Molecular scatology and next-generation sequencing identified previously unknown linkages among ecosystems in the diet of Semipalmated Sandpipers (Calidris pusilla) in the Bay of Fundy, Canada. During their annual migratory stopover, the birds consumed a wider range of prey items than previously reported, which suggests that they are not selecting for the amphipod Corophium volutator and are acting as generalist foragers. Our analysis identified several novel prey items—arachnids, crabs, bivalves, several terrestrial and freshwater insect species, ctenophores, cnidarians, and fish (likely eggs or juveniles)—indicating that Semipalmated Sandpipers consume prey from marine, freshwater, and terrestrial ecosystems. Connections between Semipalmated Sandpipers and freshwater and terrestrial ecosystems were previously unknown in the Bay of Fundy. Current conservation efforts for this species are focused on beach and intertidal habitats; however, we may also need to consider the surrounding freshwater and terrestrial habitat.
Most migratory shorebirds exhibit strong dietary flexibility (Hicklin and Smith 1979, Pienkowski et al. 1984, Skagen and Oman 1996, Skagen 2006). This is not surprising given the variety of prey items that long-distance migrants encounter. Food quality at migratory stopover sites is important because birds use these areas to accumulate fat prior to the next leg of their migration (Skagen 2006, Niles et al. 2009, Smith et al. 2012). When birds specialize on particular prey items at these stopover areas, they become vulnerable to changes in the food base (Skagen 2006, Niles et al. 2009). For example, Red Knots (Calidris canutus) that forage on horseshoe crab eggs (Limulus polyphemus; Tsipoura and Burger 1999, Karpanty et al. 2006) have experienced substantial population declines as a result of human overexploitation of horseshoe crabs (Niles et al. 2009).
The Semipalmated Sandpiper (C. pusilla) is an Arctic-breeding shorebird that undergoes a long-distance fall migration to wintering grounds in South America. Much of the population is thought to complete this migration with a single stop in the Bay of Fundy, Canada (Gratto and Cooke 1987, Hicklin 1987, Gratto-Trevor et al. 2012). While in the region, these birds feed extensively on infaunal prey found on intertidal mudflats (Hicklin 1987, Hamilton et al. 2006, Gratto-Trevor et al. 2012). Historical studies using stomach content analysis suggested that while in the Bay of Fundy, Semipalmated Sandpipers foraged preferentially on the amphipod Corophium volutator (Hicklin and Smith 1979, 1984). Thus, like Red Knots, Semipalmated Sandpipers might be vulnerable to changes in their prey base. However, recent work using stable isotope analysis and behavioral observation has revealed that Semipalmated Sandpipers consume a broader range of intertidal prey items than previously reported (MacDonald et al. 2012, Quinn and Hamilton 2012). The results of these studies suggest a diet composed of a variety of polychaete worms, biofilm, C. volutator, and ostracods, which matches what we know about the foraging ecology of Semipalmated Sandpipers in general. These shorebirds exhibit a broad diet on their breeding grounds in the Arctic (Holmes and Pitelka 1968, Baker 1977) and, over the course of their annual cycle, consume prey items from at least 25 orders and 37 families of invertebrates (Skagen and Oman 1996, Skagen 2006). Although differing methodology makes it impossible to directly compare historical and current diets, perceived changes over time in Semipalmated Sandpipers' diet in the Bay of Fundy match a general decline in abundance of C. volutator at some sites (while other sites increased or remained stable), with concurrent increases in other prey such as polychaete worms (see data in and compare the following: Yeo 1977, Gratto 1979, Peer et al. 1986, Wilson 1988, Shepherd et al. 1995, Barbeau et al. 2009, Gerwing et al. 2015). The observed temporal shift in their diet suggests that Semipalmated Sandpipers are probably feeding as generalists according to prey availability while in the Bay of Fundy, as they do elsewhere during their annual cycle.
Semipalmated Sandpipers in the Bay of Fundy typically feed when the mudflat is exposed, and rest on shore during high tide. Until now, we have assumed that their diet is fully intertidal-based; however, given their opportunism and broad diet, there is no inherent reason why they would not take prey of other origins if given the opportunity. Diet opportunism is common in other shorebirds, such as the Western Sandpiper (C. mauri), Least Sandpiper (C. minutilla), Pectoral Sandpiper (C. melanotos), Lesser Yellowlegs (Tringa flavipes), and Killdeer (Charadrius vociferus) (Davis and Smith 2001, Smith et al. 2012, Jardine et al. 2015), as well as in birds generally (Poulin et al. 1994). Thus, opportunistic foraging by Semipalmated Sandpipers could generate links between this species and multiple ecosystems.
Unfortunately, existing methods of diet assessment such as gut content analysis and stable isotope analysis do not lend themselves to identification of unexpected prey items. Morphological methods suffer from misidentification of similar prey items, overrepresentation of hard-to-digest prey items, and underrepresentation of easily digested prey items (Barrett et al. 2007, Deagle et al. 2007, Pompanon et al. 2011). Stable isotope or fatty acid analyses offer an assessment of diet integrated over an extended period but rely on a priori knowledge of an animal's diet (Deagle et al. 2005, 2007, Casper et al. 2007, Pompanon et al. 2011, Quinn and Hamilton 2012). Therefore, we employed molecular scatology, the use of molecular techniques to identify prey DNA in fecal samples (Barrett et al. 2007, Deagle et al. 2009, Pompanon et al. 2011), as a way of gaining greater insight into the breadth of the Semipalmated Sandpiper diet. Assessing diet via molecular scatology offers a noninvasive and high-resolution snapshot of a predator's diet, even when prey items are degraded (Kohn and Wayne 1997, Deagle et al. 2006, Pompanon et al. 2011).
Populations of most North American shorebirds, including Semipalmated Sandpipers, are declining (Bart et al. 2007, Gratto-Trevor et al. 2012, Galbraith et al. 2014). Because an animal's diet influences every aspect of its biology, understanding diet is essential for conservation and management plans (Fryxell et al. 2014). Full knowledge of Semipalmated Sandpipers' diet on stopover grounds is particularly important because the primary function of these stops is to refuel for the next leg of migration (Davis and Smith 2001, Skagen 2006, Smith et al. 2012). The results of the present study will enhance our understanding of dietary breadth in Semipalmated Sandpipers during their time at this very important stopover region.
Study Sites and Sample Collection
We collected fecal samples from Semipalmated Sandpipers in summer 2010 on 3 intertidal mudflats in the upper Bay of Fundy, Canada (Figure 1): Avonport (AV), Grande Anse (GA), and Mary's Point (MP). To assess temporal variability in diet, we sampled at 2 points in the birds' stopover period: August 3–5 and August 22–23. Fecal samples were collected on a receding tide approximately 1–2 hr after the birds ceased roosting and commenced foraging on the mudflat, thus ensuring that we obtained only fresh fecal samples (in that feces dropped before high tide would have been washed away) and potentially allowing us to detect prey obtained during roosting and mudflat foraging. All fecal samples were from the upper intertidal zone, which facilitated clear observation of the birds from shore and quick access to samples as soon as the birds had moved out of the area. Flocks were observed to ensure that fecal samples were collected only from flocks composed primarily of Semipalmated Sandpipers. Although other species of shorebirds were present, Semipalmated Sandpipers typically make up ~95% of shorebirds using the region (Hicklin and Smith 1979, Hicklin 1987). We collected, with sterilized tweezers, 50–75 fresh fecal samples (the brown digested matter at the center and on top of the uric acid), being attentive not to touch the mudflat surface with the tweezers. Each fecal sample was stored in 95% ethanol immediately upon collection and frozen (−20°C) within 12 hr. Samples remained frozen until DNA extraction.
Primer Design and DNA Preparation
Given that DNA extracted from feces is often degraded (Kohn and Wayne 1997, Deagle et al. 2006) and that molecular analyses based on a single locus can result in variable coverage of diet composition (Bowser et al. 2013), we used universal primer pairs that targeted short fragments (130–300 base pairs [bp]) of 2 mitochondrial genes: cytochrome c oxidase subunit 1 (COI; Meusnier et al. 2008) and the 16S ribosomal RNA subunit (16S; Deagle et al. 2007). We used a pooled massively parallel sequencing (MPS) approach described by Puritz et al. (2012). To facilitate the recovery and identification of sequences obtained from individual fecal samples, we included a 10-base multiplex identifier (MID) tag between the Lib-L 454 sequencing adapter (26 bp plus a 4 bp signal calibration key) and the universal primer (16S or COI) in our custom engineered forward and reverse primers. Details of the primer design, including the MID tags used, can be found in Bowser et al. (2013).
Ethanol was removed from the fecal samples by decanting following 30 min of centrifugation at 4°C. DNA was extracted with QIAamp DNA Stool Mini Kit (Qiagen, Toronto, Ontario, Canada) following the manufacturer's protocol. Samples with small amounts of fecal material were eluted with 75–100 μL of buffer AE instead of the recommended 200 μL to increase DNA concentration in the extracted sample. DNA was stored in 2 mL microcentrifuge tubes at −20°C until used for 16S and COI gene amplification. Fecal samples can contain significant amounts of DNA from the host species as a result of the sloughing of cells from the digestive tract (Shehzad et al. 2012). However, we did not attempt to block the amplification of the host DNA during polymerase chain reaction (PCR), because several species of shorebirds forage on the mudflats, and we used the traces of host DNA to exclude samples that did not come from Semipalmated Sandpipers (see below).
454 Library Preparation
Amplification of fecal DNA with 16S MID-tagged sequencing primers was achieved in 20 μL reactions containing 5 μL undiluted template fecal DNA, 0.2mM dNTP (New England Biolabs, Whitby, Ontario), 1X bovine serum albumin (New England Biolabs), 5 mM MgSO4 (Life Technologies, Burlington, Ontario), 0.5 μM of each primer (Integrated DNA Technologies, Coralville, Iowa, USA), 1X High Fidelity Buffer (Life Technologies), and 1 unit of Platinum Taq DNA Polymerase High Fidelity (Life Technologies). The thermocycling protocol for 16S began at 94°C for 2 min followed by 35 cycles of 94°C for 30 s, 55°C for 30 s, and 68°C for 45 s, with a final extension of 68°C for 5 min (C-1000 Thermal Cycler, Bio-Rad, Mississauga, Ontario). Amplification with COI followed similar component and cycling conditions as the 16S, but the annealing temperature was dropped to 45°C. Successful amplification of individual PCR samples was verified by electrophoresis in 1.5% agarose and visualized under ultraviolet light using SYBR Safe (Life Technologies). Amplicons were purified using the QIAEX II Gel Extraction Kit (Qiagen), and the concentration was determined with dsDNA BR Assays on a Qubit 2.0 Fluorometer (Life Technologies). Samples with amplicon concentrations <1 ng μL−1 were reamplified using the same PCR conditions. Finally, the pooled library was prepared by combining 20 ng of each of the 16S and COI amplicons and concentrating to 35 ng μL−1 in 150 PCR-grade H2O using the DNA Clean & Concentrator-100 (Zymo Research, Irvine, California, USA). The library was sequenced unidirectionally on half a pico titre plate using the Roche GS-FLX (454) platform at the Genome Quebec Innovation Centre (McGill University, Montreal, Quebec, Canada).
454 Data Sorting and MOTU Identification
Raw reads (DNA sequences) were imported to the CLC Genomics Workbench (Qiagen) for downstream sequence analyses. The library was first sorted by MID tag and then by locus based on the first 5 bases of the primer (GACGA for 16S; TCCAC for COI). Following the trimming of the MID tag and primers, reads were discarded if (1) there were more than 2 ambiguous nucleotides, (2) they had a minimum quality score >0.01 (equivalent to a Phred quality score of 20), and (3) they were <30 nucleotides. Duplicate reads within a sample were removed to increase the representation of unique reads.
The reads were assembled into contigs using the default de novo assembly parameters. Both contigs and singletons were considered “molecular operational taxonomic units” (MOTUs) and were searched in the nucleotide database of the National Center for Biotechnology Information (NCBI; U.S. National Library of Medicine, Bethesda, MD, USA) using the Basic Local Alignment Search Tool for nucleotide sequence alignments (BLASTn) through the CLC Genomics Workbench. We used the following criteria to identify candidate species in the fecal DNA samples: (1) 50-nucleotide minimum length of query sequence, (2) 90% minimum fraction on length consensus between top hit and query sequences, (3) 90% minimum identity between top hit and query sequences, and (4) 60 minimum bit score of top hit sequence. Finally, the identification of each MOTU was accomplished by matching the top-hit 16S or COI MOTU with 69,614 genera and 305,936 species that were registered to the taxonomy database of the NCBI as of July 2014. A list of all animal prey items detected within each fecal sample is presented in Supplemental Material Table S1 (AUK-15-176_Supplemental Materials.docx).
Identification of Corophium volutator in Fecal Samples
Although our main goal was to assess the breadth of the Semipalmated Sandpipers' diet, we also wanted to evaluate the prevalence of Corophium volutator in their diet. Both empirical and bioinformatics analyses indicated a mismatch between the COI primers and the sequence in C. volutator. There was an adequate match with the 16S primers, but the public database did not contain Corophium spp. Therefore, 16S sequences and our BLASTn searches failed to identify C. volutator in any fecal sample. To probe our library further, we sequenced the short universal 16S from 7 different C. volutator DNA extractions and constructed our own species-specific reference sequence for this locus. A similar database was not constructed for other infauna (e.g., the polychaetes Nereididae and Nephtyidae; Gerwing et al. 2015) because the public database contained sequences from closely related species, and no mismatches between primers and the available sequences were detected. Using the same settings in CLC Genomics Workbench described above, we aligned all demultiplexed and quality-trimmed 16S reads and 16S MOTUs (contigs + singletons) to our C. volutator 16S consensus. As a result, we found 67 of the 1,775 16S contigs and 867 of the 6,146 16S singletons (11.8% of the 7,896 16S MOTUs) matched to the 16S consensus, positively identifying >100 fecal samples as containing C. volutator DNA.
To ease interpretation and increase confidence in prey identities, DNA sequences were placed into prey bins.
Fecal Samples with and without Semipalmated Sandpiper DNA
Semipalmated Sandpipers always dominated flocks of foraging shorebirds; however, the possibility of collecting feces from other species remained. To assess this, we examined fecal samples for the presence of Semipalmated Sandpiper DNA, as well as that of other shorebirds. Among a total of 389,328 sequences, we detected 16,836 non–Semipalmated Sandpiper avian DNA sequences in 54 fecal samples; therefore, these samples were excluded from all further analysis. To determine whether it would be appropriate to include the samples that contained no DNA of any bird species in our investigation of the Semipalmated Sandpiper diet, we compared prey items found in fecal samples with Semipalmated Sandpiper DNA to those with no bird DNA, using an analysis of similarities (ANOSIM) in the statistical program PRIMER with the PERMANOVA (permutational multivariate analysis of variance) add-on (McArdle and Anderson 2001). A diet matrix was constructed using presence–absence of each prey item (in prey bins) found in feces samples for COI and 16S (individually and combined), using the Bray-Curtis coefficient.
Data Analysis of Binned Prey Items Determining Semipalmated Sandpiper Diet
Data from both 16S and COI loci were combined to maximize prey identification (not all samples were amplified for both loci; Table 1). Frequency of occurrence (FOO; number of fecal samples that a prey item appeared in, divided by total number of samples) was calculated for each prey bin.
Summary of the number of successful amplifications of prey DNA in fecal samples of Semipalmated Sandpipers collected in August 2010, for 16S and COI and the 2 loci combined, excluding all samples that contained non–Semipalmated Sandpiper avian DNA. For site names, see Figure 1.
PRIMER was also used to examine temporal and spatial variability in the Semipalmated Sandpiper diet. A PERMANOVA was conducted with site and time as random factors. The resemblance matrix was constructed from prey item presence–absence data using the 16S and COI loci as described above for the ANOSIM. We set α = 0.05 a priori to assess statistical significance. Following a significant site × time interaction (see below), we ran a one-way PERMANOVA with all site–time combinations as levels of a single independent variable. Finally, variance components (Searle et al. 1992, Anderson et al. 2008) were calculated to determine how much of the observed variation in the Semipalmated Sandpiper diet was due to spatiotemporal factors.
The ANOSIM detected no differences in diet between fecal samples with Semipalmated Sandpiper DNA and without any bird DNA (16S: Semipalmated Sandpiper DNA, n = 47; no bird DNA, n = 101; R = 0.04, P = 0.18. COI: Semipalmated Sandpiper DNA, n = 109; no bird DNA, n = 162; R = −0.008, P = 0.65. Both loci: Semipalmated Sandpiper DNA, n = 110; no bird DNA, n = 173; R = 0.001, P = 0.87). Therefore, we included in subsequent analyses all samples that contained either Semipalmated Sandpiper DNA or no bird DNA.
The pooled library generated 389,328 total reads, of which 118,253 unique high-quality sequences were used for analysis after samples containing non–Semipalmated Sandpiper avian DNA and low-quality DNA reads were removed (Table 2). We assembled 7,921 and 26,189 unique MOTUs from the 16S and COI data, respectively. BLASTn analysis of these data identified 132 unique prey taxa in the Semipalmated Sandpiper diet (Table 3), not only from marine (intertidal and pelagic) systems, but also from freshwater and terrestrial ecosystems (to compare FOO of main prey items by locus, see Supplemental Material Table S2 (AUK-15-176_Supplemental Materials.docx)). The diet varied over time and space (Table 4); however, the majority of variation was observed between fecal samples (residual in the second analysis: 80%), whereas variation in diet between sites and times accounted for 20% of the observed variation.
Total numbers of reads (with percentage of total in parentheses) and numbers of molecular operational taxonomic units (MOTUs) identified by the analysis of 389,328 raw reads sequenced from the pooled library for prey items in feces of Semipalmated Sandpipers collected from the Bay of Fundy in August 2010.
Frequency of occurrence (%) of prey items in feces of Semipalmated Sandpipers, identified by the combination of 2 DNA loci (the mitochondrial COI and 16S; Table 1) in the Bay of Fundy in August 2010. Total represents the frequency of occurrence pooled over sites and times.
Results of PERMANOVA investigating whether the diet of Semipalmated Sandpipers varied over time and space at 3 intertidal mudflats (sites AV, GA, and MP; see Figure 1) in the Bay of Fundy, Canada. Diet was assessed during 2 periods in 2010 at each site: “early” (August 2–5) and “late” (August 22–23). Prey items were identified via presence–absence in Semipalmated Sandpiper feces, using the 16S and COI mitochondrial loci. In the first analysis (A), site and time are random factors. In the second analysis (B), the random factor “site–time” combines site and time in a single factor with 6 levels (AVEarly, AVLate, GAEarly, GALate, MPEarly, MPLate); 999 permutations were conducted for this second analysis. Significant and interpretable sources of variation are in bold (α = 0.05).
Diatoms were present in every sample, and the second most common prey item was C. volutator, which was present in >65% of samples from all except one mudflat–time combination (MP; August 5, 2010). Arachnida, Ostracoda, Nemertea, and Nematoda had consistently low FOO, and the remaining prey items generally had higher but variable FOO. The FOO and number of prey items varied over time and space, though there were no systematic temporal trends evident across locations. The most notable variability was at MP during the early sampling period, where Semipalmated Sandpipers appeared to have a substantially different diet than at any other location or time: Relatively few individuals consumed C. volutator; use of copepods was much higher than anywhere else; use of various insects increased; and one-third of the Semipalmated Sandpipers consumed nematodes, which were never present in the diet elsewhere or at other times (Table 3). The number of different prey items (binned) that were consumed was relatively consistent across sites and times, though there was a trend at GA for use of previously unrecognized prey items to be less frequent, especially in the late sampling period, when the diet appeared to be more restricted. The diet at AV exhibited a higher FOO of bivalves, gastropods, and fish, during both periods, compared with MP and GA. Overall, we observed several prey items not previously identified for this species in the Bay of Fundy: arachnids, crabs, bivalves, several terrestrial and freshwater insect species, fish (likely eggs or juveniles), and cnidarians–ctenophores.
Diet Breadth and Variability
Fecal samples from Semipalmated Sandpipers staging in the Bay of Fundy contained a wide variety of prey items and varied over time and space (Tables 3 and 4). In the past, it was thought that Semipalmated Sandpipers foraged preferentially on C. volutator in this region (Hicklin and Smith 1979, 1984), and we did observe C. volutator in the majority of fecal samples. However, this does not suggest selection or preference, because C. volutator, while perhaps not as abundant as it once was (Yeo 1977, Gratto 1979), is still the most abundant infaunal resident of these mudflats (Gerwing et al. 2015). A high FOO of the most common prey item would be expected if Semipalmated Sandpipers are generalist foragers. Further, the low FOO of C. volutator in Semipalmated Sandpiper feces at MP on August 5 (see below) supports our view that Semipalmated Sandpipers are not dependent on this intertidal amphipod. The generalist foraging detected in the present study is consistent with results of recent studies of the Semipalmated Sandpiper diet in the Bay of Fundy (MacDonald et al. 2012, Quinn and Hamilton 2012), with the species' diet while breeding in the Arctic (Holmes and Pitelka 1968, Baker 1977), and with the breadth of prey items consumed over the species' annual cycle (Skagen and Oman 1996, Skagen 2006).
The fact that Semipalmated Sandpipers appear to be generalist foragers in the Bay of Fundy is encouraging in a conservation context. Populations of most North American shorebirds, including Semipalmated Sandpipers, are declining. These declines are partially a product of anthropogenic alteration of key shorebird habitat (Bart et al. 2007, Gratto-Trevor et al. 2012, Galbraith et al. 2014), especially migratory stopover sites (Davis and Smith 2001, Skagen 2006, Smith et al. 2012). The broad diet we documented may suggest that these birds will be relatively resilient to changes in their environments, given that diet generalists are more resistant to changing environmental conditions than specialists (Colles et al. 2009).
The observed variation in the Semipalmated Sandpiper diet over space and time (Tables 3 and 4) was not surprising. Jardine et al. (2015) observed that the diet of Western Sandpipers varied between foraging sites located within 100 km of each other along the west coast of Canada. Similarly, Davis and Smith (2001) reported that diets of several shorebirds varied over time within the same migratory stopover point. In our study, although there was substantial spatiotemporal diet variability, the majority of it was between individuals. Some of the observed variability in the Semipalmated Sandpiper diet is likely linked to variation among sites in prey abundance (see data in Gerwing et al. 2015) and suitability. For example, the Semipalmated Sandpiper diet at AV frequently included gastropods, but these were relatively rare prey items at MP or GA. The mudsnail Nassarius obsoletus, although present at all the sites, would not normally be considered a potential prey item for Semipalmated Sandpipers because of its size and protective shell. However, during August 2010 there was an unusually high density of very small juvenile snails in the upper intertidal of AV (T. G. Gerwing personal observation), which could easily be consumed by foraging Semipalmated Sandpipers; this may explain the higher FOO of Gastropoda at AV during August.
Conversely, other observed foraging patterns appear to be unrelated to prey availability. At MP on August 5, 2010, the FOO values of C. volutator and of copepods were very low and high, respectively, compared with other sites and with that site later in the season. This was not matched by the availability of these prey items, given that C. volutator was very abundant (~80,000 individuals m−2) and copepods were also readily available (~20,000 individuals m−2) at that time (Gerwing et al. 2015). Thus, availability of prey did not drive this observed variation in the Semipalmated Sandpiper diet. It is possible that the sampling in early August followed an unusual foraging bout for Semipalmated Sandpipers, in which they fed in a localized area that was not representative of the prey availability on the broader mudflat. More detailed prey sampling in specific areas where birds were observed feeding would be required to clarify this point. However, our findings do highlight the fact that the species' diet can vary widely within a relatively short period, again pointing to foraging flexibility.
Our analysis identified diatoms in all fecal samples, and there is other evidence that Semipalmated Sandpipers consume diatoms in this system (MacDonald et al. 2012, Quinn and Hamilton 2012). Similarly, biofilm has been identified as a major diet component in numerous other small-bodied sandpipers (Kuwae et al. 2012, Jardine et al. 2015). Although we were careful not to contact the sediment when collecting feces, we cannot exclude the possibility that diatom contamination occurred when feces contacted the sediment. Therefore, although our results are consistent with those of other studies with respect to diatom consumption, they should not be viewed as an independent confirmation.
Use of Multiple Ecosystems
Notably, we detected a variety of terrestrial, freshwater, and pelagic taxa in the diet of Semipalmated Sandpipers. Although links to nonmarine systems have not previously been identified for Semipalmated Sandpipers while in the Bay of Fundy, they may not be new. Hicklin and Smith (1979) observed unknown insects in 7.1–66.6% of Semipalmated Sandpiper gut contents in the Bay of Fundy, but these insects accounted for only 0.1–1.4% of Semipalmated Sandpiper gut content volume. Gut content analyses can miss rare prey and can skew results, depending on time of collection, digestibility, and the presence of identifiable parts that resist digestion (Deagle et al. 2005, Barrett et al. 2007, Pompanon et al. 2011). On the basis of our findings, we suggest that nontraditional prey items, including those from other ecosystems, make up a variable but nontrivial component of the Semipalmated Sandpiper diet and merit further investigation.
Freshwater prey items (mayflies, caddisflies, and damselflies) may have washed onto the mudflats through small streams that discharge into the Bay of Fundy. Semipalmated Sandpipers also frequently forage near these streams in the upper intertidal area, potentially picking up such items before reaching the mudflats. Variability in FOO (0–54%; average = 16.5%) suggests that freshwater insects are taken opportunistically and may be more accessible at some sites or times than at others, possibly depending on insect life stage. Terrestrial prey items, on the other hand, are probably consumed opportunistically during roosting (T. G. Gerwing personal observation) or as birds are moving off the beach toward mudflats as the tide starts to recede. The FOO values for terrestrial items were usually low, all <20% except for Diptera on one occasion (Table 3). Therefore, terrestrial prey items appear to be minor but consistent components of the Semipalmated Sandpiper diet, present in at least some birds at all sites and times. Pelagic prey items (Atlantic herring, flounder, unknown fish, cnidarians–ctenophores) likely consist of eggs or the smallest juveniles, along with detrital remains deposited on the mudflats as the tide recedes and then consumed opportunistically.
Novel Prey Items
Using molecular scatology, we observed several dietary items (arachnids, crabs, bivalves, several terrestrial and freshwater insect species, fish, and cnidarians–ctenophores) not previously identified as prey. Although Hicklin and Smith (1979) observed the gastropod Hydrobia totteni in Semipalmated Sandpiper gut contents, we found DNA from several gastropod taxa (Littorina spp., Hydrobia spp., Nassarius obsoletus). In some cases, snails were relatively common in the Semipalmated Sandpiper diet (FOO: 0–73.3%; average = 26.4%), which suggests that snails may occasionally comprise the bulk of prey consumed during a foraging bout. Their consumption probably depends on the availability of appropriately sized individuals, because the bill of the Semipalmated Sandpiper is too small to consume any but the smallest gastropods. Other novel prey were less common; for instance, crab DNA appeared in only 5 fecal samples, 3 with invasive green crab (Carcinus maenas) and 2 with native Cancer spp. It is likely that birds consumed crab eggs or pieces of crab in detritus. Bivalves, another novel prey item, were consumed sporadically, mostly at AV. As with snails, the availability of very small bivalves probably determined consumption; however, Semipalmated Sandpipers also ingest bivalves as detritus, or in potentially nonlethal consumption of bivalve siphons (Skilleter and Peterson 1994, Maire et al. 2010). Finally, our study was the first to detect fish (Atlantic herring, flounder, and unknown fish) and cnidarians–ctenophores in Semipalmated Sandpiper diets, albeit at low frequencies at all sites. It is likely that fish eggs or the smallest juveniles, along with detrital remains of both groups of animals, are consumed by Semipalmated Sandpipers.
Molecular scatology can often detect the prey items of prey items, a phenomenon referred to as “secondary consumption” (Deagle et al. 2005, 2007, Barrett et al. 2007, Pompanon et al. 2011). It is therefore possible that some prey items may have been detected indirectly; however, we suggest that secondary consumption did not greatly influence our results. Prey at low trophic levels (diatoms and the invertebrates that consume them, such as C. volutator) would be more likely to be detected through secondary consumption because they are food for higher consumers such as annelids. However, these low-trophic-level taxa are already well established in the diet of Semipalmated Sandpipers (Baker 1977, Hicklin and Smith, 1979, 1984, Quinn and Hamilton 2012). Further, the FOO of lower-level consumers was much higher than that of the most likely sources of secondary consumption. Finally, some of the more unusual prey items (e.g., insects, bivalves, snails) are unlikely to have been consumed, except perhaps as detritus, by animals that Semipalmated Sandpipers commonly eat.
Molecular scatology is also limited by the presence–absence data it produces, and by the reference databank available for comparison. Bay of Fundy intertidal infauna were underrepresented in the NCBI database, which forced grouping into higher taxonomic levels to ensure confidence in identification. This introduced complications for taxa like nematodes, which may have been either free-living species consumed by birds or parasites of prey items or of the birds themselves. A more complete reference library of DNA sequences in this region, especially for infaunal prey items, is required to clarify the dietary relationships reported here.
Finally, we collected feces shortly after birds completed roosting near shore. This maximized our ability to detect prey of all origins, but the overall frequency of these novel prey in the Semipalmated Sandpiper diet throughout the feeding cycle is unknown. The timing of collection may have resulted in the low FOO of annelids (polychaetes), which have been found to contribute substantially to the diet of Semipalmated Sandpipers in other studies (MacDonald et al. 2012, Quinn and Hamilton 2012). It also may have inflated the importance of some novel prey items that are most likely to be detected following periods when birds were near shore. Collecting fecal samples from throughout the tidal cycle would help clarify the relative importance of both newly observed prey items and previously recognized prey.
Notwithstanding the limitations discussed above, our results reveal a generalist diet and links between Semipalmated Sandpipers and nontraditional environments. While encouraging with respect to this species' ability to adapt to a changing prey base, use of these novel prey items also raises new ecological questions and conservation considerations. Disturbances or contamination of prey from freshwater (Kraus et al. 2014), terrestrial (Hallmann et al. 2014), or marine systems (Simmons et al. 2014) could affect Semipalmated Sandpipers. Therefore, the magnitude of the relationships observed here requires additional investigation. Conservation of stopover points is crucial for migratory shorebirds (Davis and Smith 2001, Skagen 2006, Smith et al. 2012), and current conservation plans for Semipalmated Sandpipers in the Bay of Fundy revolve around protecting beach and intertidal habitat. Our results, however, suggest that this strategy may not protect all resources used by Semipalmated Sandpipers. Further, the results of our study suggest that it may be useful to consider use of nontraditional prey and habitats by other shorebirds with similar ecologies.
We thank A. M. Gerwing, L. K. Boone, T. T. Bringloe, A. L. Einfeldt, M. A. Hebert, S. M. MacNeil, J. T. Quinn, A. M. Savoie, and The Nature Conservancy of Canada for assistance.
Funding statement: This study was funded by Natural Sciences and Engineering Research Council of Canada grants to J.A.A., M.A.B., and D.J.H.; by New Brunswick Wildlife Trust Fund and MITACS (Mathematics of Information Technology and Complex Systems) grants to M.A.B.; and by Canadian Foundation for Innovation and New Brunswick Innovation Fund grants to J.A.A.
Author contributions: J.A.A. analyzed the data. J.A.A., M.A.B., and D.J.H. provided funding and resources.