Stable isotope analyses identify trophic niche partitioning between sympatric terrestrial vertebrates in coastal saltmarshes with differing oiling histories

Study species and life histories

Seaside sparrows are the most common small passerines (∼15–25 g) found in Louisiana coastal marshes (Grenier & Greenberg, 2006). Sparrows rely on seeds, insects, and other marsh invertebrates (Post & Greenlaw, 2006). Both sparrows and rats are year-round residents with small home ranges (Cooney, Schauber & Hellgren, 2015), especially during the breeding season, making them highly susceptible to a wide range of environmental perturbations to coastal marsh ecosystems (e.g., flooding, climate change, sea rise, pollution; (Bergeon Burns et al., 2014; Kern & Shriver, 2014). Of the 1,185 sparrows captured from our study sites between 2011–2017, 79 sparrows were captured in more than one sampling year. While a majority of recaptures occurred one or two years apart (n = 43, 24 respectively), six birds were captured across a time span of three years while six were captured across a time span of four years. Notably, these sparrows were adults when initially captured, so these birds were likely older than three or four years of age. The oldest wild sparrow recorded was at least nine years old (Sykes, 1980); while this lifespan is likely uncommon, it may be possible that a small number of birds sampled in this study could have been alive during the DWH spill. No nestlings were recaptured as adults during our study.

Marsh rice rats are mid-sized (∼40–80 g), semi-aquatic cricetid rodents common in Louisiana coastal marshes (Wolfe, 1982). Rats are omnivorous and known to consume a range of prey with plants, crabs, insects and molluscs making significant contributions to their diet (Sharp, 1967; Kruchek, 2004). Rat populations are characterized by turnover rates, with rats rarely living longer than one year, although one individual was documented to live up to two years (Wolfe, 1985). At our study sites, only three rice rats were recaptured over two consecutive field seasons (A. Perez-Umphrey, unpublished data), one in 2014 then in 2015, two in 2016 then in 2017 (although, note, 2014 samples were not included in this present study). Because only subadult and adult rats were bled, and all of these individuals were juveniles in their first year of capture; only a single sample exists for each of these three individuals from when they were recaptured. No individuals sampled would have been present in 2010.

Ethics statement

All animals (birds, rats, and invertebrates) and plants were sampled on private or public land after obtaining prior permission from owners or managers, and the necessary permits. The capture and handling of all animals was approved by the United States Fish and Wildlife Service (collecting permits MB095918-0 and MB095918-1; USFWS Federal Bird Banding permit 22648; and, Louisiana Department of Wildlife and Fisheries scientific collecting permits LNHP-15-039, LNHP-16-056, LNHP-17-064, LNHP-15-033, LNHP-16-048, LNHP-17-039). All sampling protocols adhered to statutes of the Institutional Animal Care and Use Committee of the LSU AgCenter (IACUC permits: A2013-09, A2015-04, A2016-06).

Study area

Samples were collected from seven saltmarsh sites in the Barataria Bay region of coastal Louisiana, USA between April and June of 2015, 2016, and 2017 (Fig. 1). The saltmarsh in this area is either dominated by smooth cordgrass, Spartina alterniflora or less commonly co-dominated with Juncus roemerianus. This region of Barataria Bay was selected as the focus of this study because it received some of the heaviest oiling following the 2010 DWH oil spill (Michel et al., 2013; Turner et al., 2014a). We selected seven marsh sites based on their oiling histories as determined by a combination of systematic methods [Shoreline Clean-up and Assessment Technique (SCAT); (Michel et al., 2013)] and chemical analyses (Turner et al., 2014a) following the 2010 spill. These seven sites laid the basis for our experimental design in 2015, 2016, and 2017. We categorized sites as either ‘oiled’ (n = 4) or ‘unoiled’ (n = 3) based on the SCAT maps’ documented categories of oiling (categories: heavy, moderate, light, very light, trace oiling and no oiling; Turner et al., 2014a) from the marsh edge. Our sites were in Bay Batiste (O2, O3, O4- heavily oiled), Bay Sanbois (U1,U2- unoiled), and Bay Jimmy (O1- heavily oiled, U3- unoiled). Sampling plots within sites were 2.5 ha (500 m along the marsh edge and 50 m inland) and sites were a minimum of 1 km apart (Fig. 1). Site classification was confirmed through sediment analyses (Turner et al., 2014a; Turner et al., 2014b)—oiled marsh sites had initial polycyclic aromatic hydrocarbon (PAH) mean ±  standard deviation concentrations of 19,524 ±  2,158 µg kg⁻¹ (Turner et al., 2014a) and unoiled sites had initial (PAH) mean ±  standard deviation concentrations of 164 ±  21 µg kg⁻¹ (Ashton-Meyer, 2017). Although our study started five years after initial oiling when oiling effects may have attenuated, we explicitly explored these comparisons to test for chronic and/or indirect effects because oil can persist in marsh habitats over long periods of time (Turner et al., 2019).

Sample collections

Rats were caught in baited Sherman live traps set overnight. Sparrows were caught in mist nets (34 mm mesh, 12 m length). All rats and sparrows captured were weighed, sexed, and aged. Aging of sparrows and rats ensured that only adults and sub-adults were sampled (i.e., juveniles were excluded from all subsequent analyses). Blood samples were drawn from sparrows and rats via venipuncture of the brachial vein and retro-orbital sampling, respectively. All blood samples were transferred to glass vials and stored frozen until further analysis.

Suspended particulate organic matter (POM) was collected by passing ∼250mL of water collected off the marsh edge through a 105-micron screen to remove large zooplankton and inorganic particulates and onto combusted glass fibre filters (Whatman GF/F). We confirmed that we retained most of the phytoplankton community with this approach by analysing chlorophyll-a concentration on screened and unscreened samples. Chlorophyll-a concentrations on screened samples were >85% of total chlorophyll-a on unscreened samples (and over 90% in most cases; B. Roberts, unpublished data). Gulf ribbed mussels (Geukensia granossissima) were collected by hand from the marsh surface. Samples of the dominant marsh vegetation (i.e., Spartina alterniflora) were collected along with herbivorous terrestrial Hemiptera (Prokelisia spp and Ischnodemus spp) by hand and using sweep nets. These primary producers and primary consumer samples provided isotopic proxies of the dominant aquatic and terrestrial marsh energy pathways, respectively. Muscle tissue was dissected from ribbed mussels, whole bodies were retained for insects, and all samples were frozen in the field prior to analysis.

Stable isotope analysis

Blood, muscle, insect, and plant samples were freeze-dried for 24 to 48 h, homogenised into a powder, and weighed into tin capsules for stable isotope analysis. Freeze-dried POM samples were removed from filters and placed into tin capsules for stable isotope analysis. Samples were flash combusted using a Costech ECS4010 Elemental Analyzer and analysed for δ¹³C and δ¹⁵N values via a coupled to a Thermo Scientific XP or Delta V Isotope Ratio Mass Spectrometer. Stable isotope ratios are expressed in δ notation in per mil units (‰), according to the following equation:

δX = [(R_sample/R_standard) - 1] x 1000

Where X is ¹³C or ¹⁵N and R is the corresponding ratio ¹³C /¹²C or ¹⁵N /¹⁴N. The R_standard values are referenced to the Vienna PeeDee Belemnite (VPDB) for δ¹³C and atmospheric N₂ for δ¹⁵N.

Raw δ values were normalized on a two-point scale using glutamic acid reference materials with low and high values [i.e., USGS40 (δ¹³C = −26.4‰, δ¹⁵N = −4.5‰) and USGS41 (δ¹³C = 37.6‰, δ¹⁵N = 47.6‰)]. The analytical precision, based on standard deviations of repeated reference materials were 0.1‰and 0.2‰for δ¹³C and δ¹⁵N, respectively.

Data analyses

We performed a three-way analysis of similarities (ANOSIM), which permits unbalanced replication between treatments (Clarke & Gorley, 2006), to describe differences in the isotope values of basal end members and primary consumers. Taxa, year, and treatment (site) were classified as factors in all analyses and resemblance matrices were based on Euclidean distances (9999 permutations). ANOSIM, a non-metric multivariate statistical method, has no underlying assumptions about the statistical distribution of the data (e.g., normality, variance equality) and creates an overall test statistic (R) that indicates if differences among taxa, sites and years exist (Clarke & Gorley, 2006). ANOSIM results in an R-value and a p-value. The R-value is scaled from −1 to +1. R-values = 0 indicate random grouping, R ≥ 0.3 shows that groups are slightly different but overlapping, and R ≥ 0.5 indicates well separated groups (Clarke & Gorley, 2006).

We calculated trophic position (TP) of sparrows and rats using Bayesian models using mode for TP estimates in the tRophicPosition package (Quezada-Romegialli et al., 2018) in the statistical software R (version 3.5.0; R Core Team, 2019). This approach couples Markov Chain Monte Carlo simulations with stable isotope data to estimate TP. Because TPs calculated from Bayesian models display higher uncertainty with smaller sample size (Quezada-Romegialli et al., 2018), and low samples sizes were present in some species, site, and year combinations, we pooled data across all oiled and unoiled sites, respectively. In all TP calculations, we set the model parameters as follows: baseline = ‘twobaselinesfull’, iterations = 100000, number of chains (n. chains) = 5 and ‘burn-in’ (number of initial iterations discarded) = 20000. We used the stable isotope values (δ¹³C; δ¹⁵N) of ribbed mussels and herbivorous insects as representative of aquatic and terrestrial food web baselines, respectively. To validate our assumption that primary consumers were representative of aquatic and terrestrial resources, we visually compared them to the stable isotope values of POM and marsh vegetation (i.e., Spartina alterniflora; Fig. 2). We then used tRophicPosition to quantify the relative importance of terrestrial and aquatic food resources (α) to rats and sparrows, with larger α values implying greater use of terrestrial relative to aquatic food resources. We calculated TP and α separately for each species, site, and year combination and all models incorporated an assumed mean trophic fractionation (Δ¹³C = 0.39 ± 1.30; Δ¹⁵N = 3.4 ± 0.98) per trophic transfer (Post, 2002). TP and α metrics are presented as modal values. Associated 95% credibility intervals (CI) and Bayesian posterior probabilities (PP > 0.95) from pairwise comparisons of posterior distributions of TP and alpha (α), which were used to indicate statistically significant differences across species, site, and year combinations. Pairwise comparisons among years between and within species were expressed as Bayesian posterior probabilities (PP > 0.95) to test for significant differences among groups. We pooled data across all oiled and unoiled sites for sparrows and rats. We used this approach for three reasons: first, preliminary analyses revealed that there were similar trends between pooled and site-specific analyses (Table S1). Second, our analyses are sensitive to sample sizes. As such, it was necessary to combine samples to avoid bias (Jackson et al., 2011; Quezada-Romegialli et al., 2018). Third, pooling samples into ‘oiled’ and unoiled’ categories would allow us to make direct comparisons of overall differences between previously oiled and unoiled sites.

We quantified the trophic niche width of sparrows and rats using the Stable Isotope Bayesian Ellipses in R (SIBER) package version 2.1.4 (Jackson et al., 2011). We visualized trophic niches using the standard ellipse area corrected for sample size (SEA_c) represented by the bivariate standard deviation of δ¹³C and δ¹⁵N values, encompassing approximately 40% of the data (Jackson et al., 2011). Moreover, we compared the Bayesian approximation of standard ellipse area (SEA_b) between groups as a quantitative measure of isotopic trophic niche width. Because SEA_b display higher uncertainty with smaller sample size (Jackson et al., 2011), and low samples sizes were present in some species, site, and year combinations, we pooled data across oiled and unoiled sites prior to analyses similar to our analysis of TP. Values for SEA_b are reported as modal values with 95% CI and Bayesian posterior probabilities (PP>0.95) were used to test for significant differences among groups.

To quantify overlap in resource use (i.e., trophic niche overlap) among species and between sites, we quantified the proportional overlap between Bayesian-estimated standard ellipses (SEA_b) of each group using nicheROVER package 1.0 (Swanson et al., 2015). As in analysis of SEA_b area, we pooled data into oiled and unoiled sites prior to analyses. Trophic niche overlap is defined as the probability that an individual from one species is found within the total trophic niche of the other species, with values ranging from 0% (no overlap) to 100% (complete overlap). We considered trophic niche overlap to be asymmetrical when 95% CI did not overlap among reciprocal comparisons of CI, such that individuals in one group are significantly more likely to be encompassed in the ellipses of another group than vice versa.

Summary of basal endmembers and primary consumers

Biplot of δ¹³C and δ¹⁵N and ANOSIM revealed clear separations between POM and Spartina alterniflora (Fig. 2; Global R = 1.0, p = 0.001) with no significant differences in stable isotope values by treatment [oiled sites versus unoiled sites; ANOSIM, Global R = 0.1, p = 0.23] or year (ANOSIM = 0.1, p = 0.13). δ¹⁵N values of POM (range = 4.3 to 8.6‰) were higher than Spartina alterniflora (range = 1.7 to 5.8‰). Conversely, δ¹³C values in POM (range =-24.4 to −22.2‰) were lower than Spartina alterniflora (range =-13.6 to −11.8‰).

Biplot of δ¹³C and δ¹⁵N and ANOSIM also revealed clear separations between hemipterans and ribbed mussel (Fig. 2; ANOSIM, Global R = 0.9, p = 0.001). Ribbed mussels (aquatic baseline) were similar across all sites (ANOSIM, Global R = 0.3, p = 0.02) and year (ANOSIM, Global R = 0.2, p = 0.4). Similarly, hemipterans (terrestrial baseline) did not vary by site (ANOSIM, Global R = 0.0, p = 0.2) or year (ANOSIM, Global R = 0.0, p = 0.5). Stable isotope values of ribbed mussels (aquatic baseline; δ¹³C range = −23.7 to −19.2‰; δ¹⁵N range = 5.2 to 7.8‰) and hemipterans (terrestrial baseline; δ¹³C range = −14.0 to −11.2; δ¹⁵N range = 3.3 to 6.5) plotted closely to POM and Spartina alterniflora, respectively.

Trophic position [Hypothesis 1]

Trophic position (TP) for sparrows and rats ranged between 2.5 to 3.1 (modal values; Table 1). TP was higher in sparrows than rats at both oiled and unoiled sites in 2016 and 2017, and at oiled sites in 2015 (Table 1; Fig. 3). Trophic positions of sparrows did not differ between oiled and unoiled sites in any year examined (Table 1; Table S2; Fig. 2). Similarly, TP of rats did not differ between oiled and unoiled sites in any year examined (Table 1; Table S2; Fig. 3).

Resource use [Hypothesis 2]

Sparrows consistently relied on terrestrial resources to a greater degree than aquatic resources (i.e., mode α > 50%; Fig. 3; Table 1) in all three years sampled. Rats relied more on terrestrial resources than aquatic resources in 2015. Conversely, in 2016 and 2017 rats assimilated more aquatic prey than terrestrial prey. (Fig. 3; Table 1). Considering resource use between rats and sparrows, the use of terrestrial prey resources relative to aquatic prey resources (α) did not differ between sparrows and rats in 2015; however, in 2016 and 2017 rats incorporated more aquatic prey resources (i.e., lower α) relative to sparrows irrespective of the oiling history of sites (Fig. 3; Table 1; Table S3).

Within species (pairwise comparison of posterior probabilities of α; Table 1), significant differences in α were observed in some comparisons between oiled and unoiled sites in both sparrows and rats (Table 1; Table S3). Specifically, modal α values of sparrows were higher at oiled sites than unoiled sites in 2015 and 2017 but were similar in 2016 (Table 1). Modal α values of rats at oiled sites were greater than those at unoiled sites in 2015 but were similar in 2016 and 2017 (Table 1).

Trophic niche width [Hypothesis 3]

Sparrows had larger isotopic niche widths (SEA_b) than rats at unoiled sites in all three years, and at oiled sites only in 2015 (Table 1; Fig. 3). Sparrows had smaller trophic niche widths (SEA_b) at oiled sites relative to unoiled sites in 2016 and 2017 (Table 1; Fig. 3). In contrast, rats at oiled sites had consistently larger trophic niche widths (SEA_b) relative to rats at unoiled sites in all three years examined (Table 1; Fig. 3; Table S4).

Trophic niche overlap [Hypothesis 4]

Trophic niche overlap between sparrows and rats was higher in 2015 than 2016 and 2017 at both oiled and unoiled sites (Fig. 4), with 2016 and 2017 showing minimal overlap between rats and sparrows at oiled and unoiled sites. When comparing niche overlap within species (rats vs rats, sparrows vs sparrow), trophic niche overlap within species was higher in 2016 and 2017 relative to 2015 for both sparrows and rats (Fig. 4).

The trophic niches of sparrows and rats exhibited a generally asymmetric pattern of niche overlap at both oiled and unoiled sites (Table 2). Specifically, rats exhibited larger trophic niche overlap relative to sparrows than vice versa, in all three years examined at unoiled sites (Table 2). However, at oiled sites the direction and consistency of asymmetric trophic niche overlap differed across years. For example, at oiled sites in 2015, rats exhibited larger niche overlap (95% CI [71–93]%) relative to sparrows, than vice versa (95% CI [49–70]%; Table 2). In contrast, at oiled sites in 2017 sparrows exhibited larger niche overlap (95% CI [39–80]%) relative to rats, than vice versa (95% CI [34–66]%; Table 2). Trophic niche overlap between species was symmetrical at oiled sites in 2015 (Table 2).

Conclusions and additional considerations

Our study identified a general pattern of interspecific trophic niche partitioning between rats and sparrows co-occurring within saltmarshes in coastal Louisiana. While we observed differences in trophic position and resources use, trophic niches were not fully segregated between rats and sparrows, with asymmetrical overlap between rats and sparrows. When comparing between sites with differing oiling histories we found both interspecific and intraspecific differences in the trophic niche width and the consumption of terrestrial vs. aquatic prey resources, although identifying the proximate causes of these differences is beyond the scope of this study. In addition, because trophic dynamics and environmental processes may be specific to individual sites or regions, the extension of our findings to other saltmarshes should be approached with some caution (Nelson, Deegan & Garritt, 2015). Even so, our findings provide reference information on respective trophic niche of sparrows and rats that will aid in the evaluation of these species as bioindicators of oiling and other environmental stressors.