Integrating network ecology with applied conservation: a synthesis and guide to implementation

Abstract

Ecological networks are a useful tool to study the complexity of biotic interactions at a community level. Advances in the understanding of network patterns encourage the application of a network approach in other disciplines than theoretical ecology, such as biodiversity conservation. So far, however, practical applications have been meagre. Here we present a framework for network analysis to be harnessed to advance conservation management by using plant-pollinator networks and islands as model systems. Conservation practitioners require indicators to monitor and assess management effectiveness and validate overall conservation goals. By distinguishing between two network attributes, the 'diversity' and 'distribution' of interactions, on three hierarchical levels (species, guild/group and network) we identify seven quantitative metrics to describe changes in network patterns that have implications for conservation. Diversity metrics are partner diversity, vulnerability/generality, interaction diversity and interaction evenness, and distribution metrics are the specialization indices d' and [Formula: see text] and modularity. Distribution metrics account for sampling bias and may therefore be suitable indicators to detect human-induced changes to plant-pollinator communities, thus indirectly assessing the structural and functional robustness and integrity of ecosystems. We propose an implementation pathway that outlines the stages that are required to successfully embed a network approach in biodiversity conservation. Most importantly, only if conservation action and study design are aligned by practitioners and ecologists through joint experiments, are the findings of a conservation network approach equally beneficial for advancing adaptive management and ecological network theory. We list potential obstacles to the framework, highlight the shortfall in empirical, mostly experimental, network data and discuss possible solutions.

Keywords: Adaptive management; biodiversity conservation; ecological integrity; ecosystem functions; indicators; interaction networks; islands; pollination.

Figure 1.

Real-world pollination web (A; data from Kaiser-Bunbury et al. 2012), hypothetical pollination web (B) and network metrics of the hypothetical web (C). Bipartite pollination webs (A) depict quantitative relationships between pollinators (top) and plants (bottom). Species are represented by rectangles, which are linked by wedges. The width of the rectangles reflects the relative abundance of the species, and the width of the wedges shows the relative interaction frequency between species. Pollinators are coloured by taxonomic groups (e.g. red = bees and wasps, green = flies), and plants shown in pink are exotic species, to indicate potential groupings within guilds. The real-world pollination web visualizes the hierarchical levels of the network metrics proposed in Table 1. The hypothetical web (B) illustrates conceptual differences between partner diversity of plant species (species-level generality) and specialization d′_poll and between partner diversity of pollinator species (species-level vulnerability) and specialization d′_pl. Note that animal species 1–3 visit only a single plant species, thus their partner diversity is minimum (=1). On the contrary, animal species 1 is most selective (exclusive visitor of plant species A), and animal species 3 is least selective in terms of the distribution of all pollinators, hence d′ declines accordingly from species 1 to 3. For animal species with a single individual, partner diversity always equals one, whereas d′ can vary between zero and one depending on the exclusiveness of the selected plant species. Other network metrics (C) of the hypothetical web describe the diversity and the distribution of interactions. With higher generalization, the generality (G)/vulnerability (V) increases whereas complementary specialization d′ decreases (P, plants; A, animals).

Figure 2.

Pathway to implement an interaction network approach in biodiversity conservation. Practitioners undergo a multi-stage process to define conservation objectives specific to one ecosystem function. Concurrently, network ecologists determine the causal relationship between human action and network patterns, and identify suitable metrics as indicators to assess conservation management effectiveness. Selecting indicators, setting thresholds and choosing the appropriate methodology for data collection are jointly carried out between ecologists and practitioners to ensure rigorous experimental setup. Findings are used for adaptive management by practitioners and for refining network analysis by ecologists.