Human-induced alterations in the birth and mortality rates of species and in the strength of interactions within and between species can lead to changes in the structure and resilience of ecological communities. Recent research points to the importance of considering the distribution of body sizes of species when exploring the response of communities to such perturbations. Here, we present a new size-based approach for assessing the sensitivity and elasticity of community structure (species equilibrium abundances) and resilience (rate of return to equilibrium) to changes in the intrinsic growth rate of species and in the strengths of species interactions. We apply this approach on two natural systems, the pelagic communities of the Baltic Sea and Lake Vättern, to illustrate how it can be used to identify potential keystone species and keystone links. We find that the keystone status of a species is closely linked to its body size. The analysis also suggests that communities are structurally and dynamically more sensitive to changes in the effects of prey on their consumers than in the effects of consumers on their prey. Moreover, we discuss how community sensitivity analysis can be used to study and compare the fragility of communities with different body size distributions by measuring the mean sensitivity or elasticity over all species or all interaction links in a community. We believe that the community sensitivity analysis developed here holds some promise for identifying species and links that are critical for the structural and dynamic robustness of ecological communities.

The complexity–stability relation is a central issue in ecology. In this paper, we show how the sampling method most often used to parameterize an ecological community, can affect the conclusions about whether or not complexity promotes stability and we suggest a sampling algorithm that overcomes the problem. We also illustrate the importance of treating feasibility separately from stability when constructing model communities. Using model Lotka–Volterra competition communities we found that probability of feasibility decreases with increasing interaction strength and number of species in the community. However, for feasible systems we found that local stability probability and resilience do not significantly differ between communities with few or many species, in contrast with earlier studies that, did not account for feasibility and concluded that species-poor communities had higher probability of being locally stable than species-rich communities.

Ecological communities are continuously exposed to natural or anthropogenic disturbances of varied intensity and frequency. The way communities respond to disturbances can depend on various factors, such as number of species, structural characteristics of the community, stability properties, species characteristics and the nature of the disturbance. This thesis is a collection of theoretical studies on how ecological communities respond to different kind of disturbances, mainly in relation to interaction strength between species, a measure of how strongly or weakly species interact with each other.

A major disturbance for natural communities is the loss of a species. Although species extinctions is a natural process in the geological time scale, it has lately been dangerously accelerated due to human activities and interferences. Extinction of a species can have dramatic consequences for the community, can trigger a cascade of secondary species extinction and can even lead to community collapse. In Paper I, we identify species whose loss can trigger a large number of secondary extinctions (namely keystone species), species that are particularly vulnerable to become extinct following the loss of another species and mechanisms behind the sequence of secondary extinctions. We also highlight the fact that the keystone status of a species can be context dependent, that is, it is dependent on the structure of the community where it is embedded.

Although the global extinction of a species is an irreversible process, in cases of local extinction, conservation and restoration plans can include reintroduction of the species to their former location. Such reintroduction or translocation attempts often fail, due to characteristics of the reintroduced species and to changes in the community structure caused by the initial loss of the species (Paper II ). Using model communities we show that this risk of reintroduction failure can be high - even in cases where the initial species loss did not cause any secondary extinctions - and it differs between attempts to reintroduce weakly or strongly interacting species (Paper II ).

Disturbances are not always as profound as species extinction. Human activities and environmental changes can cause small and permanent changes in birth and mortality rates of species and in the strength of interaction between species. Such disturbances can change the stability properties of ecological communities making them more vulnerable to further disturbances. In Paper III, we derive analytical expressions for the sensitivity of resilience to changes in the intrinsic growth rate of species and in the strength of interaction links. We also apply the method to model communities and identify keystone species and links, i.e. species and links whose small disturbance would cause large changes in community resilience. We found that changes in the growth rate of strongly interacting species have a larger impact on resilience than changes in the growth rate of weakly interacting species, which is in line with the main findings of the species deletion study (Paper I ).

Community complexity - mainly expressed as number of species and links - was one of the first community characteristics to be related to stability properties and the way communities respond to disturbances. Many theoretical works support the hypothesis that complexity reduces stability, contradicting intuition, observation and many experimental studies. In Paper IV, we state that this could be, at least partly, due to methodological misconceptions and misinterpretations. We propose a new sampling method for parameterizing model communities and we highlight the importance of feasibility of model communities (all species densities are strictly positive), for a more proper estimation of stability probability of communities with different degrees of complexity.

Human-induced changes in the birth and mortality rates of species and in the strength of interactions within and between species can lead to changes in the resilience of ecological communities. Here we derive analytical expressions for the sensitivity of community resilience to changes in the intrinsic growth rate of species and in the strengths of interaction links. These new methods are applied on model communities to illustrate how they can be used to identify keystone species and keystone links - keystone species and keystone links in the sensethat small changes in their growth rates and strengths, respectively, will have large effects on the resilience of the community. We find that small changes in the intrinsic growth rates of rare species (e.g. top predators) and strongly interacting species have larger effect on resilience than changes in the growth rates of abundant species and weakly interacting species. Moreover, we find that small changes in the strength of weak interaction links cause larger changes in resilience than changes in the strength of strong links. We believe that thecommunity sensitivity analysis developed here holds some promise for identifying species and links that are critical for the resilience of ecological communities.

The loss of a species from an ecological community can trigger a cascade of secondary extinctions. The probability of secondary extinction to take place and the number of secondary extinctions are likely to depend on the characteristics of the species that is lost—the strength of its interactions with other species—as well as on the distribution of interaction strengths in the whole community. Analysing the effects of species loss in model communities we found that removal of the following species categories triggered, on average, the largest number of secondary extinctions: (a) rare species interacting strongly with many consumers, (b) abundant basal species interacting weakly with their consumers and (c) abundant intermediate species interacting strongly with many resources. We also found that the keystone status of a species with given characteristics was context dependent, that is, dependent on the structure of the community where it was embedded. Species vulnerable to secondary extinctions were mainly species interacting weakly with their resources and species interacting strongly with their consumers.

Ecological communities are often exposed to different kind of disturbances of varied magnitude. The response to disturbances can be affected by, among other factors, the strength of interactions between species within the community. The main purpose of this thesis is the study of the relation between interaction strength (distribution and positioning of strong and weak links) and the response of model communities to disturbances. Additionally, we wish to identify keystone species and keystone links and to recognize species vulnerable to secondary extinctions with respect to the interaction strength concept.

The first part of this study (paper I) deals with species removal from model ecological communities. The number of secondary extinctions following the loss of one species measures the response to this disturbance. The loss of the following species categories triggered, on average, the largest number of secondary extinctions: a) rare resources strongly interacting with many consumers b) abundant intermediate consumer species strongly interacting with many resources and c) abundant weakly interacting resources. Species vulnerable to secondary extinctions were mainly weakly interacting consumers and strongly interacting resources (excluding primary producers). This vulnerability to secondary extinctions was not only dependent on characteristics of the species themselves but also on which species was initially removed. Some species were facing a high probability of extinction after the deletion of a wide range of species categories, while others were vulnerable only to the deletion of few species categories.

The second part (paper II) deals with less severe disturbances; small changes in the intrinsic growth rate of species and small changes in the interaction strength of links. In this way the importance of both species and links was revealed. The response of the community was measured as the effect of these changes on the resilience of the system. Overall, the resilience of the model communities was more sensitive to changes in weakly interacting species or weak links.

The main message of this work is that keystone species and keystone links could be context dependent, both with respect to characteristics of the community and the stability criterion used. Studying the effects of severe disturbances, such as species loss, and using the number of secondary extinctions to measure the community's response, the disturbance of strongly interacting species affected the community most. On the contrary, by applying small perturbations to community parameters and measuring the response as the effect on resilience, the perturbation of weakly interacting species and weak links affected the community most. Thus, both strong and weak agents of the community can be important for stability properties under different regimes of disturbance.

The complexity - stability relation has been a central issue in ecology for a long time. Both theoretical and empirical studies have been investigating whether complexity promotes stability or not, which could be the underlying mechanisms creating a positive or negative cmnplexity-stability relation and which are the structures, characteristics and constraints that would allow complex ecological systems to persist and be stable. In this paper we approach the subject from amethodological point of view. We study the effect of parameterizing the communities in a certain way and illustrate the effect of treating feasibility (i.e. densities of all species are strictly positive) separately from stability. We observe that increasing number of species in the communities requires a more skewed interaction strength distribution toward weak interactions, in order for the communities to theoretically exist. Using model Lotka-Volterra competition communities we illustrate that probability of feasibility decreases with increasing interaction strength and number of species in the community. However, forfeasible systems we find that local stability probability and resilience do not significantly differ between communities with few or many species, in contrast with earlier studies that, did not account for feasibility and concluded that species poor communities had higher probability of being locally stable than species rich communities.

For many threatened species reintroduction of captive-reared individuals into the wild may offer the only hope of preventing global extinction. However, attempts to reintroduce a species into an ecological community from which it has once gone extinct often fail. This can be due to captivity-induced genetic and behavioural changes in the species itself. Here we argue that also changes in the structure of the community caused by the initial loss of the species can hinder its subsequent reintroduction. Due to the interdependences amongspecies in ecological communities the local extinction of a species leads to changes in the densities and strength of interactions among the remaining species in the community and sometimes even to a cascade of secondary extinctions. Such alterations in the structure of a community caused by the initial extinction of the species might make it impossible for the species to reinvade at a later stage. Here we show, using models of multitrophic communities, that the risk of reintroduction failure can be high and that it differs between species belonging to different trophic levels and between weakly and strongly interacting species. Specifically, there is a high risk of reintroduction failure for primary producer species and weakly interacting consumer species. We also find the risk that the reintroduction will cause new extinctions to be higher in communities where the initial loss triggered secondary extinctions. In the worst case extinctions caused by the reintroduction may lead to the subsequent extinction of the reintroduced species itself. This risk is not negligible implicating that reintroductions might sometimes aggravate the state of local communities.