no effect on multifunctionality.

4.2. Effect of Frequent Cutting

Except for erosion control (no effect on plant density was expected due to cutting being implemented only three weeks after sowing) and forage quality (for which we anticipated a positive effect on quality [42,43]), we hypothesized negative impacts of cutting on C sequestration, forage quantity [44], and soil fertility [45]. These expectations were based on the anticipated reductions in plant survival [46] and biomass accumulation [47] caused by cutting. Our results were largely in agreement with this hypothesis. Cutting led to a 20% reduction in multifunctionality, primarily due to its negative effects on most ecosystem services, consistent with patterns observed under heavy grazing regimes [48]. This suggests that the implemented cutting regime functionally mimicked overgrazing. By contrast, moderate or low grazing intensities have been shown to enhance multifunctionality [48]. However, some outcomes deviated from expectations:

• Erosion control was unexpectedly reduced (by 20%) due to cutting, a result aligned with reported negative effects of grazing on plant density [49] and on plant cover [50] in dryland ecosystems. Although our erosion control metric included both living and dead individuals, some plants that died following the first or second cut may have deteriorated to the point of no longer contributing effectively to soil protection at the end of the experiment.
• No effect on C sequestration was observed, despite the reduction in biomass (Table S4), which would typically imply lower C uptake. This neutral effect likely reflects a trade-off between reduced C assimilation due to biomass loss and a gain in plant survival, particularly for serradella (Table S3). Cutting alleviated drought stress by reducing evapotranspiration surface area, thereby enhancing survival [28]. Such positive impacts of grazing on plant survival under drought have been previously documented [51,52].

Interestingly, and in line with our hypothesis, forage quality increased following cutting, likely due to the regrowth of younger, more nutrient-rich leaves, which are more photosynthetically active and efficient in water and nutrient use [53]. This improvement in forage quality (Figure 2; Table S1) supports findings that selective biomass removal can enhance the nutritional profile of regrown forage [42,43].

4.3. Effect of Water Stress

Due to the sharp rise in ambient temperature, with mean daily values exceeding 24 °C during the final three weeks of the experiment (Figure 1), even the mesocosms receiving more water experienced moderate drought stress. This was evident from the complete mortality of serradella and 25% mortality of sorghum in the no cuts + moderate water stress treatment (Table S3). Under these stressful conditions, further reductions in water input (severe water stress treatments) intensified drought severity, and serradella only survived when cutting reduced its evapotranspiration surface (cuts + severe water stress). Therefore, it is important to note that all plants were exposed to some level of drought.

With the exception of erosion control (for which we anticipated no change, since water stress treatments began only after three weeks, and both living and dead plants were included in the density metric [31]) we hypothesized negative effects of severe water stress on all other ecosystem services. Specifically, we expected decreases in plant biomass [54] and survival [55], on soil fertility [45], and on multifunctionality [39]. However, the expected effect on forage quality was less clear, as the literature reports contradictory findings regarding drought impacts on shoot N concentration, with some studies reporting reductions [44] and others reporting increases [56]. Our results largely corroborated these expectations: water stress had no effect on erosion control, but reduced C sequestration, forage quantity, soil fertility, and multifunctionality (Table S1; Figure 2). Interestingly, forage quality increased under severe water stress, which aligns with the findings of Dumont et al. [57], who reported a 5% increase in forage N concentration under drought conditions in a global meta-analysis of grassland responses to climate change.

Altogether, watering was the only experimental factor for which our hypotheses were fully supported. These findings suggest that increasing aridification and more frequent drought events, as predicted by many climate change scenarios for dryland ecosystems [58], will likely lead to a decline in most ecosystem services, with the notable exception of forage quality, which may improve.

4.4. Effect of Multiple Stressors

In a rapidly changing world, improving our understanding of the cumulative ecological impacts of multiple stressors is essential for effective biodiversity conservation and ecosystem management [59]. Reflecting this urgency, a growing number of studies have quantified ecological responses to multiple stressors, consistently showing that the resistance of terrestrial ecosystem services declines as the number of stressors increases, both under field [60] and controlled microcosm conditions [61]. In line with these findings, our mesocosm experiment demonstrated that increasing the number of stressors reduced ecosystem service provision, whether assessed individually (Figure 3 and Figure S2) or simultaneously through multifunctionality (Figure 4).

Interestingly, plant cover composition did not influence ecosystem service provision when mesocosms were subjected to multiple stressors (Figure 3 and Figure 4). This result is consistent with the global synthesis by Zhou et al. [60], which analyzed over 14,000 observations and found that the combined effects of drought and grazing were more detrimental than either stressor alone. In our experiment, the legume serradella was more affected by water stress (as indicated by its low survival rate even with moderate water stress; Table S3), whereas sorghum appeared to be more impacted by cutting (as reflected in biomass reduction; Table S4). This species-specific sensitivity to different stressors helps explain why neither plant cover composition (Grass or Mixed) was able to mitigate the combined effects of both stressors (Figure 3 and Figure 4).

Finally, it is important to recognize that all mesocosms experienced additional background stressors, regardless of the cutting or watering treatments. These included (i) low soil fertility, as evidenced by a starting SOM of only 0.13%; (ii) water stress, demonstrated by the low plant survival even under moderate water stress (Table S3); and (iii) space limitation, inferred from the lack of differences in plant density between Grass and Mixed covers (Table S2), despite differences in initial sowing density. Thus, cutting and water stress should be interpreted as additional stressors imposed on an already stressed system, rather than the sole sources of disturbance. This underscores the complexity of stressor interactions in real-world dryland ecosystems and highlights the limited buffering capacity of plant functional diversity when multiple abiotic and biotic stressors co-occur.

5. Conclusions

Our mesocosm experiment examined how plant cover composition, frequent cutting, and water stress may affect ecosystem services provision in dry grasslands. Contrary to our hypothesis, the Mixed cover (sorghum + serradella) was unable to mitigate the negative effects of multiple stressors due to the legume’s higher drought sensitivity. Frequent cutting, simulating overgrazing, reduced overall ecosystem functioning, though it slightly improved forage quality. Water stress had the most consistent negative impact on plant survival, biomass, and ecosystem services, except for an increase in forage quality. Combined stressors further reduced ecosystem service provision, and growing sorghum together with serradella failed to buffer these effects due to species-specific vulnerabilities under environmental stress. Altogether, the effects of plant cover composition, frequent cutting and water stress on the dry grassland ecosystem services that we observed are within the ranges observed in other studies (

Table 1. Comparison between our study and other studies’ findings on the effects of plant cover composition, frequent cutting and water stress on dry grassland ecosystem services. ↑ and ↓ indicate an increase and a decrease on ecosystem services, respectively.

Ecosystem Service	This Study (% Change)	Other Studies (% Change)	References
Erosion control	−20% under frequent cutting	↓ by 15–30% due to grazing	[49,50]
C sequestration	−20% in Mixed vs. Grass −40% under severe water stress	Lower in legumes than in grasses ↓ by 30–50% due to drought	[33,34,54]
Forage quantity	−50% under frequent cutting −30% under severe water stress	↓ by 25–60% due to grazing/drought, depending on severity	[44,55]
Forage quality	+30% in Mixed vs. Grass cover +40% under frequent cutting +60% under severe water stress	↑ by 20–40% due to grazing ↑ by 5–15% due to drought	[42,43,57]
Soil fertility	−40% under frequent cutting −10% under severe water stress	↓ by 10–40% due to grazing and drought	[45,48]
Multifunctionality	−20% under frequent cutting −10% under severe water stress	↓ by 15–35% due to combined stressors	[39,60,61]

).

The short-term duration of this study and the potential influence of other environmental drivers may limit the generalization of our results, underscoring the need for long-term, field-based studies. From a resource perspective, our findings highlight that ecosystem service provision in dry grasslands is tightly constrained by water and nutrient availability, and that species-specific differences in resource acquisition strategies shape responses to multiple stressors. This reinforces global evidence that increasing environmental pressures exacerbate resource scarcity and, in turn, ecosystem service declines. Given the ongoing trends of aridification and land-use intensification, adaptive management strategies that prioritize resource-efficient species selection, improve soil resource retention, and optimize grazing regimes will be crucial to enhance resilience and multifunctionality in dryland ecosystems.