Empirically Characterising Trophic Networks

Abstract

Food webs in agricultural systems are complex and trophic linkages are difficult to track using conventional methodologies. Here, we review three alternative approaches that allow empirical assessment of feeding interactions: DNA-based techniques, and stable isotope and fatty acid analyses. DNA-based methods, namely multiplex PCR and next-generation sequencing, allow identification of food types and host–parasitoid linkages, resulting in taxonomically highly resolved feeding networks. Stable isotopes and fatty acids reflect the assimilation of broader categories of resources, as metabolised into the consumers’ tissue, together with the associated energy and nutrient fluxes in the food web. We discuss the strengths of the approaches but also highlight their limitations, providing practical advice on which technique is best suited to answer specific questions in examining food web interactions in agroecosystems. Future refinements of these techniques, especially when used in combination, could herald a new era in agricultural food web ecology, enabling management and environmental impact to be placed in the mechanistic context of trophic networks.

Introduction

Interest in food webs has increased considerably in recent years, with especially rapid progress being made via the development of new theoretical modelling and numerical simulation tools (e.g. Dobson et al., 2009, Ings et al., 2009, Stouffer, 2010). Although models and simulations have provided important new insights into food web structure and dynamics, it is still essential to validate trophic links empirically with real-world data (Finlay-Doney and Walter, 2012). However, it is often not known what, when and where specific trophic interactions occur in arable ecosystems, as feeding behaviour is affected by both the biotic and abiotic environment, which remains poorly understood. Consequently, our understanding of agricultural food webs is often still too fragmentary to understand fully how the community functions via the application of food web approaches. The current paucity of accurate dietary information is, in part, due to the difficulties of tracking trophic interactions in the field, and the lack of high-throughput techniques to record feeding interactions in communities is a longstanding obstacle for the construction of empirical food webs (Cohen et al., 1993, Memmott, 2009).

These difficulties apply to ecological networks in general (Ings et al., 2009), but they are particularly acute in the agricultural context (Bohan et al., 2013), where the principal actors are typically small, invertebrates with cryptic and complex trophic behaviour, which are difficult to assess using classical techniques such as direct observation or morphological identification of prey remains in gut contents or faeces (Sunderland et al., 2005). In addition, many of the most common taxa in agricultural fields, such as carabid beetles, are opportunistic feeders, consuming a wide range of animal prey and plant material, which makes their diets difficult to characterise fully for any given system (Holland, 2002). Many important consequences of generalist feeding behaviour, such as prey switching or intraguild predation, therefore remain difficult to measure in the field, despite their importance for the structure of food webs and the delivery of ecosystem services, such as pest control (e.g. Finke and Denno, 2004, Wilson and Wolkovich, 2011). Due to these difficulties, the most complex feeding networks measured in arable systems to date have dealt with relatively simple host–parasitoid communities (e.g. Bukovinszky et al., 2008, Tylianakis et al., 2007) or mutualistic, plant–pollinator webs (Burkle and Alarcón, 2011, Memmott, 2009). Although determining and quantifying host–parasitoid interactions is relatively easy, in contrast to assessing what generalist herbivores or carnivores consume, there are still significant methodological hurdles associated with the description of trophic interactions in host–parasitoid communities; for example, discriminating among cryptic interactions, such as multiparasitism and hyperparasitism, is impractical with conventional methods (Gariepy et al., 2008a, Traugott et al., 2008) and requires novel approaches. Another particular problem is that, unlike many aquatic food webs, which tend to be dominated by engulfing, gape-limited consumers (e.g. see Gilljam et al., 2011, Ledger et al., 2013, O'Gorman et al., 2012), agroecosystems have large numbers of fluid-feeding consumers, whose gut contents are unidentifiable using traditional microscopy techniques.

In recent years, significant methodological advances have been made for studying feeding interactions in the field, opening up exciting new perspectives in trophic ecology. Stable isotope and fatty acid (FA) analyses provide new means to elucidate patterns of resource allocation (Boecklen et al., 2011, Ruess and Chamberlain, 2010), while DNA-based techniques allow feeding interactions to be characterised to a high level of taxonomic resolution (Gariepy et al., 2007, Symondson, 2012). These approaches have been reviewed and their potential applications summarised elsewhere (e.g. Gariepy et al., 2007, King et al., 2008, Martínez del Rio et al., 2009, Pompanon et al., 2012, Post, 2002, Ruess and Chamberlain, 2010, Sheppard and Harwood, 2005, Symondson, 2002), yet an integrated overview of the possibilities they offer for describing trophic networks is notable by its absence. Here, we address this gap by comparing DNA-based, stable isotope and FA analyses and providing guidance as to which tools are best suited to address specific questions in agroecosystem food web ecology (Fig. 3.1).