Humans have indirectly influenced species at lower trophic levels by driving losses of apex consumers.Furthermore, humans have indirectly influenced species at higher trophic levels by driving losses of primary producers.Beyond these broad classes of apex consumers and primary producers, it remains challenging to identify minimum subsets of species that are particularly important for maintaining ecosystem structure and functioning.Here we use a novel method at the intersection of control theory and network theory to identify a minimum set of driver node species upon which ecosystem structure strongly depends.Specifically, humans could unintentionally completely restructure ecosystems (i.e., change species abundances from any initial values to any final values, including zero) by altering the abundances of these few critical driver node species.We then quantify the proportion of these driver nodes that are influenced by humans, top predators, and primary producers in several marine food webs.We find that humans could unintentionally completely restructure marine food webs while only directly influencing less than one in four species.Additionally, humans directly influence: (1) most or all of the species necessary to completely restructure each network, (2) more driver nodes than top predators, and at least as many driver nodes as primary producers, and (3) an increasing proportion of driver nodes over time in the Adriatic Sea.We conclude that humans have potentially huge impacts on marine ecosystems while directly influencing only the relatively small subset of species that are currently fished.It may be possible to reduce unintentional and undesirable cascading human influences by decreasing human impacts on driver node species in these and other food webs.
Biodiversity and ecosystem functioning research has been some of the most controversial of the last decade but rapid progress has been made by deriving hypotheses from the differing view points and challenging them with appropriate experimental and analytical tests (Loreau et al. 2001). Here we address some recent criticisms of the BIODEPTH project (Thompson et al. 2005) and show that: While legume species play an important role in the BIODEPTH results, patterns are not generally consistent with the multispecies sampling effect for legumes proposed by Huston & McBride (2002) as suggested in Thompson et al. (2005). The BIODEPTH results are also not consistent with transient biodiversity effects. Levels of species diversity were generally maintained over the 3 years of the project (i.e. little competitive exclusion) and diversity-productivity relationships in BIODEPTH generally strengthened during the experiments. Thompson et al. (2005) suggest that the effects of diversity on biomass production seen in the BIODEPTH project can be explained adequately by the mechanisms described by Huston & McBride (2002). Huston & McBride's proposed explanation is a complex multispecies sampling effect in which more diverse plots have a higher chance of simultaneously containing legume species, productive non-legume species and of occurring on nitrogen-limited patches. We cannot address initial soil nitrogen heterogeneity much in our experiment except at the Sheffield site where nutrients were evenly applied in solution on a sand substrate. Here, the presence of a positive diversity–productivity relationship (Fig. 1g) in the absence of soil nitrogen heterogeneity does not seem to be consistent with Huston & McBride (2002). To further examine the role of legume and non-legume species we briefly report the relationship between diversity and total plant biomass at the end of the BIODEPTH project for communities with and without nitrogen fixers (Supplementary Tables S1 and S2; data from Spehn et al. 2005). The BIODEPTH design is limited in how far it can distinguish between different aspects of diversity (including legume species) because these different aspects are collinear (Schmid et al. 2002). In a sequential statistical model the presence of legumes can be fitted before testing for additional effects of diversity although this runs the risk of attributing to legumes some of the effects of collinear aspects of diversity. Nonetheless, when we do this conservative analysis we usually do find additional significant effects of other aspects of diversity such as numbers of species and plant functional groups (Loreau & Hector 2001; Spehn et al. 2005). For total plant biomass (Fig. 1) there is a highly-significant interaction between the effects of species richness and legumes (F1,159 = 10·7, P < 0·001; Supplementary Table S1). Main effects are of limited interest and use in the presence of an interaction (but here it may be justified to cautiously examine them since they generally explain more variation than the interaction) but even when tested after legumes the residual main effect of species richness is of similar size to that of this group (Supplementary Tables S1a vs S1b). Individual site relationships vary (Supplementary Table S2) but only in Germany does the species richness effect depend on the presence of legumes (Fig. 1a; but see Scherer-Lorenzen et al. (2003) for consideration of effects on nitrate leaching as well as productivity). At other sites the species richness effects for communities with and without legumes are similar to each other (e.g. Ireland, Fig. 1e) and in Greece there is no effect of species richness on total biomass whether legumes are present or not (Fig. 1d). Therefore, with the possible exception of one site our results are not consistent with the multispecies sampling effect proposed by Huston & McBride (2002). Further detailed information on the complex effects of legume species in BIODEPTH can be found in many previously published papers (see Hector et al. 1999; Loreau & Hector 2001; Jumpponen et al. 2002; Mulder et al. 2002; Spehn et al. 2002; Gastine, Scherer-Lorenzen & Leadley 2003; Scherer-Lorenzen et al. 2003; Palmborg et al. 2005; Spehn et al. 2005). The relationship between sown species richness and total biomass (sum of above- and below-ground biomass at the end of the third growing season) for communities with (closed symbols, solid lines) and without legumes (open symbols, broken lines) in (a) Germany, (b) Portugal, (c) Switzerland, (d) Greece, (e) Ireland, (f) Sweden, (g) Sheffield, UK and (h) Silwood Park, UK. Lines are linear regression slopes from the statistical model with significant site-by-legume interaction that is presented in Supplementary Table S1. Note differences in Y axis limits and that symbols have been staggered to avoid overlap. More generally, a biodiversity experiment that deliberately omitted legumes produced results that were similar to the overall pattern from BIODEPTH: a loglinear relationship between diversity and productivity that strengthened over time and which was largely driven by complementarity effects (Van Ruijven & Berendse 2005). To date only one other biodiversity experiment without legumes exists. Grime (2001, p. 309) and colleagues found that a 12-species mixture, and a particular subset of three species that it contained, showed strong complementary nitrogen uptake (long-term results for productivity and other ecosystem processes are currently unpublished). If the results of biodiversity experiments are a feature of the first year or two as Thompson et al. suggest then they should weaken in the longer-term. While we could only maintain weeded communities at all sites for 3 years (but for up to eight at some) patterns were generally weakest in the first year and strengthened or maintained thereafter (Pfisterer et al. 2004; Spehn et al. 2005), as were those from comparable studies (Tilman et al. 2001; Hooper & Dukes 2004; Van Ruijven & Berendse 2005; Tilman, Reich & Knops 2006). The relationship between biodiversity and many ecosystem processes observed from these experiments therefore appears to strengthen over time, not weaken. Thompson et al. also argue that high diversity and productivity only coincide in biodiversity experiments in the first year or two before competitive exclusion erodes the high-diversity mixtures. In BIODEPTH we generally aimed to set our highest level of diversity at each site to approximate levels of diversity seen in unmanipulated 'reference' plots in neighbouring grasslands, in anticipation that these realistic levels of species richness would be maintained (Spehn et al. 2005). For four of the five sites with reference plots levels did indeed match relatively well (Fig. 2) and we did not observe substantial reductions in species richness. The main exception comes from the Portuguese experiment where diversity did decline substantially in the final year but due to an unusually extreme winter (see Caldeira et al. 2005). Another minor exception is the Swiss site where, although there were no reference plots, the highest level of diversity was apparently set slightly above that which could be maintained (at 32 species to complete the log2 diversity gradient) and declined slightly during the main weeded period of the experiment (Pfisterer et al. 2004). Target, observed and reference diversity levels at individual BIODEPTH fieldsites. Horizontal lines show the target (sown) levels for the highest level of diversity, 'Exp' shows average observed diversities of the same experimental plots in the third year of the experiments (with SEMs) and 'Ref' shows average diversity levels in neighbouring unmanipulated reference plots in the same year. Note that all experimental communities were at or below levels observed in the reference communities. We propose that the apparent conflict in the results of biodiversity experiments and observational biodiversity studies arises because they use different approaches (experimental vs correlational) to address different questions (within- vs across-habitat relationships) (Loreau et al. 2001; Schmid 2002). It is because the direct influence of diversity on ecosystem processes can be obscured by variation in environmental conditions (Fridley 2002; Mulder et al. 2002; Dimitrakopoulos & Schmid 2004) that it is necessary to perform biodiversity experiments and why it is not surprising that observational surveys and biodiversity manipulation experiments sometimes show different relationships (e.g. Wardle et al. 1997 vs. Wardle & Zackrisson 2005; Levine 2000). However, there are ways in which observational and experimental studies could be made more comparable. First, biodiversity experiments like BIODEPTH become more comparable to observational studies like the one at Bibury reported in Thompson et al. (2005) once the designed experiment is over and the weeding treatment that maintained the differences in diversity and composition is stopped, as has now happened at all of the BIODEPTH sites (Troumbis, Galanidis & Kokkoris 2002; Pfisterer et al. 2004; Dimitrakopoulos et al. 2005). Once weeding ceases (i.e. the simulation of species loss is stopped) the diversity–productivity relationship degrades as the diversity gradient decays through low-diversity plots gaining species and high-diversity plots losing species following exclusion by invaders. As biodiversity experiments generally attempt to remove underlying environmental heterogeneity (like soil depth at Bibury) no clear environmentally-driven pattern emerges in our experiments and species richness and productivity appear to be unrelated, at least in the first few years following the cessation of weeding (e.g. Pfisterer et al. 2004; Neßhöver 2005). On the other hand, Bibury could be made more closely comparable with BIODEPTH by creating an experimentally-controlled diversity gradient (by removing species from plots – cf. Symstad & Tilman 2001– or by synthesizing communities of differing diversity from the Bibury species pool). Indeed the herbicide application experiments that existed at Bibury prior to 1990 could serve as a removal experiment as long as the herbicide reduced diversity of the treatment plots relative to controls. More generally, comparison of experimental and observational studies may reveal something about the processes influencing diversity at different scales because biodiversity experiments show the effects of dispersal limitation in homogeneous environments while observational surveys show the effects of environmental heterogeneity in sites with presumably little variation in the propagule pool. If natural plant communities are mainly limited by dispersal of species from the pool into local areas, and competitive interactions amongst species are small and equal, we would expect surveys to show that plots with many species also have high biomass. At the largest scales plant diversity and productivity are often positively associated but the fact that regional surveys sometimes show negative or unimodal relationships suggests that interactions between environmental conditions and competition have played an important role in shaping these patterns. Such comparisons could inspire a new generation of biodiversity experiments in which the size of the effects of biodiversity are compared to those of other factors (fertility, soil depth, dispersal limitation, etc.). Nevertheless, biodiversity experiments may indicate the future impacts of species loss through processes that may impose dispersal limitation (habitat fragmentation, over-harvesting, climate change, etc.). The results of biodiversity experiments like BIODEPTH have turned out to be highly repeatable and have established that changes in biodiversity (both richness and composition) do have the potential to impact many different ecosystem processes (Loreau et al. 2001; Hooper et al. 2005; Balvanera et al. 2006; Cardinale et al. 2006; Worm et al. 2006). More generally, the first generation of biodiversity experiments has also led to new advances in relative-yield and covariance-based analytical methods (Loreau 1998; Loreau & Hector 2001), has added to the previously limited evidence for niche differentiation and complementarity effects in plant communities and has produced some unexpected results on the relationship between species traits and dominance (e.g. widespread negative selection effects (Hector et al. 2002)). We thank all our previous co-authors on earlier BIODEPTH papers for their past contributions to the project and an anonymous reviewer for useful comments. Diane Srivastava suggested the regional species pool and dispersal limitation perspective. Table S2a. Analysis of variance summary for the individual site analyses of species richness and legume effects fitting species richness first in a sequential statistical model. Labels as in Table S1. Table S2b. Analysis of variance summary for the individual site analyses of species richness and legume effects fitting legumes first in a sequential statistical model. Labels as in Table S1 and error terms as indicated in S2a. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
The dataset of 149 empirical food webs and the MATLAB code for simulated food webs analyzed in the paper "Will a large complex system be productive?" (published in Ecology Letters) is provided. Detailed methods see the paper.
Abstract Tree diversity has been shown to promote a broad range of ecosystem functions in forests. However, how important these effects are in driving ecosystem multifunctionality in natural forests, relative to other drivers, such as below‐ground biodiversity (e.g. soil microbial diversity), community‐level functional traits and environmental conditions, remains poorly understood. Here, we hypothesize that tree species or phylogenetic diversity (PD), stand structure, functional traits and soil microbial diversity jointly regulate temperate forest multifunctionality (FM) along a local‐scale environmental gradient. Using repeated census data from a 25‐ha old‐growth temperate forest, we first quantified eight ecosystem functions and properties related to above‐ and below‐ground nutrient cycling. We then used these to estimate ecosystem multifunctionality using both averaging and multiple threshold (50%, 75% and 95%) approaches. Finally, we used structural equation models to explore how different facets of tree (tree species, functional and PD) and soil (bacteria, fungi and nematode diversity) biodiversity influence ecosystem multifunctionality, as well as how these relationships are modulated by stand structural attributes and environmental conditions (topography and soil nutrients). Forest multifunctionality was positively related to stand structural complexity but negatively related to acquisitive traits (i.e. community‐weighted mean of specific leaf area). Plant PD had no significant direct effect on FM, but it had a significant indirect effect via increased stand structural complexity. The effect of soil microbial diversity on FM increased with increasing threshold levels of FM and outperformed tree diversity and environmental conditions at the highest threshold level (i.e. 95%). Forests on steep slopes had lower levels of ecosystem multifunctionality due to decreased stand structural complexity. Soil nutrients were responsible for regulating FM via plant trait composition and, to a lesser extent, via tree diversity, stand structure and soil microbial diversity. Synthesis . Plant PD, stand structure and soil microbial diversity jointly regulated FM, and these effects were influenced by local‐scale changes in environmental conditions. Soil microbial diversity was a key driver of highly multifunctional forests, whereas conservation of complex stand structure and conservative trait dominance could enhance mean values of multiple functions.
Abstract Spatial heterogeneity is a fundamental feature of ecosystems, and ecologists have identified it as a factor promoting the stability of population dynamics. In particular, differences in interaction strengths and resource supply between patches generate an asymmetry of biomass turnover with a fast and a slow patch coupled by a mobile predator. Here, we demonstrate that asymmetry leads to opposite stability patterns in metacommunities receiving localized perturbations depending on the characteristics of the perturbed patch. Perturbing prey in the fast patch synchronizes the dynamics of prey biomass between the two patches and destabilizes predator dynamics by increasing the predator's temporal variability. Conversely, perturbing prey in the slow patch decreases the synchrony of the prey's dynamics and stabilizes predator dynamics. Our results have implications for conservation ecology and suggest reinforcing protection policies in fast patches to dampen the effects of perturbations and promote the stability of population dynamics at the regional scale.
Abstract The dependence of humans on nature has come into focus as the human population continues to grow, resources diminish and production technology stagnates – threatening human well‐being on a global scale. Numerous previous models describe human population dynamics, in relation to a multitude of different factors. However, there are no consistent driving factors of human demography through history, which makes predicting future changes more challenging. Here, we review the literature on human population growth from empirical data and previous models, which allows us to highlight key trends in demography and land cover changes. We then establish an ecologically driven theory of demographic change that uses resource accessibility as a proxy for socio‐economic factors. The theory combines multiple concepts to represent 12 millennia of past population dynamics through simple human–nature relationships. Furthermore, the model allows us to compare different scenarios related to technological progress and land cover change, for which we find that the peak human population is highly dependent on whether technological developments continue at an exponential growth rate, or if and when there is a saturation point. Likewise, agriculture is shown to be helpful for growing the population, but nature is ultimately needed to maintain the human population. A plain language summary is available for this article.
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)