Ch 6 General Discussion Essay Example

Ch 6: General Discussion Paper Overview The effects of biodiversity on ecosystem functioning, hereafter called biodiversity-function, is a vast field of study. It connects the maintenance of diversity in species communities with the fluxes of energy and matter in ecosystems. Biodiversity-function relationships found in experiments are a major development in basic ecological science, which can bridge the gap between population, community and ecosystem scales of study. They highlighted an indirect consequence of biodiversity loss, that the functioning of entire ecosystems may be threatened, including those that provide services for humans. But to improve both our ecological understanding of the functional role of biodiversity in ecosystems and our applied understanding of how real-world ecosystems are affected by biodiversity change, we need further developments. Firstly, we need to explain what biological mechanisms drive the biodiversity-function relationships found in controlled experiments. Secondly, we need a next generation of long-term, field-scale experiments conducted in complex landscapes, which will have direct relevance to real-world ecosystems and their management. The work I have presented has an ambitious scope: beginning with the study of interactions between populations and the mechanisms for biodiversity effects, and ending with human impacts on biodiversity and how we might use this research to improve the conservation of real-world ecosystems. In chapters 2 and 3, I tested a method for measuring plant interactions in natural communities and then analysed how those interactions might determine properties of plant communities. These chapters firstly give a potential resolution to debate on what forces shape the diversity and composition of plant communities and then improve our understanding of the mechanisms by which diverse plant communities can enhance ecosystem functioning. I found that a major method for measuring competition in natural communities is flawed, and recommended that different approaches are required to truly measure the role of species interactions in structuring plant communities. I then used simulation modelling to connect experimental evidence with relevant theory, to assess when we might expect to find definitive evidence of positive biodiversity effects. These two chapters dealt with the mechanistic basis for biodiversity-function relationships and our technical ability to describe them—vital f or interpreting past experiments and as a basis for progressing into real-world ecosystems. We will write a custom essay sample on Ch 6: General Discussion specifically for you for only $16.38 $13.9/page Order now We will write a custom essay sample on Ch 6: General Discussion specifically for you FOR ONLY $16.38 $13.9/page Hire Writer We will write a custom essay sample on Ch 6: General Discussion specifically for you FOR ONLY $16.38 $13.9/page Hire Writer In the remaining chapters, the concern was firstly to push the frontier of biodiversity-function research into complex landscapes, where this research can have more applied impact. And secondly to equip ourselves with the knowledge on biodiversity change that is required to comprehend the real-world importance of this research. I analysed results from one of the first experiments in complex landscapes, where biodiversity change has been severe and the potential cost to ecosystem functions and services is great. I found initial evidence, from a long-term experiment, that diversity could help improve the effectiveness of forest restoration. I then scaled up even further to focus on global biodiversity change, rather than its functional role in ecosystems. To know how biodiversity change will impact real-world ecosystems we must know what change is occurring. I found that intensive agriculture reduces the diversity of farmland wildlife by a third overall. We might use this information to meet 6.131 global food requirements whilst minimising its cost to wildlife. Other work that I have been involved with looked at the effects of land use more widely and found that the impacts of human land use on biodiversity are severe, but there is scope for future mitigation (see Appendix III). Measuring interactions in plant communities In chapter 2, I tested the predominant method for measuring the effect of competition in natural plant communities. The method is an observational approach that assumes we can infer the effects of interactions from natural variation in the densities of co-occurring species. The effect of competition between species is estimated by predicting how the population size of each species would respond to the removal of competitor species. We performed this observational analysis on experimental mixtures and compared the prediction with our test: monocultures, where species grow without competitor species. The method consistently underpredicted the effect of competitor removal, undermining any inferences that can be made using this method regarding how interactions structure plant communities. If plant species did not interact, understanding the effects of diversity on ecosystem functioning would be simple. We would only need information on the abundances of each species in a community and their performance in monoculture. But this is clearly not the case. Even selection effects of diversity are the product of interactions. Huston (1997) described the selection probability effect—that diversity increases the probability of including any species—as an artefact of experimental design. But others showed that this probabilistic side-effect is not in itself sufficient to create any effect of diversity (Loreau Hector 2001; Cardinale et al. 2004; Weis et al. 2007). For any species to contribute to a diversity effect, they must alter the per capita performance (e.g. relative yield) of other species through interactions like competition (Loreau Hector 2001). Thus, regardless of debate on what mechanisms drive diversity effects (see Mechanisms of diversity effects), competition is key to understanding how species combine in diverse communities to influence ecosystem functions. The method tested in chapter 2 estimates the effect of competition by predicting the impact of species loss. Such a method is clearly relevant to the study of the impacts of biodiversity loss on the community-level properties that drive ecosystem functions. Understanding the role of competition in structuring plant communities and consequently driving ecosystem functions requires tried and tested methods for quantifying the strength of competition. Evidence of interactions in biodiversity experiments is clear, because the biomass of multi-species communities cannot be described as just the additive combination of monoculture biomasses (Cardinale et al. 2011). But taking biodiversity-function research out into natural ecosystems means that we need robust methods for measuring interactions in natural communities, which can reliably predict the effects of losing species. There are many methods for measuring competition—some are experimental, some are observational—but overall they do not give the same results (Gurevitch et al. 1992; Rees et al. 1996; Martorell Freckleton 2014). We therefore need to understand why different methods give different results, which methods accurately predict the effects they intend to quantify, and what causes any inaccuracies in these predictions. The flaws in the major experimental methods are well documented (Connell 1983; Freckleton Watkinson 2000), but until now the observational methods had received less scrutiny. The method tested in chapter 2 is predominantly conducted using observational data, but interestingly the approach was first taken using experimental data (Mack Harper 1977). An experiment could control for confounding variables like soil nutrients, moisture, light and temperature. Confounding variables can obscure the effects of the density of one species on the future density of another species, because correlation in the environmental responses of two species can be misconstrued as an effect of species interactions. This is one of the potential reasons why the observational method poorly predicted the effects of competitor removal. If true, it would highlight where experiments can help in the study of plant competition. However, we underpredicted the effect of competitor removal even in our semi-controlled experiment. Our primary explanation, that it is impossible to infer the fundamental role of competition in natural communities because of the ghost of competition past, suggests the problem is more profound. This would explain why experimental and observational evidence do not generally agree, but it would not yet provide a solution. Further modelling is required to confirm this idea. Perhaps such modelling will suggest new methods, but it may potentially show that we will never be able to use observational techniques to quantify whether plant communities are fundamentally competitive. Debate on the importance of interactions will continue unless this matter is resolved. Debate over methodologies and interpretation have made species interactions perhaps the most contentious topic in all of ecology. The role of local interactions in shaping the assembly and composition of plant communities has been a dividing line in many of the field’s important developments (Lewin 1983; Connor Simberloff 1984; Gilpin Diamond 1984; Abrams 1986; den Boer 1986; Roughgarden 1986; Adler et al. 2007; Vellend 2010). Much of community ecology is based on the idea that niche partitioning and competition are key to understanding the maintenance of biodiversity (Darwin 1859; Hutchinson 1957; Macarthur Levins 1967; Chesson 1991, 2000; Levine HilleRisLambers 2009; HilleRisLambers et al. 2012). But others have claimed that large-scale forces like dispersal and drift are more important (Connor Simberloff 1979; Ricklefs 1987; Ricklefs Schluter 1993; Hubbell 2001). Study of competition is important for biodiversity-function research, because much of the ecological theory invoked to explain diversity effects is based on coexistence between competitive species, often involving competition for resources (Tilman Downing 1994; Tilman et al. 1997a; Loreau 1998a, 2010b). However, just as with other topics in community ecology, the role of niche partitioning caused by competition has been hotly debated in biodiversity-function research (Tilman et al. 1996; Aarssen 1997; Huston 1997; Tilman 1997; Loreau 1998b; Hector et al. 1999; Hector 2000; Huston et al. 2000). There is still a lack of strong evidence for niche partitioning as the major cause of diversity effects, perhaps because such specific mechanisms are rarely cited when diversity effects have been measured (Cardinale et al. 2011). The early debates were addressed by quantifying complementarity and selection effects (Loreau Hector 2001), but in order to develop a mechanistic understanding of diversity effects we now need to move beyond these terms (Carroll et al. 2011, 2012; Loreau et al. 2012; Turnbull et al. 2013). Doing this will require further modelling, measuring, and testing of species interactions. By improving methods to infer competition from natural communities, we can better inform hypotheses about how diverse plant communities in nature can sustain ecosystem functions. For other work I have contributed to, that examined the predictability of plant–soil interactions rather than plant–plant interactions, see Appendix II (Mehrabi Tuck 2015). Mechanisms of diversity effects In chapter 3, I explored the modelling that will be required to understand what mechanisms drive biodiversity effects. I presented a model of the seasonal growth of species that competed for one shared limiting resource. Species differed according to a functional trade-off between their rate of resource capture and the volume of resource pool they can access. This trade-off enabled stable coexistence and complementarity effects. But it was not possible for diverse mixtures to reach a higher yield than the best monoculture of its component species without extra niche differentiation. We hypothesised that mixtures might yield more than the best monoculture if we included environmental resource leaching throughout the season. We assumed that resources not yet locked up in plant tissues could be leached from the soil profile. We simulated growth of communities with varying species richness and measured how much of the resources had been captured and leached by the end of the season. We expected mixtures to capture more and leach fewer resources, because species that could capture resources quickly would minimise leaching early in the season, whilst species that could access more resources in total would continue to exploit the remaining resource pool later on. As we expected, some species mixtures did capture more resources than the best monoculture in the presence of leaching—although this was only a small proportion of all mixtures. Resource leaching differentiated species in time by making some resources accessible only to those species that can capture resources quickly enough. Mixtures that combine the strengths of different species along the 6.136 trade-off could capture resources effectively throughout the entire season, thereby reaching a higher season-end yield. Whilst it was rare for mixtures to outperform the best monoculture, they frequently performed as well as the best monoculture due to the same mechanism. This mechanistic model offers a biological explanation for biodiversity effects observed in long-term experiments. The modelling presented was intended to draw directly from long-term biodiversity experiments, where there is greatest opportunity for measuring and testing how species interact and what effect this has on ecosystem functions. The case for complementarity effects such as resource use of species in mixtures has been demonstrated in theory. Models have shown that mixtures should commonly yield more than expected from the properties of its component species, which is known as overyielding. Experiments have verified this expectation, and some biological mechanisms for overyielding have even been elucidated. A greater number of models have been used to explore when we should expect mixtures to yield more than its highest yielding monoculture, also known as transgressive overyielding. But the mechanisms that could generate this most definitive effect of diversity have been less well explored and, as transgressive overyielding has mostly been found in long-term experiments, it is not clear how often we should expect to see it. By now much experimental evidence has shown that complementarity effects are common, even though the role of niche differentiation is still unclear (Cardinale et al. 2007, 2011). Multiple species usually contribute to the increased biomass in diverse mixtures (Tilman et al. 2001; Hector Bagchi 2007). Positive complementarity effects often increase over time (Reich et al. 2012; Ravenek et al. 2014)—sometimes with concomitant decreases in selection effects (van Ruijven Berendse 2003; Fargione et al. 2007). These patterns suggest that the complementarity effects observed, at least in these experiments, are not a transient effect of artificial species mixtures (but see Turnbull et al. 2013). Various mechanisms for complementarity have been suggested. At the Jena Experiment, some suggested mechanisms have been that species differ in rooting depth and architecture (Dimitrakopoulos Schmid 2004), but more recently that higher diversity might reduce the effects of plant–soil feedbacks (Ravenek et al. 2014). Similarly at Wageningen, diverse mixtures were shown to use nitrogen more efficiently (van Ruijven Berendse 2005) but negative density-dependent effects may also be important, for example species-specific root herbivory by nematodes (De Deyn et al. 2004; van Ruijven Berendse 2009). Other experimental work in similarly agricultural contexts found that transgressive overyielding was common and linked to having a diversity of functional traits regarding resource acquisition and growth strategies (Finn et al. 2013). Perhaps the best evidence of transgressive overyielding has been from Cedar Creek (Tilman et al. 2001), where diverse communities showed increased input and retention of nitrogen due to complementary rooting and belowground resource use, primarily between legumes and C4 grasses (HilleRisLambers et al. 2004; Fargione Tilman 2005; Fargione et al. 2007; Mueller et al. 2013). Thus, belowground rooting and resource capture mechanisms were a natural choice to explore transgressive overyielding in mechanistic models. Loreau (2010a) showed what conditions are required to find transgressive overyielding using a two species Lotka-Volterra model. Stable coexistence, and hence 6.138 overyielding, requires that the inferior competitor is limited more by itself than by the competitive effect of the superior competitor. In this case, the inferior competitor would overyield but the superior competitor would not, resulting in a lower yield than the superior competitor in monoculture. Transgressive overyielding requires the additional condition that both species are limited more by themselves than by the other species. To summarise the relationship between coexistence and overyielding: overyielding can occur even in transient mixtures and hence does not guarantee persistent diversity effects (Carroll et al. 2011, 2012; Loreau et al. 2012; Turnbull et al. 2013); if mixtures stably coexist, it follows that they will overyield—there is sufficient niche differentiation for a community to yield more than expected from the properties of its component species; but for a community to yield more than its highest yielding species in monoculture, i.e. show transgressive overyielding, even stronger niche differentiation is required than that necessary for stable coexistence (Loreau 2010a). Therefore, in simple Lotka-Volterra models, the conditions for coexistence and diversity effects are clear. Nevertheless, measuring competition in plant communities remains problematic (see chapter 2) and measuring the effects of niche differentiation has rarely been achieved (but see Levine HilleRisLambers 2009). Wi thout strong measurement it will remain difficult to know when conditions for diversity effects are fulfilled. In more complex models of plant competition, that for example allow nonlinear per capita population growth rates, the relationship between coexistence and diversity effects can be less clear (Gilpin Justice 1972; Loreau 2010a). It is still not known how frequently populations exhibit nonlinear population growth functions. And some mechanisms for species coexistence, such as transient nonequilibrium coexistence, do not predict the same functional consequences of diversity for ecosystem functions (Loreau 2010a). So question marks remain on the generality of such a simple set of conditions to necessitate persistent effects of diversity in plant communities. The model presented in chapter 3 helps identify the conditions under which we might expect to find transgressive overyielding. It appears that even when mechanisms for transgressive overyielding are at work, we should not expect mixtures to yield more than the best monoculture very often. This gives new light in which to consider the rarity of transgressive overyielding observed in experiments. Future work combining experiment and theory should continue to elucidate when and how mixtures outperform monocultures. Doing so will strengthen our understanding more broadly of the functional consequences of species diversity for ecosystems. Understanding theory and small-scale experiments is an important platform for scaling up to the experiments in complex landscapes that can inform management of real-world ecosystems. Diversity effects in complex landscapes In chapter 4, I presented initial results from one of the first long-term biodiversity-function experiments that will be conducted in complex environments, at landscape level, with real-world application. I analysed the first decade of survival and growth at the Sabah Biodiversity Experiment, which is situated in selectively logged lowland forest in South East Asia. The experiment will elucidate biodiversity-function effects in tropical forests, and will help inform the restoration of these degraded ecosystems. The experimental seedlings were planted into the background forest in a way that replicated the restoration practice. By analysing the survival and growth of these seedlings, we could estimate how many replanted trees remain and at what stem density. We found that species differed in survival and growth, following a survivalgrowth trade-off. Species also responded differently to the wide range of conditions throughout the landscape. These differing responses could create a spa tial insurance effect of diversity, thereby ensuring successful restoration throughout the complex landscape. The effect of plant diversity on ecosystem functioning has been extensively studied in small-scale, controlled conditions that usually strive to minimise environmental heterogeneity (Cardinale et al. 2011). In previous chapters I explored the methodological and theoretical developments required to refine our understanding of the mechanisms underlying the biodiversity effects already observed. These developments provide a platform for another frontier in biodiversityfunction research: taking experiments out into natural plant communities, with a more complex range of life histories, in complex landscapes, that directly relate to real-world conservation management (Duffy 2009; Hillebrand Matthiessen 2009; Brose Hillebrand 2016). The experiment in chapter 4 is one example of this growing effort. Tropical forests are important for, amongst other things, their rich diversity, the locally and globally valuable ecosystem services they provide, and economically valuable products such as timber (Sodhi et al. 2010; Maycock et al. 2012; Edwards et al. 2014). Changes in land use, particularly due to agriculture and logging, have drastically changed tropical forests and will continue to do so (Newbold et al. 2015, see   Appendix III). Much biodiversity is threatened or has already been lost. In addition to the direct costs of this biodiversity change, there may also be knock-on effects at the ecosystem level, on the functioning of tropical forest ecosystems and the services these functions provide us. But the functional consequences of biodiversity change for tropical forest ecosystems is not clear because very few experiments have explored these regions (but see Potvin Gotelli 2008; Yang et al. 2013). Because changes in tropical forest are current and due to human activity, they also present a useful setting to explore the real-world applications of biodiversity-function research. In South East Asia, vast areas have been selectively logged and already restocked by enrichment planting programmes. The enrichment planting was partially intended to aid forest restoration by helping reproduce the emergent canopy of old growth forest. But the effectiveness of this technique has not been fully tested, despite its widespread application. The history of land use, enrichment planting, and the natural variation in environmental conditions that is inherent to the systems is complex, producing a landscape that is fragmented and patchy at varying spatial scales. In chapter 4, I showed that enrichment planting with diverse mixtures of trees may spread the risk of failed restoration in complex landscapes, by utilising speciesspecific responses to variable environmental conditions. The average restorative effect of enrichment planting would be maintained throughout the whole landscape. Whereas monocultures might, in unfavourable areas, fail to achieve any restorative effects (or retain overly dense stands of trees, potentially leading to wasteful selfthinning that would undermine efficient enrichment planting across large areas). The results in chapter 4 show the potential improvements in enrichment planting, as informed by biodiversity-function research. Improving this practice may help sustain the functioning and conservation value of these forests. How much of this potential is realised will only become clear as this long-term experiment continues. Further monitoring needs to observe future survival and growth until the planted trees mature and reproduce. It will then be interesting to see how interactions between mature trees affect mixture performance relative to monocultures, and whether planting boosts recruitment of future dipterocarp generations. The restorative effect of this practice on the background degraded forest, which seedlings were planted into, is not yet clear. This is an important step for quantifying how replanting more diverse plant communities can boost functioning of the wider forest ecosystem over time. It has been shown that effects of biodiversity on ecosystem functioning not only increase over time, but also with increasing spatial scale (Dimitrakopoulos Schmid 2004; Venail et al. 2010; Cardinale et al. 2011), and increasing environmental heterogeneity (Finke Snyder 2008; Tylianakis et al. 2008). This suggests that smallscale controlled biodiversity experiments may have underestimated the impact of biodiversity loss on ecosystems, and that study in large-scale, complex environments will be needed to estimate the full effects (Cardinale et al. 2012). Such heterogeneous environments, with multiple sources of variability and more opportunities for niche differentiation, will be ideal settings to study the effects of biodiversity on multiple ecosystem processes at the same time (Duffy 2009; but see Wardle Jonsson 2010). There are many challenges for biodiversity-function research in complex landscapes. It will be harder to control external factors, reducing our power for inference. Many factors affect ecosystem processes and there is mixed evidence on their relative importance as drivers of ecosystem change (Grace et al. 2007; Hooper et al. 2012; Tilman et al. 2012). One solution could be to analyse biodiversity-function relationships within a constrained set of conditions. The constraining effect of productivity on biodiversity has been measured correlatively at large scales (Mittelbach et al. 2001; Adler et al. 2011; Fridley et al. 2012; Grace et al. 2012, 2016; Pan et al. 2012). Though this correlative work fundamentally differs from experimental work, it might aid research in complex landscapes. If external factors cannot be controlled it would help to know how they interact with the biodiversity-function relationship. Then they may be controlled post hoc and the biodiversity-function relationship can be examined, constrained within the conditions found in the landscape (Loreau 2010a). Globally distributed experiments are a more controlled way to find general patterns in forest ecosystems (Borer et al. 2014). Our experiment is part of such a network (Verheyen et al. 2015). There are also emerging methods and topics that will help extend biodiversity-function study into complex landscapes (see What is the future?). The scale of the problem In chapter 5, I quantified the effects of different farming strategies on farmland biodiversity, including plants and many other taxa. I did this by meta-analysing 30 years of published studies that compared the farmland biodiversity found on intensive conventional farms and extensive organic farms. I found that 34% of overall farmland wildlife is lost on conventional farms relative to organic farms—for plants alone, 73% of species are lost. The biodiversity experiments in grasslands, often rooted in landscapes with agricultural history, show large effects of plant species loss on ecosystem functioning (Tilman et al. 2001; van Ruijven Berendse 2003; Roscher et al. 2005). There is still debate on the relative merits of extensive and intensive farming for global biodiversity (Foley et al. 2011; Phalan et al. 2011; Tilman et al. 2011), but evidence suggests that intensive farming is especially damaging for the functioning of the ecosystems in which the farms are situated. This r esearch was successfully communicated to European policymakers (see Appendix IV). To understand the real-world relevance of biodiversity-function research, we need to know how biodiversity in real-world ecosystems is changing. Global biodiversity is undeniably changing and the predominant cause is human activity (Pereira et al. 2010; Barnosky et al. 2011; Pimm et al. 2014; Ceballos et al. 2015). The problem is so serious that rates of biodiversity loss might exceed the boundaries of a planetary â€Å"safe operating space for humanity† even more dramatically than climate change (Rockstrà ¶m et al. 2009; Mace et al. 2014). Thus, we should measure how human activity drives biodiversity change and then understand the knock-on effects of this change for natural ecosystems—I have provided such measurements of biodiversity change within agricultural systems. To understand the functional consequences of global biodiversity for ecosystems, we need to solve problems about describing biodiversity and measuring its change. Species diversity is changing in a multitude of ways: the composition and structure of species communities, the dominance of species groups, species invasions and biotic homogenisation, and species being driven to rarity and ultimately extinction (Butchart et al. 2010; Magurran 2016). Extinction is irreversible—and the effects of species loss is the domain of biodiversity-function research—so it is crucial we understand how much extinction is going on. Out of the 5–10 million species that might exist on Earth, 1.9 million have been described (Mora et al. 2011)—though misidentification and synonymies present great uncertainty in our knowledge (Goodwin et al. 2015). According to IUCN (2015), 903 known species have gone extinct since 1600. Whilst this may comprise a small fraction of global biodiversity, it represents a rate of extinction 1000 times greater than that documented in the fossil record (Pimm et al. 2014). But lacking information on the most diverse taxa means this extinction rate could be grossly underestimated (Rà ©gnier et al. 2015). Uncertainty remains on the scale of global biodiversity and how it is changing, and what types of biodiversity change most impact ecosystem functioning. But wherever truth lies within that uncertainty it seems the scale of the global biodiversity crisis is severe. Within this global context, there is a current debate on how local species diversity has responded to recent pressures (Vellend et al. 2013; Dornelas et al. 2014; Newbold et al. 2015; Gonzalez et al. 2016). Some claim that there has been no overall loss in local species richness over recent times, because most loss is countered or even reversed by influx of species (Vellend et al. 2013; Dornelas et al. 2014). But others have criticised their approach for having spatial and temporal biases toward underestimating recent loss, and measuring change against inappropriate baseline conditions (Gonzalez et al. 2016). I contributed to work by Newbold et al. (2015), who took a different approach by using spatial variation rather changes over time (spacefor-time approach), and estimated substantial losses to local biodiversity across the globe (see Appendix III). More recently, others have suggested that this space-for-time approach underestimates the impacts of human land use on local biodiversity (Franà §a et al. 2016), so the projections put forward by Newbold et al. (2015) may well be conservative. The state of biodiversity change we see may also depend on what type or 6.146 metric of biodiversity we measure (Pereira et al. 2013; McGill et al. 2015). Ecosystem functioning depends more on local biodiversity than global biodiversity (Cardinale et al. 2012; Hooper et al. 2012). The biodiversity-function relationships from experiments are important results whatever current changes in local biodiversity may be, but the broad consensus is that local diversity is declining and the functioning of ecosystems may be threatened. Even when local diversity is supplemented by an influx of other species, this may only delay local diversity loss (Gilbert Levine 2013) or it may homogenise regional biodiversity (McKinney Lockwood 1999; McGill et al. 201 5). What is the future? Emerging methods When moving into natural communities, synthetic approaches between experimental and observational study may be useful in maximising inference from complex landscapes. For example, experiments in complex landscapes inevitably impose a simplistic and discrete nature to provide more controlled study. But the discrete nature of experimental design will not reflect the landscapes in which they are situated, which may harm their ability to provide meaningful recommendations. One option may be to impose less discrete experimental treatments, for example by planting more continuous gradients of diversity in a way that fits the landscape. The landscape would then be part of the experimental design, rather than randomised away. This would pose problems for inference, as it would inevitably undermine the advantages of experiments. But the synthetic approach would be complementary to such experiments and may provide extra realism to biodiversity-function studies. This change would also benefit from emerging analytical techniques. In the chapters presented here, there are a wide range of analyses implemented: predictive testing (using Bayesian inference), data-free simulation modelling, exploratory data analysis, and meta-analysis. There are many other analytical techniques that may become part of the ecologist’s standard toolkit, often because they offer new ways to tackle the fact that ecological effects are conditional on multiple causes. Smoothing techniques such as Generalised Additive Mixed Models (GAMMs) have become very useful, and the theory behind them is becoming more complete (Wood 2006). GAMMs could utilise the continuous nature of new experiments in complex landscapes, to not only account for variation in the landscape but also capture that information for further inference. This would be useful, for example, for estimating the spatial scale of species interactions or spatial variation in diversity effects. As well as GAMMs, structural equation modelling (e.g. Grace et al. 2016) and quantile regression (e.g. Grubb 2016) are among the techniques that will be useful in elucidating effects in dynamic, realworld ecosystems. Remote-sensing is another powerful tool that has emerged in recent decades, as new satellites were launched and started releasing freely available data (Pettorelli et al. 2014). Researchers interested in any global change can use the consistently measured, global yet finely grained, remotely sensed data to ask questions that would otherwise have been impossible to answer (Pettorelli et al. 2005; Asner et al. 2008, 2014). I have produced a tool to access and use one of these archives of remotely sensed data (Tuck et al. 2014a). To date, this tool has been used by over 4000 researchers. Emerging topics Spanning such a broad swathe of research means having to decide what not to study. For example, adding trophic complexity beyond plant communities will improve assessments of the functional consequences of biodiversity change in complex landscapes. Food webs provide a quantitative framework to connect community ecology—the study of species richness, composition and interactions—with ecosystem ecology—the study of fluxes of energy and matter. Biodiversity-function research sits between these two fields of ecology, so using food webs to move beyond single-trophiclevel communities could help build a quantitative framework for the ecosystem-level consequences of biodiversity change (Worm Duffy 2003; Hillebrand Cardinale 2004; Cardinale et al. 2006; Duffy et al. 2007; Thompson et al. 2012). These are rich areas for theory and experimentation, and challenges remain for this framework to become truly predictive. There is evidence that diversity loss among trophic groups has a greater impact on ecosystem functioning than loss within trophic groups (Duffy et al. 2007; Cardinale et al. 2012; Barnes et al. 2014). But the exact structure of trophic networks between consumers and predators can alter biodiversity-function relationships, and these trophic structures are not easily predicted (Yodzis 2000; Thompson et al. 2012; Digel et al. 2014). Network complexity and the food web approach is also a frontier for species coexistence and evolutionary ecology (Chesson Kuang 2008; Allesina Levine 2011). It is possible that mechanisms of coexistence, functional traits, and trophic networks could be combined to model entire communities from the individual- up to the ecosystem-level, and assess the functional consequences of biodiversity change, analogous to how General Ecosystem Models simulate properties of the biosphere (Purves et al. 2013). Community evolution models are a promising addition to research on community and ecosystem ecology (Loreau 2010b). These emerging models could be used to study the coevolution and maintenance of diverse food webs and its ecosystem-level properties (e.g. Brà ¤nnstrà ¶m et al. 2010). They could potentially deliver a more mechanistic and predictive understanding of the structure and functioning of ecosystems (Loreau 2010a). There is a growing trend to consider the effect of biodiversity on multiple ecosystem functions at the same time, so called multi-functionality (Hector Bagchi 2007; Gamfeldt et al. 2008, 2013; Allan et al. 2013; Soliveres et al. 2016). Ecosystem multi-functionality appears to present an even stronger positive role of plant biodiversity in maintaining the functioning of ecosystems (Isbell et al. 2011). When considering only one ecosystem function there may be relatively few important aspects of species niches, so many may seem functionally redundant—although this may be a by-product of the types of short-term experiments most often conducted (Reich et al. 2012; Delgado-Baquerizo et al. 2016). However, when multiple ecosystem functions are considered more niches axes may be relevant and species differences become functionally important. Some have suggested that true redundancy might not exist (Loreau 2004). Important work is still needed to discover how ecosystem multifunctionality responds to biodiversity change and crucially whether any functions trade off with one another, such that biodiversity loss may harm one important function whilst not affecting or even benefitting another. Traditional biodiversity experiments have been conducted under controlled environmental conditions and species composition treatments. The relationships that 6.150 emerge from these experiments might differ from those where the environment can vary, due to disturbances or climatic fluctuations, and species compositions can fluctuate accordingly—but the effect of environmental variation can be unexpected and depend on the ecosystem function being examined (Craven et al. 2016; Fischer et al. 2016; Flores-Moreno et al. 2016). Environmental changes might even result in less diverse mixtures that are originally more productive, but more vulnerable to future disturbance and hence prone to collapse (e.g. MacDougall et al. 2013). Measuring the effects of nonequilibrium conditions is an important step for future research in complex landscapes. This is particularly pressing in a world where human activity is rapidly changing landscapes (Drescher et al. 2016) and environmental extremes are becoming the norm (Fischer et al. 2016; Woodward et al. 2016). Concluding remarks The research presented here has helped make the study of diverse plant communities and their role in real-world ecosystems a more predictive science, rooted in mechanistic understanding. It has combined theory, experiment and observation in a range of ecosystems to improve both our fundamental understanding and our applied impact regarding the ecosystem-level consequences of global biodiversity loss. I have suggested methodological improvements to the study of natural plant communities, and used a suite of analytical techniques to inform European conservation policy and advise restoration strategies in threatened natural ecosystems. The future of biodiversity-function research is to continue down the same path: integrating multiple fields of ecology, solidifying our basic understanding of plant diversity and its role in functioning ecosystems, and verifying its relevance for the management of real-world ecosystems. The field will need to encompass a greater diversity of taxa, trophic interactions, ecosystems, ecosystem functions, and measures of biodiversity itself. Perhaps then this research might unify ecology, from populations up to ecosystems, and become an invaluable framework for the management of our environment and global biodiversity. Previous Page   Ch 5: The Effects Of Organic Farming On Biodiversity