Introduction

Pacific salmon (Oncorhynchus spp.) attain the majority of their body mass feeding at sea and, as they return to spawn and die in freshwater habitats, their bodies subsidize terrestrial communities with marine-derived nutrients and energy (Bilby et al. 1996; Ben-David et al. 1998; Gende et al. 2002). In aquatic systems, the retention of salmon carcasses adds nitrogen, phosphorus, and other nutrients, a process that can initiate bottom-up food web effects (Wipfli et al. 1998; Zhang et al. 2003; Yanai and Kochi 2005). Live and dead salmon also support a diverse assemblage of terrestrial scavengers and predators (Reimchen 1994; Willson and Halupka 1995; Cederholm et al. 1999), although continued declines in salmon populations coast-wide threaten these salmon-dependent communities (Finney et al. 2000; Gresh et al. 2000; Stockner 2003).

Brown bear (Ursus arctos) and black bear (Ursus americanus) are important predators in salmon ecosystems and can transfer >50% of returning salmon biomass into adjacent forest habitats (Reimchen 2000; Gende and Quinn 2004). Based on tradeoffs of optimal foraging, bears will selectively feed on energy-rich portions of the salmon and distribute salmon carcasses to riparian areas in a fashion that is predicted by salmon spawning density, fish size, bear density, and habitat (Quinn and Kinnison 1999; Gende et al. 2001, 2004). Additional vectors of salmon carcasses include foraging by other vertebrates [e.g., river otters (Lutra canadensis) and wolves (Canis lupus)] and physical processes such as flooding (Ben-David et al. 1998).

The addition of salmon carcasses to terrestrial habitats can affect local ecosystem processes. For example, nutrient release from carcasses can influence plant growth and community structure (Wilkinson et al. 2005; Drake and Naiman 2007), while diet shifts to salmon in consumers can alter predator-prey dynamics (Darimont et al. 2008) and rates of detrital processing (Zhang et al. 2003). Terrestrial insect groups such as blowflies (Calliphoridae) and burying beetles (Silphidae) also increase processes of salmon decomposition and nutrient distribution in forest riparian zones (Jauquet et al. 2003; Meehan et al. 2005; Hocking and Reimchen 2006; Hocking et al. 2006). However, the complete composition and ecological role of terrestrial invertebrates using salmon carcasses is poorly known.

We surveyed terrestrial invertebrate use of natural and experimentally placed salmon carcasses in forest habitats on two remote streams on the central coast of British Columbia that support low human disturbance, high densities of spawning salmon and high total biomass of bear-transferred salmon carcasses (Hocking and Reimchen 2006). Our goals were to (1) assess the dominant consumers and principal ecological roles of invertebrates collected on salmon carcasses during autumn spawning, and (2) to test for potential seasonal and spatial variation in the patterns of salmon-nutrient uptake in this assemblage.

To quantify the ecological roles of terrestrial invertebrates, we use naturally occurring stable-isotope ratios (δ15N and δ13C) in select species. Stable-isotope analysis can quantify dietary variation, trophic structure, and sources of nutrients and energy in consumers (Kelly 2000; Post 2002; Vanderklift and Ponsard 2003), including invertebrates (Ponsard and Arditi 2000; Hocking and Reimchen 2002; Langellotto et al. 2006). An important consideration relates to tissue turnover and the seasonality of sampling, particularly when insect diets vary between larval and adult life stages (Tallamy and Pesek 1996). Isotope signatures in adult insects that indicate a salmon diet may be from a current diet of salmon or a larval diet from the previous year (Hocking et al. 2007). Further, differences in species mobility and pathway of salmon nutrient uptake may influence observed isotopic variation. We present new stable isotope findings in flies, parasitic wasps, and litter spiders where we test seasonal and spatial isotopic patterns, carcass-specific variation, and the legacy of a larval diet of salmon in adult individuals.

Materials and methods

Terrestrial invertebrate assemblage on salmon

In separate expeditions from 2000 to 2003, terrestrial invertebrates were surveyed from wildlife-transferred and experimental salmon carcass sites from two watersheds, the Clatse (52°20′15″ N; 127°50′23″ W) and Neekas (52°28′17″ N; 128°9′39″ W) rivers in coastal British Columbia, Canada. The climate of this region is cool and wet, with a mean annual temperature of ~8°C and annual precipitation ranging from 3,200 to 4,200 mm. Both watersheds occur in the Coastal Western Hemlock Biogeoclimatic Zone with forests dominated by Western hemlock (Tsuga heterophylla), Sitka spruce (Picea sitchensis), and Western red-cedar (Thuja plicata) (Green and Klinka 1994). Anadromous chum (O. keta) and pink (O. gorbuscha) salmon spawn from late August to early November within a 1-km (Clatse) or 2-km (Neekas) distance from the estuary that is terminated by an impassable waterfall. From 2000 to 2003, the average number of spawners (±STD) was 24,250 ± 12,685 pink salmon and 3,950 ± 1,666 chum salmon on the Clatse River and 43,625 ± 19,661 pink salmon and 28,750 ± 6,850 chum salmon on the Neekas River (NuSEDS: Department of Fisheries and Oceans data). From carcass surveys in 2001–2002 (Hocking and Reimchen 2006), black bears and wolves were estimated to transfer 4–6% of the pink salmon run and 16–48% of the chum salmon run into the forest riparian zone and leave >50% of this mass for invertebrate scavengers. This yields an estimated production of 21–42 million (Clatse) and 36–72 million (Neekas) fly larvae from salmon carcasses in each whole watershed (Hocking and Reimchen 2006).

In September of 2002, fresh and post-spawning chum and pink salmon carcasses with variable extent of carcass injuries (e.g., fresh whole fish, and head, brain, dorsal muscle and belly predation) were experimentally placed in the riparian zone within 100 m of the stream channel (Clatse n = 90; Neekas n = 96). Sites on the Clatse and Neekas Rivers included 36 fresh and 54 post-spawning carcasses (50 pink and 40 chum) and 20 fresh and 76 post-spawning carcasses (40 pink and 56 chum), respectively. The sequence of decay was followed for all carcasses and each were classified by their dominant (or co-dominant) consumer including: 1. Diptera; 2. Vertebrates; 3. Diptera and Vertebrates; or 4. Diptera and Coleoptera. We tested whether spawning condition or watershed influenced the dominant consumer of experimental salmon carcasses (Chi-square tests).

Over the duration of the study, invertebrates collected from salmon were classified according to their ecological role on the carcasses. This was based on field observations, literature sources (Anderson and Peck 1985; Borror et al. 1981; Furniss and Carolin 1977; Hatch 1953, 1957) and on taxonomic support provided by Dr. Monty Wood (Calliphoridae, Dryomyzidae, Drosophilidae), Dr. Steven Marshall (Sphaeroceridae), Dr. Graham Griffiths (Heleomyzidae), Dr. Andrew Bennett (Ichneumonidae), Dr. Henri Goulet (Braconidae), Dr. Matthew Buffington (Figitidae), Dr. Jan Klimaszewski (Staphylinidae), Dr. Stewart Peck (Leiodidae, Hydrophilidae), Dr. David Kavanaugh (Carabidae), Dr. Robb Bennett (Agelenidae) and Dr. Jan Addison (Enchytraeidae, Megascolecidae, Lumbricidae).

Niche variation during salmon spawning

Invertebrate diet was characterized in select species collected in September–October 2001–2003 from the Clatse and Neekas Rivers using stable isotope analysis (SIA) of δ15N and δ13C (Table 1). Whole invertebrate specimens were rinsed with distilled water, dried at 60°C for at least 48 h, and then ground into a fine homogeneous powder using a using a Wig-L-Bug grinder (Crescent Dental Co., Chicago, Ill). Invertebrate sub-samples (~1 mg) were assayed for total N, δ15N, total C and δ13C at the University of Saskatchewan Stable Isotope Facility using a Europa Scientific ANCA NT gas/solid/liquid preparation module coupled to a Europa Scientific Tracer 20/20 mass spectrometer. Isotopic signatures are expressed in delta notation (δ) as ratios relative to known isotopic standards of atmospheric N2 and V-PDB carbon. This is expressed in ‘parts per mil’ (‰) according to:

$$ \delta^{15} {\text{N}}\,{\text{or}}\,\delta^{13} {\text{C(}}\permille )= ({{\text{R}}_{\text{sample}}/{\text{R}}_{\text{standard}}} -1)*1000 $$
(1)

where R is the ratio of the heavy isotope (15N or 13C)/light isotope (14N or 12C).

Table 1 Terrestrial invertebrates collected from Pacific salmon carcasses on the Clatse and Neekas Rivers, British Columbia
Full size table

We initially compared stable-isotope ratios in all carrion taxa collected in fall by species and by watershed of collection using analysis of variance (ANOVA) and non-parametric Kruskal–Wallis tests. Watershed variation (Clatse, Neekas) was observed to be minimal for both isotopes (F 1,168 < 0.7, P > 0.4) and thus data were pooled across sites for further tests.

We determined guild placement of all salmon carrion species collected in the fall using a hierarchical cluster design of their stable-isotope ratios relative to chum and pink salmon muscle tissue and invertebrates from three terrestrial guilds. We used a single mean (±SE) for chum and pink salmon (δ15N = 12.68 ± 0.38‰; δ13C = −19.91 ± 0.38‰), which is comparable to other studies (Kaeriyama et al. 2004). Terrestrial guilds included litter-based Collembola (terrestrial detritivore) and Cybaeus spiders (terrestrial predator) collected in the fall of 2001 using pitfall traps (Christie et al. 2008), and carrion beetles (terrestrial scavenger: Nicrophorus defodiens) predicted to have a diet of shrews and songbirds (Hocking et al. 2007). Cluster classification was based on Euclidean distance from cluster centers using the within-groups linkage method. We chose five clusters (or guilds) as the most parsimonious based on standard distance coefficients in the hierarchical analysis (distance from four to five clusters = 1.010; distance from five to six clusters = 0.004).

We tested for evidence of current consumption of salmon from individual carcasses in fly and beetle larvae and adult Hymenoptera (Alysia alticola (Braconidae), Atractodes sp. (Ichneumonidae)) collected from separate carcasses (n = 3–4 carcasses; 4–6 individuals per carcass). For each guild, we used a single fixed-factor multivariate ANOVA to predict δ15N and δ13C isotopic variance by carcass of collection.

Seasonal and spatial analysis

Seasonal variation in the timing of invertebrate sampling is a consideration that may affect the isotope signatures of sampled taxa. In two separate years (2001, 2003), we surveyed dominant Diptera and Coleoptera from spring through to the fall in ground-level baited pitfall traps (2001: n = 6 traps) or hanging baited traps (2003: n = 10 traps) in sites below the falls within 20 m of the spawning channel on both the Clatse and Neekas Rivers. In 2001, traps were collected in spring (June), summer (late July and August) and fall (September, October). In 2003, collections were similar except they began earlier in spring (mid-May). We documented seasonal activity patterns of the dominant carrion species including the date of first and last capture. For the dominant Diptera (Calliphora terraenovae (Calliphoridae) and Dryomyza anilis (Dryomyzidae)), we examined isotopic (δ15N or δ13C) variation by season in separate analyses for individuals collected in 2001 (n = 49: Kruskal–Wallis tests) and 2003 (n = 224: ANOVA). In 2003, season, fly species, and watershed (Clatse, Neekas) were used in a three-way fixed-effect ANOVA of isotope variability. In a companion experiment starting in early October of 2000, we placed six post-spawning chum salmon carcasses in the riparian zone at each watershed. In June 2001, we collected fly pupae and pre-pupae of C. terraenovae and D. anilis from the soil at these carcass sites and reared them for up to a month until adult emergence. Emergent flies were not given an adult meal and instead immediately sacrificed for isotope analyses.

Spatial proximity to sites with high-density salmon spawning can predict the isotope signatures of sampled riparian consumers (Ben-David et al. 1998; Hocking and Reimchen 2002). We coupled both seasonal and spatial data to examine isotopic variability in two taxa with contrasting life-histories, the litter spider predators Cybaeus spp., and the blowfly C. terraenovae. C. terraenovae data was generated from hanging baited traps described above in 2003, including five traps located above the waterfall barrier to salmon on each watershed. In the spring and fall of 2001 we sampled Cybaeus spp. from a grid of passive pitfall traps in 10 × 10 m plots (nine traps per plot, three plots above falls, three plots below falls) within 20 m of the stream on each watershed. For each species, we tested isotope variability using a three-way fixed-effect ANOVA (fixed factors: season, location (above falls, below falls) and watershed (Clatse, Neekas)).

Sample treatment

In this dataset, we did not perform lipid extraction procedures as these can often lead to altered δ15N values (Carabel et al. 2006). As an alternative, lipid normalization equations have been devised to standardize δ13C values to a lipid content indicative of a protein metabolic pathway (Kiljunen et al. 2006). We performed a sensitivity analysis to determine the variation in percent lipid among species and the change in δ13C with lipid normalization based on equations derived in McConnaughey and McRoy (1979). Most sampled species showed minimal change in δ13C with lipid normalization (mean change in δ13C < 1‰), and were not normalized. Exceptions included fly larvae (−1.3‰), Nicrophorus investigator larvae (−2.3‰), N. investigator adults (−1.3‰), and N. defodiens adults (−1.3‰) owing to their higher lipid contents (34.7, 51.5, 32.3, and 34.6%, respectively). Lipid normalized δ13C data from these species is thus presented. All analysis was conducted using SPSS version 11.0 (SPSS Inc., Chicago, USA).

Results

Terrestrial invertebrate assemblage on salmon

Sixty species of terrestrial invertebrates from 36 families and 17 orders were collected from salmon carcasses on the Clatse and Neekas Rivers, B.C. (Table 1). This assemblage was dominated by the Diptera (ten spp.), Coleoptera (21 spp.) and Hymenoptera (six spp.), and consisted of saprophagous Diptera and Coleoptera (15 spp.), dipteran predators (eight spp.) and parasitoids (four spp.), opportunistic predators and saprophages (19 spp.), detritivores (five spp.) and groups with unknown associations with salmon carrion (nine spp. or groups).

Saprophagous Diptera were the most consistently observed group on salmon carcasses and attracted an array of predators. From experimental carcasses, the Diptera were found to be at least a co-dominant consumer on 159 of 186 carcasses (85.5%) although there were differences in consumer dominance by watershed (χ 23,186  = 80.2, P < 0.001). On the Clatse River, vertebrates wholly consumed 27 of 90 placed carcasses including a significantly greater proportion of fresh versus spawned-out fish (χ 22,90  = 30.8, P < 0.001). On the Neekas River, none of the 96 placed carcasses were consumed by vertebrates. Two fly species, Calliphora terraenovae and Dryomyza anilis, dominated adult collections in the baited traps. The seasonality of C. terraenovae emergence began in mid-May (2003), while the earliest observed emergence for D. anilis occurred in late June (2001, 2003). Both species were active through to mid-October. Other collected Diptera from salmon included Lucilia illustris (Calliphoridae) and species in the Heleomyzidae, Muscidae, Phoridae Sphaeroceridae, and Drosophilidae. Four species of parasitic wasps from the Braconidae, Ichneumonidae, and Figitidae were documented ovipositing on fly larvae on salmon.

Coleoptera species collected from salmon carcasses included predaceous and saprophagous species in the Staphylinidae, Silphidae, Carabidae, Leiodidae, and Hydrophilidae. From seasonal baited traps, Nicrophorus investigator was observed to be active beginning in late June through mid-October while Nicrophorus defodiens, a non-salmon carrion species, was first collected in mid-May (2003) and last observed in the first week of September (2001). N. investigator was observed as a co-dominant consumer with the Diptera on 14 of the 96 experimental carcasses (14.6%) on the Neekas River, although none of the experimental carcasses on the Clatse River were colonized.

Many litter species were opportunistic scavengers, detritivores, or predators on salmon. Slugs and snails, including Ariolimax, Haplotrema, Prophysaon and Vespericola, were commonly collected from fresh carcasses. Millipedes (e.g., Parajulidae) and worms (e.g., Enchytraidae) were common detritivores found within the rich humus underneath carcasses. Other opportunistic predators and scavengers included ants (Formicidae), centipedes (e.g., Geophilidae), spiders (e.g., Agelenidae), bristletails (Machilidae), cave crickets (Rhaphidiophoridae), mites (e.g., Parasitiformes), and springtails (Collembola).

Niche variation during salmon spawning

Niche differentiation in the salmon carrion community was observed within and among species. Invertebrates collected in September and October from salmon carcasses in the Clatse and Neekas watersheds differed in their δ15N and δ13C isotope signatures (Kruskal–Wallis tests for δ15N and δ13C: all χ2 > 55, all P < 0.001) (Table 2). Individuals were classified within one of five guilds (hierarchical clusters) which we define as: 1. Terrestrial: high trophic level; 2. Terrestrial: low trophic level; 3. Marine: salmon saprophage; 4. Marine: dipteran predator or parasitoid; and 5. Transitional marine and terrestrial diet (Fig. 1a; Table 3).

Table 2 δ15N and δ13C stable-isotope signatures (±SE) in insect species collected from salmon carcasses in autumn from the Clatse and Neekas Rivers, BC
Full size table
Fig. 1
figure 1

a Hierarchical cluster classification of δ15N and δ13C stable-isotope signatures in insect species collected from salmon carcasses from the Clatse and Neekas Rivers, BC (see Table 3). Included are three invertebrate guilds collected in the fall but known previously to have terrestrial-based diets. Cluster membership includes: 1 Terrestrial: high trophic level (open triangles); 2 Terrestrial: low trophic level (closed triangles); 3 Marine: salmon saprophage (closed diamonds); 4 Marine: dipteran predator or parasitoid (closed circles); and 5 Transitional marine and terrestrial diet (open squares). b Correlation between δ15N and δ13C in the parasitic wasps Alysia alticola and Atractodes sp. (closed squares) (pooled data: R = 0.778, P < 0.001) predicted to have a larval diet of fly larvae that have fed on salmon (cluster 3). Shown for comparison is the isotopic signature of salmon muscle tissue (±SE)

Full size image
Table 3 Individual cluster membership (1–5) from a hierarchical analysis of δ15N and δ13C stable-isotope signatures in insect species (a–i) collected from salmon carcasses from the Clatse and Neekas Rivers, BC (see Fig. 1a)
Full size table

The parasitic wasps Alysia alticola and Atractodes sp. had the highest isotope signatures of all species (δ15N values 6‰ higher than salmon), and were classified as having a diet of fly larvae consuming salmon (cluster 4) or were transitional between marine and terrestrial food chains (cluster 5). δ15N and δ13C signatures in the parasitic wasps were strongly correlated (Fig. 1b; R = 0.778, P < 0.001), with no difference in the relationship between δ15N by δ13C by wasp species (F 1,36 < 0.5, P > 0.5). δ15N and δ13C values for wasps also did not differ by their carcass of collection (F 3,18 < 0.5, P > 0.7).

All fly larvae and beetle N. investigator larvae rearing on salmon were grouped in cluster 3, consistent with a salmon diet. Most N. investigator adults (85%) were also classified into cluster 3. In contrast to results for parasitic wasps, carcass of collection explained 67 to 94% of the variation in δ15N and δ13C in fly and beetle larvae (Fly larvae: F 3,14 > 9.5, P < 0.001; N. investigator larvae: F 2,11 > 24, P < 0.001). The staphylinid beetle Anthobium fimetarium had high but variable signatures within both marine and terrestrial food chains. In comparison, the staphylinid Bisnius seigwaldi had signatures indicating a terrestrial predatory diet (cluster 1) overlapping with previously sampled terrestrial species all classified within clusters 1 or 2.

Seasonal and spatial diet variation

δ15N and δ13C signatures in adult flies C. terraenovae and D. anilis varied by season of collection with the highest values observed in the spring, followed by individuals collected in the summer and fall (Fig. 2). Similar results were observed across 2 years. A seasonal decline in δ15N and δ13C was observed in 2001 (Kruskal–Wallis: all χ2 > 4.6, all P < 0.031), with the highest values observed in flies collected from salmon carcass sites in early spring and raised until adult emergence. In 2003, season was an important predictor of both δ15N and δ13C (F 2,219 > 27, P < 0.001) with no significant variation observed between watersheds (F 1,219 < 2.2, P > 0.14). Adult D. anilis had higher δ15N (F 2,219 = 44.5, P < 0.001), but not δ13C (P = 0.60), than C. terraenovae. In the fall, adults of both fly species were largely classified into cluster 1, indicating a high trophic level terrestrial diet (Table 3).

Fig. 2
figure 2

Seasonal variation of δ15N and δ13C stable isotope signatures (±SE) in two species of flies Calliphora terraenovae (Calliphoridae) (filled symbols) and Dryomyza anilis (Dryomyzidae) (open symbols) collected from the Clatse and Neekas Rivers, BC. Raised flies (square symbols) were collected as pupae from salmon carcass sites in early spring and reared to adult emergence in the lab. Remaining flies were caught using baited traps in May/June (spring: triangles), July/August (summer: diamonds) and when salmon return to spawn in September/October (fall: upside-down triangles). Data pooled across 2 years (2001, 2003) and two watersheds (Clatse, Neekas)

Full size image

Contrasting seasonal and spatial isotopic patterns were observed between two species with different life-histories, the litter spiders Cybaeus spp., and the blowflies C. terraenovae (Fig. 3). In Cybaeus spp., strong spatial variation but no seasonal variation in δ15N was observed (watershed: F 1,86 = 77.2, P < 0.001; falls: F 1,86 = 595.1, P < 0.001; watershed*falls: F 1,86 = 63.8, P < 0.001; season: F 1,86 = 1.7, P = 0.2). δ13C signatures in Cybaeus spp. showed no significant main effects (all F 1,86 < 2.2, P > 0.13), but a significant season*falls interaction term (F 1,86 = 8.7, P = 0.004). For C. terraenovae, no significant spatial variation in δ15N or δ13C was observed (watershed or falls F 1,128 < 2.3, P > 0.13), but both δ15N and δ13C were more enriched in the spring than the fall (F 1,128 > 11.2, P < 0.001).

Fig. 3
figure 3

Temporal and spatial variation in δ15N and δ13C stable-isotope signatures (±95% CI) in litter spiders (Cybaeus spp.) (a, c) and blowfly adults (Calliphora terraenovae) (b, d) collected in spring (closed triangles) and fall (open squares) from the Clatse and Neekas Rivers, BC. Samples were collected from both above and below waterfall barriers to salmon migration on each watershed within 20 m of the stream channel

Full size image

Discussion

Pacific salmon carcasses provide a predictable source of protein for hundreds of species throughout coastal areas of the North Pacific, with most research focusing on consumption of salmon by vertebrates or aquatic invertebrates (Bilby et al. 1996; Ben-David et al. 1997; Wipfli et al. 1998). In this study, over 60 terrestrial invertebrate species from 36 families were documented from salmon carcasses from two watersheds on the central coast of British Columbia with the highest abundance represented within the Diptera, Coleoptera, and Hymenoptera. Many species documented here have also been collected from salmon in watersheds from the Queen Charlotte Islands (Reimchen 1994; Hocking et al. 2006), Vancouver Island (Reimchen et al. 2003) and Puget Sound (Jauquet et al. 2003). The Diptera, C. terraenovae, D. anilis, and L. illustris, have also been collected from salmon carcasses in Alaska (Meehan et al. 2005), emphasizing their widespread interaction with spawning salmon and their potential importance as a diet resource for further consumers.

Factors responsible for the dominance of flies, carrion beetles, or vertebrates on salmon carcasses likely relate to factors such as salmon spawning density, carcass energy density and size, and features of the environment such as stream habitat and ambient temperature (Reimchen 1994; Gende et al. 2001, 2004; Hocking et al. 2006). When salmon are abundant and accessible, bears can afford to be selective and will feed preferentially on the brain, eggs, and dorsal musculature (Reimchen 2000). In contrast, when salmon are scarce, vertebrates will consume a much larger proportion of individual carcasses. For example, the Neekas River supports higher chum salmon density than the Clatse River and provides higher access for vertebrates to flooded carcasses. In this study, vertebrates consumed none of the experimental carcasses at Neekas compared to 30% of experimental carcasses at Clatse. At Clatse, vertebrates also preferred fresh versus spawned-out fish indicating that carcass condition and energy density are important predictors of vertebrate scavenging (Gende et al. 2004). In the absence of vertebrate scavenging, the Calliphoridae are favored on large carcasses due to their high reproductive output, growth rates, and overall competitive ability, while other fly species are more competitive on smaller remnants (Hocking and Reimchen 2006).

Invertebrate species collected from salmon carrion are partitioned among ecological roles on the carcass including saprophages, dipteran predators and parasitoids, and opportunistic predators, scavengers, and detritivores from the terrestrial food web. Evidence for a diet of salmon or salmon consumers was found in larval flies and larval N. investigator, adult parasitic wasps Alysia alticola and Atractodes sp., and adult beetles N. investigator and A. fimetarium. In comparison, the staphylinid B. siegwaldi was classified as a terrestrial predator based on its stable--isotope composition and, along with groups such as spiders, millipedes, snails and carabid beetles, are likely to be opportunistic predators on the dipteran eggs and larvae or opportunistic scavengers on salmon tissues. The population-level consequences of potential reductions in salmon abundance for these opportunists remain unclear as the annual contribution of salmon to their diet is predicted to be small. For example, no seasonal isotopic variation was observed in the litter spiders Cybaeus spp.

In contrast to opportunist species, insects that breed directly on salmon carcasses are predicted to exhibit more seasonal variation in diet (and isotope signatures) comparable to that seen in bears, mustelids, and other vertebrates (Ben-David et al. 1997; Hilderbrand et al. 1999). We observed strong correlations between δ15N and δ13C in flies and parasitic wasps, with some individuals showing transitional isotope signatures (cluster 5) between marine and terrestrial food chains. We suggest that tissue δ15N and δ13C in individuals reared from salmon carcass sites will likely shift from marine to terrestrial signatures over time as adults feed on terrestrial sources. Seasonal shifts in isotope signatures in adult flies C. terraenovae and D. anilis provide evidence. High δ15N and δ13C values were observed in flies in the spring, with signatures declining through the summer and fall. This implies a legacy of a larval diet of salmon from the previous fall, similar to that observed in burying beetles (Hocking et al. 2007). Further, flies that were raised in our laboratory rearing experiment from pupae at salmon carcass sites and were processed upon emergence (and not given an adult meal) had the highest δ15N and δ13C. δ15N signatures from these individuals may be higher than expected based on a 3‰ trophic shift from salmon because of an ontogenetic shift in δ15N during Diptera metamorphosis (Tibbets et al. 2008).

Individual flies with relatively depleted isotope signatures were likely to have been raised on carrion from terrestrial-based sources such as mammals and birds with δ15N and δ13C ranging from 0 to 10‰ and −30 to −23‰, respectively, rather than salmon (Hocking et al. 2007). This may be particularly true in C. terraenovae as ~20 individuals had isotope signatures lower than 10‰ for δ15N (and overlapping with the terrestrial carrion feeder Nicrophorus defodiens), while the lowest δ15N signature observed in D. anilis across all seasons was 11.10‰.

Spatial variation in δ15N and δ13C is common in biota that feed across environmental gradients (e.g., Anderson and Polis 1998), although this was not observed for the Diptera presented here. Stable-isotope signatures in C. terraenovae did not differ from collections above versus below a waterfall barrier to salmon in both spring and fall on two watersheds. In contrast, the litter spider predators Cybaeus spp. showed strong spatial variation in δ15N (none in δ13C) and no seasonal variation in either isotope. These patterns highlight the different dispersal abilities and pathways of salmon nutrient uptake in these two consumers. C. terraenovae feed directly on salmon and can easily disperse along riparian corridors to alternate sites such as above the falls. In contrast, Cybaeus spp. have reduced mobility, do not feed extensively on salmon, and thus are site-specific indicators of the fertilization pathway of salmon-derived nitrogen in riparian zones (Ben-David et al. 1998; Hocking and Reimchen 2002).

The parasitic wasps Alysia alticola and Atractodes sp. had the highest isotope signatures of all species placing them at trophic positions 5.0–6.0 in the marine food chain (adult salmon are 3.0–4.0), comparable to that of sea lions and orca whales (Hobson et al. 1997; Pauly et al. 1998). Analyses of individuals collected from separate carcasses indicate that fly larvae have carcass-specific signatures, while parasitic wasps do not. This means that the high marine trophic positions of the parasitic wasps originated from their larval diet of flies from salmon carcasses the previous fall (1 year prior to sampling). As a whole, the life-history of these wasps is not well known, and it is possible that local populations extend diapause to time their emergence to the predictable and energy-rich pulse of fly larvae on salmon. This life-history strategy would involve trade-offs as increased diapause duration can reduce fitness (Ellers and van Alphen 2002).

Cataloguing invertebrate associations with salmon may be useful for developing indices of intact salmon runs, bear foraging, and subsequent nutrient transfer in coastal watersheds. Insects are commonly used as indicators (Spellerberg 1993; Rainio and Neimelä 2003), such as in forensic studies (Greenburg 1991). We show that species mobility and the pathway of salmon nutrient uptake influences both seasonal and spatial stable-isotopic patterns and is an important consideration for future analyses. The high marine trophic position of parasitic wasps illustrates their potential as ultimate indicators of salmon-forest nutrient transfers since they rely on larval Diptera that have fed on salmon. In turn, the abundance of the Diptera is directly dependent on the biomass of salmon carcasses transferred to the riparian zone (Hocking and Reimchen 2006), a process that varies with salmon spawning density and bear access to fish (Reimchen 2000; Gende et al. 2001, 2004). Declines in salmon populations may ultimately erode the complex freshwater and terrestrial food webs that act to couple marine and terrestrial ecosystems (Finney et al. 2000; Gresh et al. 2000; Stockner 2003, Christie and Reimchen 2008). Setting conservation targets for wild Pacific salmon that reduce these potential losses in biodiversity remains an important goal.