INTRODUCTION
Soil carbon is highly relevant for climate change mitigation (Canadellet al. 2007) because it constitutes the largest carbon pool in terrestrial ecosystems (Batjes 1996). However, there is still much uncertainty regarding the relationship between soil carbon storage and ecosystem biota, including the role of biodiversity patterns and human activities (Schulze 2006; Soussana & Lemaire 2014; Ali & Yan 2017). According to current conceptual models, ecosystem goods and services -counting soil carbon sequestration- are expected to depend strongly on biodiversity components (Hooper et al. 2005; Diaz et al.2007; Hector & Bagchi 2007; Kirwan et al. 2009; Suter et al. 2015; Connolly et al. 2018). However, most evidence comes from diversity function experiments, which might suggest important underlying mechanisms (Fornara & Tilman 2008; Prommer et al.2019), but do not encompass the whole complexity encountered in real ecosystems. Although small-scale experiments indicate that the relationship between biodiversity and several ecosystem services might exist (Hector et al. 1999; Ribas et al. 2015), evidence about a positive association between soil carbon storage and plant diversity in nature is scarce (Hollingsworth et al. 2008; Wardle 2016).
In recent years, attempts have been made to clarify how plant taxonomic and functional diversity drive soil carbon at broad scales (Manninget al. 2015; Chen et al. 2018). In addition, plant functional types (PFT, Steneck, 2001; Blondel, 2003) were also found relevant for explaining soil carbon content in experiments (Fornara & Tilman 2008; Lange et al. 2015) and local studies (Wiesmeieret al. 2019).
There is an unsolved discussion about the relevance of plant functional types versus plant traits (or functional diversity) on ecosystem function (Lavorel & Garnier 2002; Ricotta et al. 2016). Furthermore, some authors postulate that ecosystem function depends on the dominant plant species or functional type (Grime 1977) (e.g., grasses in grassland ecosystems (Strömberg & Strömberg 2011). However, several studies indicate an important role of PFT diversity and non-dominant plant functional types on grassland functioning (Debouket al. 2015), including legumes (Spehn et al. 2002), and even in the absence of legumes (Cong et al., 2014).
Plant functional types can be described as guilds when defined in terms of resource use (Sebastià 2007) sensu Root (Root 1967), for animals). Furthermore, resource use can also be considered a relevant plant functional trait. Although every plant species can be considered to occupy a specific and unique point in the multivariate space of functional traits (Wayne & Bazzaz 1991), and therefore their role in the ecosystem to be unique, plant guilds provide good summary representations for biodiversity-function analysis. Grasses are efficient in terms of light capture because their leaves are at vertical angles (Sebastià 2007), while the architecture of their roots made them efficient capturing soil N. Legumes can have access to symbiotically-fixed atmospheric N. Non-legume forbs present a variety of ports and (flat to obtuse) leaf angles, and cannot fix atmospheric N (Canals & Sebastià 2000; Sebastià 2007; Sebastià & Puig 2008). However, recently, legumes were found to be more active in terms of CO2 exchange per biomass unit compared with other guilds, including highly dominant grasses, and non-legume forbs (Ibañezet al. 2020).
Plant guild effects on soil carbon storage need to be studied at broad scales to understand how they work independently of broad scale abiotic variables, as it is being done for plant taxonomical and functional diversity (Manning et al. 2015; Carol Adair et al. 2018). Ecologists and modelers need this information to validate conceptual paradigms and generate new hypotheses contributing to the refinement of global mechanistic models. Land managers and policy-makers need it to establish priorities for conservation objectives.
Here, we aim to disentangle plant guild effects on soil organic carbon (SOC) in the Pyrenees, by modelling data from an extensive database generated from a survey of 98 natural grasslands. The survey included a variety of climates; different landscape positions; and a range of grazing management regimes. We applied the diversity-interaction modelling (DIM) approach (Kirwan et al. 2009), which allowed us to separate the identity and interaction effects of plant guilds. Taking into consideration that previous studies, carried out at narrower scales, found a crucial but contrasting role of legumes within the plant guilds (p.e. Fornara & Tilman, 2008b; Lange et al., 2014), we aim at answering the following questions:
Do the effects of plant guilds on SOC in natural grasslands mirror those found in experimental systems?
Are enhancement effects of legumes on SOC stable across the range of proportions commonly found in natural grasslands (10-50%)?
Do the effects of legume proportions depend on other plant guild proportions, including forbs and grasses?