The food web of Potter Cove (Antarctica): complexity, structure and function

Knowledge of the food web structure and complexity are central to better understand ecosystem functioning. A food-web approach includes both species and energy flows among them, providing a natural framework for characterizing species’ ecological roles and the mechanisms through which biodiversity influences ecosystem dynamics. Here we present for the first time a high-resolution food web for a marine ecosystem at Potter Cove (northern Antarctic Peninsula). Eleven food web properties were analyzed in order to document network complexity, structure and topology. We found a low linkage density (3.4), connectance (0.04) and omnivory percentage (45), as well as a short path length (1.8) and a low clustering coefficient (0.08). Furthermore, relating the structure of the food web to its dynamics, an exponential degree distribution (in- and out-links) was found. This suggests that the Potter Cove food web may be vulnerable if the most connected species became locally extinct. For two of the three more connected functional groups, competition overlap graphs imply high trophic interaction between demersal fish and niche specialization according to feeding strategies in amphipods. On the other hand, the prey overlap graph shows also that multiple energy pathways of carbon flux exist across benthic and pelagic habitats in the Potter Cove ecosystem. Although alternative food sources might add robustness to the web, network properties (low linkage density, connectance and omnivory) suggest fragility and potential trophic cascade effects.


0 1
The first thing to note about Potter Cove FW is that most of the species (47%) were at connected (9 species with degree > 15) than others, according to the total number of trophic 2 0 5 interactions they have (e.g. fish and echinoderms). The main properties of the network complexity for Potter Cove FW included 307 total 2 0 7 interactions and a linkage density of 3.4. As a consequence, a connectance of 0.04 was reported 2 0 8 (Table 1). the species in the FW, the basal species were also numerous (Table 1). In addition, almost half of 2 1 5 the species were omnivorous (45%), similar to the percentage observed in intermediate species. The mean trophic level (TL) for Potter Cove FW was 2.1, which was supported by the relatively 2 1 7 high proportion of basal species that tend to lower the average. Network topological properties, characteristic path length (ChPath) and clustering coefficient 2 1 9 (CC) were 1.8 and 0.08, respectively.

0
The degree distribution for the Potter Cove FW (Fig. 3) showed that the exponential model best 2 2 1 fitted the data, according to nonlinear regression and AICc analyses ( (echinoderm, 33 links) and Gondogeneia antarctica (amphipod, 20 links). shown. Best fit is the exponential model. The competition graph derived from Potter Cove FW is highly connected. It includes 60 species 2 2 7 and 478 indirect interactions (Fig. 4) and shows that several pairs of predators share many prey. For instance, all trophic species of sponges form a more connected group than with the rest of the prey species. Furthermore, some species of echinoderms, amphipods and demersal fish are   Link width and colors: number of shared prey.

3 3
To study these potential species interactions, specific competition graphs for the latter two Harpagifer antarcticus, presented highly overlapping diets. Moreover, N. coriiceps shares many 2 3 7 of the same prey species, which may or may not involve any competition (Fig. 5 a). On the other Gondogeneia antarctica and Prostebbingia gracilis have many prey in common (Fig. 5 b). and colors: number of shared prey (see Fig. 4).

4 2
The common-enemy graph shows a hyperconnected structure, where the majority of the species   Degree distribution of links in the competition and common-enemy graphs ( Fig. 7) fit best to an 2 5 5 exponential model (Table 2). graphs. Best fit is the exponential model for both distributions. Comparison between the Potter Cove FW and other marine webs showed that linkage density to the rest of the FWs, the clustering coefficient was one order of magnitude lower (Table 3).  and out-links), competition (only predators) and common-enemy (only prey) overlap graphs. Potter Cove FW properties of complexity and structure showed several singularities that make 2 7 6 the web unique in terms of species-richness, link configuration and topological characteristics. Network complexity was mainly assessed by linkage density (L/S) and connectance (C). Both of 2 7 8 these properties were found to be relatively low in the Potter Cove web: L/S=3.4 and C=0.04.

7 9
Nevertheless, direct comparisons of linkage density and connectance values suggest that marine FWs are unusual, and that extreme connectances may sometimes be artifacts of assembly network and not a source-web since it was not constructed upward from one or two basal species 2 9 5 but it is characterized by > 30% basal species. Thus there is no evidence we know of which 2 9 6 1 9 suggests that our low values of linkage density and connectance were a consequence of the 2 9 7 assembly procedure of the network. In turn this implies that the assembly-connectance 2 9 8 relationship in FWs is not as strong as previously thought (Dunne et al. 2002b).

9 9
Whether ecological networks display low or high L/S and C values is crucial to gain insight in to understand Potter Cove ecosystem functioning, since there seems to be a species-specific 3 1 6 selective consumption (Barrera-Oro and Casaux 1990, Iken et al. 1997, Iken et al. 1998. low when compared to other marine webs, but close to values for Antarctic FW as reported by de majority of species acting as predators and prey also feed on more than one trophic level 3 2 7 (omnivorous). The importance of omnivory for the structure and dynamics of FWs is a long- Antarctica a recent study suggests that omnivory is a beneficial trait as it allows for more  The mean trophic level for this FW (2.1) is also relatively low, which is the result of several Applications. Oxford University Press, Oxford, pp. 37-44. 15. Burns, T.P., 1989. Lindeman's contradiction and the trophic structure of ecosystems. Antarctica. Polar Biology 6, 207-213. 558-567. Sciences 99, 12917-12922.  the Antarctic fish Notothenia coriiceps: evidence for selective feeding on macroalgae.