Herbivore-Plant-Soil Feedbacks: Their Role on Plant Resource Partitioning and the Organization and Functioning of a Grazing Ecosystem

Herbivore-Plant-Soil Feedbacks: Their Role on Plant Resource Partitioning and the Organization and Functioning of a Grazing Ecosystem Eduardo Medina-Rold´an Lancaster Environment Centre Lancaster University A thesis submitted for the degree of Doctor of Philosophy September 2011 Declaration I herewith declare that this work has been originally produced by myself without having other external asistance than that specified in the chapters of the document. This work has not previously been presented in order to obtain a similar degree in any form. Lancaster, United Kingdom. September 2011. ii Abstract The aim of this thesis is to investigate how different land management practices, including grazing by large herbivores and herbivores exclusion, affect belowground biological properties related to soil nutrient and carbon (C) cycling in nutrient-poor soils of British upland semi-natural grassland and moorland. A field (Chapter 2) and microcosm experiments (Chapter 3) were carried out to test how defoliation and grazing by herbivores affect key soil properties and ecosystem services through their effects on plant functional traits and the composition of the vegetation, respectively. Furthermore, indirect effects of herbivores (through their effect on soil properties) on competition of dominant upland graminoid were assessed (Chapter 4), including those related to plant competition for different chemical forms of nitrogen (N) (Chapter 5). Seasonal comparisons between acidic nutrient-poor grasslands grazed by large herbivores and herbivore-excluded areas (re-wilding) showed that grazing had similar effects to those observed in the microcosms experiment, increasing N mineralization rates and soil microbial activity. Such differences in soil properties between grazed and re-wilding areas were related to changes in plant functional groups, with re-wilding increasing the abundance of dwarf-shrubs which are known to slow-down nutrient cycling. Additionally, it was found that defoliation had consistent effects on soil properties, iii decreasing soil microbial biomass C, and increasing soil N availability and rates of N mineralization. These general effects of defoliation on soil properties were independent of how different grass species varying in life-history traits responded to defoliation, suggesting that traits in this functional group (graminoids) are of less importance to mediate the effects of defoliation on soil properties in the nutrient-poor soils utilised. On the other hand, differences in soil properties driven by contrasting grazing management affected plant competition between Eriophorum vaginatum (abundant in long-term ungrazed areas) and Nardus stricta (dominant in long-term grazed areas), so that the latter increased its negative effects on the former when grown in the soil coming form the grazed grassland. In contrast, the competitive effect of E. vaginatum on N. stricta was reduced in the grazed soils. This suggests the existence of a herbivore-plant-soil feedback mediated by soil microbes affecting plant competition. It was found that indirect effects of grazing on plant competition in these species were not explained by how different soil microbial communities affected competition for different forms of N (including dissolved inorganic N and dissolved organic N), but rather for the competitive traits of the superior competitor and the stimulating effects of grazing on N cycling. Overall, results of this work show the importance of herbivores in driving soil process though their effects on plant composition and soil microbes. More research is necessary to evaluate whether effects of grazing on nutrient cycling can have negative effects on other ecosystem services such as soil C sequestration. Effects of grazing on soil properties can, in turn, affect competition of plant species, which iv showed differential abundances along gradients of grazing and soil physical properties. Whereas the effect of herbivore-plant-soil feedbacks on this plant species coexistence might be less important that direct effects of herbivores, my results suggest possible mechanisms by which plant species coexist in these poor-nutrient ecosystems. v To Greta and Martha. “Believe those who are seeking the truth. Doubt those who found it” Andr´e Gide. vi Acknowledgements This thesis work was funded by a doctoral scholarship granted by the National Council for Research and Technology of Mexico (CONACYT), of which I am deeply grateful. I am grateful to my supervisor, Professor Richard D. Bardgett for the experience and advice he shared with me. To Jorge Paz-Ferreiro for the time spent in the field and in the lab making science come true. I am thankful to Natural England and Colin Newlands for having granted access to the experimental field sites and provided information about their land use history. My gratitude to the UK Meteorological Office for the access to meteorological datasets (MIDAS) for the study area as well. To the members (current and present) of the Soil and Ecosystem Ecology Lab in Lancaster University, and people in the Centre for Ecology and Hydrology in Lancaster which helped me out with field or lab work, engaged with me in deep and abstruse scientific discussions, assisted me to clarify technical details on how to survive a PhD being a Latin American in the UK (language issues included), or just shared time with me drinking a beer. They include: Helen Quirk, Simon Oakley, Gerlinde De Deyn, Nick Ostle, Sarah Dale, Dario Fornara, Melanie Hartley, Sue Ward, Franciska De Vries, Victor van Velzen, Catherine Turner, Iain Gould, Kate Orwin, Kate Harrison, Will Mallot, Anna Wilkinson and Helen Gordon. Outsiders include Ian Dodd et al., and Jos Barlow et al. vii To the people in OpenWetWare who maintain the website where the L A T EX template of this thesis was taken from. To all people who contribute to GNU-licensed software and similar projects such as R, Rlplot, Kile, GIMP, Inskape, Ubuntu, etc. In general, these people are barely mentioned heroes, but this thesis is full with open-source spirit. I am also grateful to friends on this and other sides of the ocean who supported me and nourished me (sometimes in indescribable and unspeakable ways). Following a convention of Cort´azar’s hopscotch, from This Side: Ricardo Gallego, Felipe Otondo, Alberto Cortijo, Rosario Ponce, Arthur, Xico, Miriam, Rosal´ıa, Sof´ıa, and Maca. From the Other Side: Griselda Quiroz, Gabriela Uribe, Leonardo Hern´andez, and Abel Hern´andez. A special mention goes to Raymundo Ram´ırez-Delgadillo. Ray, the demons of entropy decided you are not among us any longer, so this thesis is in memoriam. To my beloved wife Greta who has a good share of this work. viii Contents List of Figures xiii List of Tables xix 1 General Introduction 1 1.1 Introduction................................... 2 1.2 Background................................... 6 1.2.1 Herbivore-mediated changes on ecosystem functioning: The abovebelowground perspective . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.2 Herbivore-plant soil systems: Feedback on the plant community and plant species coexistence . . . . . . . . . . . . . . . . . . . . . 10 1.2.3 Dissolved organic nitrogen: Their role in low productivity habitats, plant nutrition and plant species coexistence . . . . . . . . . 11 1.2.4 Extensive grazing systems: Semi-natural mountain grasslands and the United Kingdom study case . . . . . . . . . . . . . . . . . . . . 15 1.2.5 Background summary . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.3 Thesis layout: Aims and hypotheses . . . . . . . . . . . . . . . . . . . . . 19 2 Influence of re-wilding practices versus sheep grazing on aboveground and belowground properties of upland grasslands 23 2.1 Abstract..................................... 24 2.2 Keywords .................................... 25 2.3 Introduction................................... 25 2.4 MaterialandMethods ............................. 27 2.4.1 Site description and experimental design . . . . . . . . . . . . . . . 27 2.4.2 Soil and vegetation sampling . . . . . . . . . . . . . . . . . . . . . 29 2.4.3 Aboveand belowground biomass, and size of litter horizons . . . . 30 2.4.4 Soilanalysis............................... 30 ix LIST OF FIGURES 4.1 Comparative indirect effects of grazing and direct effects of competition on: a) total shoot biomass, b) crown biomass, c) root biomass, d) shoot: belowground (crown+root) ratio of Eriophorum vaginatum (Ev) and Nardus stricta (Ns). Plants were grown in a glasshouse microcosm experiment with soil coming from a grazed Nardus-dominated acidic upland grassland (G+) or a Eriophorum-dominated ungrazed area (G−) in the Yorkshire Dales, England. E. vaginatum plants grew either in monocoltures (C−) or with inter-specific competition by N. stricta (C+). N. stricta plants grew either in monocoltures (C−) or with inter-specific copetition by E. vaginatum (C+). ................................ 83 4.2 Comparative indirect effects of grazing and direct effects of competition on biomass partitioning of a) Eriophorum vaginatum, b) Nardus stricta. AlllegendsasinFig.4.1............................ 85 4.3 Comparative indirect effects of grazing on competition of: a,b) Eriophorum vaginatum; c,d) Nardus stricta. Surface responses of plant individual biomass data fitted with the hyperbolic competition model (see equation 4.1) which is explained in the text (parameter estimates are presented in Table 4.2). Legends as in Fig. 4.1 . . . . . . . . . . . . . . . . . . . . . . . 87 4.4 Comparison in properties between grazed and ungrazed soils from upland ecosystems and the effect of plant species on these properties: a) soil microbial biomass C, b) microbial biomass C:N ratio, c) soil microbial biomass N, d) soil basal respiration, e) soil NH+ 4concentration, f) soil NO− 3concentration. Soil was planted with E. vaginatum plants in monocoltures (Ev), N. stricta plants in monocoltures (Ns), both species under competition in a range of plant densities (Both), and soil without plants at all (None). Values are means ±standard errors. All other legends as inFig.4.1.................................... 90 5.1 Effect of microbial inoculum from a grazed Nardus-dominated acidic seminatural grassland (G+) and a Eriophorum-dominated ungrazed grassland (G−), and inter-specific competition on: (a) E. vaginatum shoot biomass; (b) N. stricta shoot biomass; (c) E. vaginatum root biomass; and (d)N. stricta root biomass. Data show the effects of inter-specific competition when both plants were grown without (C−) or with (C+) inter-specific competition. Gray bars (NA) indicate the effect of competition on each species biomass component on non-inoculated soils. Values are means ±se.109 xvi LIST OF FIGURES 5.2 Regression analysis and parameter estimates of log transformed data of 15N against 13C mol excess for shoot (a) and root (b) biomass of plants treated with dual-labelled (13C and 15N) glycine (Gly) + unlabelled NH+ 4. 111 5.3 Effects of microbial inoculum source (G+vs G−) and inter-specific competition on Eriophorum vaginatum 15N and 13C isotopic enrichment (expressed as molar excess) of: (a-d) shoot; and (e-h) root biomass. Data show the effects of inter-specific competition when plants were grown without (left panels) or with inter-specific competition (right panels) after applying glycine-ammonium solutions with isotopic dual-labelled (13C and 15N) glycine (Gly), 15N-labelled NH+ 4(Din) or both compounds unlabelled (None). Values are means ±se. Legends as in Fig. 5.1. . . . . . . . . . . 112 5.4 Effects of microbial inoculum source (G+vs G−) and inter-specific competition on Nardus stricta 15N and 13C isotopic enrichment (expressed as molar excess) of: (a-d) shoot; and (e-h) root biomass. Data show the effects of inter-specific competition when plants were grown without (left panels) or with (right panels) inter-specific competition after applying glycine-ammonium solutions with isotopic dual-labelled (13C and 15N) glycine (Gly), 15N-labelled NH+ 4(Din) or both compounds unlabelled (None). Values are means ±se. Legends as in Fig. 5.1. . . . . . . . . . . 113 5.5 Effects of inter-specific competition on proportional N uptake (based on shoot 15N mol excess) of glycine (Gly) vs NH+ 4(Din) for E. vaginatum (a,b) and N. stricta (c,d). Left and right panels correspond to plants without or with inter-specific competition respectively . . . . . . . . . . . 114 5.6 Effects of microbial inoculum source (G+vs G−) and pot-type (a proxy for plant species) on: (a) soil microbial biomass C; (b) soil microbial biomass N; and (c) microbial biomass C:N ratio. Soils were planted with E. vaginatum plants in monocultures (Ev), N. stricta plants in monocultures (Ns), and both species under competition (Both). Values are means ±se. Legends as in Fig. 5.1. . . . . . . . . . . . . . . . . . . . . . . . . . 116 xvii LIST OF FIGURES A.1 Principal component analysis based on the correlation matrix of transformed plant variables (shoot biomass, root biomass, root length and root: shoot ratio). The first and second eigenvalues explained 78 and 20 % of the variance respectively. The Box-plots indicate data distribution. Top, sampling units identified by defoliation treatment. Bottom, sampling units identified by species. PC1 = first principal component, PC2 = second principal component. Nardus= N. stricta, Festuca = F. rubra, Poa = P. pratensis, Holcus = H. lanatus, Anthoxanthum = A. odoratum, Agrosostis = A. capillaris, Lolium = L. perenne. CN = control, MC = two-weekly defoliation, HC = one-weekly defoliation . . . . . . . . . . . . 136 B.1 Indirect effects of grazing and direct effects of density of heterospecifics on: a) total shoot biomass, b) crown biomass, c) root biomass of Eriophorum vaginatum (Ev) and Nardus stricta (Ns). Plants were grown in a glasshouse microcosm experiment with soil coming from a grazed Nardusdominated acidic upland grassland (G+) or an Eriophorum-dominated ungrazed area (G−) in the Yorkshire Dales, England. E. vaginatum plants grew in a range of densities of N. stricta shown in the horizontal axis. N. stricta plants grew in a range of densities of E. vaginatum..........138 xviii List of Tables 3.1 Some ecological traits of the species selected for the experimental study. Rmax= maximum relative growth rate (week−1), Rmean= mean relative growth rate (week−1). See references for details. ..................................... 54 4.1 Differences in selected soil properties between soil-sand mixtures from a grazed (G+) versus ungrazed (G−) seminatural upland ecosystems. Cmic and Nmic = soil C and N in microbial biomass (mg kg dry soil−1) respectively. NH+ 4av, NH+ 4min, NO− 3av, NO− 3min are ammonium and nitrate measures where av = soil extractable concentration (mg kg dry soil−1) and min = net potential mineralization rates (mg kg dry soil−1day−1) respectively. Csoil and Nsoil = total soil C and N (%) respectively. Analytical techniques are described in the methods section except for total soil C and N (ground soil, oven-dried at 105 ◦C, and analysed in a Vario EL Elemental Analiser, Elementar Inc, Germany), and net mineralization rates described in Harrison and Bardgett (2010). Values are means (se). ............. 77 xix LIST OF TABLES 4.2 Indirect effects of grazing (G+) versus herbivore exclusion (G−) on intra-specific (αi) and inter-specific (αij and j) competition between E vaginatum and N. stricta as estimated by data from a microcosm experiment fitted to the hyperbolic competition model (see equation 4.1 for more details). Values are parameter estimates, their standard errors (Std. error) and values of tand associated probability (P) after fitting the model. G+/G−stands for the proportional effect of grazing and herbivore exclusion, so that G+/G−>1 indicates grazing increasing competition intensity, or <1 decreasing it. ....................... 88 5.1 Differences in selected soil properties between air-dried-autoclaved (S) versus air-dried-non-autoclaved (NS) soil-sand mixtures. Cmic and Nmic =soil C and N in microbial biomass (mg kg dry soil−1) respectively. DOC and DON are dissolved organic C and N respectively (mg kg dry soil−1). Basresp = soil microbial basal respiration (µL CO2kg dry soil−1h−1). |t|=absolute value for the Welch Two Sample t-test and its associated probability (P). n= 3. Values are means (se). .......................103 xx Chapter 1 General Introduction 1 1. General Introduction 1.1 Introduction Ungulate herbivores are almost ubiquitous in terrestrial ecosystems where they have a key role in regulating vegetation dynamics and ecosystem functioning (McNaughton, 1985, 1986; Milchunas et al., 1988; McNaughton et al., 1988; Milchunas and Lauenroth, 1993; Bardgett and Wardle, 2003, 2010). The livestock production systems based on domestic relatives of these animals constitute an important industry, both in terms of its economic and environmental impacts worldwide (Fig.1.1). The global livestock population is estimated to be almost 2.5 billion head (only cattle and sheep, GLIPHA, 2011), with an increasing trend globally. In relation to this increasing trend in livestock numbers, Delgado et al. (2000) reported that from the 1970’s to the middle of the 1990’s, the consumption of livestock-related products in developing countries increased by 175 million metric tons, and the market value of such an increase was 155 billion dollars (1990 US dollars). Economic growth has been recognised as one of the main causes for the increased livestock numbers and the consequent higher inclusion of livestock products into people diets in a number of emergent economies (Smil, 2000). In many areas of the world, such increasing trends of consumption of livestock products translate into an incentive to increase livestock numbers or intensify production, thereby placing a great pressure on ecosystems and the services that they provide. Just to cite an example, livestock production was considered the main anthropogenic source of methane in the middle 1990’s, showing an increasing trend (Stern and Kaufmann, 1996). 2 1.1 Introduction Figure 1.1: Grazing density worldwide as modelled between the relationship of herds size and environmental variables. Taken from FAO (2011a) - 3 1.1 Introduction Whereas much of this production is carried out under intensive farming conditions, a non-negligible proportion of world livestock production is carried out under extensive, nomadic or less intensified regimes (Delgado, 1999). Extensive grazing regimes usually take place in low productivity ecosystems such as in the high mountains and grasslands of the world, including those of upland temperate regions of the United Kingdom (Fig. 1.2). The study of factors affecting these areas is important since such ecosystems are highly vulnerable to disturbances (Holden et al., 2007), yet they have a high conservation value given the assemblage of species and ecosystem services provided by them, such as regulation of water quality and soil carbon (C) storage (Holden, 2009; Worrall and Evans, 2009). Therefore, the aim of this thesis is to examine direct and indirect effects of grazing by domestic herbivores on ecosystem functioning and structure in semi-natural mountain grasslands in the United Kingdom. Given the contemporary interests in linking abovebelowground processes (Wardle et al., 2004; van der Putten et al., 2009; Bardgett and Wardle, 2010), I focus on direct effects of herbivores on soil properties and processes, and indirect effects via interactions between key plant species and soil using a model upland ecosystem where contrasting grazing management has taken place. In the next subsections of this introduction, I offer background material related to the main research topics covered in this thesis. First, I describe the main mechanisms by which large ungulate herbivores affect soil properties and processes and how, by doing so, herbivores can potentially affect plant-plant interactions through the operation of herbivore-plant-soil feedbacks. Then, I consider dissolved organic nitrogen (DON) as an N source for plant nutrition and its potential role in regulating plant-plant interactions in grazed grassland. I also outline the semi-natural mountain ecosystems which most 4 1.1 Introduction Figure 1.2: Grazing production systems in the world. The system is based on FAO and the International Livestock Research Institute classification. LGT in the legend corresponds to production systems in tropical and temperate highlands and uplands as those found through the United Kingdom uplands. Taken from FAO (2011b) - 5 1.2 Background Figure 1.4: Diagrammatic illustration of plant soil feedbacks. A plant species A conditions the soil where it grows (SA). In turn, SAsoil affects the performance of A (αA) and another plant species B (αB). The same for plant species B, the soil it conditions (SB), and the effect of SBon A (βA) and B (βB). Depending on the effects of each soil on each plant species, species competition (CA,CB) and competitive exclusion can be intensified or reduced. Taken from Bever (2003). - 12 1.2 Background by the balance between processes of consumption and production (Kalbitz et al., 2000; Chen and Xu, 2006, Fig. 1.5). Main production processes include: 1) decomposition of soil organic matter, plant residues and rhizodeposits carried out by soil microorganisms (Schimel and Weintraub, 2003; Schimel and Bennett, 2004); 2) microbial biomass turnover (Schimel and Weintraub, 2003; Chen and Xu, 2006); and 3) direct deposition (through wet deposition or throughfall, Chen and Xu, 2006). On the other hand, the main consumption processes are 1) direct uptake by microbes (including mycorrhizae) and plants (Barraclough, 1997; Schimel and Bennett, 2004); 2) ecosystem loss trough leaching (Harriman et al., 1998; Cleveland et al., 2004); and 3) stabilization with organic matter or the soil mineral phase through a number of physicochemical processes (Jones and Hodge, 1999). Since factors regulating DON production-consumption processes vary hugely across environmental gradients, one would expect to find differences in the size of this N pool among different ecosystem types. Among these differences, it has been suggested that low productivity ecosystems should have a greater proportion of this N pool relative to DIN (Christou et al., 2005; Farrell et al., 2011), perhaps as a result of conditions limiting microbial activity and N mineralization (Schimel and Bennett, 2004, but see Jones and Hodge 1999; O’Dowd et al. 1999; Jones and Kielland 2002). Evidence showing that terrestrial plants can directly take up several DON forms is accumulating (for a detailed list of studies see Lipson and N¨asholm, 2001; Weigelt et al., 2005; N¨asholm et al., 2009b). Not only can plants take up simple DON forms of low molecular weight (e.g., amino acids, amides, polypetides), it has been shown that plants can satisfy their N requirements exclusively from complex organic N sources (proteins) 13 1.2 Background Figure 1.5: Diagram showing main steps involved in dissolved organic N (SON in the figure) cycling in a forest ecosystem. a = decomposition, b = cell uptake, c= microbial release,d = microbial biomass turnover, e = root uptake, f = root exudation, g = humification, h = leaching, i = microbial predation. SIN = soil inorganic N. Taken from Chen and Xu (2006) - 14 1.2 Background without the assistance of microbial symbionts (Paungfoo-Lonhienne et al., 2008). Such evidence has made it clear that previous assumptions about how plants rely completely on release of dissolved inorganic N (DIN) - through N mineralization or microbial biomass turnover - (Schimel and Bennett, 2004) might not be valid. Utilization of DON by plants might have important ecological implications. In this way, it has been shown that plants differ in their preferences of different chemical forms of N (McKane et al., 2002; Harrison et al., 2007, 2008). For example, McKane et al. (2002) showed that the graminoid Eriophorum vaginatum utilized glycine and NH+ 4, whiche were the most abundant N forms, whereas the subordinate species Carex bigelowii preferred the least available form, namely NO− 3, in a tundra ecosystem in Alaska. This example, as with many others (e.g., Lipson et al., 1999; Bardgett et al., 2003), suggests that differential uptake of N chemical forms among plant species could be a mechanism regulating plant coexistence and plant-soil microbial interactions. Since different microbial communities can differ in their effects on a number of soil processes (Bardgett et al., 1996; Strickland et al., 2009), factors affecting microbial communities could affect plant species N uptake preferences, which is an aspect of plant-soil feedbacks still poorly explored (Ashton et al., 2008). 1.2.4 Extensive grazing systems: Semi-natural mountain grasslands and the United Kingdom study case The Cambridge Advanced Learner’s Dictionary (CD, 2011) defines upland as: “area of land that is situated high up, such as on a hill or mountain”. However, for environmental management purposes, the United Kingdom department of rural affairs (DEFRA, 2011b) equates the concept of upland with that of Less Favoured Areas as: “land which is 15 1.2 Background suitable for extensive livestock production but not, owing to the geography of the area, other agricultural production” (compare Figs. 1.2 1.6). In this thesis, I use the terms upland, semi-natural mountain grassland, and moorland interchangeably. In the United Kingdom, grazing as a management activity in the uplands has long historical roots (Simmons, 2003). However, post-war land use policies provided incentives for farmers to increase livestock numbers in upland habitats over the last part of the twentieth century (Simmons, 2003). For instance, trends of sheep numbers in the UK showed dramatic increases since the 1950s until 1990’s when herd sizes stabilised or decreased slightly (Fuller and Gough, 1999). This increase in sheep numbers has been put forward as one the main drivers of a large scale change from dwarf-shrub dominated moors into acidic grasslands across the British uplands (Bardgett et al., 1995). Direct effects of herbivores on vegetation composition in upland habitats are relatively well studied. Thus, it is thought that the mentioned replacement of moorland vegetation by grasses is driven by direct grazing effects on the competitive balance among plant species (Welch, 1984, 1986; Hartley and Amos, 1999; Hartley and Mitchell, 2005), although other environmental factors, such as soil moisture content and organic layer depth, might be involved (Ratcliffe, 1959; Edgell, 1971; Rodwell, 1992; Genney et al., 2002). A consequence of herbivory is that plant species of low palatability to domestic herbivores, such as Nardus stricta, might have a high invasibility and persistence over a large part of upland ecosystems as a result of the negative effects of herbivory on their competitors (Fenton, 1937; Rawes, 1961; Welch, 1986). Indirect effects of herbivores (those caused by herbivore-plant-soil feedbacks) can also influence the organization of upland habitats, but are far less understood and studied. 16 1.2 Background Figure 1.6: Geographical distribution of uplands in England. Taken from DEFRA (2011a). - 17 1.2 Background The recent introduction of the reform in the common agricultural policy within the European Union would certainly have a large effect on the way upland areas are managed (Condliffe, 2009). As this reform decouples subsidy payments from production, there might be an incentive for certain farmers to supply other ecosystem services (Oglethorpe, 2005), such as those related to soil C sequestration and the maintenance of supporting services such as soil nutrient cycling. This might imply reductions in livestock population in upland areas with concomitant changes in the structure and composition of the vegetation (referred as to re-wilding). Estimating the effects of re-wilding on ecosystem processes is therefore important to determine trade-offs in different ecosystem services related to different land use management options. 1.2.5 Background summary In summary, the impact of herbivores on plant community composition and functional traits can affect soil microbial communities and nutrient cycling. However, whether these changes occur across a range of plant functional traits is still unclear (Chapter 3). Secondly, contrasting effects of herbivores on soil nutrient cycling and soil microbial communities are expected in ecosystems with contrasting productivity, although more information is necessary on this subject. Thus, it is important to test the effects of herbivores on soil properties in low productivity ecosystems such as those found in the British uplands (Chapter 2). Furthermore, since changes in grazing management policies are expected in these ecosystems, it is also important to evaluate how grazing management can affect ecosystem services in these habitats (Chapter 2). Thirdly, plant soil feedbacks are key mechanisms in the organization of vegetation through their effects on 18 1.3 Thesis layout: Aims and hypotheses plant species performance and plantplant competitive interactions. However, whether plant-soil feedbacks can be caused by effects of herbivores on soil microbial communities and soil processes is poorly understood (Chapter 4). Finally, plants are able to take up different chemical forms of N, and plant species can differ in their preferences for DON or DIN. These plant species N preferences could have consequences for plant species coexistence and competition. However, if herbivore-plant-soil feedbacks are able to alter plant competition through N uptake preferences has been rarely addressed (Chapter 6). Next, the aims and hypotheses of this research are presented together with the organization of the chapters. 1.3 Thesis layout: Aims and hypotheses As stated above, this thesis aims to test effects of herbivores on soil properties and plantplant interactions in model upland ecosystem in northern England, where contrasting grazing management has taken place. To address this aim, a series of specific objectives were tested through a combination of field and glasshouse studies, as detailed below. •1) To determine the effects of different grazing management on soil properties in poor-nutrient semi-mountain ecosystems in northern England. •2) To determine the effects of defoliation on soil properties across a range of temperate grass species differing in functional traits. •3) To asses how the effects of grazing on soil properties (herbivore-plant-soil feedback) affect competition dynamics in E. vaginatum and N. stricta. •4) To link herbivore-plant-soil feedbacks effects on plant competition to effects of 19 1.3 Thesis layout: Aims and hypotheses soil microbes on N uptake preferences of E. vaginatum and N. stricta. Chapter 2 aimed to determine, in the field, the effects of grazing on soil properties and ecosystem services related to C and N cycling in semi-natural mountain grasslands in northern England. Since herbivores promote plant functional groups with traits related to higher litter quality than those occurring in ungrazed areas, it was hypothesised that grazing would cause higher rates of nutrient cycling. Soil biological properties between a continuously sheep-grazed acidic grassland dominated by N. stricta,Festuca spp., and Agrostis capillaris, and a 10-year fenced-off area (re-wilding) were contrasted, taking into account seasonal variability. It was observed that re-wilding changed the composition of functional plant groups, as well as pools of plant biomass. Herbivores promoted nutrient cycling, microbial activity and reduced the proportion of DON:DIN in comparison to re-wilding conditions, suggesting that grazing is associated with faster nutrient cycling. Despite these results, no differences in total soil C or N storage were detected between different grazing managements, suggesting that low levels of sheep grazing are compatible with soil C storage. Chapter 3 aims to to determine the effects of defoliation on soil properties across a range of temperate grass species differing in functional traits. In this chapter, a number of grass species from a range of life-story and trait strategies were subjected to different levels of defoliation to evaluate soil responses to plant defoliation. It was hypothesised that soil responses would mirror plant species responses. Defoliation caused consistent effects on all plant species reducing plant performance overall. Despite the wide range of life-history traits included, only subtle intra-specific responses to defoliation were observed, but they indicated differences in defoliation resistance among plant species. 20 1.3 Thesis layout: Aims and hypotheses Defoliation caused consistent effects on soil properties as well, such as decrease in soil microbial biomass C and increases in N availability. Again, despite the broad range of life history used, no consistent effects of defoliation were observed through its effects on plant species. It is concluded that, under certain circumstances such as poor nutrient soils, defoliation is an important mechanisms affecting soil properties, but independently from plant species identity to a large extent. Chapter 4 assessed how grazing effects on soil properties affect competition balance between E. vaginatum and N. stricta, two graminoids with contrasting abundances along grazing and soil properties gradients in semi-natural mountain grasslands. Based on differences on soil properties between grazed and re-wilding areas previously observed, it was hypothesised that species would perform better on the soils where they are dominant. Then, a microcosm experiment was set-up to test effects of grazing-induced plant-soil feedbacks on intraand inter-specific competition in these plant species. It was observed that soil from the grazed area increased intra-specific competition in both plant species, since this soil showed higher microbial activity and N mineralization rates. However, when it came to inter-specific competition, the grazed-conditioned soil increased the negative effect that the plant species dominating the grazed area (N. stricta) had on E. vaginatum. Grazed-conditioned soil also reduced to a half the negative effect of E. vaginatum on N. stricta. These grazing-mediated plant-soil feedbacks effects on intraand inter-specific competition had not been described yet, but they could contribute to the persistence of N. stricta in long-term grazed upland habitats and its invasibility in other terrestrial ecosystems. Chapter 5 examined how competitive interactions between E. vaginatum and N. stricta could be affected by herbivore-plant-soil feedback effects on plant N uptake pref21 2.4 Material and Methods Reference Number SD 763762). The climate is temperate maritime, with a mean annual precipitation of 1840 mm (averaged for 10 years, UK Meteorological Office 2010, see Fig. 2.6b). Several vegetation types are located within the nature reserve, but large areas are covered by sheep-grazed, acidic grassland dominated by the graminoids Nardus stricta L., Festuca ovina L., Agrostis capillaris L. and Eriophorum vaginatum L (Rodwell, 1992). Soils are derived from carboniferous sandstones in the Yoredale group (Waltham, 2008) with a pH of 4.5 (based on a triplicate measure in 1:2.5 soil to water suspensions w/v) and an organic surface horizon of 20-30 cm depth. This grassland type and its variants are widespread across western parts of upland Britain where it forms the mainstay vegetation type of the sheep farming industry (Rodwell, 1992). Our study commenced in 2007, when we selected 2 adjacent areas of moorland of similar topography and altitude with contrasting recent grazing management (Fig. 2.1). One of them (hereafter the re-wilding area) is a 170 ha area which was previously grazed by sheep at stocking rates of 1-3 ewes ha−1, but which was fenced off in 2000 to exclude livestock as part of a landscape-scale re-wilding experiment set-up to restore it into a heather dominated community (Colin Newlands, Natural England, pers. comm.). The second area (hereafter the grazed area) is an immediately adjacent 58 ha acidic grassland of similar topography and soils, which has had a very similar grazing history to the re-wilding area. However, since 1996 stocking rates have been maintained at 4 ewes ha−1, with some cattle grazing for 10 weeks from since mid-July onwards. The winter grazing regime in this area includes a maximum stocking rate of 1.5 ewes ha-1 from early November until mid-March. As far as we are aware, neither area has received artificial fertilisers, manures or supplementary feeding. Given that re-wilding was done at the landscape level, pseudo-replication was un28 2.4 Material and Methods avoidable in that similar areas were not available in the area. However, in order to minimize the sources of error associated with this problem, we used multiple, randomly located plots within the grazed and re-wilding areas. For this, 6 experimental plots, each 4 X 4 m, were randomly positioned along a NW-SE transect within each of the experimental areas. Individual sampling plots within the re-wilding and grazed areas were of similar topography with an average soil depth of 40 cm, and were representative of vegetation and soil conditions at the landscape-scale on both sites (Fig. 2.1). Once the experimental plots were established, each replicated experimental plot was split into 16 1 X 1 m sub-plots wherein all soil and vegetation samples were randomly taken. 2.4.2 Soil and vegetation sampling We sampled soil and vegetation 5 times from mid spring 2007 until early summer 2008 in order to capture seasonal variability in responses of soil properties to re-wilding. On each sampling date, a turf that included vegetation and the organic peat horizon (20 X 20 cm area, 20 cm depth) was excavated from each experimental sub-plot on each replicated plot by using a shovel. Samples were kept immediately within coolers after sampled and transported back to the laboratory where they were then stored at 4 ◦C until the analytical procedures took place within 5 days after collection. Additionally, root biomass estimation was determined from two soil cores taken to a depth of 10 cm within the organic horizon on each sup-plot and sampling date. We inserted perforated PVC pipes (21 mm diameter X 1000 mm length) on each experimental plot in order to record the water table depth (Oechel et al., 1998) on approximately two-weekly intervals from mid July until mid November 2007. 29 2.4 Material and Methods 2.4.3 Aboveand belowground biomass, and size of litter horizons Dry mass pools in the sampled turves were divided in 4 main components following Ward et al. (2007): 1) aboveground plant biomass; 2) a litter (L) horizon underlying the green vegetation but well differentiated from; 3) a horizon with unidentifiable vegetation residues and humified plant material (F and H horizon not recorded on the second sampling period); and 4) the organic horizon (O horizon). Aboveground plant biomass was sorted into plant functional groups (Ward et al., 2009), namely graminoids, dwarfshrubs, forbs and a group of non-flowering plants which included mosses and species in the division Lycopodiophyta, i.e., Lycopodium sp, oven-dried at 70 ◦C for 48 hours, and weighed to determine standing crop biomass and plant growth form relative abundance. The L and F and H horizons were oven-dried, as described above, to estimate their dry biomass. Roots were recovered by washing and sieving (minimum mesh size 0.5 mm) soil cores, oven-dried and weighed to determine root biomass. 2.4.4 Soil analysis Before the O horizon was perturbed, we took a core from the soil turves with a soil sampler (35 mm diameter, 100 mm depth) in order to measure bulk density and soil moisture content by standard procedures. The remaining soil in the turves was passed through a 2 mm sieve prior to laboratory analysis. Sieved soil from each sampling date was analysed in the laboratory for a number of biological properties that are described below. Measures of total soil C and N content for two different sampling dates were done in a Carlo Erba EA-1108 elemental analyser in SAI-UDC University of Coru˜na. A Wilcoxon signed-rank test showed that soil % C did not vary across sampling dates 30 2.4 Material and Methods (W= 71, n= 12, P= 0.97); therefore, we calculated the total soil C and N for both management regimes assuming no change over the sampling period. 2.4.5 Soil N availability and dissolved organic C and N Soil ammonium (NH+ 4) content was measured by extracting 10 g of fresh soil in 1 M KCl, stirring the extracts during 1 hour on an orbital shaker, filtering them in Whatman paper No.1, and determining NH+ 4concentration by continuous-flow colorimetry using the sodium nitroprusside reaction in a Bran and Luebbe AutoAnalyzer 3. Nitrate (NO− 3) was measured in the same way as NH+ 4, but through the sulfanilamide method. Net NH+ 4 and NO− 3mineralisation rates were estimated as the difference between non-incubated and incubated soil samples (Campbell and Rochefort, 2003). Ten g of fresh soil were incubated at 25 ◦C during 14 days after which they were extracted as described above. Non-incubated samples were extracted shortly after sampling. Ammonium and NO− 3 concentrations were measured colorimetrically as already described. Dissolved organic C (DOC) and N (DON) were measured by shaking 10 g of soil in 70 mL of distilled water during 10 minutes and filtering the soil-water extracts in Whatman paper No. 1. DOC was measured by the difference between total C and inorganic C in the soil-water extracts as analysed by a Shimadzu 5000A TOC analyser (Shimadzu Inc., Japan). DON was measured by continuous-flow colorimetry as the difference in the soil-water extracts between total N after digestion in potassium persulfate and inorganic N. 31 2.4 Material and Methods 2.4.6 Microbial biomass and activity Microbial biomass C and N were assayed by the chloroform fumigation extraction technique. Microbial biomass C was estimated according to Vance et al. (1987) by extracting 5 g of fresh soil in 0.5 M K2SO4, shaking the soil-extract for 30 min in an orbital shaker, filtering the soil extract in Whatman paper No. 1. Microbial biomass C was calculated as the difference between fumigated and non-fumigated soils after analysing the extracts for C content as described for DOC using an extraction efficiency of 0.45 (Sparling et al., 1990). Microbial biomass N was assayed by digesting the soil extracts with potassium persulfate (Cabrera and Beare, 1993) and determining N contents with flow colorimetry as described above. Microbial biomass N was calculated as the N difference between fumigated and non-fumigated soils using an extraction efficiency of 0.54 (Brookes et al., 1985). Soil basal respiration was determined following Bardgett et al. (1999b) as the production of CO2by soils after a 24-h period of incubation at 25 ◦C of 1 g dry mass equivalent soil within sealed McCartney bottles. The CO2production after the incubation period was measured by injecting 1 mL of sample from the bottles’ head-space into an ADC 225 MK3 IRGA (ADC Bioscientific Ltd., Hoddesdon, UK) and using respective blanks and CO2standards. 2.4.7 Statistical analysis We tested for significant statistical effects of re-wilding on all dependent variables by using repeated measures ANOVA through the linear mixed models routine in R (Pinheiro et al., 2008). Land management (re-wilding vs grazing) and date of sampling (seasonal variation) were considered fixed effects (between and within-subjects factors 32 2.5 Results respectively). On the other hand, the experimental plots (subjects) within each sampling date were considered as the random factor. Differences in the time length intervals among sampling events were taken into account by specifying a linear correlation structure for the dates factor. Total soil C differences between land management regimes were tested with a t-test. For most variables we analysed factor effects on a per square meter basis using bulk density and a soil depth of 10 cm. All variables were transformed to satisfy normality criteria but results and figures are presented for untransformed variables using means ±se. All analysis were carried out with the R package for Linux (R Development Core Team, 2010). Given the high number of zeros in the plant functional groups dataset, the effect of re-wilding on the proportion of plant functional groups was tested by using a nonparametric multivariate ANOVA (Anderson, 2008) implemented in R (Oksanen et al., 2010). Land management (re-wilding vs grazing) and sampling date were used as the factors. If the overall model result was significant, the importance of each single plant functional group was estimated by removing one of the groups from the model and comparing the resultant pseudo-Fand R-square values (a decrease in R-square and lack of significance were used as the criteria to decide whether certain plant functional group was important). 2.5 Results 2.5.1 Vegetation composition Re-wilding resulted in significant changes in plant functional group abundances - abundance is expressed in terms of relative biomass - (F1,48 = 8.0, P<0.005). When averaged across sampling dates, re-wilding caused a significant increase in the abundance of 33 2.5 Results Figure 2.1: The field site at Ingleborough National Nature Reserve, northern England, showing the main vegetation characteristics of a re-wilding area where domestic herbivores have been excluded over 7 years (G0) and an adjacent grazed acidic grassland (G+). Images obtained from http://maps.google.com/ and authors’ own collection - 34 2.5 Results dwarf-shrubs (20.4 ±1.1 vs 2.9 ±3.4 % in the re-wilding and grazed area, respectively, Fig. 2.2a) and a decrease in the proportion of graminoid species (23.0 ±4.8 vs 43.0 ±4.7 %), despite high between-plot variation in plant abundances. Mosses and other non-flowering plants (mainly Lycopodium sp) did not differ in abundance between the different management regimes, although they were an important component of the vegetation on both areas (52.0 ±4.2 and 54.0 ±5.2 % in the re-wilding and grazed areas, respectively, Fig. 2.2a). Seasonal differences in vegetation were of marginal significance (F4,48 = 1.9, P= 0.08), although across treatments there was a lower relative abundance of dwarf-shrubs in May 2007 and June 2008 (Fig. 2.2a), and of graminoids in July 2007 and January 2008. 2.5.2 Aboveand belowground biomass, and size of litter horizons There was a significant interaction between land management and season (F4,38 = 14.6, P<0.0001) on aboveground biomass, as the higher values observed in the re-wilding area in May and July 2007 were not recorded in the following sampling dates (Fig. 2.3a). Root biomass was not affected by re-wilding (F1,10 = 1.1, P= 0.32), but it varied significantly with season across both sites (F4,36 = 10.6, P<0.0001), being maximal in early summer (May 2007) and minimal in winter (January 2008) (4280 ±570 vs 1710 ± 328 g DM m−2respectively, averaged across land management, Fig. 2.3c). Re-wilding increased the size of the litter (L) horizon (980 ±124 vs 590 ±67 g DM m−2in the re-wilding and grazed area, respectively, Fig. 2.3b), although this difference was only marginally significant (F1,10 = 4.3, P= 0.06). The size of the L horizon also varied seasonally across management treatments (F4,38 = 15.5, P<0.0001), although this was mainly down to the lower dry mass recorded in July 2007 (Fig. 2.3b). The F and H 35 2.5 Results Figure 2.2: Effects of re-wilding (G0) vs sheep grazing (G+) across seasons on: a) plant functional groups abundance; b) total soil C content; c) total soil N content. Forb = non-legume herbaceous dicots; graminoids = plants in Poaceae and Cyperaceae (mainly Eriophorum vaginatum); shrubs = Calluna vulgaris and Vaccinum myrtilus; and nsper = mosses and other non-flowering plants. In a) horizontal axis (day/month/yr). In b) and c) data are mean ±se, n= 12 - 36 2.5 Results horizon was not affected by re-wilding (F1,10 = 0.4, P>0.50), and it did not vary seasonally (F3,30 = 2.2, P>0.10; pooled estimates across management and dates 2330 ±171 g DM m−2, Fig. 2.3d). Re-wilding did not affect total soil C, measured as both % C or C content (g C m−2)(t= 0.7, df = 17.9, P= 0.50 and t= 0.7, df = 20, P>0.45 respectively; Fig. 2.2b), or total soil N (neither N % nor N content t= 0.4, df = 8.3, P>0.60 and t= 0.2, df = 7.4, P>0.80; Fig. 2.2c). Figure 2.3: Effects of re-wilding (G0) vs sheep grazing (G+) across seasons on: a) plant aboveground biomass; b) mass of the litter horizon; c) belowground biomass; d) F+H horizon. All data are mean ±se, per data point n= 6. This dry mass pools division follows that in Ward et al. (2007). Horizontal axis (day/month/yr) - 2.5.3 Soil N availability and dissolved organic matter There was a significant interaction between land management and season for potential net NH+ 4mineralisation rate (F4,38 = 4.1, P<0.01): rates of NH+ 4mineralisation tended 37 2.6 Discussion spatial variability in aboveground biomass has been observed in other upland ecosystems (Marrs et al., 1989; Hill et al., 1992; Milne et al., 2002), and although re-wilding has been associated to increases in standing crop biomass, response of aboveground biomass to grazing exclusion in temperate uplands is generally highly variable (Welch and Rawes, 1964). Re-wilding also had a strong effect on the litter (L) horizon, increasing the amount of litter accumulated on the soil surface by some 70 %. Welch and Rawes (1964) reported similar increases in the size of the L horizon after exclusion of grazing at Moor House in northern England, and after a similar time interval as that elapsed in our study. Moreover, in a reassessment of the same areas in Moor House, Marrs et al. (1989) observed that after 30 years differences in the size of the L horizon between grazing-exclusions and grazed areas became larger than those observed here. The increase in the size of the L horizon might be related to high biomass allocation to litter by the dwarf-shrubs that became dominant in the re-wilding area, and also the generally lower decomposability of dwarf-shrub litter which is know to be higher in concentration of condensed tannins and phenolics (van Vuuren et al., 1992). Furthermore, it is well known that grazing increases plant biomass turnover rates through consumption (Cebrian, 1999), so its absence in the re-wilding area might translate into a build-up of litter. Although we did not measure C content in the L horizon, Ward et al. (2007) observed that grazing-exclusion lead to a 35 % accumulation of C in litter in Moor House over 60 years, which highlights the potential of re-wilding to increase this C pool in moorland ecosystems in the longer term. Re-wilding also resulted in changes in some soil properties that are typically associated with slow nutrient cycles (Bardgett and Wardle, 2003) and which have been widely observed across gradients of grazing-intensification in temperate upland habitats 44 2.6 Discussion (Bardgett et al., 1993, 2001). In particular, re-wilding caused significant reductions in soil basal respiration, a measure of heterotrophic microbial activity, and rates of NH+ 4 mineralisation, which are also indicative of a reduction in microbial activity in soils. As mentioned, these findings are consistent with other studies of temperate grasslands that have shown that cessation grazing reduces biological activity and rates of N cycling in soils (Bardgett et al., 1993, 1997, 2001). This response is most likely due to the lack of input of animal excreta, which is known to stimulate microbial activity and nutrient cycling in grassland soil (McNaughton et al., 1988; Augustine et al., 2003; Bardgett and Wardle, 2003), and to changes in vegetation composition that alter the quality and quantity of organic matter entering soil. For example, the increase of litter accumulation in the re-wilding area was mainly from dwarf-shrubs which, as already mentioned, are known to have lower chemical quality (e.g., higher concentration of condensed tannins and phenolics) and decomposability (Berendse, 1998; van Vuuren et al., 1992). Then, lower chemical quality may have contributed to the decrease in soil microbial activity and N mineralisation in the re-wilding area. The increase in the ratio of DON to DIN and reduction in microbial biomass N due to re-wilding is also potentially indicative of an overall slowing down of N cycling. As shown recently by Farrell et al. (2011), the ratio of DON to DIN is typically higher in low compared to high productivity grassland ecosystems, which has been attributed to constraints on microbial activity which reduce DON turnover rates in low productivity ecosystems (Farrell et al., 2011; Schimel and Bennett, 2004). In our case, the use of N in microbial biomass as an indicator of changes in ecosystem N status is more complicated as we did not measure concomitant changes in plant N pools (e.g., Bardgett et al., 2002). Nevertheless, we did detect a positive correlation between microbial N and NH+ 4 45 2.6 Discussion mineralisation; and between microbial biomass N and DON. These correlations indicate that both inorganic and organic N pools might be regulated by microbial biomass as found by Zeller et al. (2001) in the case of microbial biomass N and total N in a subalpine grassland. Some of the changes we observed between the grazed and the re-wilding areas could also be a result of an increase in soil moisture content in the latter. However, soil humidity was extremely high on both areas and re-wilding did not affect overall the depth of the soil water table, which suggests that moisture was not a major factor explaining differences in microbial activity and N cycling between sites. As mentioned earlier, re-wilding of overgrazed degraded moorlands has been given a high conservation priority in the United Kingdom (Pakeman et al., 2003; Milligan et al., 2004; Littlewood et al., 2006; Mitchell et al., 2008) in order to restore the main vegetation attributes and associated fauna of these ecosystems (Reed et al., 2009). We enquired whether re-wilding could also influence other ecosystem services, particularly the potential to sequester C and N in soils, which is a key aspect in ecosystem services valuation (Dominati et al., 2010; Millennium Ecosystem Assessment, 2005), and is of high relevance to current land management policies (Sutherland et al., 2006). Despite re-wilding resulting in a slowing down cycling of N cycling and a build up of litter on the soil surface, we did not observe a concomitant change in total soil C or N. This might imply that, so far, not enough time has been elapsed for changes on these soil properties to be affected by re-wilding, which is consistent with other studies of moorland ecosystems that likewise show a lack of response of soil C to removal of grazing, even after more than 30 years (Marrs et al., 1989; Garnett et al., 2000; Ward et al., 2007). These results collectively suggest that re-wilding of moorlands via the removal of domestic herbivores is unlikely to reap benefits for soil C sequestration, at least in the short term 46 2.6 Discussion (i.e., decades). More research is needed to corroborate this conclusion as high variation in grazing effects on soil C has been reported (Conant et al., 2001; Reeder and Schuman, 2002; Cui et al., 2005; Bardgett and Wardle, 2010). Together with total soil C and N, re-wilding did not modify other key soil properties including DOC, DON and microbial C. Dissolved organic matter (including DOC and DON) is a key component in the C and N fluxes in systems with highly organic soils such as those of our study (Freeman et al., 2001). We did not record any effect of re-wilding on these properties, but they showed strong seasonal differences with higher values associated to periods characterised by warm temperatures. Likewise, Ward et al. (2007) observed that season was more important that grazing in explaining the variation in DOC concentrations in moorlands, and they attributed DOC concentration control to a combination of environmental factors (warm temperatures and reduction in soil water saturation) which lead to higher microbial activity. On the other hand, although it has been shown that bottom-up controls on soil microbial biomass are rather idiosyncratic (Bardgett and Wardle, 2010), the lack of response of microbial biomass C to re-wilding could be explained by the fact that re-wilding did not affect root biomass. 2.6.1 Conclusion In conclusion, after almost a decade of re-wilding, we detected significant changes in vegetation and some signs of a slowing down of C and N cycling in moorlands. However, despite this, and a significant build up in surface litter as a result of re-widling, this has has not yet translated into any change in stocks of C and N in soil, which is a key goal of moorland restoration management. 47 Chapter 3 Plant and soil responses to defoliation: A comparative study of grass species with contrasting life history strategies This chapter has been published in Plant and Soil, authored and under the title of: “Medina-Rold´an, E., and R.D. Bardgett. 2011. Plant and soil responses to defoliation: a comparative study of grass species with contrasting life history strategies. Plant and Soil 344:377-388.” Superficial changes were done here for the sake of the thesis format. Co-author R.D.B is the supervisor of this PhD work. 48 2. Plant and soil responses to defoliation: A comparative study of grass species with contrasting life history strategies 3.1 Abstract The overall aim of this study was to test for inter-species variation in plant and soil responses to defoliation among a broad range of temperate grass species and life-history strategies. We used a microcosm experiment where a range of grass species differing in life history traits were subjected to different intensities of defoliation, and a range of aboveground and belowground plant and soil responses were measured. All plant attributes, including accumulated shoot biomass, root biomass and root length, showed a strong negative response to defoliation, although plant species exhibited subtle differences in the way that they responded to increased severity of defoliation. Defoliation also exerted a strong influence on soil properties, decreasing soil microbial carbon (C) and the soil microbial C:nitrogen (N) ratio, and increasing inorganic N availability and potential N mineralisation across all species. Despite the wide range in life history strategies, plant species did not differ in their influence on most of the soil variables, except for the rate of nitrate mineralisation, which was lowest under plant species that displayed the least relative detrimental responses to defoliation. Collectively, our results suggest that plant and soil responses to defoliation are reasonably consistent across a broad range of grass species, with only subtle inter-specific differences among species. 49 3.2 Keywords 3.2 Keywords Defoliation, microbial biomass, grassland, nitrogen cycling, herbivory, life history strategies. 3.3 Introduction The last few decades have witnessed an increasing attention being devoted to understanding the role that grazers play in regulating soil biogeochemical processes (Bardgett and Wardle, 2010). This research has shown how grazing can modify a number of ecosystem properties, all of which ultimately impact on rates of soil nutrient cycling with feedback consequences for primary production. For example, grazing-induced increases in soil compaction (Cumming and Cumming, 2003), alterations of soil water regimes (Medina-Rold´an et al., 2007), plant species composition (Ritchie et al., 1998; Wardle et al., 2001), and aboveground and belowground primary productivity (Milchunas and Lauenroth, 1993; Burke et al., 1998) are regularly cited as drivers of soil nutrient dynamics. Also, grazing animals can affect the spatial distribution of nutrients within ecosystems as a result of their movements and patchy return of nutrients in excreta, which in turn influences vegetation patterns (de Mazancourt et al., 1998; Bardgett and Wardle, 2010). Over time scales from days to months, plant responses to biomass removal, or defoliation, act as a key mechanism by which grazing affects soil nutrient dynamics. Defoliation has been shown to alter plant allocation patterns, in terms of both root biomass (Mikola et al., 2001a) and root exudation (Paterson and Sim, 1999, 2000). These changes in plant allocation can, in turn, have cascading effects on below50 3.3 Introduction ground food-webs, modifying microbial abundance and activity (Guitian and Bardgett, 2000; Mikola et al., 2001a; Hamilton III and Frank, 2001) and the abundance of microbial predators (Mikola et al., 2001a, 2009; Hokka et al., 2004). All together, the effects on belowground food-webs mediated by defoliation have been linked to increases in soil nutrient mineralisation and availability (Hamilton III and Frank, 2001; Mikola et al., 2001a), and this is thought to contribute to the compensatory response of plant growth to herbivory (Hamilton III and Frank, 2001; Mikola et al., 2009). Past studies suggest that plant and soil responses to defoliation vary across plant species and with the frequency of defoliation (Chapin III and Slack, 1979; Briske, 1996; Guitian and Bardgett, 2000; Klironomos et al., 2004; Ilmarinen et al., 2005; Gastal et al., 2010). Such differential responses of plant species to defoliation have been explained on the basis of plant functional traits which influence soil-resource dynamics, such as root exudates entering into the decomposers food-web (Holland et al., 1996; Mawdsley and Bardgett, 1997; Mikola and Kyt¨oviita, 2002), and traits directly linked to the uptake of nutrients and competition for these with soil microbes such as root biomass and root length density (Chapin III and Slack, 1979; Oesterheld, 1992). Most studies on this topic, however, have focused on a handful of plant species, and none, as far as we are aware, have comprehensively and simultaneously tested the responses of a broad range of plant species to defoliation. As a result, little is known about the differential response of grassland plant species and their associated soil microbial communities to defoliation, and this hampers our ability to draw generalizations about plant and soil responses to grazing in grassland ecosystems. Here, we redress this lack of knowledge by testing for inter-species variation in plant and soil responses to defoliation among a broad range of temperate grass species and life-history strategies. Specifically, we test 51 3.4 Material and methods the hypotheses that soil biological properties related to N cycling would be correlated with the particular response of a plant species to defoliation (i.e., soil responses could be predicted from the defoliation response of the plants). This was done by measuring aboveground and belowground plant responses of a range of common UK grass species to different intensities of defoliation, and associated changes in soil microbial biomass and rates of N cycling, in a glasshouse microcosm experiment. 3.4 Material and methods 3.4.1 Experimental set-up A range of seven common British grass species occurring on different habitat types and representing a spectrum of life histories including response to grazing and defoliation (see Table 3.1 for a summary of RGRmax, RGRmean, C-S-R strategy, Ellenberg numbers, and grazing and defoliation response), were grown in a glasshouse experiment during the spring of 2007 at Lancaster University, U.K. Briefly, seeds (Ermorsgate Seeds, Norfolk, UK) of the grass species Nardus stricta L., Anthoxanthum odoratum L., Festuca rubra L., Poa pratensis L., Agrostis capillaris L., Lolium perenne L., and Holcus lanatus L. (nomenclature follows Clapham et al., 1987), were germinated on a 1 : 1(w/w) mixture of commercial acidic sand and Levingtons M3 growing media. The slow growing species N. stricta was sown 6 weeks before the rest of the grasses in order to reduce differences in plant size which are inherently dependent on the wide range in RGRmax employed in this study (Grime and Hunt, 1975) (Table 3.1). Established seedlings were transplanted into pots (11 x 11 cm, height and diameter) following a similar design as that used by Guitian and Bardgett (2000). Pots’ substratum was composed of a nutrient-poor 52 3.4 Material and methods soil (N % = 0.6) coming from and acidic (pH in H2O = 4.5) soil taken from a seminatural Festuca ovina-Agrostis capillaris grassland (National Vegetation Classification: U4b, Rodwell, 1992) located in Littledale, Lancashire, UK (Bardgett et al., 2003). Two weeks later, each grass species planted in the experimental pots was randomly assigned to 1 out of 3 defoliation treatments simulating different intensities of grazing, namely: 1) control plants without defoliation (CN); 2) light defoliated plants, clipped each 2 weeks (MC); and 3) heavy defoliated plants clipped at weekly intervals (HC). The defoliation treatment was imposed over an 8-week period and consisted of the removal of aboveground tissue at 4 cm above the soil surface. Treatments were applied in a randomised block design with 4 replicates per plant species and defoliation treatment combination, yielding a total of 84 pots (7 species x 3 defoliation x 4 blocks). Plants were watered on every other day and remained in a glasshouse in Lancaster University with the following settings: 12 hours light-night periods providing a mean irradiance of 430 Wm−2, mean day temperature of 22 ◦C, and a mean night temperature of 18 ◦C. 3.4.2 Measurement of aboveground variables Before applying the defoliation treatments, initial shoot biomass was calculated in order to use this measure as a covariate for statistical analyses, and hence rule out size-specific effects of different plant species on response variables. This calculation was based on species-specific allometric relationships between leaf length and shoot biomass of a subset of grass seedlings (N. stricta,y=−6.79x1.1871,R2= 0.87, P<0.001, n= 17; A. odoratum,y=−6.3x1.2768,R2= 0.97, P<0.001, n= 6; F. ovina,y=−4.64x0.9617 ,R2= 0.96, P<0.001, n= 18; H. lanatus,y=−5.37x1.1667,R2= 0.78, P<0.001, n= 20; P. pratensis,y=−3.53x0.9433,R2= 0.95, P<0.001, n= 17; A. capillaris,y= 53 3.5 Results the individual qstatistics of the Tukey’s test not included) in the order H. lanatus > A. odoratum >A. capillaris =N. stricta =L. perenne >P. pratensis >F. rubra. We detected a significant defoliation by species interaction for the root to shoot mass ratio (R:S) (F12,55 = 3, P= 0.004) which provides additional evidence that plant species responded differently across the two defoliation frequencies (Fig. 3.1c). No clear relationship could be observed for the reductions in R:S in response to defoliation or in relation to species life history strategy, at least based on published values of RGRmax. Total root length (TRL) decreased in response to defoliation (Fig. 3.1d), and this reduction varied among the plant species with defoliation frequency as shown by the species by defoliation interaction (F12,52 = 2, P<0.05). A capillaris and P. pratensis showed similar reductions in TRL across the two defoliation frequencies (A. capillaris = 71 % reduction between the undefoliated control and light defoliation, and 55 % between light defoliation and the heavy defoliation treatment; P. pratensis = 65 and 57 % for the same treatments respectively). On the other hand, the reduction in TRL of L. perenne was higher when defoliation frequency increased (33 % reduction between undefoliated controls and light defoliation and 55 % reduction between light and heavy defoliation respectively). A. odoratum,F. rubra,H. lanatus and N. stricta were relatively insensitive to heavy defoliation, displaying only a low reduction in TRL from the low to the high intensity treatment. Individual ANOVA’s for each single species (data not shown) showed basically the same patterns as those already described, i.e., a group of species which responded consistently negative to increasing frequency of defoliation (A. capillaris,H.lanatus,L. perenne and P. pratensis) and other which was sensitive to low but insensitive to high frequency defoliation (as shown by the means differences between these treatments, A. odoratum,F. rubra and N. stricta). Since we used root 60 3.5 Results mass in our species-specific estimations of TRL, statistical analysis of specific root length would produce more or less the same qualitative results as those for root mass. However, results among species can portray some insight on the extent of the costs species incurred in terms of root production (with lower values indicating higher construction costs). Specific root length followed the sequence L. perenne (110 ±10) <F. rubra (137 ± 11) <P. pratensis (141 ±14) <N. stricta (143 ±22) <H. lanatus (169 ±14) <A. odoratum (172 ±11) <A. capillaris (172 ±12 mg−1respectively). We used principal component analysis (PCA) on transformed response variables (aboveground and belowground biomass, root to shoot ratio and total root length) to summarize plant species responses to defoliation. The first PCA axis explained 78 % of the variance in the data and was associated with defoliation frequency, with belowground plant attributes (i.e., root mass and total root length) having the highest loads on this first component (see Fig. A.1). 3.5.3 Soil responses When data were integrated across all species, light and heavy frequency defoliation reduced microbial biomass C in soil (F2,53 = 7.5, P<0.01, Fig. 3.2a). Microbial biomass C was reduced from 2040 ±147 (mean ±se) mg C kg−1in soil of control plants to 1480 ±71 and 1465 ±102 mg C kg−1in soil of plants under moderate and heavy defoliation frequency respectively, and this reduction in microbial C was correlated to reductions in root biomass caused by defoliation (r= 0.36, t74 = 3.5, P= 0.001). Posthoc analyses showed, however, that microbial C means across defoliation frequencies were significantly different only for H. lanatus and A capillaris; and for A. odoratum between the undefoliated and the clipped-treatments (Fig. 3.2a). In contrast, defoliation did not 61 3.5 Results Figure 3.2: Responses to different intensities of defoliation across a broad range of common grass species of soil microbial biomass carbon (a); soil microbial nitrogen (b); and the microbial C: N ratio (c). UN = undefoliated controls; LD = light-defoliation (defoliation each two weeks); HD= heavy defoliation (weekly). N.st = Nardus stricta, A.od = Anthoxanthum odoratum, F.ru = Festuca rubra, P.pr = Poa pratensis, L.pe = Lolium perenne, A.ca = Agrostis capillaris, H.la = Holcus lanatus. All values are means ±se. - 62 3.5 Results affect microbial biomass N (F2,54 = 1.5, P= 0.26, Fig. 3.2b); hence, across all plant species, microbial C:N ratio was reduced by 20.0 and 25.6 % under moderate and heavy defoliation, respectively, when compared with the undefoliated control (F2,53 = 8.5, P<0.001, Fig. 3.2c). Soil N availability was also affected by defoliation treatments (Fig. 3.3a). Across all plant species, heavy defoliation increased soil NH+ 4(F2,54 = 3, P= 0.05) and total inorganic N (F2,54 = 3, P= 0.05) concentrations by 35 % and 29 % respectively compared to the undefoliated control. Mean comparison tests showed significant differences in N stricta and A. capillaris between undefoliated controls and the defoliated treatments but not in the rest of the species. Soil NO− 3concentrations were not significantly affected by defoliation (F2,54 = 3, P>0.05, Fig. 3.3b), although potential net nitrification rate increased by almost 200 % under heavy defoliation (F2,54 = 3.5, P<0.05) relative to the undefoliated controls (Fig. 3.3c). The rate of net nitrate mineralisation also varied significantly among plant species (F6,54 = 3.5, P<0.01, Fig. 3.3c), being greatest in soil planted with N. stricta (2.9 ±1.9 mg NO− 3kg soil−1day−1) and lowest in soil of A. capillaris (0.21 ±0.22 mg NO− 3kg soil−1day−1). The sequence of NO− 3 mineralisation after post-hoc tests was, in descending order, N. stricta,L. perenne,F. rubra,P. pratensis,H. lanatus,A. odoratum, and A. capillaris. Some plant species at the fast-growing end of the life history gradient had the lowest values of net NO− 3 mineralisation, but the overall trend was erratic (non significant species x defoliation interaction). Post-hoc tests showed that means differences between heavy defoliation and the other treatments were significant for A. odoratum,F. rubra,H. lanatus and P. pratensis. Neither net potential NH+ 4nor total N mineralisation rates (Fig. 3.3d) were affected by the experimental treatments and grass species identity did not, in general, 63 3.6 Discussion influence soil variables, aside potential net NO− 3mineralisation rate as already stated. 3.6 Discussion In this study, we tested whether plant responses to defoliation vary across a range of grassland plant species representing a broad range in life-history strategies, and whether soil microbial and biogeochemical responses to defoliation were related to such interspecies variation. We detected a general detrimental effect of defoliation on shoot and root growth among all plant species tested, and subtle inter-specific differences in the response to defoliation frequency. In particular, A. capillaris and H. lanatus showed consistently higher absolute values in shoot and root biomass and root length and relatively lower losses in root attributes across defoliation treatments, but their performance was reduced when defoliation was intensified (from light to heavy defoliation frequencies). The same reductions in response to defoliation were observed in P. pratensis and L. perenne, but these later species displayed low absolute values in plant measures. Yet another group of species, A. odoratum,F. rubra and N. stricta, showed no additional decreases in performance between light and severe defoliation in most of our plant measures which we interpret here as a sign of resistance to defoliation. Abundance of N stricta is often high in grazed grasslands (Welch, 1986), which is attributed to its unpalatable shoot tissue (Massey et al., 2007), and Hartley and Amos (1999) report that defoliation (a less severe regime than that used here) did not cause reductions in root length of N. stricta plants, adding evidence that this species is resistant to defoliation. There is evidence of F. rubra showing compensatory growth responses to grazing as well (Berg et al., 1997; van der Graaf et al., 2005). Several other studies have likewise found declines in root productivity in response to defoliation (Guitian and Bardgett, 2000; Mikola et al., 64 3.6 Discussion Figure 3.3: Soil responses to different intensities of defoliation across a broad range of common grass species, including ammonium availability (a), mineral soil N (ammonium+nitrate) (b), net potential nitrate mineralisation rate (c) and the total N mineralisation rate (ammonium+nitrate) (d); U= undefoliated controls; LD = light defoliation (defoliation each two weeks); HC= heavy defoliation (weekly). N.st = Nardus stricta, A.od = Anthoxanthum odoratum, F.ru = Festuca rubra, P.pr = Poa pratensis, L.pe = Lolium perenne, A.ca = Agrostis capillaris, H.la = Holcus lanatus. All values are means ±se. - 65 3.6 Discussion 2001a), although evidence is mixed. For example, fenced exclusion studies on Serengeti grasslands show that mammalian grazers do not necessarily inhibit root biomass and productivity (McNaughton et al., 1998), and in a global literature synthesis Milchunas and Lauenroth (1993) reported both enhancements and reductions in root biomass as a result of herbivore exclusion. More recently in a meta-analysis on graminoids, Ferraro and Oesterheld (2002) showed that the effects of defoliation are less acute on root than on shoot biomass. Nevertheless, our analysis showed that root mass and root length are important attributes in describing inter-specific differences in response to defoliation. Root length is a measure of plant foraging scale (Kembel et al., 2005) and therefore it might be associated with the response of grasses to defoliation. In general, the detrimental response of plant growth to defoliation was mirrored in the soil biological properties measured, although inter-specific differences were not detected. Across all plant species, defoliation was found to reduce microbial biomass and its C: N ratio, and to increase NH+ 4availability and the rate of NO− 3mineralisation. The negative response of microbial biomass C to defoliation that we observed contrasts sharply with results of a number of experimental studies, which have found that defoliation stimulates soil microbes (Mawdsley and Bardgett, 1997; Bardgett et al., 1998; Mikola et al., 2001b). Increases in microbial biomass following defoliation have been attributed to the stimulation of root exudation (Holland, 1995; Holland et al., 1996; Mawdsley and Bardgett, 1997; Hamilton III et al., 2008), and this stimulation of root exudates as a result of clipping has been reported for some of the plant species used in our experiment, namely L. perenne and F. rubra (Paterson and Sim, 1999, 2000). Our findings indicate, however, that defoliation caused soil microbes to become limited by C, as evidenced by the decline in the microbial C:N ratio, which is indicative of increase 66 3.6 Discussion in C relative to N limitation (Kaye and Hart, 1997). This decline is also likely to be related to the reduction in root biomass across all species as a result of defoliation, a view supported by the positive correlation of root biomass with microbial biomass C. The lack of effect that increasing the intensity of defoliation had on microbial biomass C is likely explained by the relatively large effect that light defoliation had on plant performance relative to the undefoliated controls. The decline in microbial biomass C could also be partly due to increased predation by soil animals, given that previous studies have shown that defoliation enhances the abundances of microbial-feeding faunal groups in soil (Mikola et al., 2001a,b). However, in our study, it is most likely that reductions in root C allocation under defoliation, and hence C supply to soil, is the main cause of the consistent decline in microbial biomass across all defoliated plant species (Mikola et al., 2001a; Bazot et al., 2005; Hamilton III et al., 2008; Sankaran and Augustine, 2004). Defoliation increased NH+ 4and total inorganic N availability, as well as the potential rates of NO− 3mineralisation across all species tested. A number of studies have documented a stimulatory effect of defoliation (Hamilton III and Frank, 2001; Mikola et al., 2001a; Ayres et al., 2007) and ungulate grazing (Seagle et al., 1992; Hamilton III et al., 2008) on soil N availability and mineralisation, and this response is thought to be a key mechanism contributing to compensatory growth in grazed grassland (Owen, 1980; Ritchie et al., 1998; Hamilton III and Frank, 2001; Bardgett et al., 2003). Such enhanced soil N mineralisation has been attributed to a variety of mechanisms, including the return of N-rich plant litter and animal wastes to soil (Day and Detling, 1990; de Mazancourt et al., 1998), and the stimulation of microbial activity and N mineralisation in the root zone due to enhanced root exudation in defoliated plants (Hamilton III and Frank, 2001; Mikola et al., 2001a; Ayres et al., 2007). In our experiment, however, 67 3.6 Discussion we propose that the defoliation-induced increase in soil N availability and mineralisation was due to the previously mentioned switch to C limitation of the microbial biomass, as indicated by the reduction in microbial C:N across all defoliated plants. Indeed, it is well established that under conditions when microbial growth is C limited, microbes use the C to support their energy needs and they excrete plant available ammonium (NH+ 4) as a waste product into soil i.e., N is mineralised by the microbial biomass (Kaye and Hart, 1997). Microbial C limitation together with the strong negative impact of defoliation on plant size and the concomitant reduction in total plant N might be the cause for the increase in N availability. As previously mentioned, it is also possible that higher rates of microbial predation in soils of defoliated plants contributed to the stimulation of soil N availability via the microbial-loop (Clarholm, 1985), although this was not measured in this study. Despite the wide spectrum in ecological traits in the plant species we used, and the subtle differences in plant growth responses to defoliation, few inter-specific differences were observed in the response of soil properties. Only for the rate of NO− 3mineralisation did we detect inter-specific differences in the responses to defoliation. Here, we found that the rate of NO− 3mineralisation was significantly lower in soils planted with the grasses which showed the highest biomass values, namely A. capillaris,A. odoratum and H. lanatus, than in soils planted with N. stricta. However, we did not detect a defoliation by species interaction, indicating that such inter-specific differences in NO− 3mineralisation were independent of defoliation. This trend of lower rates of NO− 3mineralisation in soils planted with those species which exhibited higher biomass values across defoliation treatments is difficult to interpret given that no concomitant changes in NH+ 4availability were detected for the same set of species. The absence of inter-species differences in other 68 3.6 Discussion soil properties across the species tested is in contrast to previous studies which show that plant species, and even genotypes, can have markedly different effects on soil biological properties, acting as major determinant of microbial communities in soil (Bardgett et al., 1999b; Innes et al., 2004; Bezemer et al., 2006; Markham et al., 2008; Harrison and Bardgett, 2010; Orwin et al., 2010). We do not know the reason for the absence of such inter-species variation in soil properties in our experiment. However, given that interspecific differences in most soil biological properties were apparent in the undefoliated controls, but not in defoliated plants, it appears that defoliation has cancelled out any differences at the species level. Secondly, it has been shown that the effects of plant species on soil properties are dependent on soil type (Innes et al., 2004; Marschner et al., 2004; Bezemer et al., 2006). In this way, it seems that our soil might have restricted the expression of strong plant effects on soil biological properties, suggesting that other factors, such as low pH and nutrient availability, might have been primary determinants of these measures. Despite this, defoliation was found to consistently and strongly promote soil nutrient availability in soil across all species tested. In conclusion, our results show that grassland plant species representing a broad range of life history strategies respond in a consistent way to defoliation. Across all species tested, we found that defoliation reduced plant growth, especially of root mass and length, but stimulated N availability in soil. We did not measure the consequences of this defoliation-induced stimulation of N availability, but we propose that it would, in the long term, positively feedback to the plant in terms of improved N acquisition and, potentially, improved growth. Surprisingly, we found only subtle differences in the response of different plant species to defoliation, and no inter-species variation in the response of soil properties to this treatment. This suggests that, in these soils, effects 69 4.4 Material and methods Park, northern England (54.18◦N, 2.36◦E, UK National Grid Reference Number SD 763762). One area is an acidic upland grassland dominated by N. stricta (Rodwell, 1992) and subjected to continuous grazing by sheep, whereas the other is an ungrazed adjacent site that was fenced-off in 2000 in order to exclude domestic herbivores (Medina-Rold´an et al. in review-a, Chapter 2 for more information about the site characteristics). Exclusion of herbivores has resulted in an increase in the abundance of E. vaginatum and ericaceous shrubs, and decreases in soil microbial activity, soil N availability, and an increase in soil moisture content (Medina-Rold´an et al. in review-a, Chapter 2). We sampled soil from 10 randomly chosen points within both the grazed and ungrazed areas in order to collect sufficient material for the experiment. Soils were passed through a 2 mm mesh to remove roots and plant residues and stored at 4 ◦C. Since our aim was to test how the effects from the contrasting soils built-up as the experiment progressed, we eliminated confounding effects caused by initial soil nutrient contents by mixing soils with sand in a 1:5 ratio. This practice is common in plant-soil experiments (Bever, 1994; Frank et al., 2003). Most soil properties including extractable inorganic N did not differ significantly in the soil-sand mixtures between the soils taken from the grazed and ungrazed areas at the beginning of the experiment (Table 4.1), except for a higher net potential NH+ 4mineralization rate in the grazed than ungrazed soil, and a slightly higher total C content in the ungrazed than grazed soil (Table 4.1). The mixture was used to fill 1-L experimental pots (10 x 12 cm) where plants were allowed to establish. Competitive interactions between N. stricta and E. vaginatum were assessed with a full-factorial design with 4 combined densities of each plant species in order to separate intraand inter-specific components of competition. This design is commonly used 76 4.4 Material and methods Table 4.1: Differences in selected soil properties between soil-sand mixtures from a grazed (G+) versus ungrazed (G−) seminatural upland ecosystems. Cmic and Nmic = soil C and N in microbial biomass (mg kg dry soil−1) respectively. NH+ 4av, NH+ 4min, NO− 3av, NO− 3min are ammonium and nitrate measures where av = soil extractable concentration (mg kg dry soil−1) and min = net potential mineralization rates (mg kg dry soil−1day−1) respectively. Csoil and Nsoil = total soil C and N (%) respectively. Analytical techniques are described in the methods section except for total soil C and N (ground soil, oven-dried at 105 ◦C, and analysed in a Vario EL Elemental Analiser, Elementar Inc, Germany), and net mineralization rates described in Harrison and Bardgett (2010). Values are means (se). Variable G+G−tvalue (P) Cmic 127 (13) 159 (26) 1.0 (NS) Nmic 49 (2.5) 48 (1.4) -0.1 (NS) NH+ 4av 16.0 (0.4) 16.0 (0.4) -2.0 (†) NO− 3av 2.1 (0.25) 1.6 ( 0.17) 1.5 (NS) NH+ 4min 1.19 (0.07) 0.26 (0.09) -8.0 (***) NO− 3min -0.15 (0.01) -0.11 (0.01) 1.5 (NS) Csoil 6.0 (0.15) 7.3 (0.30) 4.5 (**) Nsoil 0.27 (0.005) 0.26 (0.006) 1.3 (NS) NS = no significant difference, †= 0.1 ≤P≤0.05, ** = P<0.01, *** = P<0.001. to study plant-plant competitive interactions because it overcomes many of the disadvantages present in the additive or substitution series (Snaydon, 1991; Watkinson and Freckleton, 1997; Inouye, 2001). Intact soil turves taken from the grazed and ungrazed areas, with plants of N. stricta and E. vaginatum respectively, were sampled on the same date as soil, and taken to Lancaster where they were placed in a glasshouse for 6 weeks. Once the plants had grown fully-green tissues, they were split into individual tillers with 2-fully developed leaves and roots. Tillers of each species were randomly assigned to 1 out of 32 combinations of the full-factorial arrangement of soil source (grazed vs ungrazed) and 4 tiller densities for each plant species (4 densities for N. stricta and 4 densities for E. vaginatum), with 5 replicates per treatment (2 x 4 x 4 x 5 = 160). We used the following plant densities for each plant species: no plants, low density = 77 4.4 Material and methods 2 tillers; medium density = 4 tillers; and high density = 6 tillers. Tillers were planted equidistantly, and in the case of both species mixtures, tiller position was assigned randomly. After planting, tillers were allowed to establish for another 6 weeks. During this period, unsuccessful tillers were replaced until full establishment was achieved. At this time, the experiment was allowed to run for 7 months with a 12/12 hrs light-dark cycle, a mean irradiance of 430 Wm−2(std. dev. = 190 Wm−2), mean average temperature of 20 ◦C, and mean atmospheric humidity of 55 %. Experimental pots were watered every day with deionized water to reach approximately 60 % of the water holding capacity (65 % gravimetric soil content). 4.4.2 Plant measures In late September 2009, experimental pots were harvested by cutting the shoot material. Crowns (defined as the transitional tissue between shoots and roots) and roots were separated from the soil manually and sorted by species, except for a root fraction (0.36 ±0.03 g per pot) which could not be assigned with certainty to any of the species, and which was not included in the analyses. All plant material was oven-dried at 70 ◦C for 48 hours and weighed per species to determine individual tiller weight, total shoot, crown, and root biomass, and to partition of total biomass among these 3 plant components. 4.4.3 Soil measures The soil collected after plants were harvested was passed through a 2 mm sieve and stored at 4 ◦C until laboratory analysis took place. Soil was analysed for concentrations of extractable ammonium (NH+ 4) and nitrate (NO− 3), and for soil microbial biomass carbon 78 4.4 Material and methods (C) and N, and basal respiration, as a measure of microbial activity. Soil NH+ 4content was measured by extracting 10 g of fresh soil in 1 M KCl, stirring the extracts during 1 hour in an orbital shaker, filtering them in Whatman paper No.1, and determining NH+ 4 concentration by continuous-flow colorimetry using the sodium nitroprusside reaction in a Bran and Luebbe AutoAnalyzer 3. Nitrate was measured as for NH+ 4, but through the sulfanilamide method. Microbial biomass C and N were assayed by the chloroform (CHCl3) fumigation extraction technique. Microbial biomass was estimated according to Vance et al. (1987) by extracting 5 g of both non-fumigated and 24-h CHCl3fumigated fresh soil in 0.5 M K2SO4, shaking the soil-extract for 30 min in an orbital shaker, and filtering the soil extract in Whatman paper No. 1. Microbial biomass C was calculated as the difference between fumigated and non-fumigated samples after analysing the extracts for C content in a Shimadzu 5000A TOC analyser (Shimadzu Inc., Japan), and using an extraction efficiency of 0.45 (Sparling et al., 1990). Microbial biomass N was assayed by digesting the soil extracts with potassium persulfate (Cabrera and Beare, 1993) and determining N contents with flow colorimetry as described above. Microbial biomass N was calculated as the N difference between fumigated and non-fumigated soils using an extraction efficiency of 0.54 (Brookes et al., 1985). Soil basal respiration was determined following Bardgett et al. (1999b) as the production of CO2by 24-h incubated at 25 C of 1 g dry mass equivalent soil within sealed McCartney bottles soils. The CO2production after the incubation period was measured by injecting 1 mL of sample from the bottles’ head-space into an ADC 225 MK3 IRGA (ADC Bioscientific Ltd., Hoddesdon, UK), and using respective blanks and CO2standards. Soil moisture content was standardized in all samples at 30 % for the basal respiration determinations. 79 4.4 Material and methods 4.4.4 Data analysis 4.4.4.1 Plant biomass responses Total shoot, crown and root biomass, shoot to belowground (crown + roots) ratio, and the proportion of total biomass partitioned to shoots, roots or crowns per species, were analysed by ANOVA following two approaches. First, soil source and type of competition treatment (monocultures vs mixtures) were used as the ANOVA factors in order to evaluate how both species responded to the presence of intraversus interspecific competitors. Second, as a complementary approach, we explored how the species responded to the increasing levels of inter-specific competition by analysing the ANOVA models with soil source and density of the inter-specific competitor as factors. For brevity, in the results we use grazed and ungrazed to make reference to the soils from the two areas, competition effect for the type of competition (monocultures vs mixtures), and density effect for the density of the inter-specific competitor effect. 4.4.4.2 Intraand inter-specific competition Shoot biomass per individual tiller for each species and soil source (4 models) was analysed by the hyperbolic competition model (Mead, 1970). This model has the form: wi=wmi[1 + αi(xi+jxj)]−1(4.1) which, on a natural logarithms basis, can be expressed as: log(wi) = log(wmi)−log(1 + αixi+αijxj) (4.2) 80 4.4 Material and methods where wiis the weight of an individual of species iwhich is modelled as a function of the density of conspecifics (xi) and heterospecifics (xj), wmi is the mean weight of an individual of plant species iexperiencing no competition, αi,αij are parameters which measured the strength of intraand inter-specific competition respectively, and jcan be considered as a plant species equivalence term (how many individuals of the species jare necessary to have a completive effect equivalent to a conspecific individual) (Freckleton and Watkinson, 1997). Parameter estimates for the model were obtained by using the non-linear least squares estimation routine in the R statistical package. Effect of soil source on intraand inter-specific competition was determined by comparing the values of the parameters for each species in the model. 4.4.5 Soil responses We used our non-planted treatment (density of both species = 0) to test the effects that plant species had on soil properties in the grazed and ungrazed soils at the end of the experiment. Soil extractable NH+ 4and NO− 3, microbial biomass C and N, and basal soil respiration were analysed with ANOVA using soil source (grazed and ungrazed) and type of pot (non-planted, E. vaginatum and N. stricta monocultures, and both species mixtures) as the main factors. Although we added a fixed water volume to the pots during the experiment duration, there was pot to pot variation in soil water content at the end of the experiment. Thus, final soil water content was used as a covariate for most of the analyses, except for basal respiration as moisture was standardized for this measure (see soil measures). All variables were transformed to meet the criteria of normality and variance homogeneity when necessary but results are presented for 81 4.5 Results untransformed values (means ±se.), except in the case of the competition model. All analyses were carried out with the statistical package R for linux (R Development Core Team, 2010). 4.5 Results 4.5.1 Plant biomass responses Shoot biomass of E. vaginatum in monocultures was approximately two fold greater in plants grown in grazed than ungrazed soil, but it was similar in both soils when plants experienced competition with N. stricta (Fig. 4.1a, soil source x competition interaction, F1,108 = 45.0, P<0.001). The reduction in shoot biomass of E. vaginatum in response to increasing densities of N. stricta was stronger in grazed than ungrazed soil (see Fig. B.1, soil source x density interaction F3,104 = 17.5, P<0.001). In contrast, the negative response of N. stricta shoot biomass to E. vaginatum (Fig. 4.1a, competition effect, F1,111 = 25.0, P<0.001) was independent from soil source (soil source x competition interaction, F1,111 = 0.5, P= 0.4). E. vaginatum crown biomass was reduced by competition with N. stricta only in the grazed soil (Fig. 4.1b, soil source x competition interaction, F1,107 = 8.5, P<0.01), and this reduction was particularly strong in response to low densities of N. stricta (Fig B.1, soil source x density interaction F3,107 = 3.0, P<0.01). Competition with E. vaginatum likewise reduced N. stricta crown biomass (Fig. 4.1b, competition effect, F1,108 = 24.4, P<0.001), but this response was independent of soil source (soil source x competition interaction, F1,108 = 0.3, P= 0.5). Nardus stricta crown biomass was on average 25 % greater when grown in grazed than in ungrazed soil (Fig. 4.1b, soil source effect, F1,108 = 2.6, P<0.01). 82 4.5 Results Negative effects of N. stricta on E. vaginatum root biomass were greater in grazed than in ungrazed soil (Fig. 4.1c, soil source x competition interaction, F1,107 = 2.0, P= 0.06). Averaged across soils, competition with E. vaginantum decreased N. stricta root biomass by 40 % (competition effect, F1,107 = 30.8, P<0.001), and this reduction did not differ with soil source (Fig. 4.1c, soil source x competition interaction, F3,107 = 2.6, P= 0.1). Figure 4.1: Comparative indirect effects of grazing and direct effects of competition on: a) total shoot biomass, b) crown biomass, c) root biomass, d) shoot: belowground (crown+root) ratio of Eriophorum vaginatum (Ev) and Nardus stricta (Ns). Plants were grown in a glasshouse microcosm experiment with soil coming from a grazed Nardus-dominated acidic upland grassland (G+) or aEriophorum-dominated ungrazed area (G−) in the Yorkshire Dales, England. E. vaginatum plants grew either in monocoltures (C−) or with inter-specific competition by N. stricta (C+). N. stricta plants grew either in monocoltures (C−) or with inter-specific copetition by E. vaginatum (C+). - 83 4.5 Results 4.5.2 Biomass partitioning When competing with N. stricta,E. vaginatum allocated 15 % lower (competition effect, F1,106 = 7.1, P<0.01) and 22 % greater (competition effect, F1,106 = 5.9, P<0.05) biomass to shoots and crowns respectively (Fig. 4.2a). Nardus stricta competition did not affect biomass allocated to roots by E. vaginatum (competition effect, F1,106 = 0.03, P= 0.8), but the shoot: belowground ratio of E. vaginatum was reduced in competition with N. stricta (Fig. 4.1d, competition effect, F1,106 = 6.5, P<0.05). Eriophorum vaginatum biomass partitioning did not respond to soil source in any case (soil source effect for proportion in: shoots, F1,106 = 1.9, P= 0.1; crowns, F1,106 = 0.7, P= 0.3; roots, F1,106 = 0.06, P= 0.8; and shoot: belowground ratio F1,106 = 2.1, P= 0.1). For N. stricta, biomass allocated to shoots and crowns was 13 % greater (soil effect, F1,105 = 13.6, P<0.001) and 20 % lower (soil source effect, F1,105 = 26.9, P<0.001), respectively, when grown on grazed compared to ungrazed soil (Fig. 4.2b). There was no effect of soil source on biomass allocation to roots (soil effect, F1,105 = 0.8, P= 0.3), but the shoot: belowground ratio of N stricta was greater in grazed than in ungrazed soil (Fig. 4.1d, soil effect, F1,105 = 13.2, P<0.001). Competition with E. vaginatum did not modify N. stricta biomass partitioning (competition effect for proportion in: shoots, F1,105 = 1.2, P= 0.2; crowns, F1,105 = 0.3, P= 0.5; roots F1,105 = 1.3, P= 0.3; and shoot: belowground ratio F1,105 = 1.1, P= 0.2). 4.5.3 Intraand inter-specific competition Biomass of an individual plant with no competition (represented by the parameter wmi) was larger in the grazed than in the ungrazed soil for both species (Table 4.2, a ratio for log[wmi] between grazed and ungrazed soils >1; compare the intercepts with the 84 4.5 Results Figure 4.2: Comparative indirect effects of grazing and direct effects of competition on biomass partitioning of a) Eriophorum vaginatum, b) Nardus stricta. All legends as in Fig. 4.1 - 85 4.6 Discussion considered an important trait in plant competition (Wedin and Tilman, 1993). Although we did not condition soil with targeted plant species, our approach was broadly similar to a plant-soil feedback experiment (reviewed in Kulmatiski and Kardol, 2008) in that we used soils where our plant species showed contrasting abundances. The occurrence of positive plant-soil feedbacks in N. stricta has been observed in other studies (Kardol et al., 2006; Markham et al., 2008), and has been interpreted as being mediated by soil microbes, including mycorrhizal fungi, that enhance access to growth limiting nutrients, such as N and phosphorous (Kardol et al., 2006). Although we did not measure the occurrence of symbionts on plant roots, this interpretation was supported by our finding of lower N at the end of the experiment when N. stricta was present as stated before. Not many empirical studies have looked at the effects of plant-soil-feedback on plant competition and its components (intraand inter-specific). In a serpentine grassland in Pennsylvania, Casper and Castelli (2007) showed how a negative plantsoil feedback translated into larger biomass of grass species when they grew in soil conditioned by heterospecific plants, but these plant-soil feedback effect was cancelled when plants experienced inter-specific competition. Since negative effects of intraand inter-specific competition on biomass were the same independently of soil conditioning, Casper and Castelli (2007) suggested that plant species show no niche differentiation in plant-soil feedbacks and competition interactions. We observed that the larger biomass of E. vaginatum in grazing-conditioned soil was cancelled out when this species grew in competition with N. stricta. However, unlike Casper and Castelli (2007), we interpret this finding as a result of a greater ability of N. stricta to exploit the increased soil N availability attributed to grazing. This interpretation is supported by the fact that competition with E. vaginatum did not influence soil effects on N. stricta. As far as we 92 4.6 Discussion are aware, this differential effect on inter-specific competition induced by grazing is an aspect of plant-soil feedbacks that has not yet been described. Do changes in competitive interactions between these species driven indirectly by grazing effects on soil have any significance for the dynamics of plant communities in temperate semi-natural mountain grasslands? Although artificial (i.e., it did not include important aspects of plants auto-ecology necessary to predict vegetation dynamics or changes in other soil conditions such as moisture or more plant competitors), we can use our results to make some conjectures on coexistence between our plant species. The heuristic approach in Bever et al. (1997) and Bever (2003) suggests that for plants that show strong competitive interactions to coexist, a strong negative plant soil feedback should operate. Conversely, strong competition and the occurrence of a positive feedback should lead to competitive exclusion in pair-wise competitive arenas. As said before, N. stricta presented a positive plant-soil feedback and its negative effect on E. vaginatum was increased in grazing-conditioned soil. On the other hand, E. vaginatum presented a negative feedback and the negative influence of N. stricta on it was reduced in soil where grazing was excluded. This implies that, under equilibrium conditions, E. vaginatum should be excluded from the grazing-induced soils by N. stricta; and that the later should be able to invade E. vaginatum dominated areas where grazing has been excluded, but in lower densities. Eriophorum vaginatum actually occurs in low densities in grazingconditioned soils in the field where we collected our soils. Thus our results suggest that a non-equilibrium process such as gaps created by grazing or by heterogeneity in soil conditions promote N. stricta-E. vaginatum co-existence on the grazed acidic grassland. In this way, Ejankowski (2008) observed that experimental gap creation increased E. vaginatum seedling recruitment in an open bog habitat in Poland. 93 4.6 Discussion Changes from N. stricta dominated upland grasslands into more dwarf-shrub vegetation are thought to be caused by interactive effects of reduced grazing intensity and altered edaphic conditions, such as higher soil moisture content and increased surface organic matter accumulation (Ratcliffe, 1959; Edgell, 1971; Welch, 1986; Rodwell, 1992). In our field site, the reduction of N. stricta after grazing exclusion was also associated with an increase in soil moisture content and a decrease in N availability (Medina-Rold´an et al. in review-a, Chapter 2). Reduction of N. stricta after grazing cessation has been interpreted as a result of other upland plants outcompeting N. stricta when they are released from herbivory (Welch, 1986). Our results suggest that the reason might also lie on the effect of other environmental variables, such as soil moisture, on N. stricta which we did not take into account in our microcosm experiment. As an heuristic example, our model parameters are analogous as those in the Lotka-Volterra competition model (Damgaard, 1998). Apart from competition coefficients, key parameters which determine equilibrium plant species densities in the later are the intrinsic rate of growth (r) and the carrying capacity of a particular habitat (K). It seems likely that the changes in soil properties brought about by grazing exclusion are beyond N. stricta optimal ecological response curve (i.e., an habitat where N. stricta rand Kare low). This view agrees with experimental results (Genney et al., 2002) which showed that N. stricta was not a superior competitor when its roots were not exposed to soil mineral layers (i.e., it was exposed to sub-optimal conditions). In summary, our results showed that grazing can indirectly alter competitive interactions of plants from semi-natural mountain grasslands, and those changes might be mediated by processes regulated by soil microbial communities and other components of the soil food-web. It is well established that grazers directly influence plant competitive 94 4.6 Discussion interactions via selective grazing, and our results point to an additional indirect mechanisms by which grazers might alter plant-plant interactions via plant-soil feedbacks. Further studies are needed in more realistic situations to test the significance of such plant-soil feedback mechanisms as regulator of plant community dynamics in grazed ecosystems. 95 Chapter 5 Inter-specific competition, but not soil microbial community affects uptake of different chemical forms of nitrogen by graminoids of semi-natural, low productivity grassland 96 5. Inter-specific competition, but not soil microbial community affects uptake of different chemical forms of nitrogen by graminoids of semi-natural, low productivity grassland 5.1 Abstract Interest in resource-based plant competition has increased in recent years with growing evidence that plants differ in their ability to take up different chemical forms of nitrogen (N) from soil, including organic and inorganic N. However, whether patterns of plant uptake of inorganic and organic N can be modified by other factors (such as differences in soil microbial community composition), and the consequences of this for plant competition for soil resources is less explored. Here, we report results from a plant competition microcosm experiment designed to test the hypothesis that soil microbial communities from areas of differing grazing management can modify N uptake patterns and N uptake competition in two graminoids. This was done by inoculating sterilised soil with contrasting microbial communities taken from long-term grazed and ungrazed grasslands, combined with the addition of 15N labelled ammonium and dual 13C, 15N labelled glycine, to test for effects of soil microbes and plant competition on isotopic enrichment of plant tissue. The species used were Eriophorum vaginatum and Nardus stricta, which are common graminoids of nutrient-poor, acidic semi-natural grasslands 97 5.2 Keywords with differential responses to long-term grazing. We found that the different microbial communities had no effect on N uptake of either N form by these graminoids, which might suggest functional equivalence of the two microbial communities on N uptake. However, the two species differed in their preference of the N forms, E. vaginatum took up more organic N than did N. stricta, which is consistent with its field dominance in soils with a higher proportion of organic to inorganic N. In accordance with the results on plant biomass, N. stricta, the superior competitor, altered E. vaginatum N uptake patterns, reducing the proportion of organic N taken up by the later, thus increasing niche overlap in N usage. Whereas local abundances of these species in semi-natural low productivity grasslands are mainly controlled by gradients in grazing and soil moisture content, our results suggest that other soil properties such as proportion of organic N in the dissolved N pool can play a role in determining local abundances as well. Additionally, our results suggest that observed coexistence of these species in the field is not based on resource complementarity mechanisms so that other mechanisms based on non-equilibrium dynamics (such as disturbance) or soil heterogeneity might operate in order to maintain coexistence of these two plant species. 5.2 Keywords 15N, 13C, plant nitrogen uptake, organic nitrogen, plant competition, plant species coexistence, uplands, soil amino acids. 98 5.3 Introduction 5.3 Introduction Plants’ ability to directly take up organic nitrogen (ON) forms is spread across a broad range of plant taxa from such contrasting habitats, including tropical and temperate forests, arctic tundra, shrublands, and grasslands (Lipson et al., 1999; Schimel and Bennett, 2004; Jones et al., 2005). Besides questioning previous assumptions on the nitrogen (N) cycle (Schimel and Bennett, 2004), it has been proposed that this ability for plants to take up ON might constitute an important mechanism regulating plant species competition (Lipson et al., 1999; McKane et al., 2002; Harrison et al., 2007; Hill et al., 2011). For instance, if plants species show differential N uptake preferences for ON or inorganic N (IN), niche overlap and competition intensity could decrease (McKane et al., 2002). Since the importance of ON increases as primary productivity decreases (Schimel and Bennett, 2004; Farrell et al., 2011), there might be some degree of niche differentiation in N use among plants that exhibit contrasting abundances across environmental gradients (Weigelt et al., 2005). However, studies on plant N chemical form uptake have shown conflicting results, with plant species from different habitats displaying preference for one N form (McKane et al., 2002; Kahmen et al., 2006), or no preference at all (Harrison et al., 2007, 2008; Paungfoo-Lonhienne et al., 2008). Whereas no preference for N chemical forms among plants that differ in habitat or plant traits might suggest weak niche differentiation, other factors could alter uptake of N chemical forms among different plant species. Inter-specific competition (Miller et al., 2007; Ashton et al., 2008, 2010) and soil microbes (Kaye and Hart, 1997; Dunn et al., 2006) are some of these factors still poorly explored, but could shed light on plant species coexistence through their effects on plant resource competition. The role of plant competition has recently 99 5.3 Introduction been addressed (Miller et al., 2007), but little is known about how soil microbes affect plant N uptake preferences (Dunn et al., 2006), despite the fact that microbes are key agents in N cycling. A key factor in grasslands that is known to strongly modify microbial communities and nutrient cycling is grazing by large herbivores, which in turn can feedback to influence plant species performance and competition (Bardgett et al., 1997; Ritchie et al., 1998; Hamilton III and Frank, 2001; Bardgett and Wardle, 2003; Sørensen et al., 2008a). For instance, Frank et al. (2003) attributed increased performance of a dominant temperate grass species in Yellowstone National Park, USA, to changes in soil microbial communities (particularly mychorrhizal fungi) induced by grazing by large herbivores, which might have allowed plants to have improved access to soil nutrients. Similarly, Medina-Rold´an et al. (in review-b) (see Chapter 4) found that grazing-induced changes on soil properties, especially higher microbial activity and N availability, in semi-natural, acid grasslands increased the competitive ability of Nardus stricta, which could contribute, in part, to its dominance in grazed grassland. Such observations point towards a role of grazing effects on soil microbes in mediating plant competition for soil resources, including ON and IN, either by modifying plant niche overlap (Reynolds et al., 2003), or by modifying rates of nutrient cycling. Here, we report on a glasshouse experiment that was designed to test how grazing induced changes in soil microbial communities modify N uptake patterns for ON and IN of two dominant graminoids of semi-natural, acidic grassland that show differential responses to grazing, namely Eriophorum vaginatum and N. stricta. We also tested how patterns of ON and IN uptake by these two graminoids were affected by competitive 100 5.4 Material and methods interactions of these two species. Specifically, we tested three hypotheses. First, we hypothesized that the soil microbial community of grazed grassland, which was found to be more metabolically active in previous studies (Medina-Rold´an et al. in review-a,b; see Chapters 3 and 4), would facilitate increased N uptake in both plant species, although the increase would be larger for N. stricta due to its greater competitive ability. Second, in terms of uptake patterns of ON and IN, we hypothesised that E. vaginatum would show a greater preferences for ON than N. stricta, since the former is known to grow in soils with higher proportion of ON to IN (Medina-Rold´an et al. in review-a, Chapter 3). Finally, we hypothesised that, when in competition, N. stricta would alter the N uptake patterns of E. vaginatum because of its higher competitive ability. These hypotheses were tested in a plant competition glasshouse experiment, using soil inoculum from grazed and ungrazed semi-natural grasslands previously shown to differ in soil biological properties, and the use of 15N labelled IN and dual-labelled 15N-13C amino acids to measure how experimental treatments affected the uptake of N compounds by the two plant species. 5.4 Material and methods 5.4.1 Experimental design 5.4.1.1 Soil substratum and inoculum preparation Our study area is located in the Ingleborough National Nature Reserve, Yorkshire Dales, northern England (54.18◦N, 2.36◦E, UK National Grid Reference Number SD 763762). This area is part of a landscape re-wilding experiment where acidic grassland dominated by N. stricta,Agrostis capillaris and Festuca spp., and which is subjected to continuous 101 5.5 Results 5.5 Results 5.5.1 Plant biomass At the end of the experiment, inter-specific competition with N. stricta caused an 85 % reduction in E. vaginatum shoot biomass (F1,47 = 50.0, P<0.001), and a 63 % decrease in its root biomass (F1,43 = 26.0, P<0.001) in comparison with monocultures (Fig. 5.1a,c). There was a weakly significant inoculum source x competition interaction for E. vaginatum root biomass (F1,43 = 4.5, P= 0.04), in that the negative effect of inter-specific competition on this measure was slightly stronger in the ungrazed soil (Fig. 5.1c). No effects of soil microbial communities were detected for either shoot (F1,47 = 0.06, P>0.8) or root biomass (F1,43 = 0.5, P>0.5) of E. vaginatum, except for the already described interaction for roots. In contrast, N. stricta was not affected by inter-specific competition either in terms of shoot (F1,50 = 0.004, P>0.9) or root biomass (F1,48 = 0.9, P>0.4), and inoculum source had no influence on the growth of this species either (F1,50 = 1.6, P>0.2; F1,48 = 0.6, P>0.4 for shoot and root biomass receptively, Fig. 5.1b,d). 5.5.2 Plant N uptake Linear regressions of log transformed data of 13C against 15N molar excess in the glycine (Gly) labelled treatment showed that it was likely that both species took up intact organic N (E. vaginatum shoots: F1,10 = 19.0, P<0.01; roots: F1,8= 38.5, P<0.001; N. stricta shoots: F1,10 = 5.5, P<0.05; roots: F1,10 = 4.0, P= 0.07; Fig. 5.2). There was no evidence of 13C or 15N enrichment in either species in the unlabelled control, suggesting that contamination of plant tissue though incorporation of respired 13C was negligible (Figs. 5.3 and 5.4). Soil microbial communities from grazed or 108 5.5 Results Figure 5.1: Effect of microbial inoculum from a grazed Nardus-dominated acidic semi-natural grassland (G+) and a Eriophorum-dominated ungrazed grassland (G−), and inter-specific competition on: (a) E. vaginatum shoot biomass; (b) N. stricta shoot biomass; (c) E. vaginatum root biomass; and (d)N. stricta root biomass. Data show the effects of inter-specific competition when both plants were grown without (C−) or with (C+) inter-specific competition. Gray bars (NA) indicate the effect of competition on each species biomass component on non-inoculated soils. Values are means ±se. - 109 5.5 Results ungrazed soils did not have a significant effect on any of the root or shoot isotopic enrichment variables (15N or 13C) for either species. On the other hand, a marginal species (within competition) x labelled N solution interaction for 15N enrichment in shoot biomass (F2,34 = 3.0, P= 0.06) provided evidence that our plant species displayed different N uptake preferences. Thus, enrichment of E. vaginatum shoots with 15N was 63 % greater when supplied with labelled glycine (Gly) than NH+ 4(Fig. 5.3a), indicating a preference for ON over IN source by this species. The opposite was true for N. stricta, in that 15N enrichment of shoot tissue of this species was 66 % greater when supplied with labelled NH+ 4than Gly (Fig. 5.4a), suggesting a preference for IN over ON. With respect to the effects of inter-specific competition on N compounds uptake, shoot 13C enrichment was 35 % less (averaged across plant species) in plant mixtures than in monocultures in the Gly treatment (F1,15 = 6.2, P<0.05; Figs. 5.3c,d; and 5.4c,d). Inter-specific competition also reduced shoot 15N enrichment by 27 % (averaged across species and N sources) in comparison with monocultures (marginally significant effect: F1,34 = 3.1, P= 0.08; Figs. 5.3a,b; and 5.4a,b). However, inter-specific competition did not affect either 13C (F1,15 = 1.8, P= 0.19; Figs. 5.3g,h; and 5.4g,h), or 15N enrichment in roots (F1,37 = 1.5, P= 0.22; Figs. 5.3e,f; and 5.4e,f) for either species. When N uptake preference per species was expressed as total 15N enrichment in shoots, interspecific competition did not alter N uptake preferences of either species (competition x labelled N solution interaction, F1,34 = 0.33, P= 0.56). However, when preference was expressed as proportion of 15N enrichment of a single compound in relation to total enrichment in shoots (e.g., preference for NH+ 4:mol excess NH+ 4/[mol excess Gly + mol excess NH+ 4]), inter-specific competition did affect N preferences of E. vaginatum, 110 5.5 Results Figure 5.2: Regression analysis and parameter estimates of log transformed data of 15N against 13C mol excess for shoot (a) and root (b) biomass of plants treated with dual-labelled (13C and 15N) glycine (Gly) + unlabelled NH+ 4.- 111 5.5 Results Figure 5.3: Effects of microbial inoculum source (G+vs G−) and inter-specific competition on Eriophorum vaginatum 15N and 13C isotopic enrichment (expressed as molar excess) of: (a-d) shoot; and (e-h) root biomass. Data show the effects of inter-specific competition when plants were grown without (left panels) or with inter-specific competition (right panels) after applying glycineammonium solutions with isotopic dual-labelled (13C and 15N) glycine (Gly), 15N-labelled NH+ 4(Din) or both compounds unlabelled (None). Values are means ±se. Legends as in Fig. 5.1. -112 5.5 Results Figure 5.4: Effects of microbial inoculum source (G+vs G−) and inter-specific competition on Nardus stricta 15N and 13C isotopic enrichment (expressed as molar excess) of: (a-d) shoot; and (e-h) root biomass. Data show the effects of inter-specific competition when plants were grown without (left panels) or with (right panels) inter-specific competition after applying glycine-ammonium solutions with isotopic dual-labelled (13C and 15N) glycine (Gly), 15N-labelled NH+ 4(Din) or both compounds unlabelled (None). Values are means ±se. Legends as in Fig. 5.1. -113 5.5 Results but not of N. stricta (Fig. 5.5). More precisely, inter-specific competition with N. stricta reduced the proportion of ON taken-up by E. vaginatum in comparison with E. vaginatum monocultures (Fig. 5.5a,b). Figure 5.5: Effects of inter-specific competition on proportional N uptake (based on shoot 15N mol excess) of glycine (Gly) vs NH+ 4(Din) for E. vaginatum (a,b) and N. stricta (c,d). Left and right panels correspond to plants without or with inter-specific competition respectively - 114 5.6 Discussion 5.5.3 Soil microbial properties There were no differences between grazed and ungrazed inoculated soils in microbial properties, namely microbial biomass C (F1,70 = 0.007, P>0.90), N (F1,70 = 0.06, P>0.70), and C:N ratio (F1,66 = 0.15, P>0.70) (Fig. 5.6). Also, none of these soil microbial variables responded to plant species identity, either as monocultures or when both plants grew in competition (F1,70 = 0.15, P>0.80; F1,70 = 0.6, P>0.50; and F1,66 = 0.75, P>0.70, for soil microbial C, N and C:N ratio, respectively). 5.6 Discussion We investigated how grazing-induced differences in microbial communities and plant competitive interactions influenced patterns of organic (ON) and inorganic (IN) N uptake by the graminoids E. vaginatum and N. stricta of semi-natural, low productivity grasslands. Abundances of these two plant species in semi-natural grasslands vary across gradients of grazing and soil properties (Ratcliffe, 1959; Edgell, 1971; Grant et al., 1985; Welch, 1986; Rodwell, 1992), with E. vaginatum being more abundant under ungrazed conditions, and higher soil moisture contents (Medina-Rold´an et al. in review-a, Chapter 3); and N. stricta being dominant in grazed grasslands. Furthermore, the ratio of ON to IN is higher in soils where E. vaginatum is more abundant (Medina-Rold´an et al. in review-a, Chapter 3). Therefore, we predicted that these two plant species would show different N uptake preference; with E. vaginatum showing a preference for ON over IN, and N stricta showing the opposite. Consistent with this notion, we observed that E. vaginatum displayed greater uptake of ON than N. stricta, even when both N chemical forms were added in similar concentrations, and N stricta showed a higher preference for 115 5.6 Discussion Figure 5.6: Effects of microbial inoculum source (G+vs G−) and pot-type (a proxy for plant species) on: (a) soil microbial biomass C; (b) soil microbial biomass N; and (c) microbial biomass C:N ratio. Soils were planted with E. vaginatum plants in monocultures (Ev), N. stricta plants in monocultures (Ns), and both species under competition (Both). Values are means ±se. Legends as in Fig. 5.1. - 116 5.6 Discussion IN. Preferential use of ON over IN by E. vaginatum has been shown elsewhere (Chapin III et al., 1993). However, in contrast to Weigelt et al. (2005) who showed that N. stricta preferred to uptake the amino acid serine over IN, we found that this species showed a preference for IN over glycine. However, other studies, done both as single species in glasshouse conditions (Harrison et al., 2008) and in mixed communities in situ (Harrison et al., 2007) have not detected greater ON utilization over IN by this grass species. Also, the greater capacity of N. stricta to uptake IN compared to E. vaginatum in our study is consistent with the findings of Havill et al. (1974), who found a greater nitrate reductase activity in N. stricta in response to the addition of nitrate than for E. vaginatum. Previous studies have interpreted differentiation in uptake of chemical forms of N as a mechanism for plant species local coexistence (McKane et al., 1990, 2002; Kahmen et al., 2006). Unlike those studies, we interpret differentiation in preferences for chemical forms of N in our plant species as a result of their differentiation in occurrence along gradients of grazing and soil properties (Medina-Rold´an et al. in review-a, Chapter 3). Thus, contrasting abundances of our species in the field seem to be determined mainly by their inter-specific differences to soil properties (Edgell, 1971; Wein, 1973; Rodwell, 1992), and grazing resistance (Grant et al., 1985; Welch, 1986; Hartley and Amos, 1999; Medina-Rold´an and Bardgett, 2011, see Chapter 2). Our main hypothesis was that soil microbial communities from grazed an ungrazed grassland would influence the uptake preferences for inorganic (IN) and organic N (ON) by E. vaginatum and N. stricta. As already stated, these two graminoids are common in these upland habitats where they exhibit a contrasting response to grazing (N. stricta increases and E. vaginatum decreases) and different plant traits related to competition for soil resources (Medina-Rold´an et al. in review-b, Chapter 4). Given these differences 117 6.2 Direct effects of grazing on soil properties in poor-nutrient semi-natural grasslands ferences in plant traits related to nutrient conservation and turnover between graminoids and dwarf-shrubs. In this way, lower nutrient use efficiency, nutrient resorption and leaf life span in graminoids in comparison to dwarf-shrubs is related to production of higher quality litter which could, in turn, positively affect microbial-mediated litter decomposition, thus increasing nutrient cycling (Aerts, 1996; Berendse, 1998; Garnier and Aronson, 1998). I did not measure differences in plant functional traits related to nutrient use efficiency (as described above) between continuously-grazed and ungrazed areas in the field experiment directly. However, my results of higher microbial activity, N mineralization, microbial biomass N, and higher dissolved inorganic N (DIN) to dissolved organic N (DON) ratio in the continuously-grazed area suggest that the well-known differences in traits among plant functional groups affecting litter decomposition are a plausible explanation for the effects of grazing on soil properties. This was true even against the high backdrop noise produced by inter-annual variability. Additional evidence that grazing increased rates of nutrient cycling through its effects on plant functional groups composition comes from the fact that stocking rates in the continuously-grazed area are relatively low, in the order of 4 ewes ha−1during the growing season and 1.5 ewes ha−1during the season of low plant productivity (Colin Newlands, Natural England, pers. comm.), so that input of N-enriched animal excreta might be low (note that estimates of N and other nutrient inputs by herbivores and their spatial variability in the field are scarce, Sheldrick et al., 2003; Jewell et al., 2007). Furthermore, although grazing cessation increased gravimetric soil moisture content slightly (20 % in relation to continuouslygrazed grasslands on average), both areas showed very high soil moisture content values (above 450 % ), and grazing management did not affect water table depth. These facts 124 6.3 Plant defoliation recreates observed direct effects of grazing, differences in plant traits are less important make it improbable that differences in soil properties between the continuously-grazed and the ungrazed areas were due entirely to their differences in hydrological properties affecting microbial processes (Fenchel et al., 1998). It has been suggested that it is more likely for grazing to increase soil nutrient cycling in rich-nutrient environments (Bardgett et al., 2003; Bardgett and Wardle, 2010), since these conditions are associated more often with plant compensatory responses to herbivory and plant grazing tolerance strategies (Maschinski and Whitham, 1989, but see Wise and Abrahamson 2005). Despite low-productivity of semi-natural mountain grasslands, continuous grazing increased N mineralization and soil microbial activity in comparison with exclusion of grazing in the field. Since grazing cessation promotes the dominance of dwarf-shrubs like Calluna vulgaris and Vaccinium spp. over that of grasses like Nardus stricta,Festuca spp. and Agrostis spp. in these semi-naturalmountain grasslands, it seems that grazing by large herbivores can increase nutrient cycling when contrasting functional groups show differential response to herbivory, which adds up to our knowledge of grazing effects on soil processes. 6.3 Plant defoliation recreates observed direct effects of grazing, differences in plant traits are less important As shown in Chapter 2, large herbivores can affect nutrient cycling and soil microbial properties through their effects on abundance of plant functional group composition with contrasting plant traits, even in low-nutrient ecosystems such as semi-natural mountain grasslands. Could these effects of herbivores on soil properties be explained entirely by plant tissue removal (defoliation), and the subsequent physiological and morphological 125 6.3 Plant defoliation recreates observed direct effects of grazing, differences in plant traits are less important changes in plants? I set up a defoliation microcosm experiment with a range of grass species differing in occurrence across productivity gradients (and putatively in functional traits) in order to respond to this question. Thus, in Chapter 3, I set the specific objective of determining the effects of defoliation on soil properties across a range of temperate grass species differing in life-history traits (specific objective 2, see Chapter 1). As plant species showing differential abundance across productivity gradients are well-known to vary in response to different environmental factors, including disturbance such as defoliation (Grime, 1977, 2001), I hypothesised that defoliation effects on soil would be mediated by the different plant species response to defoliation. Effectively, the microcosm experiment showed that defoliation can consistently recreate those stimulatory effects of grazing on soil processes observed in the field, including N availability and mineralization rates. However, unlike the field experiment, defoliation reduced microbial biomass C and did not affect microbial biomass N. Reduction in microbial biomass C was related to reductions in root biomass, which suggests a strong donor-controlled food web that is characteristic of short-term pot experiments. However, no effects of defoliation on soil microbial N might indicate that defoliation did not mimic effects of grazing on soil microbial community composition (increase in bacteria in relation to fungi), which are reported when grazing stimulates soil nutrient cycling (Bardgett et al., 1998, 2003; Bardgett and Wardle, 2010). On the other hand, increases in N availability were possibly related to reduction in plant root biomass and total root length, which limited plant capacity to take up soil nutrients. Contrary to my expectations, inter-specific differences in response to defoliation across the grass species were of small or no importance for most soil processes measured. Lack of species effects on soil properties observed in this experiment (Chapter 126 6.4 Indirect effects of grazing alter plant competition 3) is in contrast to other studies where plant species have been shown to differ in their effects on soil properties (e.g., Innes et al., 2004; Bezemer et al., 2006; Harrison and Bardgett, 2010; Orwin et al., 2010). Since my experiment included a broad variation in life history traits, it is likely that this variation included differences in traits such as grazing response (Pakeman, 2004), litter quality (Niemann et al., 1992), root exudates production (Bais et al., 2006), susceptibility to mycorrhizal infection (Johnson et al., 1997), all of which could be affected by defoliation, and affect soil micro-organisms and the processes driven by them. Therefore, how can these results be reconciled with those in the field study where variation in plant functional traits might explain effects of grazing on soil processes? I suggest that other factors might be more important than inter-specific variation in plant traits in mediating the effects of defoliation on soil processes under the conditions of the microcosm experiment. Namely, the short-term nature of the experiment that might not simulate appropriately long-term effects of grazing, which include replacement of some functional groups by others, and the inclusion of only one functional group (graminoids) might explain why different plant traits were of no importance, although these ideas have yet to be tested. 6.4 Indirect effects of grazing alter plant competition Direct effects of grazing on plant-plant interactions, particularly competition, have been widely studied in order to understand the non-random plant assemblages, which characterise grazing environments (Louda et al., 1990). However, whether herbivores can indirectly affect plant-plant interactions (an herbivore-pant-soil feedback) through their effects on soil properties, as those described in Chapter 2, is a much less explored topic. 127 6.4 Indirect effects of grazing alter plant competition Therefore, in Chapter 4, I assessed how the effects of grazing on soil properties affect competition dynamics in dominant plant species of semi-natural grassland (specific objective 3, see Chapter 1). This was carried out through a microcosm plant competition experiment, which allowed me to separate the intraand inter-specific components of plant of competition, and included the effect of soil source from the contrasting grazing regimes described in Chapter 2. Selected plant species were E. vaginatum and N. stricta, a sedge and a grass respectively, which exhibit contrasting responses to long-term grazing (E. vaginatum is abundant in ungrazed conditions whereas N. stricta does in long-term grazed conditions). Overall, it was found that grazing can exert indirect effects on plant competition through its effects on soils. Thus, the competitive effect of Nardus stricta increased twofold when this species was grown in soils from continuously-grazed grasslands where it is dominant. On the other hand, when Eriophorum vaginatum (dominant in ungrazed habitats) was grown in soils from the continuously-grazed grasslands, its competitive effect on N. stricta was reduced by half. These results are consistent with those found in other studies where grazing effects on soil properties or microbial communities have been shown to affect plant performance (Frank et al., 2003; Sørensen et al., 2008a,b). However, as far as I am concerned, this is the first proof that grazing induced platsoil feedbacks can affect differentially the components of plant competition (intraand inter-specific). Differential effects of soil from sources differing in grazing management coincided with both differences in soil properties previously observed in the field study (Chapter 2) and plant traits of the superior competitor (Chapter 3 and Chapter 4). In this way, the grazing-conditioned soil showed higher rates of microbial activity and N availability in comparison with the ungrazed-conditioned soil. These differences in 128 6.4 Indirect effects of grazing alter plant competition soil properties between grazedand ungrazed-conditioned soils might be explained by differences in soil microbial communities or in the quality of soil organic matter between the continuously-grazed and the ungrazed areas. With respect to plant traits, traits of the superior competitor N. stricta related to its allocation of biomass into exploitative structures might have contributed to its higher competitive ability in comparison with E. vaginatum. Thus, N. stricta showed higher proportional biomass in roots, unlike E. vaginatum which did in storage structures (crowns). This is consistent with what I observed in Chapter 3 , where N. stricta capacity to withstand defoliation was strongly related to its high root biomass allocation, and with those findings of Hartley and Amos (1999) where N. stricta higher competitive ability in comparison with dwarf-shrubs was associated to its root traits. Other potential mechanisms by which soil microbes can alter plant-plant interactions include a decrease in plant performance as a result of soil-borne pathogens, and mechanisms related to the role of soil microbes in regulating plant niche overlap and competition for soil resources (Reynolds et al., 2003; Bever et al., 2010). For instance, with respect to soil micro-organisms effects on plant niche overlap, continuous grazing might have affected the arbuscular mycorrhizal fungi community, so that plant species might have had differential access to soil nutrients across grazed and ungrazed sites. It is well known that mycorrhizae can increase plant performance under nutrient limiting conditions (Bever et al., 2001), and that sedge species like E. vaginatum do not rely on these fungal symbionts (Wein, 1973; Brundrett, 2002). This might have given N. stricta a competitive advantage on E. vaginatum. Whether changes on soil mycorrhizae and soil microbial communities are related to the competitive balance between upland plant species is something which remains to be clarified. 129 6.5 Indirect effects of grazing are not explained by soil microbes effects on soil resource competition 6.5 Indirect effects of grazing are not explained by soil microbes effects on soil resource competition I explored into more detail whether the differential effects of herbivore-pant-soil feedbacks observed in Chapter 4 and described above could be caused by soil micro-organisms effects on plant niche overlap of soil resources, namely inorganic N (IN) and organic N (ON). To do this , I set up another microcosms experiment with the specific objective of linking herbivore-plant-soil feedbacks effects on plant competition to effects of soil microbes on N uptake preferences of E. vaginatum and N. stricta (specific objective 4, see Chapter 1). This experiment included the use of stable 13C and 15N isotopes in a plant competition experiment. I hypothesised that microbial communities from areas subjected to contrasting grazing regimes would affect plant competition for organic and inorganic N. I based this hypothesis on previous results of grazing-induced feedbacks on plant competition (Chapter 4, see above), and because microbial communities differing in activity or composition (as those between continuously-grazed and ungrazed areas, Chapter 2 and Chapter 4) could alter plant N competition by modifying N cycling. In contrast to my hypothesis, N competition between N. stricta and E. vaginatum was not explained by different microbial communities associated with different grazing regimes. Lack of effects of microbial communities that differed in activity on plant N uptake preferences is in contrast to results of a previous study which showed that the stimulation of microbial activity led to changes on plant N uptake preferences (Dunn et al., 2006). Soil microbial biomass C and N measured at the end of the plant competition experiment (Chapter 5) were much lower than those found in the field (Chapter 2), and than those observed in the microcosm experiment where herbivore-plant soil 130 6.6 Ecological and management implications feedbacks affected plant competition (Chapter 4). These difference in soil microbial properties among the experiments, and the fact that I did not measure the fraction of N taken-up by soil microbes in the N-competition experiment, make it difficult to interpret whether soil microbes from soil with different grazing management were functionally equivalent (Wertz et al., 2006, 2007), or whether the lack of effect was due to a small size of the microbial community. Studies testing effects of microbial communities on N uptake patterns and plant competition for N are scarce. Therefore, more research on this topic would shed light on how soil microbial composition can affect plant niche overlap, and thus plant species competition and coexistence. As mentioned above, it is likely that N. stricta out-competed E. vaginatum in grazedconditioned soils more easily based on its root traits and the higher activity of microbial communities to mineralize N in these soils. Other plant trait potentially related to higher competitive ability unexplored so far might be the capacity to modify preferences of IN and ON uptake in competitors, as N. stricta did on E. vaginatum. This N uptake preference change caused by N. stricta on E. vaginatum is likely to have increased N niche overlap in these plant species, with potential implications for mechanisms by which these species coexist in semi-natural mountain ecosystems, and which are discussed next. 6.6 Ecological and management implications What are the ecological and applied implications of the results observed in this thesis? On one side, taken cautiously, my results (Chapter 2) have implications on total soil C and N storage policies and land use management in upland ecosystems. Since these systems are among the major reservoirs of C on terrestrial ecosystems (Milne and Brown, 131 6.6 Ecological and management implications 1997), attention is being devoted to understanding the regulation of C dynamics as C sequestration in organic soils as an important ecosystem service (Dominati et al., 2010) delivered by these areas. The grazed grassland had higher rates of nutrient cycling (higher N mineralization rates),and microbial activity, suggesting that grazing could lead to higher decomposition rates and less C sequestration in soils. Nevertheless, this was not the case, and the continuously-grazed grassland had similar total soil C values as those in the ungrazed area. This agrees with results found in other studies in nutrient-poor semi-natural mountain ecosystems showing that long-term grazing exclosure does not lead to increases in total soil C, at least in the superficial organic soil horizons (Garnett et al., 2000; Ward et al., 2007, but see Bardgett et al. 2001). Therefore, grazing as a land use activity could be compatible with soil C sequestration in these ecosystems. However, more extensive sampling and experimental studies varying grazing intensity are necessary to prove this assertion. Interestingly, it has been shown recently that grazing can increase the temperature sensitivity of soil C in the rich-organic soils of semi-natural mountain areas (Paz-Ferreiro et al. in preparation). This last finding suggests that grazing could interact with climate warming to effectively increase soil respiration in these ecosystems. Additionally, whereas grazing cessation might not increase soil C in the order of a few decades, grazing cessation is still associated with an increase in the size of the litter horizon, whose loss under grazing by livestock represents a considerable C loss (Ward et al., 2007). My results also have implications for understanding the mechanisms that determine plant species coexistence across gradients of grazing intensity and soil properties in temperate semi-natural mountain grasslands (Chapter 4 and Chapter 5). Nardus stricta and E. vaginatum have marked differences in their ecological optima across the gradients 132 6.6 Ecological and management implications in grazing intensity and soil properties (particularly moisture), which characterise seminatural mountain grasslands. Whereas E. vaginatum occurs preferentially in areas with higher soil moisture (Edgell, 1971; Wein, 1973), N. stricta tends to dominate in longterm grazed areas. Higher occurrence of N. stricta in long-term grazed conditions is thought to result from the avoidance of this species by herbivores (Welch, 1986) caused by its low palatability as a result of high contents of silica in leaves (Massey et al., 2007, , although note that less selective herbivores such as cattle can graze heavily on it, and that my results in Chapter 3 suggest that N. stricta also exhibits tolerance). Despite their different ecological optima, these two plant species coexist along the gradients in grazing and soil moisture content occurring in different pairwise densities (Chadwick, 1960; Wein, 1973; Rodwell, 1992)). My results on plant competition in Chapter 4 and 5 showed that N. stricta is the superior competitor, at least in the conditions tested in the microcosm experiments. Combined, this higher competitive ability of N. stricta, the strong positive soil-feedback present in this species in grazed-conditioned soils (Chapter 4), and the higher niche overlap in N uptake preferences when both species competed (Chapter 5), suggest that mechanisms different to those based on niche complementarity must operate in order for these plant species to coexist in areas where they co-occur. These other coexistence mechanisms might be related to those generating spatial heterogeneity (i.e. in soil conditions such as moisture), and/or those that promote non-equilibrium species densities (i.e. gap creation by herbivores) which reduce competitive exclusion of E. vaginatum. 133