Agriculture, Ecosystems and Environment 356 (2023) 108601 Contents lists available at ScienceDirect Agriculture, Ecosystems and Environment journal homepage: www.elsevier.com/locate/agee Sown diversity effects on yield and resistance to weed invasion: Clues to improve mixture design under climatic change in the Mediterranean Àngela Ribas a, b, Alba Llovet a, b, c, *, Rosa Llurba c, d, John Connolly e, Maria-Teresa Sebastià c, d a BABVE, Edifici C, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain b CREAF, Cerdanyola del Vallés, 08193 Catalonia, Spain c Centre Tecnològic Forestal de Catalunya (CTFC), 25280 Solsona, Spain d Group GAMES & Dept HBJ, ETSEA, Universitat de Lleida, 25198 Lleida, Spain e School of Mathematics and Statistics, University College Dublin, Dublin 4, Ireland A R T I C L E I N F O A B S T R A C T Keywords: With the aim to improve mixture design, particularly in regions vulnerable to climate change, we tested several Agroecology forage communities following the biodiversity–ecosystem function (BEF) framework. We sowed monocultures Climate change adaptation and 4-species mixtures from a pool of 7 forage species in a sub-Mediterranean region (Eastern Pyrenees) and Diversity effects assessed the diversity effects on yield and resistance to weed invasion. The tested species included two grasses Forage persistence and five legumes with contrasting temporal patterns and different climatic amplitudes. The communities differed Mediterranean mountain systems Polycultures in their specific composition (mixture types) and the relative abundance of the components, following a simplex design, which allowed us to estimate separately the two components of the diversity effect: the individual species effects and that due to species interactions. Whereas monocultures performed in a highly variable way within and across harvests, both in relation to yield and weed suppression, mixture variability was narrower. Both functions increased in mixtures (with significant interaction effects between 24% and 57% for yield and 13% and 96% for weed suppression), especially in those mixtures including Mediterranean species, which showed the highest diversity effects that persisted over the three experimental years. Extreme climatic events during the experimental period might have affected not only the species’ individual performances but also the strength of species interactions. Both components of diversity, identities and interactions, were key in maintaining high performances. We conclude that, under the current climate change scenario, it is important to include species in mixtures that increase resistance or resilience not only at the species level but also at the community level, through enhanced interaction effects. 1. Introduction reduced nitrogen (N) losses through leaching (Hauggaard-Nielsen et al., 2003; Szumigalski and Van Acker, 2006) and emissions (Cummins et al., In the current global change scenario, agricultural systems have the 2021; Pappa et al., 2011; Ribas et al., 2015). Diversity may also be an challenge to increase their potential for mitigation and adaptation important adaptation strategy against climatic risks in an uncertain (IPCC, 2022) and to decrease their dependency on external inputs future, through enhanced resistance and resilience (Altieri et al., 2015; (Kirschenmann, 2007; Niggli et al., 2009). Increasing the diversity of Isbell et al., 2015; Lüscher et al., 2022; Volaire et al., 2014), after agroecosystems may be a tool to respond to these challenges (Brooker extreme climatic events (Grant et al., 2014; Haughey et al., 2018; Sunil et al., 2015; Haughey et al., 2018). Diverse grassland systems have et al., 2020) or under sub-optimal environmental growing conditions resulted in enhanced yield (Finn et al., 2013; Hooper et al., 2005; Tilman (Porqueddu et al., 2016; Psyllos et al., 2022). et al., 2002, 2014), lower fertilizer (Nyfeler et al., 2009; Suter et al., The changes in environmental patterns derived from the current 2021; Weisser et al., 2017) and agrochemical needs linked to reduced climate change may represent an important challenge in Mediterranean incidence of weeds, pests and diseases (Connolly et al., 2018; Van areas. These are expected to be exposed to increased frequency of Ruijven et al., 2003), and reduced environmental impact, including extreme temperatures, expanded drought periods and more irregular * Corresponding author at: BABVE, Edifici C, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain. E-mail address: Alba.Llovet@uab.cat (A. Llovet). https://doi.org/10.1016/j.agee.2023.108601 Available online 2 June 2023 0167-8809/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 precipitation patterns (IPCC, 2022), thus being especially vulnerable to Swiss pastoral landscapes). For simplicity, hereafter we refer to all seven climate change. Agricultural systems will have to overcome a multitude forages as species, even though the two T. pratense varieties correspond of new biotic and abiotic stressors. The multi-faceted and uncertain to the same species. All the species are commonly found in the Iberian challenge that climate change poses on agriculture requires other stra­ Peninsula as cultivated species or in natural meadows except for the tegies besides breeding. In this context, mixtures may allow the simul­ prostrate T. pratense, however, in the Pyrenees similar varieties can be taneous incorporation of several resistance traits at the community level found naturally in grasslands. The selected species are adapted to (Dooley et al., 2015; Ehrmann and Ritz, 2014). different climatic conditions and have different ecological/climatic In forage systems, the difficulty of establishment and lack of persis­ amplitudes and range from typical Mid-European forage species to the tence may be particularly important, affecting not only the stability of most typical Mediterranean forages (species climatic preferences are forage supply but also weed invasibility and thus forage quality. As both shown in Table 1 along with their establishment/persistence patterns establishment and persistence are expected to be highly affected by and biogeographic distribution). environmental conditions, under increased unpredictablility, the sus­ Medicago sativa and Onobrychis viciifolia can be classified as typical tainability of these systems in the Mediterranean will be exceptionally Mediterranean legumes regarding their climatic preferences and threatened (Porqueddu et al., 2016). adaptability, mainly due to their drought-tolerant character (see The use of a broader range of cultivated species or varieties to Table 1) and will be hereafter referred to as “Mediterranean legumes”. guarantee a minimum output under uncertain conditions has been Indeed, Medicago sativa is one of the most widely cultivated forage known since ancient times (Altieri, 1999; Altieri et al., 2015; Vander­ legume species in the area (40% of all forage monocultures in Catalonia; meer, 1989). This positive relationship between diversity and produc­ Statistics of the Spanish Ministry of Agriculture, 2012). It is grown in tivity is underpinned by two major components: the identity effect and irrigated and rainfed systems as a monoculture but also in combination the interaction effect. The former corresponds to individual species with grasses (Delgado, 1984). On the other hand, Onobrychis viciifolia is contributions and is essentially equivalent to the portfolio effect (i.e., usually grown above 600 m of altitude, substituting Medicago sativa in increase the probability that at least one of the included species will mountain rainfed areas (Delgado, 1984; Delgado et al., 2001, 2014), as a respond favourably to the given conditions) (Campbell et al., 2012; Díaz monoculture or in association with a grass, e.g., Dactylis glomerata and Cabido, 2001). The latter is related to species interactions owing to (Sebastià et al., 2011). Nevertheless, its use is decreasing in the Iberian complementarity mechanisms (niche differentiation and facilitation) Peninsula (Delgado et al., 2001; Pujol, 1984). and is defined as the increase in a given function observed on top of the expected value according to the proportional contribution of the com­ 2.2. Site description ponents (McIntire and Fajardo, 2014; Kirwan et al., 2007; Nyfeler et al., 2011). The experiment was carried out in the municipality of Gósol, in the The optimum composition and management of mixtures for a given Eastern Pyrenees, Catalonia (42◦ 13’N, 1◦ 39’E, 1410 m.a.s.l.). Gósol is agro-climatic system, or under unpredictable conditions, remain located in a valley surrounded by mountains on a relatively flat water­ important questions of mixture design. This gap of knowledge has shed, and the study site is placed in the montane altitudinal belt and has resulted in diverse forage systems still occupying a limited area (around a sub-Mediterranean climate. The soil is clay textured (22.4% sand; 10% in Catalonia; Statistics of the Spanish Ministry of Agriculture, 2012) 37.4% silt; 40.2% clay), it presents a high organic matter content while the use of forage monocultures has been generalized (Peeters (7.93%) and an alkaline pH of 8 (Kirwan et al., 2014); whereas et al., 2006). Nevertheless, recent advances in methodological ap­ extractable nutrients accounted for 26 mg kg− 1 Olsen phosphorus, 303 proaches, such as the one developed in the Agrodiversity experiment mg kg− 1 potassium (flame photometer determination) and 174 mg (Connolly et al., 2013; Kirwan et al., 2009), are setting the basis to kg− 1magnesium (measured by atomic absorption spectrophotometry). optimize the design of mixture composition, as they test different levels Natural vegetation includes deciduous forests with oaks and mixed ri­ of evenness, plant functional traits and sown density. Additionally, this parian forests by the river bank, but in many places the original forests approach allows to segregate between identity and interaction effects. have been substituted by Scots pine trees. The valley bottom is covered With the aim to develop forage mixtures best adapted to the current by forages and other crops while semi-natural grasslands occupy slopes climatic conditions in Mediterranean mountains, we manipulated sown and marginal sites. diversity using a range of species proportions in four-species mixtures, and repeated this design with five mixture types, differing in the specific 2.3. Climatic conditions forage species composition. We wanted to know: 1) if sown diversity provides advantages in terms of forage yield and resistance to weed According to models based on long-term data (1940–2005) for the invasion for a range of mixture types within a site; 2) if sown diversity study area, mean annual temperature (MAT) and mean annual precipi­ provides stability of forage supply and resistance to weeds in the context tation (MAP) are respectively 7.1ºC and 1126 mm, estimated from the of rainfed systems under experimental conditions; 3) if there is a mixture database of the Confederación Hidrográfica del Ebro (CHE). Neverthe­ type best adapted to the current changing conditions in these Mediter­ less, CHE data for the experimental period (2000–2005) estimates 8.0ºC ranean mountains in which the experiment was developed; and 4) the and 877.4 mm for MAT and MAP, respectively (9.2 ºC and 812.1 mm relative importance of identity and species interactions effects in the according to the closest meteorological station-Gisclareny), thus exem­ studied responses for the different tested mixtures in order to maintain plifying the trends of temperature increase and precipitation decrease or increase crop performance. for that region (which is indeed warming and drying at a faster rate than other areas in Catalonia according to Calbó et al., 2012) (see the Sup­ 2. Materials and methods plementary Material section 1.1: climatic conditions, for more detail). It is worth noting that during the experimental years, both temper­ 2.1. Selected forage species ature and precipitation suffered anomalies. According to the Gisclareny meteorological station, the monthly average of maximum (daily) tem­ In this study we used six forage species: two grasses (Lolium perenne peratures reached in 2003 values 3.9ºC higher than those reached on L. var. Lacerta and Dactylis glomerata L. var. Accord) and four legumes average across the 3 previous years. Similarly, the monthly average of (Trifolium repens L. var. Milo, T. pratense L., Onobrychis viciifolia Scop. minimum temperatures was 4 and 6ºC lower in February 2003 and 2005, var. Fakir and Medicago sativa L. var. Aragó), one of which (Trifolium compared to that in the 2000–2002 period which, in addition to low pratense L.) was represented by two varieties, including an erect precipitation during this period (with a total between November and (T. pratense L. var. Merivot) and a prostrate form (wild variety native to March less than 50% of the average in the 2000–2002 period), left the 2 À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 Table 1 Biogeographic distribution (not including areas where the species have been introduced for cropping) and origin of the studied species. Species are also compared within each functional group (grasses and legumes) in terms of climatic amplitude (temperature growth conditions and water requirements) and establishment and persistence traits.*T. pratense cv prostrate characteristics are not shown as considered similar to T.pratense (erect variety). Functional Species Biogeographic distribution/ Origin Temperature growth Water requirements Establishment/ group conditions persistence Grasses Lolium perenne Moist-temperate/ Europe, Asia and northern Limited growth when > Minimum annual rainfall = Fast-establishing/ L. Africa 25ºC stopping at 35ºC. 457–800 mm moderate persistence Tolerant to moderate cold Not tolerant to drought. temp. Dactylis Mediterranean and temperate/ Optimum temperature = Minimum annual rainfall = 380 Slow-establishing/ glomerata L. Eurasia 4.3–23.8 ◦ C. mm. persistent Wide climatic range, Tolerant to drought. tolerant to high temp. Legumes Trifolium repens Moist-temperate and Mediterranean/ Optimum temperature = 24 Minimum annual rainfall = 750 Slow-establishing/ L. Mediterranean region ºC / limited growth when > mm. persistent 35 ºC. Not tolerant to drought, needs Moderate tolerance to cold regularly distributed water supply. temp. Trifolium Moist-temperate and Mediterranean/ SE Optimum temperature = Minimum annual rainfall = 550 − Fast-establishing/ pratense L. Europe and Asia Minor 18–25 ºC. 700 mm for persistent growth, 350 limited persistence Tolerant to cold temp. mm for survival. Moderate tolerance to drought. Medicago sativa Mediterranean/ Optimum temperature = 25 Minimum annual rainfall = 350 Slow-establishing/ L. Mediterranean Basin and SW Asia ºC mm persistent Wide climatic range, Tolerant to drought. Tolerant to high and low temp. Onobrychis Mediterranean and temperate (typical of Optimum = 20/0 ◦ C (day/ Minimum annual rainfall = 330 Variable establishment/ viciifolia Scop. montane Mediterranean areas)/SE Europe night temperature) mm. moderate persistence and west and south-central Asia Tolerant to intense cold Tolerant to short drought periods. temp. but not high temp. References Alemayehu et al. (2020); Cooper and Tainton (1968); Delgado (1984); Delgado et al., (2001, 2014); Frame (2005); Hannaway et al. (1999); Heuzé and Tran (2015); Heuzé et al. (2019); Hybner and Winslow (2012); Jahufer et al. (2001); Lamp et al. (1990); Lattimore and McCormick (2012); Lloveras (1999); McKenna et al. (2018); Montserrat and Capdevila (1964); Montserrat and Fillat (1984); Mora-Ortiz and Smith (2018); POWO Plants of the World Online (2019); UPNA (2019). plants exposed to extreme temperatures without a protective snow mixtures included: four communities dominated in turn by each of the cover. Indeed, total precipitation recorded in the climatic station was four component species (with 70% of the dominant species and 10% of around 30% lower during the experimental period than the 6-years each of the other three); six communities co-dominated in turn by spe­ period, overall, it was 50% lower than the long-term average from CHE. cies pairs (40% of the two dominant species and 10% of each of the other two); and the centroid mixture with an even proportion (25%) of each of the four species. These communities were established at two sowing 2.4. Experimental design densities, the lower being a 60% of the total applied density (Table 2), which was determined following the agricultural recommendations of This study was established in the context of a multisite diversity each location (Kirwan et al., 2014). Replication in experiments using the experiment, the Agrodiversity experiment (Kirwan et al., 2014), simplex design is across the set of continuous species proportions. In this following the biodiversity-ecosystem function modelling framework design, estimation is based on regression methods whereas inference is (Connolly et al., 2013), where species composition and evenness of based on the residual variation around the regression model fitted forage mixtures were manipulated to assess diversity effects on several (Draper and Smith, 1998); thus, replication of each community is not ecosystem functions. The experimental design of the Agrodiversity needed. experiment (Finn et al., 2013; Kirwan et al., 2007, 2009) was based on In Gósol, the experiment was extended and a multi-experiment was the variation of the relative abundances of four species, which corre­ established, comprising five mixture types each including four species sponded to four functional types, according to their nitrogen fixation out of the pool of seven. All mixture types included two grasses: Lolium ability (legumes versus grasses) crossed with different temporal de­ perenne (fast-establishing) and Dactylis glomerata (persistent). In four out velopments (fast-establishing versus persistent patterns). of the five mixture types, the third species was the legume Trifolium The common experimental design of the Agrodiversity followed a repens. In those four mixture types the fourth component was: a) simplex design (Cornell, 1990; Ramseier et al., 2005). In particular, the Table 2 Identity and specific composition of the sown mixtures and sowing density for each species at full rate (scaled for each species and mixture according to the included proportion, e.g., in centroid mixtures sowing densities are multiplied by 0.25). Tp-mix Tpp-mix Ov-mix Ms-mix OvMs-mix Sowing density (kg seed ha− 1) Lolium perenne Lolium perenne Lolium perenne Lolium perenne Lolium perenne 30 Dactylis glomerata Dactylis glomerata Dactylis glomerata Dactylis glomerata Dactylis glomerata 25 Trifolium repens Trifolium repens Trifolium repens Trifolium repens 20 Trifolium pratense 20 (erect) Trifolium pratense 40 (prostrate) Onobrychis viciifolia Onobrychis viciifolia 100 Medicago sativa Medicago sativa 30 3 À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 Trifolium pratense (Tp-mix) erect variety; b) Trifolium pratense (Tpp-mix) power of θ (Connolly et al., 2013). Sowing density, essentially a nominal prostrate variety; c) Onobrychis viciifolia (Ov-mix); d) Medicago sativa variable with two levels, was scaled to have a zero mean so that all other (Ms-mix). The fifth mixture type included Onobrychis viciifolia and terms could be read directly as the effects at average density. The tem­ Medicago sativa, plus the two grasses (OvMs-mix; Table 2). poral variability across harvests and its interaction with diversity terms Overall, the complete experimental design included the mono­ were assessed by considering it as a repeated variable associated to each cultures of the seven experimental species sown at two densities plot as a subject (through a mixed model). (totalling 14 communities), and the 11 mixtures of each of the five In the tested models, we can distinguish two groups of response mixture types sown at two densities (totalling 11 ×5×2 = 110 com­ variables: the individual species’ effects and their interactions. The munities). Thus, we established a total of 124 plots with a completely former, corresponds to species identity effects, while the latter expresses randomized distribution. the effects of mixing or interaction effects, which can be studied sepa­ rately and thus can be compared, or aggregated as a single term (as a 2.5. Sward management and sampling global interaction or evenness effect, which in our analysis corresponded to the sum of all pairwise interactions) if differences between interaction The experiment was established in April 2003 and lasted until the terms are statistically negligible. autumn harvest in October 2005. Prior to sowing the field was har­ We restricted the number of three-way interactions included in the rowed. The communities were hand-sown in plots measuring 8.75 m2 model by using the Bonferroni correction. The models finally selected (3.5 ×2.5), whereas distances between plots had a minimum of 20 cm. included the separated pairwise interactions and θ = 1. For the model Sowing rates corresponded to those recommended for the study area including all harvests (Ms-mix and OvMs-mix), two out of the 12 (Table 2). Weeds were not removed over the duration of the experiment possible three-way interactions emerged as important after the Bonfer­ to allow for the natural succession in the sward. The system was rainfed. roni corrections. Plots were not fertilized except in 2004, when an isotopic label was To illustrate the differences in species performances (percentual incorporated as NH4NO3 at a dose of 40 kg N ha− 1. relative performances) and the importance of the diversity effects, we Yield and species composition were measured in all harvests. Total estimated the responses from the model for the different mixture types. yield was estimated at the same dates in all plots by harvesting 4.32 m2 We estimated the responses for each of the seven monocultures, and for in the centre of the plots to a height of approximately 5 cm following the the centroid community of each mixture type, characterized by an even protocol of Kirwan et al. (2014). Fixed 0.5 × 0.5 m2 quadrats were cut proportion of all included species. We also estimated the expected and total biomass was separated into the four sown components and centroid responses assuming no diversity effect, which correspond to the weeds, to estimate the sown and unsown (i.e., weeds) species fractions. average monoculture performances. All analyses were conducted with Samples were then oven-dried to constant mass at 60ºC. In the estab­ SAS (9.1; SAS Institute Inc., Cary, NC, USA). lishment year, the vegetation was harvested twice after sowing, in August and November 2003. The first harvest, often considered as a 3. Results cleaning cut, eliminates annual weeds, giving competitive advantage to the forage components. In the following years, the vegetation was har­ 3.1. Species performance and persistence patterns vested three times in 2004 and twice in 2005. Harvest frequency mirrored the local farmer’s common practice in rainfed forage systems. There was a higher stability in the response of mixtures compared to Namely, depending on the climatic conditions of a particular year and monocultures for sown species yield and resistance to weed invasion, the concomitant vegetation development, farmers perform one or two with monocultures showing a higher variability than mixtures for both harvests per year before the aestival drought period. The autumn responses in the three main harvests (spring and summer 2004 and regrowth, generally small, is typically grazed by livestock (an effect that summer 2005 harvests, Figs. 1, 2, 3 and Supplementary Tables 4 and 5). here we simulated also by harvesting) (specific harvest dates can be Sown-species yield of the seven forage species ranged from 381 to 3290 checked in Supplementary Fig. S1). DM (dry matter) kg ha− 1 in the most productive harvest (spring 2004) Because of the climatic anomalies and combined extreme events (Fig. 2). Monoculture responses in terms of sown species yield (Fig. 2) during the experimental period, most plots in three out of the five and resistance to weed invasion (Fig. 3) changed throughout the mixture types were lost in the second harvest of the final experimental different harvests (see also Supplementary Figures 2 and 3 for sown year. For this reason, in the autumn of 2005 we only harvested plots species yield and weed biomass evolution per treatment and year). For from the Ms-mix and the OvMs-mix mixture types. instance, the most productive species in the most productive harvest (spring 2004) was the erect variety of Trifolium pratense, but this species 2.6. Data analysis was one of the least productive components in 2005 (Fig. 2). Conversely, Medicago sativa showed the lowest productivity in spring 2004, yielding We modelled total forage dry matter yield (sown and unsown spe­ only 10.2% of the erect T. pratense; nonetheless, its productivity grad­ cies), and the segregated dry matter yield of the sown and unsown ually increased, rating among the best-performing species in 2005 (weeds) fractions, including the five mixture types and the 3 main (Fig. 2). These temporal patterns reflect differences in establishment harvests: 2 (spring and summer) harvests in 2004, and one (summer) ability and persistence among the seven species. Overall, the following harvest in 2005, leaving the autumn harvest which simulates the grazed species acted as fast-establishing: Lolium perenne, Trifolium pratense biomass and, moreover, presented high plant mortality for many plots, (both varieties) and Onobrychis viciifolia, while Dactylis glomerata and especially in 2005. For the Ms-mix and the OvMs-mix mixture types, we Medicago sativa acted as persistent species. Trifolium repens was initially modelled total yield in the seven harvests, including the cleaning cut: sown as a persistent legume, nonetheless, all Trifolium species performed two in 2003, three in 2004 and two in 2005. The most relevant models very poorly in the last year of the experiment (Fig. 2), after a very dry discussed in the paper are detailed in Supplementary Tables S1, S2, and and cold winter (detail in Supplementary Fig. S1). Seasonal patterns S3. within a year could also be recognized in sown species distributions: Using the Generalized Diversity-Interaction (GDI) modelling within 2004, Onobrychis viciifolia attained maximum yield in the spring approach (Kirwan et al., 2009; Connolly et al., 2013), the response harvest, while Medicago sativa performed better in the summer harvest variables were modelled as a function of sowing density (M), the sown (Fig. 2). proportions of the species (Pi), and the species pairwise (PiPj) and, eventually, three-way (PiPjPk) interactions. We also tested the pro­ pensity of species to interact by raising the species product PiPj to the 4 À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 best-performing component in the mixture (Trenbath, 1974), emerged only for the OvMs-mix in summer 2004 (P = 0.03; Fig. 2, Supplementary Table S7). Although transgressive overyielding was not generalized, in spring 2004 all other centroid mixtures showed yields around that of the best-performing monoculture, which in 2004 was T. pratense (NS for all comparisons; Fig. 2, Supplementary Table S8). In summer 2004, the same trend was found, except for the Tpp-mix centroid, which gave lower yields than the best monoculture, namely the erect T. pratense (P = 0.010; Fig. 2, Supplementary Table S8). In 2005 only the centroids including O. viciifolia or M. sativa, or both species gave yields in the range of the best monoculture (thus non-significantly different), but this was then Dactylis glomerata (Fig. 2, Supplementary Table S8). In the spring harvest of 2004, the difference in yield between the most and the least productive monoculture was as high as 89% (relative to the maximum); however, the difference between the most and the least productive centroid of the studied mixture types was of only 24%. This suggests a higher stability of mixtures compared to monocultures. In summer 2004 and in 2005, the difference between monocultures reached 94% and 99.9%, but the difference between centroids only increased with time up to 34% in summer 2004 and 56% in 2005. Moreover, for the Ov-mix, Ms-mix and OvMs-mix mixture types the difference between centroids was consistently below 20%. The best-performing monoculture averaged across harvests was Trifolium pratense with 2198 DM kg ha− 1 (non-significantly different from Dactylis glomerata, Lolium perenne and Onobrychis viciifolia). When averaged across harvests, the OvMs-mix centroid showed transgressive overyielding, that is, the yield of this community (2890 DM kg ha− 1) was higher than that of the best monoculture also considered across harvests, Trifolium pratense (P = 0.038). The Ov-mix and Ms-mix centroids gave average yields across harvests in the range of the best monoculture (T. pratense, P > 0.20). Fig. 1. Dry matter yield of the sown fraction (Mg ha− 1) and weed proportion 3.3. Species diversity benefits on resistance to weeds (%), boxplots for the three main harvests and for all monocultures (n = 14, corresponding to the 7 species at two densities) and mixtures (n = 110; including all mixtures of the 5 studied mixture types). The solid and dashed The advantages of sown diversity for yield described above emerged lines show the mean and median, respectively. Dots show the 5th and 95th also in terms of resistance to invasion (Figs. 2 and 3). On one hand, as percentiles. observed for yield, monocultures showed high variability in terms of weed invasion, with mixtures showing a decrease in the variability of 3.2. Species diversity benefits on yield and stability for the studied mixture weeds proportion in comparison (Fig. 1). T. pratense (erect) showed the types lowest weed biomass (Fig. 3) and proportion (14%, 0% and 54% in the two 2004 and in the 2005 harvests respectively). Generally, all Trifolium Along the three main harvests mixtures consistently provided higher species performed better in terms of resistance to weeds at the beginning total and sown-species yields than expected from the relative contribu­ of the experiment than at the end in 2005 (Fig. 3). The most invaded tion of the component species, as a result of a positive and significant monocultures were those of M. sativa and O. viciifolia, particularly in global interaction effect on yield (P < 0.0001 for the total and sown- 2004 (Fig. 3) reaching a weed proportion of 56% and 66%, respectively. species yield models, with all pairwise interactions grouped as a single There was a general negative interaction effect (with all interaction term). For each mixture type, the interaction effect on sown-species terms grouped as a single term) on weed biomass (P < 0.0001), which yield persisted along the three main harvests (P ≤ 0.0004, Table 3) resulted in a decrease of weed invasion in mixtures relative to the ex­ with the exception of the Tp-mix, for which it was significant in 2004 pected one according to monoculture weed biomass (Fig. 3). Moreover, (P < 0.0001 and P = 0.031 in spring and summer respectively) but not this effect was significant individually for most mixture types, either significant in 2005 (P = 0.95; Fig. 2, Table 3). As a result of the inter­ when comparing with the average of the mixture’s corresponding action effect, within mixture types, the sown-species yield of the monocultures (except for Tp- and Tpp-mix; Fig. 3) or with the average of centroid communities (mixtures with maximum evenness and thus a all seven monocultures (Supplementary Table S9). As observed for maximum expression of this effect) was higher than that expected from sown-species yields (Fig. 2), in spring 2004 all centroid mixtures showed averaging the corresponding monoculture yields, for all mixture types similar weed biomasses (maximum 32% difference), in spite of the and harvests except the one mentioned above (Tp-mix in 2005; Fig. 2). considerable range in invasibility existing for monocultures (75%; In addition, for the three harvests all centroid communities performed Fig. 3). In summer 2004, there were considerable differences in weed better than the average of the seven tested monocultures overall the biomass within monocultures (99%) and within centroids (83%). multi-experiment (P ≤ 0.0071 for all comparisons; Fig. 2, Supplemen­ Nevertheless, in centroids, weed biomass was below 63 DM kg ha− 1 in tary Table S6) except the Tp-mix in 2005. This mixture type shifted from all cases, while in monocultures weed biomass reached 671 DM kg ha− 1. being the highest yielding in spring 2004 to the least productive one in In 2005, differences were 48% and 36% for monocultures and centroids summer 2005. Even so, it still gave yields around the seven mono­ respectively. We did not observe transgressive over-suppression in cultures average (non-significantly different; Fig. 2, Supplementary centroid mixtures, i.e., weed biomasses in mixtures lower than the least Table S6). invaded of all tested monocultures, but within each mixture type, weed Transgressive overyielding, i.e., centroid mixture yielding above the biomass of the centroid didn’t exceed that of the least invasible mixture component (Supplementary Tables S10 and S11). 5 À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 Fig. 2. Dry matter yield of the sown species fraction (Mg ha− 1) in the three main harvests. Estimates (+SE) at average density for each monoculture (open bars) and for each centroid of the 5 mixture types, without including the interaction effect (average of corresponding monocultures; light grey) and including it (dark grey). The interaction effect is expressed rela­ tive to the total centroid value (%). Significant effects are expressed in bold. Dashed lines indicate the global averages across mono­ cultures and across centroid mixtures (including the interaction effect). Lp: Lolium perenne; Dg: Dactylis glomerata; Tr: Trifolium repens; Tp: erect Trifolium pratense; Tpp: pros­ trate Trifolium pratense; Ov: Onobrychis viciifolia; Ms: Medicago sativa. See Table 2 for mixture type composition. 3.4. Differences in diversity benefits among mixture types mix), showed the strongest effects on weed invasion reduction (through a negative and significant interaction effect) when compared Despite the general beneficial diversity effect on the studied func­ with the other mixture types (against Tp-, Tpp-, and Ms-mix, P < 0.013 tions, there were differences in the strength of the tested species in­ for all comparisons; Table 3, Supplementary Table S12), while this teractions that resulted in differences in the total interaction effect for species (O. viciifolia) showed the highest weed invasion in monoculture the studied mixture types (Table 3), as witnessed by a significance of the (Fig. 3). χ2 comparison test of segregating the global interaction term in indi­ vidual species interactions, with P = 0.01 and P = 0.003 for the sown- 3.5. Persistence of diversity benefits in two Mediterranean mixture types species yield and weed biomass models. Mixtures including at least one Mediterranean legume (O. viciifolia and M. sativa) showed the A detailed analysis including all seven harvests across the three highest interaction effects across harvests resulting in the best perfor­ years, for the Ms-mix and OvMs-mix mixture types, reveals that the mances (Figs. 2, 3 and Table 3). For the sown-species yield, the mixture interaction effects on total yield (sown species plus weeds) emerged including the erect T. pratense showed the highest interaction strength in after the first harvest (cleaning cut) and persisted until the last harvest the year after establishment (first harvest of 2004), when this species (Fig. 4), all being positive and significant (P < 0.041; Fig. 4, Supple­ also attained the best performances, but it dropped after that moment, mentary Table S13). In this model, two out of the twelve three-way being nil in 2005 (NS; Fig. 2, Table 3). As a result, considered globally interaction terms were significant, namely, L.perenne×D.glomerata×O. across harvests, this mixture showed the minimum interaction effect viciifolia and L.perenne×O.viciifolia×M.sativa. The interaction effects for (Table 3). For the rest of the mixture types the changes in the interaction the two mixture types were singularly high in August 2005 (Supple­ effect across harvests were less important (P > 0.001 for all PiPj mentary Table S13), with a significant evenness×harvest interaction ×harvest interactions, excluded by the Bonferroni correction). (P = 0.0016), although in relative terms they contributed to total yield Regarding weed invasion resistance, the mixtures with Mediterranean particularly in autumn (Fig. 4) with a contribution above 48% for all legumes, and specifically those including O. viciifolia (Ov- and OvMs- autumn harvests and for both mixture types. 6 À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 Fig. 3. Unsown species (weeds) biomass (Mg ha− 1) in the three main harvests. Estimates (+SE) at average density for each monoculture (open bars) and for each centroid of the 5 mixture types, without including the interac­ tion effect (average of included monocultures, light grey) and including it (dark grey). The interaction effect is expressed as the decrease relative to the expected centroid value without the interaction effect (%). Significant effects are expressed in bold. Dashed lines indicate the global averages across monocultures (in the left-hand side panels) and across centroid mix­ tures (including the interaction effect) in the right-hand side panels. Lp: Lolium perenne; Dg: Dactylis glomerata; Tr: Trifolium repens; Tp: erect Trifolium pratense; Tpp: prostrate Trifolium pra­ tense; Ov: Onobrychis viciifolia; Ms: Medicago sativa. See Table 2 for mixture type composition. 4. Discussion the differential species responses to the harsh climatic conditions that took place during the experiment, seem to be behind the huge differ­ 4.1. Insurance effects and species interactions ences in species performances (white bars in Figs. 2 and 3). This was the case of the difference in the performance of T. pratense and O. viciifolia, 4.1.1. Species establishment and persistence patterns both acting as fast-establishing, but O. viciifolia showed a higher In our experiment, the used species generally followed the predicted persistence in 2005 compared to T. pratense, which thus showed less temporal patterns selected in the design, including fast-establishing adaptation to the given harsh conditions. species, which suffered from a limited persistence, and persistent spe­ Despite these differences in monoculture performances, mixtures cies, which tended to establish slowly (Fig. 2). But apart from these showed a stabilizing effect (light grey bars in Figs. 2 and 3). The basic specific trends, the extremely harsh climatic conditions given during the effect behind this stability is the portfolio effect, i.e., that of averaging experiment, seem to have affected both the species establishment and species with contrasting responses, traditionally used as an insurance persistence patterns, with a stronger effect on the Trifolium species, even strategy under unpredictable conditions (Altieri et al., 2015). This is though they are naturally found in the study area. These effects must essentially a probabilistic effect, but a positive effect can also be targeted have been particularly important for T. repens, commonly showing a through the inclusion of specifically aimed traits. In this sense, the in­ slow-establishment and persistent development (Finn et al., 2013; clusion of species with different temporal patterns (i.e., different sea­ Hycka, 1973). In Mediterranean irrigated systems this species has been sonal or inter-annual optimums) has been used to balance the changes in found to dominate the mixtures one or two years after establishment forage supply due to temporal variability (Fukai and Trenbath, 1993; (Hycka, 1973), but it is known to be highly affected by water deficit Lithourgidis et al., 2011; Yu et al., 2015). Indeed, in experiments (Neal et al., 2009), which occurred during the experimental years (mean assessing grass-legume binary mixtures, where the temporal comple­ annual precipitation across the 3 years being 50% lower than the mentarity axis was missing (both being fast or slow-establishing), long-term average, as explained above). mixture performance suffered a temporal variability associated with Thus, the different temporal patterns of the studied species, but also that of the included species (Hycka, 1975). Other interesting contrasting 7 À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 Table 3 Interaction effects expressed as the sown-species yield increase or weed biomass decrease (both in Mg ha− 1) for each mixture type. We show estimates ± standard errors for the maximum interaction effects, given for the centroid communities (maximum evenness). sown DMY increase (Mg ha− 1) weed biomass decrease (Mg ha− 1) estimate ± SE P estimate ± SE P Tp-mix 0.53 ± 0.20* < 0.0001 a -0.058 ± 0.052 0.26 a Tpp-mix 0.69 ± 0.19 0.0004 a,b -0.10 ± 0.052 0.054 a Ov-mix 0.98 ± 0.19 < 0.0001 b,c -0.27 ± 0.051 < 0.0001 b Ms-mix 1.09 ± 0.19 < 0.0001 c -0.14 ± 0.051 0.0066 a OvMs-mix 1.13 ± 0.19 < 0.0001 c -0.29 ± 0.051 < 0.0001 b * The interaction effect for the different mixture types did not change significantly with harvest after the Bonferroni correction, except for the Tp-mix. For this reason, for Tp-mix we give an average value of the interaction effect across the three studied harvests. For Tp-mix, the estimates±SE and P-values for the three studied harvests are respectively: 1.19 ± 0.28 and P < 0.0001 in June 2004; 0.43 ± 0.20 and P = 0.031 in Aug 2004; and − 0.02 ± 0.26 and P = 0.95 in Aug 2005. patterns, such as resistance versus survival traits (Lüscher et al., 2022), because of shifts in competitive advantage under different environ­ could be included simultaneously to widen the community response mental conditions. Contrastingly, in addition to this substitution effect, plasticity. the interaction effect can reflect an advantage associated with other complementarity and facilitation mechanisms, which are the basis of the 4.1.2. Interaction effects ecological theory regarding species coexistence (Homulle et al., 2022; Li The advantages of mixing go beyond the portfolio or identity effect, et al., 2014). exceeding the monocultures average due to species interactions, which The final effect of mixing will depend on the sum of identity and in our study resulted in an extra effect above (below for weed biomass) interaction effects, and the possible trade-off between the two groups of the expected monoculture average (Figs. 2, 3 and 4). The biological terms. Interestingly, in our study, the highest interaction potentials mechanism behind this positive interaction effect could be simply the occurred between species with modest performances, thus with a trade- substitution effect of the less persisting species by the surviving species, off between the two diversity components. Probably, under the climatic anomalies suffered by the swards, the inclusion of species with a wide climatic amplitude, and especially those adapted to cope with drought (Mediterranean legumes), was key in the persistence of the mixture components and thus its interaction potential, even though they didn’t exhibit the highest responses individually. Contrastingly, M. sativa has been reported to yield and suppress weeds with almost a total reliance on identity effects (Picasso et al., 2008). Regarding diversity advantages under drought conditions, some authors report functional group complementarity as the main driver of resistance and resilience in ley grasslands (Finn et al., 2018; Grange et al., 2021) while other studies point to both identity and functional diversity as key factors (Komainda et al., 2020). On another direction, Ergon et al. (2018) concluded that biodiversity at all levels is crucial as an adaptation strategy to drought. 4.1.3. Differences in strength and persistence of species interactions across mixture types As reported above, despite the generalized beneficial interaction effect across mixture types (on both studied functions), mixtures including at least one Mediterranean legume exhibited higher advan­ tages due to mixing. As we have explained, these differences are derived from the range in strengths of the tested species interactions (Figs. 2, 3 and 4; Table 3). Regarding interaction effects persistence across harvests, except for Tp-mix, these persisted even for the mixture including T. repens and the prostrate T. pratense, even though both species presented a major decline by the end of the experiment. Although ecologically speaking most mixtures were consistently advantageous in terms of a significant interaction effect above the expected monoculture average, from an agronomical point of view only the mixtures equalling or overyielding the best monoculture or mixture would be considered of interest. For the Tpp-mix, although the interaction effect persisted across harvests, it was not enough to compensate the low identity effect (basically of the prostrate T. pratense variety), and thus offered less agronomical interest. Fig. 4. Total (sown species + weeds) dry matter yield (Mg ha− 1) for the centroid mixtures of the Ms-mix and the OvMs-mix mixture types, for the seven harvests (h1-h7) along the three years. The grey fraction, which is only shown 4.2. Selecting mixtures components under climatic uncertainty when significative (p < 0.05), corresponds to the interaction effect added to the expected centroid value according to monoculture performances (white frac­ 4.2.1. How do environmental conditions affect diversity effects? tion). Note that h2, h5 and h7 correspond to autumn harvests. See Table 2 for The harsh climatic conditions that took place during the experiment mixture composition. might have affected not only the individual species’ performances but 8 À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 also the species’ interactions, probably resulting in the differences in the varieties to be used in monoculture (Volaire et al., 2014), and its interaction strengths observed across mixture types (Section 4.1.3). This selection should be done in mixture (Atwood and Garber, 1942). effect could also explain the differences in diversity effects of the same mixtures across sites and studies. Indeed, the Tp-mix showed important 5. Conclusions and sustained diversity effects resulting in transgressive overyielding in most sites of the Agrodiversity experiment (Finn et al., 2013; Kirwan In this study we provide sound evidence on the forage mixtures’ et al., 2007), while in our site it was important in the first 2004 harvest, advantages over monocultures in terms of increased yield, resistance to but it disappeared in the following (Table 3). In experiments carried out weed invasion and the stability of both functions. This effect may be key in irrigated systems in the Ebro Valley, this mixture type was found to be for rainfed production systems, which will be particularly threatened by the best mixture (Hycka, 1973). extreme temperatures and irregular precipitation patterns, such as those In Gósol, with a historical precipitation above 1000 mm, we ex­ experienced during the development of our experiment. Nevertheless, pected a good performance and persistence of both Tp-mix and Ms-mix, our results suggest that, although interaction effects tend to occur for but under the drought and frost experienced during the experimental most mixture compositions, not all of them overyield the best-adapted years, mixtures including species with broader climatic amplitudes, such species. Moreover, there can be a trade-off between identity and inter­ as M. sativa, performed better than mixtures with Trifolium species, action effects, thus, we should select species that maximize the sum of particularly T. repens, which may become relegated to irrigated systems both diversity components, as happened in our experiment with mix­ even in mountain areas. Nonetheless, it must be noted that Finn et al. tures including at least one Mediterranean legume. In the context of (2018) found out that the tested mixture including T. repens and climate change, it is crucial to explore species interactions and related T. pratense outperformed monocultures when an experimental drought mechanisms that have a special role under harsh conditions. The se­ was imposed in two temperate regions (however, T. repens proportion in lection of the best mixture composition and management for a particular mixture declined in one of the two locations). In Mediterranean areas, agro-climatic environment will require the study of identity and inter­ where water scarcity is forecast to increase, drought resistance is a target action effects, through GDI models, under different environmental trait to pursue. In this regard, and accordingly to our results, M. sativa conditions, and for multiple functions. and O. viciifolia have been found to outperform several species (including the ones in the present study) in terms of drought resistance (Mueller-Harvey et al., 2019). Indeed, in early experiments in the se­ Declaration of Competing Interest venties, grass-legume mixtures with M. sativa and O. viciifolia, were considered appropriate for the extremely harsh conditions of a conti­ The authors declare that they have no known competing financial nental Mediterranean climate in the Ebro Valley, overyielding the interests or personal relationships that could have appeared to influence best-performing species (M. sativa), but O. viciifolia showed a lack of the work reported in this paper. adaptation below 600 m (Hycka, 1974), as found in several studies (Delgado, 1984; Delgado et al., 2014; Montserrat and Fillat, 1984). In Data Availability the lowlands, O. viciifolia could be substituted by T. pratense, particularly under irrigated systems. In the Agrodiversity experiment, the grass-legume Data will be made available on request. mixture including these two legumes (T. pratense and M. sativa) gave significant diversity effects in two out of three sites in which it was sown Acknowledgements (Kirwan et al., 2007) but it failed in France, probably due to the extreme climatic events experienced in most European countries in 2003 (Olesen We would like to thank all people collaborating in field work and and Bindi, 2004). Binary grass-legume mixtures including M. sativa sample processing. The work was conducted in relation to the EU COST outperformed those including T. pratense in Greek mountain systems Action 852, and was finished thanks to the projects BIOGEI (Hadjigeorgiou, Thanapoulos, 2004), while the reverse was found in (CGL2013–49142-C2–1-R) and IMAGINE (CGL2017–85490-R) and is Mid-European conditions (Kleen et al., 2011), according to a better supported by a Margarita Salas grant under the European Union- adaptation of M. sativa in Mediterranean conditions. Contrastingly, NextGeneration EU funds. We also acknowledge the Catalan Govern­ Ergon et al. (2018) advocate for both T. pratense and M. sativa as key ment for some additional funding. components in Mediterranean mixtures to avoid drought-related yield reductions. Appendix A. Supporting information The differences in diversity effects listed above may be explained by the environmental variability underlying the studies. Ecological theory Supplementary data associated with this article can be found in the predicts that species interactions may change under stressful conditions online version at doi:10.1016/j.agee.2023.108601. (Kikvidze and Armas, 2010). Notably, Lüscher et al. (2022) have also identified the severity of stressful events such as droughts as a key factor References to understand the importance of species interactions, concluding that complementary effects are maximized at moderate drought levels Alemayehu, S., Ayana, E.K., Dile, Y.T., Demissie, T., Yimam, Y., Girvetz, E., Aynekulu, E., whereas identity effects play a major role at severe droughts. Solomon, D., Worqlul, A.W., 2020. Evaluating land suitability and potential climate change impacts on alfalfa (Medicago sativa) production in Ethiopia. Atmosphere 11 (10), 1124. https://doi.org/10.3390/atmos11101124. 4.2.2. From selected adapted cultivars to selected adapted mixtures Altieri, M.A., 1999. The ecological role of biodiversity in agroecosystems. Agric. Ecosyst. The development and use of adapted cultivars are key in Mediter­ Environ. 74, 19–31. https://doi.org/10.1016/S0167-8809(99)00028-6. Altieri, M.A., Nicholls, C.I., Henao, A., Lana, M.A., 2015. Agroecology and the design of ranean regions (Hycka, 1974), especially considering that many forage climate change-resilient farming systems. Agron. Sustain. Dev. 35, 869–890. https:// species have been developed in temperate regions (Volaire et al., 2009). doi.org/10.1007/s13593-015-0285-2. Nevertheless, the trade-off between identity and interaction effects Atwood, S.S., Garber, R.J., 1942. Evaluation of individual plants of white clover for observed in our study stresses the importance of selecting species that yielding ability in association with bluegrass. J. Am. Soc. Agron. 34, 1–6. https:// doi.org/10.2134/agronj1942.00021962003400010001x. maximize the sum of the two effects (Connolly et al., 2013; Kirwan et al., Brooker, R.W., Bennett, A.E., Cong, W.-F., Daniell, T.J., George, T.S., Hallett, P.D., 2007, 2009; Lüscher et al., 2022). If for monocultures we would select Hawes, C., Iannetta, P.P.M., Jones, H.G., Karley, A.J., Li, L., McKenzie, B.M., species with high identity effects, for mixtures we should include species Pakeman, R.J., Paterson, E., Schöb, C., Shen, J., Squire, G., Watson, C.A., Zhang, C., Zhang, F., Zhang, J., White, P.J., 2015. Improving intercropping: a synthesis of with high individual, but also with high interaction potential. As a research in agronomy, plant physiology and ecology. N. Phytol. 206 (1), 107–117. result, the genetic material to be included in mixtures may differ from https://doi.org/10.1111/nph.13132ç. 9 À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 Statistics of the Spanish Ministry of Agriculture, 2012. https://www.mapa.gob.es/es/ grassland species in the face of drought and heavy rainfall. Ecol. Res. 29, 991–1001. estadistica/temas/publicaciones/anuario-de estadistica/2012/default.aspx https://doi.org/10.1007/s11284-014-1187-5. (Accessed 15 March 2022). Hadjigeorgiou, I., Thanapoulos, R., 2004. A comparative study of five sown "grass- Calbó, J., Sanchez-Lorenzo, A., Barrera-Escoda, A., Cunillera, J., 2012. Climate change legume" mixtures and the indigenous vegetation when grown on a rain-fed mountain projections for Catalonia (NE Iberian Peninsula). Part II: integrating several area of Greece. Land Use Systems in Grassland dominated Regions. Proceedings of methodologies. Tethys 9, 13–24. https://doi.org/10.3369/tethys.2012.9.02. 20th EGF General Meeting. Campbell, A., Biesmeijer, J., Varma, V., Wäckers, F., 2012. Realising multiple ecosystem Hannaway, D., Fransen, S., Cropper, J., Teel, M., Chaney, M., Griggs, T., Halse, R., Hart, services based on the response of three beneficial insect groups to floral traits and J., Cheeke, P., Hansen, D., Klinger, R., Lane, W., 1999. Perennial ryegrass (Lolium trait diversity. Basic Appl. Ecol. 13 (4), 363–370. https://doi.org/10.1016/j. perenne L.). Technical Bull. No. PNW503. https://ir.library.oregonstate.edu/concer baae.2012.04.003. n/administrative_report_or_publications/9306sz60j (Accessed 5 March 2022). Connolly, J., Bell, T., Bolger, T., Brophy, C., Carnus, T., Finn, J.A., Kirwan, L., Isbell, F., Hauggaard-Nielsen, H., Ambus, P., Jensen, E.S., 2003. The comparison of nitrogen use Levine, J., Lu¨scher, A., Picasso, V., Roscher, C., Sebastià, T., Suter, M., Weigelt, A., and leaching in sole cropped versus intercropped pea and barley. Nutr. Cycl. 2013. An improved model to predict the effects of changing biodiversity levels on Agroecosyst 65, 289–300. https://doi.org/10.1023/A:1022612528161. ecosystem function. J. Ecol. 101 (2), 344–355. https://doi.org/10.1111/1365- Haughey, E., Suter, M., Daniel Hofer, D., Hoekstra, N., McElwain, J., Lüscher, A., Finn, J. 2745.12052. A., 2018. Higher species richness enhances yield stability in intensively managed Connolly, J., Sebastià, M.T., Kirwan, L., Finn, J.A., Llurba, R., Wang, M., Collins, R.P., grasslands with experimental disturbance. Sci. Rep. 8, 15047. https://doi.org/ Porqueddu, C., Helgadóttir, A., Baadshaug, O.H., Bélanger, G., Black, A., Brophy, C., 10.1038/s41598-018-33262-9. Čop, J., Dalmannsdóttir, S., Delgado, I., Elgersma, A., Forthergill, M., Frankow- Heuzé, V., Tran, G., 2015. Cocksfoot (Dactylis glomerata). Feedipedia, a programme by Linderg, B.E., Ghesquiere, A., Golinski, P., Grieu, P., Gustavsson, A.M., Höglind, M., INRAE, CIRAD, AFZ and FAO. https://www.feedipedia.org/node/466 Last updated Huguenin-Elie, O., Jørgensen, M., Kadziuliene, Z., Lunnan, T., Nykanen-Kurki, P., on May 11, 2015, 14:33. Ribas, A., Thumm, U., De Vliegher, A., Lüscher, A., 2018. Weed suppression greatly Heuzé, V., Tran, G., Hassoun, P., Lebas, F., 2019. White clover (Trifolium repens). increased by plant diversity in intensively managed grasslands: a continental-scale Feedipedia, a programme by INRAE, CIRAD, AFZ and FAO. https://www.feedipedia. experiment. J. Appl. Ecol. 55 (2), 852–862. https://doi.org/10.1111/1365- org/node/245 Last updated on April 10, 2019, 14:04. 2664.12991. Homulle, Z., George, T.S., Karley, A.J., 2022. Root traits with team benefits: Cooper, J.P., Tainton, N.M., 1968. Light and temperature requirements for the growth of understanding belowground interactions in intercropping systems. Plant Soil 1–26. tropical and temperate grasses. Herb. Abst 38 (3), 167–176. https://doi.org/10.1007/s11104-021-05165-8. Cornell, J.A., 1990. Experiments with Mixtures: Designs, Models, and the Analysis of Hooper, D.U., Chapin, F.S., Ewel, J.J., Hector, A., Inchausti, P., Lavorel, S., Lawton, J.H., Mixture. Data, 2nd ed. John Wiley & Sons, New York. Lodge, D.M., Loreau, M., Naeem, S., Schmid, B., Setälä, H., Symstad, A.J., Cummins, S., Finn, J.A., Richards, K.G., Lanigan, G.J., Grange, G., Brophy, C., Vandermeer, J., Wardle, D.A., 2005. Effects of biodiversity on ecosystem Cardenas, L.M., Misselbrook, T.H., Reynolds, C.K., Krol, D.J., 2021. Beneficial effects functioning: a consensus of current knowledge. Ecol. Monogr. 75, 3–35. https://doi. of multi-species mixtures on N2O emissions from intensively managed grassland org/10.1890/04-0922. swards. Sci. Total. Environ. 792, 148163 https://doi.org/10.1016/j. Hybner, R.M., Winslow, S., 2012. Orchargrass Dactylis glomerata L.: An Introduced scitotenv.2021.148163. Conservation Grass for Use in Montana and Wyoming. Plant Materials Technical ́ del Duero Delgado, I., 1984. La alfalfa en pastoreo, in: Cultivo de la alfalfa en los regadios Note No. MT-82. United States Department of Agriculture, Natural Resources y Ebro. Servicio de Extensió n Agraria, MAPA, Madrid (Spain), pp. 45–59. Conservation Service. https://www.nrcs.usda.gov 〉 mtpmctn11287 (Accessed 28 ́ Delgado, I., Andueza, D., Muñoz, F., Martinez, N., 2001. Effect of nitrogen fertilisation on March 2022). alfalfa (Medicago sativa L.) regrowth and production. Options Méditerranéennes. Sér. Hycka, M., 1973. Veza común, su cultivo y utilización. Pastos 3, 240–247. A: Sémin. Méditerranéens 45, 141–143. Hycka, M., 1974. Praderas artificiales en los secanos de condición extrema. Pastos 4, Delgado, I., Muñoz, F., Andueza, D., 2014. Evaluation of the competition between alfalfa 170–190. and sainfoin sown in mixture, in: Baumont, R., Carrère, P., Jouven, M., Lombardi, G., Hycka, M., 1975. Praderas artificiales en los secanos del Mediterráneo. Pastos 5, López-Francos, A., Martin, B., Peeters, A., Porqueddu, C., (Eds.), Forage resources 188–203. and ecosystem services provided by Mountain and Mediterranean grasslands and IPCC, 2022: Climate Change 2022: Impacts, Adaptation and Vulnerability. Contribution rangelands. Zaragoza: CIHEAM / INRA / FAO / VetAgro Sup Clermont-Ferrand / of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel Montpellier SupAgro, pp. 139–142 (Options Méditerranéennes: Série A. Séminaires on Climate Change [Pörtner, H.-O., Roberts, D.C., Tignor, M., Poloczanska, E.S., Méditerranéens; n. 109). Mintenbeck, K., Alegría, A., Craig, M., Langsdorf, S., Löschke, S., Möller, V., Okem, Díaz, S., Cabido, M., 2001. Vive la difference: plant functional diversity matters to A., Rama, B. (Eds.)]. Cambridge University Press. Cambridge University Press, ecosystem processes. Trends Ecol. Evol. 16, 646–655. https://doi.org/10.1016/ Cambridge, UK and New York, NY, USA, 3056 pp. https://10.1017/ S0169-5347(01)02283-2. 9781009325844. Dooley, A., Isbell, F., Kirwan, L., Connolly, J., Finn, J., Brophy, C., 2015. Testing the Isbell, F., Craven, D., Connolly, J., Loreau, M., Schmid, B., Beierkuhnlein, C., Bezemer, T. effects of diversity on ecosystem multifunctionality using a multivariate model. Ecol. M., Bonin, C., Bruelheide, H., de Luca, E., Ebeling, A., Griffin, J.N., Guo, Q., Lett. 18, 1242–1251. https://doi.org/10.1111/ele.12504. Hautier, Y., Hector, A., Jentsch, A., Kreyling, J., Lanta, V., Manning, P., Meyer, S.T., Draper, N.R., Smith, H., 1998. Applied Regression Analysis, Wiley Series in Probability Mori, A.S., Naeem, S., Niklaus, P.A., Polley, H.W., Reich, P.B., Roscher, C., and Statistics, p 736. Seabloom, E.W., Smith, M.D., Thakur, M.P., Tilman, D., Tracy, B.F., van der Ehrmann, J., Ritz, K., 2014. Plant: soil interactions in temperate multi-cropping Putten, W.H., van Ruijven, J., Ewigelt, A., Weisser, W.W., Wisley, B., Eisenhauer, N., production systems. Plant Soil 376, 1–29. https://doi.org/10.1007/s11104-013- 2015. Biodiversity increases the resistance of ecosystem productivity to climate 1921-8. extremes. Nature 526 (7574), 574–577. https://doi.org/10.1038/nature15374. Ergon, Å., Seddaiu, G., Korhonen, P., Virkajärvi, P., Bellocchi, G., Jørgensen, M., Østrem, Jahufer, Z., Rogers, H., Rogers, M., 2001. White clover. Report No. AG0705. Department l., Reheul, D., Volaire, F., 2018. How can forage production in Nordic and of Natural Resources and Environment, Victoria. Mediterranean Europe adapt to the challenges and opportunities arising from Kikvidze, Z., Armas, C., 2010. Plant Interaction Indices Based on Experimental Plant climate change? Eur. J. Agron. 92, 97–106. https://doi.org/10.1016/j. Performance Data, in: Pugnaire, F. (Ed.) Positive Plant Interactions and Community eja.2017.09.016. Dynamics. CRC Press, Fundación BBVA, Bilbao, Spain, pp. 17–39. Finn, J.A., Kirwan, L., Connolly, J., Sebastià, M.T., Helgadottir, A., Baadshaug, O.H., Kirschenmann, F., 2007. Potential for a new generation of biodiversity in agroecosystems Bélanger, G., Black, A., Brophy, C., Collins, R.P., Čop, J., Dalmannsdó ttir, S., of the future. Agron. J. 9, 373–376. https://doi.org/10.2134/agronj2006.0104. Delgado, I., Elgersma, A., Fothergill, M., Frankow-Lindberg, B.E., Ghesquiere, A., Kirwan, L., Lu¨scher, A., Sebastià, M.T., Finn, J.A., Collins, R.P., Porqueddu, C., Golinska, B., Golinski, P., Grieu, P., Gustavsson, A.M., Ho¨glind, M., Huguenin- Helgadottir, A., Baadshaug, O.H., Brophy, C., Coran, C., Dalmannsdó ttir, S., Elie, O., Jørgensen, M., Kadziuliene, Z., Kurki, P., Llurba, R., Lunnan, T., Delgado, I., Elgersma, A., Fothergill, M., Frankow-Lindberg, B.E., Golinski, P., Porqueddu, C., Suter, M., Thumm, U., Lu¨scher, A., 2013. Ecosystem function Grieu, P., Gustavsson, A.M., Ho¨glind, M., Huguenin-Elie, O., Iliadis, C., enhanced by combining four functional types of plant species in intensively managed Jørgensen, M., Kadziuliene, Z., Karyotis, T., Lunnan, T., Malengier, M., Maltoni, S., grassland mixtures: a 3-year continental-scale field experiment. J. Appl. Ecol. 50, Meyer, V., Nyfeler, D., Nykanen-Kurki, P., Parente, J., Smith, H.J., Thumm, U., 365–375. https://doi.org/10.1111/1365-2664.12041. Connolly, J., 2007. Evenness drives consistent diversity effects in intensive grassland Finn, J.A., Suter, M., Haughey, E., Hofer, D., Lüscher, A., 2018. Greater gains in annual systems across 28 European sites. J. Ecol. 95, 530–539. https://doi.org/10.1111/ yields from increased plant diversity than losses from experimental drought in two j.1365-2745.2007.01225.x. temperate grasslands. Agric. Ecosyst. Environ. 258, 149–153. https://doi.org/ Kirwan, L., Connolly, J., Finn, J.A., Brophy, C., Lüscher, A., Nyfeler, D., Sebastia, M.-T., 10.1016/j.agee.2018.02.014. 2009. Diversity–interaction modeling: estimating contributions of species identities Frame, J., 2005. Forage legumes for temperate grasslands. Rome: Food and Agriculture and interactions to ecosystem function. Ecology 90, 2032–2038. https://doi.org/ Organization of the United Nations, first ed. CRC Press, Boca Raton. https://doi.org/ 10.1890/08-1684.1. 10.1201/9780429187667. Kirwan, L., Connolly, J., Brophy, C., Baadshaug, O., Belanger, G., Black, A., Carnus, T., Fukai, S., Trenbath, B.R., 1993. Processes determining intercrop productivity and yields Collins, R., Čop, J., Delgado, I., De Vliegher, A., Elgersma, A., Frankow-Lindberg, B., of component crops. Field Crops Res 34 (3–4), 247–271. https://doi.org/10.1016/ Golinski, P., Grieu, P., Gustavsson, A.-M., Helgadóttir, Á., Höglind, M., Huguenin- 0378-4290(93)90117-6. Elie, O., Jørgensen, M., Kadžiulienė, Z., Lunnan, T., Lüscher, A., Kurki, P., Grange, G., Finn, J.A., Brophy, C., 2021. Plant diversity enhanced yield and mitigated Porqueddu, C., Sebastia, T., Thumm, U., Walmsley, D., Finn, J., 2014. The drought impacts in intensively managed grassland communities. J. Appl. Ecol. 58 agrodiversity experiment: three years of data from a multisite study in intensively (9), 1864–1875. https://doi.org/10.1111/1365-2664.13894. managed grasslands. Ecology 95, 2680. https://doi.org/10.1890/14-0170.1. Grant, K., Kreyling, J., Heilmeier, H., Jentsch, A., 2014. Extreme weather events and Kleen, J., Taube, F., Gierus, M., 2011. Agronomic performance and nutritive value of plant–plant interactions: shifts between competition and facilitation among forage legumes in binary mixtures with perennial ryegrass under different 10 À. Ribas et al. Agriculture, Ecosystems and Environment 356 (2023) 108601 defoliation systems. J. Agric. Sci. 149 (1), 73–84. https://doi.org/10.1017/ Picasso, V.D., Brummer, E.C., Liebman, M., Dixon, P.M., Wilsey, B.J., 2008. Crop species S0021859610000456. diversity affects productivity and weed suppression in perennial polycultures under Komainda, M., Küchenmeister, F., Küchenmeister, K., Kayser, M., Wrage-Mönnig, N., two management strategies. Crop Sci. 48 (1), 331–342. https://doi.org/10.2135/ Isselstein, J., 2020. Drought tolerance is determined by species identity and cropsci2007.04.0225. functional group diversity rather than by species diversity within multi-species Porqueddu, C., Ates, S., Louhaichi, M., Kyriazopoulos, A.P., Moreno, G., Del Pozo, A., swards. Eur. J. Agron. 119, 126116 https://doi.org/10.1016/j.eja.2020.126116. Ovalle, C., Ewing, M.A., Nichols, P.G.H., 2016. Grasslands in ‘Old World’and ‘New Lamp, C.A., Forbes, S.J., Cade, J.W., 1990. Species descriptions. In: Lamp, C.A., World’Mediterranean-climate zones: past trends, current status and future research Forbes, S.J., Cade, J.W. (Eds.), Grasses of temperate Australia, a field guide. Inkata priorities. Grass Forage Sci. 71 (1), 1–35. https://doi.org/10.1111/gfs.12212. Press, Melbourne, pp. 194–197. POWO (Plants of the World Online), 2019. Facilitated by the Royal Botanic Gardens, Lattimore, M.A., McCormick, L., 2012. Pasture varieties used in New South Wales 2012- Kew. Published on the Internet; http://www.plantsoftheworldonline.org/ (Accessed 13. Published jointly by NSW Department of Primary Industries and the Grassland 2 September 2022). Society of NSW Inc, New South Wales, New South Wales. Psyllos, G., Hadjigeorgiou, I., Dimitrakopoulos, P.G., Kizos, T., 2022. Grazing land Li, L., Tilman, D., Lambers, H., Zhang, F.S., 2014. Plant diversity and overyielding: productivity, floral diversity, and management in a semi-arid Mediterranean insights from belowground facilitation of intercropping in agriculture. N. Phytol. landscape. Sustainability 14 (8), 4623. https://doi.org/10.3390/su14084623. 203 (1), 63–69. https://doi.org/10.1111/nph.12778. Pujol, M., 1984. Producción de forrajes en Cataluña. Pastos 14 (1), 1–34. Lithourgidis, A.S., Dordas, C.A., Damalas, C.A., Vlachostergios, D., 2011. Annual Ramseier, D., Connolly, J., Bazzaz, F., 2005. Carbon dioxide regime, species identity and intercrops: an alternative pathway for sustainable agriculture. Aust. J. Crop. Sci. 5 influence of species initial abundance as determinants of change in stand biomass (4), 396–410. https://doi.org/10.3316/informit.281409060336481. composition in five-species communities: an investigation using a simplex design Lloveras, J., 1999. El cultivo de la alfalfa y su relación con el Medio Ambiente. Pastos 19 and RGRD analysis. J. Ecol. 93, 502–511 https://doi.org/10.111 1/j.1365- (2), 145–167. 2745.2005.00999.x. Lüscher, A., Barkaoui, K., Finn, J.A., Suter, D., Suter, M., Volaire, F., 2022. Using plant Ribas, A., Llurba, R., Gouriveau, F., Altimir, N., Connolly, J., Sebastià, M.T., 2015. Plant diversity to reduce vulnerability and increase drought resilience of permanent and identity and evenness affect yield and trace gas exchanges in forage mixtures. Plant sown productive grasslands. Grass Forage Sci. 77 (4), 235–246. https://doi.org/ Soil 391, 93–108. https://doi.org/10.1007/s11104-015-2407-7. 10.1111/gfs.12578. Sebastià, M.T., Palero, N., De Bello, F., 2011. Changes in management modify agro- McIntire, E.J.B., Fajardo, A., 2014. Facilitation as a ubiquitous driver of biodiversity. diversity in sainfoin swards in the Eastern Pyrenees. Agron. Sustain. Dev. 31, N. Phytol. 201, 403–416. https://doi.org/10.1111/nph.12478. 533–540. https://doi.org/10.1007/s13593-011-0008-2. McKenna, P., Cannon, N., Conway, J., Dooley, J., 2018. The use of red clover (Trifolium Sunil, S., Akshit, R., Sheoran, R.S., Satpal, S., Harender, D., Deepak, L., Sushil, K., Paras, pratense) in soil fertility-building: a review. Field Crops Res 221, 38–49. https://doi. 2020. Impact of climate change on forage and pasture production and strategies for org/10.1016/j.fcr.2018.02.006. its mitigation -a review. Forage Res 46, 105–113. Montserrat, P., Capdevila, M., 1964. La esparceta en el nordeste español. V Reunión Suter, M., Huguenin-Elie, O., Lüscher, A., 2021. Multispecies for multifunctions: Cient. De la S.E.E.P. Jaca-Pamplona 131–138. combining four complementary species enhances multifunctionality of sown Montserrat, P., Fillat, F., 1984. La esparceta en los climas continentales pirenaicos. Pastos grassland. Sci. Rep. 11 (1), 1–16. https://doi.org/10.1038/s41598-021-82162-y. 14 (1), 133–141. Szumigalski, A., Van Acker, R., 2006. The agronomic value of annual plant diversity in Mora-Ortiz, M., Smith, L.M.J., 2018. Onobrychis viciifolia; a comprehensive literature crop-weed systems. Can. J. Plant. Sci. 86 (3), 865–874. https://doi.org/10.4141/ review of its history, etymology, taxonomy, genetics, agronomy and botany. Plant P05-074. Genet. Resour. 16 (5), 403–418. https://doi.org/10.1017/S1479262118000230. Tilman, D., Cassman, K.G., Matson, P.A., Naylor, R., Polasky, S., 2002. Agricultural Mueller-Harvey, I., Bee, G., Dohme-Meier, F., Hoste, H., Karonen, M., Kölliker, R., sustainability and intensive production practices. Nature 418, 671–677. https://doi. Lüscher, A., Niderkorn, V., Pellikaan, W., Salminen, J.-P., Skøt, L., Smith, L., org/10.1038/nature01014. Thamsborg, S., Totterdell, P., Wilkinson, I., Williams, A., Azuhnwi, B., Baert, N., Tilman, D., Isbell, F., Cowles, J., 2014. Biodiversity and ecosystem functioning. Annu. Grosse Brinkhaus, A., Copani, G., Desrues, O., Drake, C., Engström, M., Fryganas, C., Rev. Ecol. Evol. Syst. 45, 471–493. https://doi.org/10.1146/annurev-ecolsys- Girard, M., Huyen, N., Kempf, K., Malisch, C., Mora-Ortiz, M., Quijada, J., 120213-091917. Ramsay, A., Ropiak, H., Waghorn, G., 2019. Benefits of condensed tannins in forage Trenbath, B., 1974. Biomass productivity of mixtures. Adv. Agron. 26, 177–210. https:// legumes fed to ruminants: importance of structure, concentration, and diet doi.org/10.1016/S0065-2113(08)60871-8. composition. Crop Sci. 59 (3), 861–885. https://doi.org/10.2135/ UPNA, 2019. Flora Pratense y Forrajera Cultivada de la Península Ibérica. Herbario de la cropsci2017.06.0369. Universidad Pública de Navarra. https://www.unavarra.es/herbario/pratenses/ Neal, J.S., Fulkerson, W.J., Lawrie, R., Barchia, I.M., 2009. Difference in yield and htm/inicio.htm (Accessed 23 July). persistence among perennial forages used by the dairy industry under optimum and Van Ruijven, J., De Deyn, G., Berendse, F., 2003. Diversity reduces invasibility in deficit irrigation. Crop Pasture Sci. 60 (11), 1071–1087. https://doi.org/10.1071/ experimental plant communities: the role of plant species. Ecol. Lett. 6 (10), CP09059. 910–918. https://doi.org/10.1046/j.1461-0248.2003.00516.x. Niggli, U., Fließbach, A., Hepperly, P. and Scialabba, N., 2009. Low Greenhouse Gas Vandermeer, J.H., 1989. The Ecology of Intercropping. Cambridge University Press, Agriculture: Mitigation and Adaptation Potential of Sustainable Farming Systems. Cambridge, UK. https://doi.org/10.1017/CBO9780511623523. FAO, April 2009, Rev. 2 – 2009. Volaire, F., Norton, M.R., Lelièvre, F., 2009. Summer drought survival strategies and Nyfeler, D., Huguenin-Elie, O., Suter, M., Frossard, E., Connolly, J., Lüscher, A., 2009. sustainability of perennial temperate forage grasses in Mediterranean areas. Crop Strong mixture effects among four species in fertilized agricultural grassland led to Sci. 49, 2386–2392. https://doi.org/10.2135/cropsci2009.06.0317. persistent and consistent transgressive overyielding. J. Appl. Ecol. 46, 683–691. Volaire, F., Barkaoui, K., Norton, M., 2014. Designing resilient and sustainable grasslands https://doi.org/10.1111/j.1365-2664.2009.01653.x. for a drier future: adaptive strategies, functional traits and biotic interactions. Eur. J. Nyfeler, D., Huguenin-Elie, O., Suter, M., Frossard, E., Lu¨scher, A., 2011. Grass-legume Agron. 52, 81–89. https://doi.org/10.1016/j.eja.2013.10.002. mixtures can yield more nitrogen than legume pure stands due to mutual stimulation Weisser, W.W., Roscher, C., Meyer, S.T., Ebeling, A., Luo, G., Allan, E., Beßler, H., of nitrogen uptake from symbiotic and non-symbiotic sources. Agric. Ecosyst. Barnard, R.L., Buchmann, N., Buscot, F., Engels, C., Fischer, C., Fischer, M., Environ. 140, 155–163. https://doi.org/10.1016/j.agee.2010.11.022. Gessler, A., Gleixner, G., Halle, S., Hildebrandt, A., Hillebrand, H., de Kroon, H., Olesen, J.E., Bindi, M., 2004. Agricultural impacts and adaptations to climate change in Lange, M., Leimer, S., Le Roux, X., Milcu, A., Mommer, L., Niklaus, P.A., Europe. Farm Policy J. 1 (3), 36–46. Oelmann, Y., Proulx, R., Roy, J., Scherber, C., Scherer-Lorenzen, M., Scheu, S., Pappa, V.A., Rees, R.M., Walker, R.L., Baddeley, J.A., Watson, C.A., 2011. Nitrous oxide Tscharntke, T., Wachendorf, M., Wagg, C., Weigelt, A., Wilcke, W., Wirth, C., emissions and nitrate leaching in an arable rotation resulting from the presence of an Schulze, E.D., Schmid, B., Eisenhauer, N., 2017. Biodiversity effects on ecosystem intercrop. Agric. Ecosyst. Environ. 141, 153–161. https://doi.org/10.1016/j. functioning in a 15-year grassland experiment: patterns, mechanisms, and open agee.2011.02.025. questions. Basic Appl. Ecol. 23, 1–73. https://doi.org/10.1016/j.baae.2017.06.002. Peeters, A., Parente, G., Le Gall, A., 2006. Temperate legumes: key-species for sustainable Yu, Y., Stomph, T.J., Makowski, D., van der Werf, W., 2015. Temporal niche temperate mixtures. Grassl. Sci. Eur. Vol. 11. http://hdl.handle.net/10532/3679. differentiation increases the land equivalent ratio of annual intercrops: a meta- analysis. Field Crops Res 184, 133–144. https://doi.org/10.1016/j.fcr.2015.09.010. 11