Seasonal fluctuations in food availability can affect diets of consumers, which in turn may influence the physiological state of individuals and shape intra- and inter-specific patterns of resource use. High-elevation ecosystems often exhibit a pronounced seasonal “pulse” in productivity, although few studies document how resource use and energetic condition by avian consumers change in relation to food-resource availability in these ecosystems. We tested the hypothesis that seasonal increases (pulses) in food resources in high-elevation sagebrush ecosystems result in 2 changes after the pulse, relative to the before-pulse period: (1) reduced diet breadth of, and overlap between, 2 sympatric sparrow species; and (2) enhanced energetic condition in both species. We tracked breeding-season diets using stable isotopes and energetic condition using plasma metabolites of Brewer's Sparrows (Spizella breweri), Vesper Sparrows (Pooecetes gramineus), and their food resources during 2011, and of only Brewer's Sparrows and their food resources during 2013. We quantify diet breadth and overlap between both species, along with coincident physiological consequences of temporal changes in resource use. After invertebrate biomass increased following periods of rainfall in 2011, dietary breadth decreased by 35% in Brewer's Sparrows and by 48% in Vesper Sparrows, while dietary overlap decreased by 88%. Energetic condition of both species increased when dietary overlap was lower and diet breadth decreased, after the rapid rise of food-resource availability. However, energetic condition of Brewer's Sparrows remained constant in 2013, a year with low precipitation and lack of a strong pulse in food resources, even though the species' dietary breadth again decreased that year. Our results indicate that diet breadth and overlap in these sparrow species inhabiting sagebrush ecosystems generally varied as predicted in relation to intra- and interannual changes in food resources, and this difference in diet was associated with improved energetic condition of sparrows at least in one year.
Species coexistence within a given patch may be facilitated by a number of ecological processes, including resource partitioning (Hutchinson 1959). Whereas combined effects with other ecological drivers, such as predator–prey dynamics (Martin 1988) or limiting resources (Holt 1977, Holt 1984), are difficult to rule out, evidence of differences in sympatric species' niches are observable. Competition theory predicts that sympatric species avoid interspecific competition during periods of low resource availability by reducing the diversity of prey resources in their diet (hereafter, diet breadth) and by decreasing overlap in habitat use, time of activity, or diet with sympatric species (MacArthur 1958, Hutchinson 1959). Conversely, during times of high resource availability, niche breadth and overlap of sympatric species can increase, because resources are no longer limiting to consumers. Among other things, optimal foraging theory (OFT) predicts that a species' diet breadth should increase during times of low resource availability, because a consumer might otherwise not procure enough prey resources to meet its energy demands (MacArthur and Pianka 1966). In contrast, when resource availability is high, OFT predicts a narrowing of species' diet breadth to achieve foraging requirements. This narrowing allows sympatric species to coexist in environments with fluctuating resources by modulating their dietary niches to maximize energy gains and minimize energy loss (Pyke 1984, Perry and Pianka 1997). The dietary niche includes interactions among species within a community for shared food resources that vary in availability (Rotenberry 1980b). One way to quantify to what degree these factors influence sympatric species involves collecting data on their dietary breadth and overlap across a seasonal gradient that encompasses a food-resource pulse (Beaulieu and Sockman 2012, Correa and Winemiller 2014). In seasonal environments, food availability can fluctuate dramatically, producing a “resource pulse” in availability wherein food availability increases during a low-frequency, short-duration, large-magnitude event (figure 1 from Yang et al. 2008). Classic examples of environments with resource-pulsed systems where species are also monitored include desert environments with monsoonal rains (Warne et al. 2010), oak forests with mast production (McShea 2000), and old-growth forests with irruptions of insects (Haney 1999).
When viewed over longer time scales (multiple years to decades) in sagebrush (Artemisia spp.) dominated ecosystems, resource pulses predictably occur from spring to summer. However, over shorter time scales, the resource pulse in any given year may be larger or smaller, or occur earlier or later. Thus, interannual variation exists in the timing, duration, and magnitude of the spring-to-summer resource pulse during a bird's breeding season (Rotenberry and Wiens 1989, Rotenberry and Wiens 1991, Bates et al. 2006, Lowe et al. 2010). Total insect abundance in sagebrush ecosystems of the Intermountain West often peaks during June (Wenninger and Inouye 2008, Lowe et al. 2010). When combined with early-spring variability in temperature and precipitation, such abundance can strongly influence an individual's ability to gain pre-breeding body condition (Arizmendi-Mejía et al. 2013), maintain pair bonds (Chalfoun and Martin 2007), and achieve reproductive success (Rotenberry and Wiens 1989) and inclusive fitness (Luck 2003). Due to a paucity of empirical data for sagebrush-ecosystem birds, it remains largely unknown to what degree this seasonal change in resource availability affects diet breadth and overlap of sympatric species (Rotenberry 1980a).
Recent advances in stable isotope ecology and Bayesian modeling of these isotope data provide a means by which to determine resource use of animals (Parnell et al. 2010, Jackson et al. 2011). When combined with plasma metabolites, these advances can reveal how resource use of individuals affects their instantaneous physiological state (MacDade et al. 2011). For animals living in dynamic environments with different carbon and nitrogen stable isotope background in their food resources, stable isotope analyses can identify to what degree diets of sympatric species differ under various food conditions (Pearson et al. 2003, Beaulieu and Sockman 2012, Correa and Winemiller 2014, Ma et al. 2014). Concomitant measures of plasma metabolites can predict direction of body-mass change in birds captured only once (Jenni-Eiermann and Jenni 1994, Jenni-Eiermann and Jenni 1997, Smith and McWilliams 2010, Stutchbury et al. 2011) and can thus potentially indicate the physiological consequences of changes in resource use over time.
Sagebrush ecosystems of the Intermountain West constitute one of the 22 most highly endangered major ecosystems of the United States (Noss and Peters 1995) due to climate change, land use, and invasive species (Bradley 2010). Within these systems, although both sparrow species inhabit sagebrush-dominated communities, Vesper Sparrows (Pooecetes gramineus) typically nest on or near the ground, whereas Brewer's Sparrows (Spizella breweri) more commonly nest in sagebrush canopy (Best 1972). Upon their arrival at the breeding grounds, both species have high energy demands and commonly consume invertebrate prey and grass seeds (Best 1972, Rotenberry et al. 1999, Jones and Cornely 2002).
Our objectives in this study were to use stable isotopes and plasma metabolites from 2 species of sagebrush-inhabiting sparrows to investigate each species' diet breadth, the dietary overlap between the 2 species, and their physiological state as resource availability substantially changed over the spring–summer seasons. We hypothesized that seasonal increases in food resources would result in the following changes, relative to the before-pulse period: (1) reduced within-species diet breadth in Vesper and Brewer's sparrows, in a given year; (2) decreased dietary-niche overlap between the 2 sparrow species; and (3) higher energetic condition in both species. To address these predictions, we quantified plant phenology, invertebrate biomass consumed by sparrow species, and both stable isotopes and metabolites in sparrow blood plasma during May–July on the western edge of the Greater Yellowstone Ecosystem. We addressed partitioning of dietary niches between Vesper and Brewer's sparrows in 2011, and compared this to a within–Brewer's Sparrow change in dietary niche breadth investigated in 2013.
Brewer's Sparrows and Vesper Sparrows inhabit sagebrush ecosystems across the western United States (Knick and Rotenberry 2002). North American populations of Brewer's Sparrows declined by an estimated 68% during 1966–1996, whereas Vesper Sparrows declined by an estimated 28% during 1966–1998 (Rotenberry et al. 1999, Sauer et al. 2000). Best (1972) found that Brewer's and Vesper sparrow diets (as determined by gut content analysis) consisted mainly of beetles (Coleoptera) during June. Diets during July consisted mainly of beetles, grasshoppers (Orthoptera), Hemiptera, and grass seeds for Brewer's Sparrows, and beetles, grasshoppers, and grass seeds for Vesper Sparrows.
Our study sites were in stands of basin big sagebrush (Artemisia tridentata spp. tridentata) across the 3,200-ha Centennial Sandhills (hereafter, Sandhills) in the Centennial Valley of southwestern Montana (44.68°N, −111.74°W). The average elevation of the Sandhills is 2,014 m. We divided the Sandhills, which are oriented east to west, into 3 sampling regions: western, central, and eastern (see Appendix Figure 5). We randomly selected 2 locations within each region to establish 500 m × 500 m study plots in 2011. We added one additional 500 m × 500 m randomly selected plot to each region in 2013 (ntotal = 3 plots per region).
In high-elevation temperate ecosystems, vegetation grows rapidly following a lengthening of photoperiod during spring. This growth corresponds to complex biogeochemical changes, including increases in air and soil temperatures and moisture availability (the latter due to melting snowpack). For the purpose of this study, we used seasonal change in flower abundance to define these complex interactions. We used forb flower abundance as a proxy for grass seed development during sparrow breeding season, similarly as applied by Wood and Pidgeon (2015) for tree flower counts to reflect seasonal abundance change in food availability to migratory passerines, because both functional groups increase their abundance following the melting of snow and increases in air temperature during spring (Appendix). In addition to grass-seed counts being less accurate and less repeatable, our experience in this ecosystem is that grass seeds from the previous year vary dramatically across elevations and microsites in their detectability. We recorded abundance of flowering plants within our study plots on 5 separate survey occasions in each study-area region, once every 8 days during May 13, 2011, through June 23, 2011 (Figure 1A). We randomly selected one sampling point in a randomly selected plot distributed across each of the 3 regions and placed a permanent belt transect (25 m × 2 m). During each flower survey, we counted total number of open, non-senescent flowers from individual forb plants located along belt transects.
We used grids of pitfall traps to quantify relative biomass of sparrow invertebrate prey through time, in both years of study. Though no single collection device will equally capture all invertebrate species available to foraging birds, pitfall traps collect both ground-surface– and foliage-dwelling invertebrates commonly found in the shrub canopy of sagebrush communities (Lowe et al. 2010). Lowe at al. (2010) demonstrated that non-baited pitfall traps can successfully quantify relative abundance of invertebrates consumed by sparrows (Orthoptera, Hemiptera, Coleoptera, Hymenoptera, Lepidoptera; Best 1972) over short time periods (3–5 days) in May and June. Our pitfall technique may have underestimated certain foliage-dwelling insects consumed by sagebrush sparrows, but the specificity of our trapping method is somewhat less important, because our basic objective was to test whether there were more prey items available after the food-resource pulse than before it. We placed pitfall traps at the first 6 locations of bird capture starting on May 13, 2011, and used these same (permanently marked) 6 pitfall grids in 2013, beginning on May 18, 2013. Individual pitfall grids were spaced >200 m apart. Each pitfall grid consisted of 5 plastic cups (11 cm in diameter, 14 cm deep) placed at the axis and endpoints of a † shape with 25 m spacing between cups. Saline solution was added to each cup (∼3 cm deep), followed by a drop of glycerin to reduce surface tension. Invertebrate samples were collected from cups every 4–6 days, May 18 through June 29, 2011, and May 22 through July 3, 2013. Invertebrate samples were emptied from cups and preserved in 95% ethanol solution. We defined catch per unit effort (mg trap-day−1) by adjusting invertebrate biomass by the number of trap nights an individual trap (cup) was deployed. We sorted invertebrate samples to taxonomic order and size class (we used only individuals of ≤10 mm total body length, based on size of prey typically consumed by these 2 sparrow species; Best 1972, Rotenberry et al. 1999, Jones and Cornely 2002). We dried samples of each invertebrate order at 60°C for 48 hours, and weighed them to the nearest 10 mg.
Sparrow Plasma Isotope and Triglycerides Sampling
Male Brewer's and Vesper sparrows were captured within the randomly selected plots distributed across the 3 regions in 2011 and 2013. We visited each plot once per day until all plots were surveyed for birds to capture. After all plots were assessed during a sample period, sampling recommenced back on the first plot; this resulted in 3–4 sample periods per year. Male Brewer's (n = 55 in 2011, n = 88 in 2013) and Vesper (n = 32 in 2011) sparrows were netted by luring males into one mist net (polyester mesh, each 6 m × 3 m, 30-mm cell size) using playback broadcasts of conspecific songs and calls (Sogge et al. 2001). Only male sparrows that exhibited territorial defense aggression were captured because in territorial species, males generally arrive earlier than (non-territorial) females in order to select a high-quality territory that maximizes the pair's fitness (Morbey and Ydenberg 2001). All captured sparrows were marked with a U.S. Geological Survey aluminum leg band.
A 27-gauge needle was used to puncture the brachial vein, allowing free flow of blood directly into heparinized micro-capillary tubes, to collect ∼200 μL whole blood from each captured bird for both stable isotope and triglyceride analyses. Due to the constriction of the brachial vein upon capture, we could not always collect our target goal of ∼200 μL of blood from each bird captured; this precluded us from analyzing both triglyceride concentrations and isotope values from every bird we captured. Each blood sample was centrifuged for 10 min at 12,000 rpm to segregate plasma from red blood cells. After tubes were centrifuged, plasma was separated from red blood cells, flash-frozen in liquid nitrogen dry shipper (Taylor Wharton CX-100, Worthington Industries, Columbus, Ohio, USA) and stored at −80°C until analysis.
Plant and Invertebrate Isotope Sampling
A sample of invertebrates from different taxonomic orders and seeds was collected to isotopically characterize diet items available for consumption to Vesper and Brewer's sparrows. However, dietary estimates are sensitive to the temporal changes in isotopic composition of their prey (e.g., Beaulieu and Sockman 2012). We therefore minimized the effect of potential temporal changes in the isotopic composition of sparrow prey (i.e. plants and insects) by timing collections to coincide with the spring and summer seasons of the study. We collected 87 individual invertebrates on June 5, 2011, and 104 invertebrates on May 23 and July 11, 2013, to characterize different prey resources available to sparrows. During the same dates, we collected grass seeds from 21 individual plants in 2011, and from 29 individual plants in 2013. Because isotope values of invertebrates submerged in saline solution for hours become degraded (Krab et al. 2012), we did not use invertebrates captured in pitfall traps to characterize the isotopic values of sparrow prey. Instead, we collected a broad array of invertebrates from potential foraging substrates by visually sampling sagebrush using forceps to collect surface-, canopy-, and foliage-dwelling invertebrates. Upon returning from the field, we immediately froze these prey samples.
Stable Isotope and Triglyceride Laboratory Analyses
All field-collected samples (i.e. vegetation, invertebrates, bird plasma) were freeze-dried and powdered with a mortar and pestle before isotopic analyses. Specimens were weighed to the nearest 1.0 ± 0.2 mg, and compressed into tin capsules. Samples were analyzed for stable-nitrogen (δ15N) and carbon (δ13C) isotopes using continuous-flow isotope-ratio mass spectrometry (following Pearson et al. 2003) at the University of New Mexico Stable Isotope Laboratory. Stable isotope values were reported in parts per thousand (‰), relative to atmospheric (AIR) nitrogen for δ15N and relative to Vienna PeeDee Belemnite for δ13C. Estimated analytical error was ±0.3‰ for δ15N and ±0.1‰ for δ13C, based on n = 145 replicate within-run measurements of organic standards.
Bird plasma triglycerides were assayed on a microplate spectrophotometer (Bio-Tek Powerwave X340, Bio-Tek Instruments, Winooski, Vermont, USA) using commercial kits modified for small volumes following Smith and McWilliams (2010). Triglycerides were measured sequentially by endpoint assay (Sigma, St. Louis, Missouri, USA: 5μL plasma, 240 μL reagent A, 60 μL reagent B) by first measuring free glycerol and then subtracting free glycerol concentration from measured total triglyceride concentration (Smith and McWilliams 2010). Samples were diluted 3-fold with 0.9% NaCl before they were assayed.
Linear mixed-effects (LME) models were used to investigate temporal changes in flower abundance and relative biomass of invertebrate prey in 2011 and 2013. Sampling location was included as a random effect to account for repeated sampling of the same site through time, with date added as a fixed-effect parameter within the model. Data on flower abundance in 2011, and invertebrate biomass in both years, were log-transformed to correct heteroscedasticity. Confidence intervals on predicted flower abundance and invertebrate biomass were estimated using the delta method (Seber 1982). Start of the seasonal transition in resource availability for our study area was determined by searching for the date that most strongly discriminated the 2 different rates of change in flower abundance during the season in 2011 (hereafter, termed the “food-resource pulse”). We used a change-point analysis to determine this date using the R package changepoint (Killick and Eckley 2014). The change-point analysis was performed on the predicted values for flower abundance based on the LME.
For each sparrow species, we quantified both within-species dietary breadth and between-species dietary-niche overlap using a standard ellipse area corrected for small sample size (SEAc) (Jackson et al. 2011) in Stable Isotope Analysis in R (SIAR; Parnell et al. 2010). Stable isotope ellipses help elucidate foraging behavior of individual species within a community. Standard ellipse areas typically contain 40% of the prey-item data for an individual species. Ellipses are also unbiased with respect to sample size, and their estimation using Bayesian inference allows for comparison within or between sparrow species containing different sample sizes. We compared overlap (expressed in ‰2) in SEAc within and between species from the periods before and after the food-resource pulse.
We used a one-way analysis of variance (ANOVA) to compare mean δ15N values among invertebrate orders and grass seeds. Multiple comparisons among invertebrate orders and grass seeds were made using Holm pairwise comparisons, to maintain experiment-wise error rates at our selected alpha level. Clades of sparrow prey items that had statistically indistinguishable δ15N values were lumped and treated as one group in all subsequent analyses, because stable isotope mixing models cannot estimate diet contributions of prey groups based on similar isotope values. We classified diet items only to the level of taxonomic order, as we were interested in comparing major patterns in sparrow dietary breadth and overlap before versus after a food pulse, rather than finely differentiating selected prey items from their relative availability. Using these ANOVA results, we then assessed temporal changes in invertebrate biomass within each of these taxonomic groups. The contribution of isotopically distinct dietary endpoints, as identified by the above ANOVA analysis, to the diets of sparrows was then investigated further using a concentration-dependent, six- (number of source contributions from 2011) or three-endpoint (number of source contributions from 2013), two-isotope Bayesian mixing model (SIAR) without informative priors to estimate relative contribution of foods from distinct invertebrate taxonomic orders to diets of Vesper and Brewer's sparrows before and after the food-resource pulse (Parnell et al. 2010). Isotopic endpoints in the mixing model included spiders (Araneae), ants (Hymenoptera), beetles, caterpillars (Lepidoptera), combined grasshoppers and leafhoppers (Hemiptera), and grass seeds in 2011. In contrast, in 2013 the model included (1) a combined ants, spiders, and beetles category, (2) a combined caterpillars, grasshoppers, and leafhoppers category, and (3) grass seeds. This model accounts for variation in isotopic values, carbon and nitrogen concentrations of prey sources, and a predictable shift in isotopic values of a sparrow's diet when assimilated to consumer tissues. A concentration-dependent model accounts for different ratios of carbon to nitrogen in grass seeds compared to insect prey, and is suitable for quantification of dietary contributions to omnivorous animals (Phillips and Koch 2002). Discrimination values were derived for bird plasma from controlled experiments (Pearson et al. 2003, Ogden et al. 2004) that fed birds a plant- or invertebrate-based diet, which included (mean ± 1 SD) 0.43 ± 0.21‰ for δ13C and 3.0 ± 0.26‰ for δ15N for invertebrates, and −0.9 ± 0.92‰ for δ13C and 2.6 ± 0.07‰ for δ15N for grass seeds.
Triglyceride concentrations of blood plasma were compared for Brewer's and Vesper sparrows before and after the food-resource pulse using a generalized linear model (with normally distributed errors) with a Gaussian link. All statistical analyses were conducted using R 3.2.2 (R Core Development Team 2015). Means are presented with ± 1 SE, unless stated otherwise. We determined significance with an alpha level of P ≤ 0.05 for statistical tests.
Abundance of Flowers and Invertebrates
Log-transformed total abundance of flowers was positively predicted by date (LME: flower abundance = e−29.813+0.219*ordinal date; t11 = 9.750, P < 0.001). Flower abundance at the beginning of the study (May 13) was 0, and it increased to a maximum count (1,415 ± 560) on June 23 (Figure 1A). The change-point model estimated the date at which flower abundance changed during the breeding season as being June 2 (ordinal date 153).
The average daily rate of increase of log-transformed total invertebrate biomass from pitfall traps was positively predicted by date in 2011 (LME: b2011 Total Invertebrate = 0.08 ± 0.02 mg trap-day−1, t27 = 3.292, P = 0.003), but not in 2013 (LME: b2013 Total Invertebrate = −0.01 ± 0.009 mg trap-day−1, t40 = −1.090, P = 0.282; Figures 1B and 1C). Spiders increased with date in 2011 (b = 0.17 ± 0.02 mg trap-day−1, t26 = 7.377, P < 0.001), as did ants (b = 0.08 ± 0.03 mg trap-day−1, t26 = 3.163, P < 0.001) and beetles (b = 0.12 ± 0.04 mg trap-day−1, t26 = 3.25, P < 0.001; Figures 1B and 1C). In contrast, caterpillars (b = 0.06 ± 0.03 mg trap-day−1, t26 = 1.8311, P = 0.08) and combined grasshoppers and leafhoppers (b = 0.03 ± 0.03 mg trap-day−1, t26 = 0.926, P = 0.36) in 2011 increased only slightly throughout the study. No change in log-transformed invertebrate biomass from any invertebrate group was strongly predicted by date in 2013 (all invertebrate groups: t40 = −0.361 to 0.531, all P > 0.05). Insect biomass at the beginning of the season in 2013 was 10.1 mg trap-day−1, whereas in 2011 insect biomass did not reach this amount until toward the end of the season.
Diet Overlap between Sympatric Bird Species
In 2011, the dietary niche of Brewer's and Vesper sparrows overlapped to a larger extent before the resource pulse (1.67‰ overlap of ellipses) than after (0.20‰ overlap; Figures 2A and 2B). Additionally, after the food-resource pulse, Brewer's Sparrow diet breadth narrowed by 35.4% (from 3.65‰ to 2.36‰) while Vesper Sparrow diet breadth narrowed by 48.2% (2.55‰ to 1.23‰).
In 2013, dietary niche breadth of Brewer's Sparrows was wider before compared to after the food-resource pulse. The Brewer's Sparrow diet breadth narrowed by 34.8% (1.12‰ to 0.73‰) from before to after the resource pulse (Figure 2C).
Dietary Contributions to Sympatric Species
The δ15N values of sparrow prey varied across 6 distinct prey groups in 2011 (F = 131.9, df = 6 and 104, P < 0.001) and 3 distinct prey groups in 2013 (F = 31.0, df = 5 and 97, P < 0.001; Table 1). All Holm pairwise comparisons tests of δ15N values among sparrow prey categories showed significant differences between the 2 sparrow species (P < 0.05), except for grasshoppers and leafhoppers (P = 0.99; Table 1). In 2011, the 6 distinct prey groups included (1) spiders, (2) ants, (3) beetles, (4) caterpillars, (5) combined grasshoppers and leafhoppers, and (6) grass seeds. In 2013, Holm pairwise comparisons identified only 3 distinct groups based on δ15N values, which included (1) combined beetles, ants, and spiders, (2) combined grasshoppers, leafhoppers, and caterpillars, and (3) grass seeds.
Carbon and nitrogen isotopic values (mean ± 1 SE) of sparrow prey (n = sample size) from 6 prey groups in 2011 and 3 prey groups in 2013. Groupings based on results from ANOVA and Holm pairwise comparisons. None of these samples are adjusted for isotopic discrimination. All samples were collected in the Centennial Sandhills, Centennial Valley, Montana, USA.
In 2011, results from isotopic mixing models suggested high dietary-niche overlap between sympatric sparrows before the food-resource pulse (Figure 3). During this time, spiders, ants, grasshoppers, leafhoppers, and caterpillars dominated the diets of both sparrow species. These prey items collectively contributed >80% of the total prey to diets of both sparrow species.
After the food-resource pulse in 2011, sparrow species exhibited both greater trophic segregation and narrowing of dietary niche breadth, as evidenced in the isotopic ellipses (Figure 2). Following the food-resource pulse, Brewer's Sparrows shifted their diet toward grass seeds and beetles (Figure 3), which comprised a majority (55%) of their diet. In contrast, Vesper Sparrows' diet consisted mainly (≥70%) of spiders, ants, and caterpillars, both before and after the food-resource-pulse (Figure 3).
Before the food-resource pulse in 2013, Brewer's Sparrows consumed mainly spiders, ants, and beetles (48%) and grass seeds (35%; Figure 3B). After the food-resource pulse, grass seeds constituted a slightly larger portion (44%) of Brewer's Sparrow diets, whereas the contribution of the 2 other prey-item groups remained relatively unchanged.
Relative to concentrations before the food-resource pulse, triglyceride concentrations were 1.41× larger after the pulse for Brewer's Sparrows (Figure 4; GLM: t46 = 2.58, P = 0.013) and 1.78× larger for Vesper Sparrows (t35 = 3.21, P = 0.003; Figure 4). In contrast, Brewer's Sparrow triglyceride concentrations did not change across the season in 2013 (t73 = −0.22, P = 0.67). Brewer's Sparrow triglyceride concentrations were similar between years before the food-resource pulse (t44 = −0.71, P = 0.48), whereas triglyceride concentrations were higher after the food-resource pulse in 2011 compared with 2013 (t93 = −4.78, P = <0.001).
The tracking of food resources by birds across seasons influences their energetic condition and eventually their survivorship and fecundity. In our system, we observed a dramatic seasonal pulse in resource availability, and this produced a sharp post-pulse decline in both the dietary breadth and overlap of 2 sympatric sparrow species in one year (2011), along with more gradual declines in diet breadth of Brewer's Sparrows in the second year (2013). When diet of sparrows became more specialized, amidst greater availability of food items, their energetic condition improved in 2011. This was not observed in 2013, a year without a marked change in availability of food resources, when energetic condition of Brewer's Sparrows remained stable throughout the (2013) study period. Stable isotope mixing models revealed that the seasonal declines in diet breadth and overlap between sparrow species were associated with changes in the dominant prey consumed.
Abundance and Phenology of Flowers and Invertebrates
The timing of life-cycle events of invertebrates in montane sagebrush ecosystems has been shown to be strongly influenced by temperature and precipitation (Hodkinson 2005, Bates et al. 2006, Lowe et al. 2010). At our study site, we found that abundance of invertebrates increased exponentially throughout the breeding season in 2011, but declined during 2013 after a brief, early-season peak. This difference between years in intra-annual trend in invertebrate abundance is consistent with between-year differences in accumulated winter and spring precipitation. Namely, accumulated precipitation was 2.11 times higher in 2011 than in 2013 (Appendix Figure 6). Wenninger and Inouye (2008) experimentally demonstrated direct effects of water availability on insect abundance and distribution in a sagebrush ecosystem in Idaho that is slightly drier than our system. Additionally, the date of snowmelt in high elevations strongly influences the start of the growing season, as well as the timing of flower (Inouye 2008) and insect emergence (Hodkinson 2005). The disappearance of snowpack at our study sites occurred approximately 3 weeks later in 2011 than in 2013 (K.A.C. personal observation), which could partially explain the delayed increase in food-resource abundance in 2011. Effects of earlier snowmelt on food-resource abundance is particularly insightful, in light of widespread advancement of timing in spring snowmelt in the Intermountain West during the 20th century (Mote 2003, Mote et al. 2005), predictions for even earlier snowmelt in the future (Leung et al. 2005), and predicted changes in precipitation (Hamlet et al. 2005). Although it is difficult to distinguish their individual contributions on number of distinct prey classes between the 2 years (6 in 2011 vs. 3 in 2013), these interannual differences in physical conditions (e.g., much-higher precipitation in 2011 than 2013 in a water-limited ecosystem, disappearance of snowpack in 2013 earlier than when most of our sampling occurred) may each have played some role in determining numbers of prey classes.
Feeding Niches of Sparrows: How Diet Breadth and Overlap Change with Resource Availability
OFT and competition theory have differing predictions about diet breadth and degree of niche overlap of sympatric species in prey-limited and prey-rich periods. We therefore were able to distinguish which of these 2 theories potentially influence the coexistence of sagebrush birds through niche partitioning at our study sites. In competition theory, one challenge is that for 2 species to be truly syntopic (i.e. strongly competing), a resource (e.g., prey) must be strongly limiting for interspecific competition to be reduced by greater niche segregation. One prediction of OFT is that a consumer's diet breadth should decrease during times of high resource availability to maximize returns per unit of energy expenditure, resulting in increased resource specialization (MacArthur and Pianka 1966, Perry and Pianka 1997, Correa and Winemiller 2014). We predicted that an increase in food resources during summer in high-elevation sagebrush communities would result in reduced diet breadth and dietary niche overlap for 2 sympatric sparrow species, Vesper and Brewer's sparrows. Consistent with this prediction, diet breadth and overlap of both sparrow species in our study decreased markedly across the food-resource pulse in 2011. Specifically, Brewer's Sparrow diets shifted to more seed- and beetle-based resources after flower availability increased dramatically, whereas Vesper Sparrow diets shifted to contain higher proportions of spiders, ants, and caterpillars. Brewer's Sparrows exhibited the same narrowing of their diet breadth across the season in 2013. Given the taxonomically coarse filter to which our isotopic analyses were constrained, we were still able to detect general reductions in dietary breadth in both sparrow species across a food-resource pulse. Krebs et al. (1977) similarly showed support for optimal foraging in Great Tits (Parus major), which also reduced dietary breadth during periods of high resource availability in experimental feeding trials. However, the observed dietary shifts in our sparrow species are not only influenced by prey quantity, but are likely also affected by prey quality, foraging and processing efficiency, and bird abundance (Weimerskirch et al. 2003, Owen et al. 2005, Cerasale and Guglielmo 2010).
In sagebrush communities of the northern Great Basin, Rotenberry (1980a) and Rotenberry and Wiens (1998) showed that erratic interannual precipitation influenced annual variation in net primary productivity (NPP). Accordingly, Brewer's Sparrows tracked the changes in vegetation structure and composition that accompany changes in NPP, and the associated changes in food resources, and correspondingly adjusted their foraging-patch selection from year to year (Rotenberry and Wiens 1998). We found substantial year-to-year differences in the magnitude of this seasonal change in diet breadth. Specifically, grass seeds contributed more to the early-summer diet of Brewer's Sparrow during the dry year (2013), whereas spiders, ants, grasshoppers, leafhoppers, and caterpillars contributed more to their early-summer diets during the wet year (2011). Such interannual variability in winter and spring accumulated precipitation clearly affects the extent and timing of the resource pulse in high-elevation sagebrush ecosystems (Rotenberry and Wiens 1998, Beaulieu and Sockman 2012), and this in turn influences the foraging behavior of avian consumers such as Brewer's and Vesper sparrows.
Energetic Consequences of Seasonal Changes in Feeding Niche
We predicted that energetic condition of both sparrow species would increase with corresponding decreases in dietary breadth and overlap. Changes in food resources available to sparrows during the breeding season corresponded with increased plasma triglyceride concentrations in 2011, but not in 2013. This inconsistent response in energetic condition is corroborated by studies on seabirds that showed that changes in plasma triglycerides were associated with a corresponding diet shift during the breeding season in one year, but not in the next year (e.g., Franci et al. 2015). Additionally, breeding Southwestern Willow Flycatchers (Empidonax traillii extimus) that reduced their diet breadth and specialized on less-abundant but higher-quality prey items improved their energetic condition (Owen et al. 2005). Based on our admittedly short-term dataset, it appears that this dietary niche pattern carries little energetic penalty and thus may have little ecological relevance other than to indicate that when resources are abundant, the different sparrow species reduce their dietary breadths to specialize on consuming different prey items, potentially reflecting different optima. For example, comparison of portions left of the change-point arrow in Figures 1B and 1C showed that total prey abundance before the food pulse was markedly higher in 2013 than in 2011, and correspondingly, dietary-niche breadth was much smaller in 2013 than in 2011. Clearly, other factors such as bird abundance, prey quality, and foraging efficiency can influence changes in foraging patterns that lead to changes in energetic condition in some years but not others (Owen et al. 2005, Cerasale and Guglielmo 2010, MacDade et al. 2011, Franci et al. 2015). Further work on sagebrush bird communities is needed to assess whether other niche dimensions besides prey choice show similar intra-annual and interannual changes, and their consequences for the ecology, conservation, and management of this bird community.
We thank S. Ismar, J. Rotenberry, and an anonymous reviewer, as well as S. B. Correa and E. R. Pianka for critical reviews of earlier drafts of the manuscript. G. Pasini, E. Trusz, J. Saba, and S. Essig provided field assistance. B. West and staff of Red Rock Lakes National Wildlife Refuge provided logistical support. W. F. Cross provided lab access. R. Levine and J. A. Dullum assisted with database design. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. The authors declare no conflict of interest.
Funding statement: This research was funded by the Wildlife Health Lab of the U.S. Fish and Wildlife Service, and the Montana Association of Conservation Districts.
Ethics statement: This research was conducted in compliance with federal (06266), state (2011-53), and animal care and use (017-12MAWEST-042412) permits.
Author contributions: K.C., M.A., and E.B. conceived the idea, design, and experiment. K.C., M.A., S.S., N.K., E.K. and S.M. performed the experiments. K.C. with inputs from co-authors and others analyzed the data. K.C., M.A., E.B., and S.M. with inputs from co-authors wrote the paper.
Air Temperature and Precipitation
Methods. To assess the transition from spring to summer, we examined hourly air temperatures and precipitation data from a RAWS weather station in the Centennial Sandhills (http://raws.wrh.noaa.gov/cgi-bin/roman/meso_base.cgi?stn=RRDM8&time=GMT). The weather station is located 2 km west of the study area, at an elevation of 2,027 m. We quantified between-winter differences in accumulated precipitation from October 1 to May 13 during both winters of study (i.e. 2010 to 2011 and 2012 to 2013).
Results. During both years, air temperature increased considerably following the onset of flower emergence (Figures 1A and 1B). Overwinter accumulated precipitation from October 1 to May 13 was 2 times greater during the winter of 2010 to 2011 than during the winter of 2012 to 2013.
In 2011, total accumulated precipitation at the start of the study was stable in periods before the pulse (May 13–21) and after the pulse (June 19–29), with a total of <1 cm of precipitation falling during these periods. However, 45% of the study duration's total accumulated precipitation fell during the middle portion (May 20 to June 4, 2011) of the study timeframe. In contrast, precipitation during the study was 75% lower in 2013 than in 2011. Most of this fell during the before-pulse period (3.8 cm of accumulated precipitation from May 16 to 29), and only 0.4 cm additional precipitation accumulated after the pulse (Appendix Figure 6).