Infectious food webs




Parts of a complex food web found in lake Takvatn, Norway. The common food web features of predator (the red-breasted merganser, Mergus serrator) and prey (the three-spined stickleback, Gasterosteus aculeatus), are joined by the less frequently observed, but very influential parasite (a tapeworm, Schistocephalus solidus). Parasites greatly increase this food web's complexity.

P.-A. Amundsen, K. D. Lafferty, R. Knudsen, R. Primicerio, A. Klemetsen & A. M. Kuris (2009) Food web topology and parasites in the pelagic zone of a subarctic lake. Journal of Animal Ecology, 78, 563–572.

Food webs, descriptions of who is eating whom in an ecosystem, are one of the most enduring and influential concepts in ecology. An increasing number of studies are including parasite–host feeding interactions in food webs, each providing evidence that parasites alter our perception of food web structure. Amundsen et al. in this issue report intriguing details on the role of parasites in the complexity of an arctic food web. They highlight the role of links generated by trophically transmitted parasites – those transmitted via a predator–prey interaction between two hosts. These data show the type of natural history knowledge necessary to advance our understanding of food web complexity, structure and dynamics.

Food webs remain one of the most easily recognized descriptions of terrestrial and aquatic communities. Voracious, large and charismatic mammalian predators feed on small herbivores, which in turn feed on verdant fields of grass and herbs; lithe, fast-swimming, large-toothed killers hunt hapless smaller fish, which in turn feed on crustaceans of all sizes, themselves feeding on the tiniest of algae and other phytoplankton. No undergraduate textbook on biology is complete without at least one picture displaying an array of complex connections between consumers and resources. Connectance (C) is one of the easiest and most-studied summaries of food web complexity and is defined as the proportion of all possible links in a food web that are realized. Taking S as the number of species, connectance is defined as L/S2, where L is the actual number of links in the web.

For decades, researchers have attempted to generalize about and explain the patterns of connectivity (C) in food webs. Specific attention has been given to the relationship between C and S. Theory suggests that C may be constant or decline exponentially with S (Warren 1994; Dunne, Williams & Martinez 2002), and the latest data seem to support the latter (Montoya & Sole 2003; Beckerman, Petchey & Warren 2006). Simultaneously, numerous models have successfully taken connectance as a primary parameter to predict additional structural characteristics of food webs, such as the amount of omnivory and the length of food chains (Cohen, Briand & Newman 1990; Williams & Martinez 2000; Cattin et al. 2004). Even the most simple rules, such as randomly feeding on species lower in a trophic hierarchy, reproduce many of the characteristics we see in real food webs.

The most recent models aiming to predict structure, the niche model and the nested hierarchy model (Cattin et al. 2004; Williams & Martinez 2000), do so by simulating connections using probablistic rules of diet choice (e.g. feeding below your own position in the food web) and constraining the number of links to levels of connectivity found in the real food web. More recently, research has shown that body size, specifically predator–prey body size ratios, are intimately linked to these structural characteristics (Brose et al. 2006a) and that this may explain the stability of food web (Otto, Rall & Brose 2007). Following from this, optimal foraging linked to body size has enabled the prediction of connectivity and structure from species-level traits (Beckerman et al. 2006; Petchey et al. 2008).

This potted history of examining connectance in food webs hides something rather extraordinary. The majority of research into these questions has focused on aquatic and terrestrial predator–prey, plant–herbivore and plant–pollinator interactions. Clearly missing are the pathogens, parasites and parasitoids (the 3 Ps, see Lafferty et al. 2008). This is not to say that these features of real food webs have been ignored by food web biologists, as there are many food web studies detailing the many direct and indirect effects of the 3 Ps (e.g. Hudson, Dobson & Newborn 1998; Lewis et al. 2002; Bukovinszky et al. 2008). Yet the fundamental question about patterns of and mechanisms driving complexity (connectance) in real food webs has effectively ignored them.

A small but growing number of studies do include parasites in food webs (Huxham & Raffaelli 1995; Memmott, Martinez & Cohen 2000; Thompson, Mouritsen & Poulin 2005; Lafferty, Dobson & Kuris 2006; Hernandez & Sukhdeo 2008). These few studies show how parasites influence measures of species richness, links, connectance, the number and length of trophic levels, and nestedness. Lafferty and colleagues have been instrumental at defining the various ways in which parasites can influence webs (Lafferty et al. 2008) and Amundsen et al. in this issue explore the effects of parasites on connectance in an incredibly well-resolved food web of the pelagic zone of a lake. Their food web contains 37 free-living species (eight phytoplankton, 18 zooplankton, three fish, and nine bird species) and 13 parasite categories (six cestode, one nematode, one copepod, one monogenean, and four fungi categories). They then compared the structure of the food web with only free-living species to the structure of the entire food web, including free-living and parasitic organisms. Inclusion of parasites increased connectance (from 0·145 to 0·173, see Fig. 1), linkage density, and food-chain length. Arguably of more interest are the patterns in two sub-webs which seem to drive these changes. The first is the parasite–host sub-web that documents which parasites infect which free-living hosts – the connectance of this subweb is 0·156. The other sub-web describes the consumption of parasites by free-living organisms; this can happen when a free-living consumer eats a parasite-infected free-living resource. Connectance of this sub-web was a whopping 0·331. This results from the finding that on average, each of the parasite types is consumed by 12·2 of the 37 free-living species. This level of vulnerability of parasites to predation is extraordinary. As Amundsen et al. point out, ‘parasites and pathogens are especially integrated in food webs when they are trophically transmitted’.

Figure 1.

The addition of parasites (red) to the arctic food web visibly increases the links among species. In these data, parasites increase connectance (complexity) because trophically transmitted species generate a doubling of links, for only a 26% increase in species richness. (Figure generously produced by Jennifer Dunne with Network3D software written by R. J. Williams; contact for more details.)

As emerging theory shows, identifying traits of species and their roles in food webs adds a great deal of value and predictability to food web complexity and structure (Beckerman et al. 2006; Brose et al. 2006a; Brose, Williams & Martinez 2006b; Petchey et al. 2008). Life cycles, body size and foraging biology, all highlighted by Amundsen et al., bear heavily on our understanding of the effects of aggregating species, the mechanisms giving rise to complexity and structure and the interplay between dynamics and structure in food webs. The resolution of the food web examined by Amundsen et al. has allowed them to show how parasites transmitted by hosts eating each other, a predator–prey interaction, produced a higher number of links and contributed more to connectance and linkage density than did free-living parasites (nontrophically transmitted sensu Amundsen et al.). While this is perhaps not a surprising result, it is precisely this type of natural history that will motivate informed data analysis and modelling. We face the interesting challenge now of determining how parasite ‘diet breadth’ arises as a function of host foraging, parasite behaviour and parasite effects on hosts. We have been guided effectively towards the features of parasites and pathogens that underpin their influence, and towards comparisons among predators and parasites, whose foraging biology are intimately linked by transmission rates and functional responses.

One further result that bears comment is that parasites had an important role to play in the nestedness of this web. Nestedness characterizes how specialist species interact with subsets of those species interacting with generalists (Bascompte et al. 2003; Bascompte, Jordano & Olesen 2006). Nestedness is a food web characteristic intimately tied to the stability and dynamics of species interactions. It describes pockets of interactions in a web, drawing focus to the asymmetry and strength of interactions (Bascompte et al. 2006). Amundsen et al. show that parasites in this web double levels of nestedness, organizing pockets of interaction in the community cohesively around a central core. This nestedness, combined with knowledge about asymmetry and the role of trophically transmitted parasites in host demography, again suggest mechanisms defining structure and dynamics that are deeply rooted in classic community ecology.

Ecologists will hopefully soon stop talking about integrating parasites into food webs. Not because we have forgotten about parasites or chosen to ignore them, but because demonstrations of their ubiquity and importance are now so numerous and clear that we must always consider them. However, there are a number of emerging opportunities that take advantage of well-resolved food webs containing parasite data. First, consider that Amundsen et al. report no evidence of trophic interactions between parasites (their parasite–parasite sub-web is empty). There is, however, increasing evidence that parasites can and do have strong effects on each other (Lello & Hussell 2008; Pedersen & Fenton 2007). Filling in this part of the web and understanding the consequences of these interactions will be quite challenging but also very illuminating.

Second, how can we better integrate the process of transmission that defines infection and the process of predation that defines consumption? Consider the extraordinary vulnerability we highlight in this food web. On face value, one might consider this to arise randomly in systems with many trophically transmitted parasites. But, in fact, there is every reason to believe data and theory which suggest that being parasitized influences an organism's hunting success or susceptibility to predation (Hatcher, Dick & Dunn 2006). We expect that these are not random patterns; they are probably driven by the behaviour of the species in the food web. There are now models that specify foraging and consumption processes underpinning the complexity and structure of food webs (Beckerman et al. 2006; Petchey et al. 2008), and we are optimistic that such a philosophy can be applied to webs with parasites as well.