The diets of 99 pumpkinseed sunfish Lepomis gibbosus from a pair of small, adjacent lakes in Ontario, Canada, were estimated from their stomach contents, trophically transmitted parasites and stable isotopes of carbon and nitrogen in fish tissue. The three methods provided virtually unrelated information. There was no significant correlation in the importance of any prey item across all three methods. Fish with similar diets according to one method of estimating diet showed no tendency to be similar according to other methods. Although there was limited variation in fish size and the spatial scale of the study was small, both fish size and spatial origin showed comparatively strong associations with diet data obtained with all three methods. These results suggest that a multidisciplinary approach that accounts for fish size and spatial origins is necessary to accurately characterize diets of individual fish.
Collection of L. gibbosus and its prey
Between 28 July and 9 August 2010, 99 L . gibbosus were collected from seven localities in Lake Opinicon (n = 64) and four localities in Rock Lake (n = 35), Ontario, Canada, by hook and line (Fig. 1). Fish were killed upon capture by cerebral percussion using a small mallet, weighed and measured (total length, L T, ±1 mm and mass, M T, ±1 g), kept on ice and dissected within 36 h. The gastrointestinal tract and viscera of freshly killed fish were examined under a stereomicroscope, and stomach contents and parasites were preserved in ethanol. Only parasites that are trophically transmitted, i.e . gastrointestinal helminths, and certain helminths from the viscera, were retained. A sample of lateral muscle tissue was frozen.
Samples of prey items were collected using dip nets or an Ekman dredge. In Lake Opinicon, these were zebra mussels, gastropods (Viviparidae, Physidae and Planorbidae), amphipods and isopods, which were the four most frequently observed items in the stomachs of fish from this lake, and Odonata, which were not observed in many fish. In Rock Lake, only gastropods and Trichoptera were collected.
Estimation of L. gibbosus diet using stomach contents
Items in stomachs that clearly corresponded to an individual prey (e.g . whole prey, heads of insects, septa of bivalves and columellae of gastropods) were identified to the species, family or order level. The proportion of each prey item in the diet of each fish was estimated by dividing the number of individual prey in a given taxon by total prey abundance.
Estimation of diet using trophically transmitted parasites
All trophically transmitted parasites were either cleared in glycerol and examined in wet mounts (nematodes) or stained in acetocarmine, cleared in clove oil and mounted on slides for identification using the keys of Arai (1989), Gibson (1996) and Hoffman (1999). Adult Leptorhynchoides thecatus in the gastrointestinal tract and larvae (cystacanths) in the body cavity were considered separate species in similarity‐based analyses described below (Steinauer et al ., 2007). In cases where parasite life cycles included alternate transmission pathways and implicated different possible prey, the distribution of the parasite and potential transmitting prey species were examined in more detail. For example, L. gibbosus may become infected with enteric L. thecatus after consuming infected fish (paratenic host) or amphipods. Because fishes were never observed in stomachs of adult L. gibbosus in Lake Opinicon (Keast, 1978a) enteric L. thecatus were assumed to have been acquired from amphipods. For similar reasons, other acanthocephalans and Camallanus oxycephalus were assumed to have been acquired from invertebrates rather than paratenic fish hosts, and Phyllodistomum cf. superbum was assumed to have been acquired from insects rather than other potential hosts.