Introduction
The cessation of grazing is a common feature of the European landscape and is expected to rise sharply over the next decade (Eurostat 2018), especially in low-productivity, mountainous areas where previously extensively grazed lands are increasingly being taken out of agricultural production (MacDonald et al. 2000; Cramer et al. 2008; Kuemmerle et al. 2016; Lasanta et al. 2017; Eurostat 2018). Extensively managed, semi-natural grasslands are widespread across Europe, often grazed since Roman or even pre-Roman times (Prins 1998; Hejcman et al. 2013), and support an important component of regional biodiversity, delivering multiple ecosystem functions and services (Hector et al. 1999; Schröter et al. 2005; Hautier et al. 2018). This has resulted in grassland ecosystems with spatially heterogeneous vegetation (Adler et al.2001). Based on studies focused on plants, there is widespread concern that the cessation of grazing in these ecosystems is causing biotic homogenization due to a loss of rare specialist species and an increase in common generalists, as well as overall declines in plant biodiversity (McKinney & Lockwood 1999; Olden et al. 2004; Clavel et al. 2011; Newman et al. 2014; Epelde et al. 2017; Oggioniet al. 2020). Further, biotic homogenization and associated loss of biodiversity resulting from grazer exclusion is likely to impact ecosystem functioning (Plieninger et al. 2014; Johansen et al. 2019; Oggioni et al. 2020).
Despite the prevalence of grazer removal from historically grazed ecosystems, major uncertainties exist regarding its impact on biodiversity and ecosystem functioning. One particular uncertainty concerns its impact on different components of biodiversity, which have been observed to operate independently of each other. Species richness at the local, plot scale (ɑ-diversity) is likely driven by changes in land management (Socolar et al. 2016), whereas compositional (between plot) variation (β-diversity), which includes variation in the taxonomic composition of communities across sites, is driven by a range of factors that operate from small to larger scales (Anderson et al. 2011). Therefore, while ɑ-diversity may be stable or increasing in some areas, β-diversity could be decreasing due to biotic homogenization, i.e. communities from different sites become more similar in composition (Dornelas et al. 2014; McGill et al. 2015; Beauvais et al. 2016). There is mounting evidence that the cessation of livestock grazing influences these different attributes of biotic homogenisation of aboveground communities, including plants (Bühler & Roth 2011; Newman et al. 2014; Beauvais et al.2016) and insects (Carvalheiro et al. 2013; van Noordwijket al. 2017), but far less is known regarding the effects on communities of belowground organisms. Soil biodiversity regulates a number of key ecosystem functions and services, for instance organic matter decomposition, plant nutrient availability, nutrient leaching, and soil structural stability (Bais et al. 2006; Mendes et al. 2011; Berendsen et al. 2012; Philippot et al. 2013; Turner et al. 2013; Schrama & Bardgett 2016). While some studies have examined the effects of cessation of livestock grazing on belowground communities in grasslands (Heyde et al. 2017; Oggioniet al. 2020), these studies generally focussed on specific groups of soil organisms (but see (Bardgett et al. 1997; Bardgettet al. 2001; Epelde et al. 2017), short time spans since grazing removal (Epelde et al. 2017) or a narrow range of climate and soil conditions (Bardgett et al. 2001). Given this, there is a clear need for an improved understanding of the long-term impact of the cessation of grazing on the composition and diversity of belowground communities.
Here, we explore how cessation of grazing impacts ɑ- and β-diversity of both plants and belowground communities, by analysing resulting changes in vegetation and a wide range of soil faunal and microbial groups. We used a series of 12 montane grassland sites positioned along an 800-km north-south gradient of the United Kingdom, and covering several of the UK’s main montane grassland regions, each with several paired plots that were either subject to historical grazing by sheep or had livestock grazers excluded by fencing for 10-65 years. We focussed on montane grasslands because they are a prominent feature of the European landscape and have been grazed by sheep for centuries, forming the backbone of the sheep farming industry across Europe (Rodwell 1990). Further, the cessation of livestock grazing is commonplace in mountain regions of Europe, including the United Kingdom, and is recognised as a key aspect of land abandonment (Eurostat 2018) and rewilding (Pereira & Navarro 2015), with the potential to have multiple, but largely unknown, effects on local diversity and compositional variation in belowground communities among sites (Figure 1; after (van Noordwijk et al.2017)).
Changes in ɑ- and β-diversity can occur simultaneously and have positive, neutral or negative relationships. As such, we tested a range of hypotheses, namely that: (1) increased ɑ-(plot-based) diversity occurs when cessation of grazing results in a higher degree of local environmental variation, and an increased availability of niches supports more species (scenario C, E and H in Figure1); (2) decreased ɑ-diversity happens when grazer exclusion results in a reduction in local environmental heterogeneity, causing loss of rare species and/or gain of generalist species (scenario A, D and F in Figure 1); (3) a decrease in β-(site-based) diversity may occur independently of changes in ɑ-diversity, which is expected when the removal of grazing has a homogenizing effect on the composition of local communities within a given area, independent of the effect on local soil community species richness (scenarios A, B and C in Figure 1); (4) if current land management causes strong homogenization of soil communities, we expect that cessation of grazing will increase β-diversity through a gradual divergence of communities, which may happen independently of changes in ɑ-diversity by changing communities in various directions through differential species losses and gains (scenario F, G and H in Figure 1). We tested these hypotheses using our unique dataset of different components of the belowground food web from long term paired grazed and ungrazed exclosures across the United Kingdom.