Sterols and stanols as novel tracers of waterbird population dynamics in freshwater ponds

Abstract

With the expansion of urban centres in the mid-twentieth century and the post-1970 decrease in pesticides, populations of double-crested cormorants (Phalacrocorax auritus) and ring-billed gulls (Larus delawarensis) around Lake Ontario (Canada and USA) have rapidly rebounded, possibly to unprecedented numbers. Along with the use of traditional palaeolimnological methods (e.g. stable isotopes, biological proxies), we now have the capacity to develop specific markers for directly tracking the presence of waterbirds on nesting islands. Here, we apply the use of lipophilic sterols and stanols from both plant and animal-faecal origins as a reliable technique, independent of traditional isotopic methods, for pinpointing waterbird arrival and population growth over decadal timescales. Sterol and stanol concentrations measured in the guano samples of waterbird species were highly variable within a species and between the three species of waterbirds examined. However, cholesterol was the dominant sterol in guano, and phytosterols were also high in ring-billed gull guano. This variability highlights a specialist piscivorous diet for cormorants compared to a generalist, omnivorous diet for gulls, which may now often include grain and invertebrates from agricultural fields. A ratio that includes cholesterol and sitosterol plus their aerobically reduced products (cholestanol, stigmastanol) best explained the present range of bird abundance across the islands and was significantly correlated to sedimentary δ¹⁵N. Overall, we demonstrate the use of sterols and stanols as a direct means for tracking the spatial and temporal presence of waterbirds on islands across Lake Ontario, and probably elsewhere.

1. Introduction

Lake sediments provide an important and reliable archive of long-term environmental changes at both local and regional landscapes [1]. Traditionally, tracking bird inputs to lake sediments has been accomplished using indirect proxies such as stable nitrogen isotopes (δ¹⁵N), biogenic metal concentrations, and biological subfossil morphological or biogeochemical remains (e.g. diatoms, algal pigments) [2–4]. However, in temperate regions, close to urban centres and agricultural runoff, these same proxies may be influenced by anthropogenic pollution and thus less reliable in directly tracking the impacts of birds. For example, an increase in δ¹⁵N may indicate sewage pollution [5,6] rather than inputs from birds of a high trophic position. Although ¹⁵N is enriched in seabird and waterbird guano relative to other sedimentary nitrogen (N) sources [7–9] and is usually a good indicator of ornithogenic N inputs to lake sediments [2,10], it can also be influenced by post-depositional nitrification and denitrification [11,12] and so is not always able to distinguish between sources of animal waste. Therefore, it is important to develop organic markers that are source-specific and independent of traditional palaeolimnological proxies. As specific sterols and stanols are both produced and excreted by birds, there is an opportunity to expand from their traditional use in identifying human and livestock faecal contamination of waterbodies [13,14] to tracking avian inputs to lakes and ponds over time.

Sterols are natural, unsaturated steroid alcohols that represent an important group of compounds for the functioning and structure of biological membranes [15]. Stanols are reduction products of sterols by hydrogenation, usually by microbes either in animal gastrointestinal tracts or in the environment [16,17]. As they are both hydrophobic compounds with a high affinity for sorption onto organic matter and suspended particles, they typically become rapidly incorporated into sediments of aquatic ecosystems [18]. Once in the sediments, the degradation of faecal sterols is minimal [19,20]. In incubation experiments, cholesterol and cholestanol (the 5α-isomer of coprostanol) showed little decay over 450 days in lake sediments with anoxia occurring in the system after 45 days [21]. Coprostanol is also persistent over time in anoxic sediment [22,23].

Given their resistance to degradation in sediments, sterols and stanols may therefore be used as markers in palaeoecological studies, and several of the more common sterols and stanols are already known to have associations with specific sources. For example, sitosterol, stigmasterol, and campesterol are important constituents of higher plants. Sitosterol is a common sterol in plant tissue and wax [24], and it is widely reported in phytogenic material [25,26], including algae [27]. Several sterols and stanols have been identified in the faeces of different animals, and their composition and concentration are dependent on the balance of dietary intake, the metabolic production of sterols, and the biota resident in the digestive tract [28]. Cholesterol is one of the most commonly occurring sterols in organisms, especially animals, and thus is also common in faecal matter [28,29]. Coprostanol is widely used to trace municipal sewage input in aquatic environments [6,30]. However, it is also reported in bird guano at a lower concentration [28,31,32].

It can often be challenging for conservation biologists to know the historical population sizes of threatened, migratory birds, or how past population declines correspond to human influence versus climate change and other stressors. Protection is frequently directed towards extant colonies, but if the birds require a matrix of islands for long-term persistence this may change how we think about and designate protected areas. Waterbirds, especially double-crested cormorants (Phalacrocorax auritus, hereafter ‘cormorants’) and ring-billed gulls (Larus delawarensis) have established and abandoned nesting colonies on islands across Lake Ontario at different time periods over the past approximately 100 years (D. Tyerman 2016, personal communication) [33]. For example, ring-billed gull colonies near urban centres are known to have rapidly expanded post-World War II, as landfills became an important gull food source [34,35] and are now the most numerous waterbird species in the Laurentian Great Lakes [36]. Similarly, after a period of near-extinction caused by pesticides and other contaminants between the 1950s and the 1970s, cormorants have rebounded and now number in the hundreds of thousands in the Great Lakes region [37]. Ponds located on islands across eastern Lake Ontario provide a natural spatial and temporal experimental design for testing the use of sterols and stanols in tracking waterbird population dynamics. Detailed observational records and nest counts across all of our selected study islands provide an important opportunity to examine and compare the efficacy of sterols, diagnostic ratios, and the more traditional bird palaeo-marker in sediments, δ¹⁵N [2,38,39], in tracking birds nesting around lakes and ponds. Here, we demonstrate how palaeolimnology can support conservation practices by shedding light on the length of time birds have been using nesting habitats, changes in population sizes related to human activities, and the time required for waterbird species to recover following twentieth-century contaminant use.

2. Site descriptions

Eight small islands in eastern Lake Ontario (figure 1) were selected for study based on a gradient of waterbird impacts from ponds located on islands that have received no significant waterbird impact (i.e. reference sites) to highly impacted ponds (electronic supplementary material, table S1). The study ponds are shallow with mean depths (Z_m) ranging between approximately 0.10 m and approximately 1.0 m. Based on the known nesting histories and field estimates of waterbird numbers currently nesting on each island (given in the electronic supplementary information), we identified that five of the study sites have some degree of impact from either double-crested cormorants or ring-billed gulls, as well as lesser abundances of herring gulls (Larus argentatus) and Caspian terns (Hydroprogne caspia), and three study sites have little-to-no waterbird impact.

Figure 1.

Location of the eight ponds (solid circles) sampled on different islands in eastern Lake Ontario. A bold box within the map inset highlights the study region in Lake Ontario relative to the North American Great Lakes.

(a). Five impact sites

High Bluff Pond (Z_m = 0.15 m) and Gull Island Pond (Z_m = 0.30 m) are located in Presqu'ile Provincial Park near the northern shore of east-central Lake Ontario (figure 1). There are currently approximately 4000 cormorant nests on High Bluff Island and a colony of approximately 30 000 ring-billed gulls on Gull Island (D. Tyerman 2016, personal communication). Little Galloo Pond (Z_m = 0.40 m) and Island hosts multiple nesting species (electronic supplementary material, table S1). Ring-billed gulls have nested here since around 1940 [40], and from the mid-1970s onwards they have consistently numbered 40 000–70 000 nests (I. Mazzocchi 2016–2017, personal communication) [41]. Cormorants began nesting on Little Galloo in 1971 [42], had peak numbers (approx. 8400 nests) in 1996, and have been managed by the New York State Department of Environmental Conservation (NYSDEC) since the mid-1990s [33]. East Brother Pond (Z_m = 0.50 m) is surrounded by cormorants, with approximately 1300 nests between 2003 and 2012 [43]. Pigeon Island Pond (Z_m = 0.10 m) has a history of mixed occupation between cormorants and gulls, but currently hosts a colony of approximately 2100 cormorants [42]. The high-impact ponds dominated by cormorant presence (e.g. East Brother, Pigeon, and High Bluff islands) have virtually no remaining terrestrial vegetation in their catchment, except dead trees and a few shrubs and vines that can tolerate the high levels of ammonia from the excessive guano deposition.

(b). Three reference sites

Calf Island Pond (Z_m = 1.0 m) is approximately 2 km southeast of Little Galloo Island in Jefferson County, NY (figure 1). Calf Island currently hosts no colonies of waterbirds, as it is actively managed by the NYSDEC. However, records show 170 nests in 2008 and 24 nests in 2009, but cormorants have since been absent from the island (I. Mazzocchi 2016–2017, personal communication). Calf Island is vegetated with live trees and is dominated in area by Calf Pond. False Duck Pond (Z_m = 0.45 m) has the evidence of some modest cormorant activity in its catchment with approximately 100 cormorant nests first noted in 1986. Main Duck Pond (Z_m = 1.0 m) has no evidence of nesting cormorants in its catchment and is part of a marsh system.

3. Material and methods

Sediment cores were retrieved using a Glew & Smol [44] shallow-water push corer and sectioned into 0.5 cm intervals using a Glew [45] extruder. For all eight sediment cores, freeze-dried sediments were dated using ²¹⁰Pb and ¹³⁷Cs radiometric dating techniques, with radioisotopic activity measured using an Ortec high-purity germanium gamma spectrometer (Oak Ridge, TN, USA). Sediment chronologies were calculated using the Constant Rate of Supply (CRS) model [46] with ScienTissiMe software (Barry's Bay, Ontario, Canada). The radioisotope activities and dating profiles for East Brother, Pigeon, Main Duck and False Duck have been previously published by Stewart et al. [43]. For the sediment records from the ponds Gull, Calf, High Bluff and Little Gallo, approximately 10 intervals evenly spaced through each core were analysed for total Pb concentrations (0.05 µg g⁻¹ limit of detection) using inductively coupled plasma mass spectrometry at SGS Canada (Lakefield, Ontario), certified by the Canadian Association for Laboratory Accreditation Inc. (CALA). In the Great Lakes region, there were steady increases in atmospheric Pb emissions owing to industrialization since the approximate 1830s, as well as the onset of leaded gasoline in the 1920s [47]. Atmospheric Pb deposition subsequently decreased with the discontinuation of leaded gasoline in the 1970s [48]. Here, we confirmed the accuracy of our total Pb profiles by matching trends with those of a 1991 core from eastern Lake Ontario [49]. Sediment chronologies were used when verified total Pb trends had reasonable agreement with CRS-derived dates (electronic supplementary material, figures S1 and S2) and the timing of historical deposition.

To examine whether modern-day sterol composition within the sediments differed from past sterol composition, we used a palaeolimnological approach that compares two discrete time intervals across a spatial gradient (the ‘before and after’ approach) [1]. These types of regional assessments can efficiently address broad-scale questions about environmental change. The bottom samples provide an integrated sample of the sterols and stanols present in the pre-impact or pre-bird environment. By contrast, the top surface sediment samples provide an integrated sample of the sterols and stanols present in modern-day aquatic habitats (post-impact). Additionally, for one reference site (Calf) and two impact sites (High Bluff, Little Galloo), we completed sterol and stanol analysis at a higher temporal resolution.

For δ¹⁵N analysis, freeze-dried sediments (unacidified) were weighed into tin capsules with two parts tungsten trioxide, with analytical details provided in the electronic supplementary material. Organic matter content of the sediment samples was determined using the Loss-On-Ignition method [50].

In addition to analysing sterols and stanols in sediment cores, a total of three guano samples were collected from cormorants, ring-billed gulls and herring gulls that defaecated while being handled by ornithologists, and analysed independently (i.e. nine unique guano samples total) for sterol and stanol composition. The nine sterols and stanols that were examined in all samples include: coprostanol (5β-cholestan-3β-ol), epicoprostanol (5β-cholestan-3α-ol), coprostanone (5β-cholestan-3-one), cholesterol (cholest-5-en-3β-ol), 5α-cholestanol (5α-cholestan-3β-ol), cholestanone (5α-cholestan-3-one), stigmastanol (5α-stigmastano-3β-ol), and sitosterol (β-sitosterol). Analytical methods for analysing sterols and stanols in sediments were modified from Birk et al. [51] and Cheng et al. [52] and detailed in the electronic supplementary material, along with detection limits and quality assurance/quality control measures.

The use of different ratios for determining a range of waterbird influence on the different ponds was explored both spatially and temporally. One of these ratios was an index for determining seabird impact on High Arctic ponds at Cape Vera (Nunavut) previously published by Cheng et al. [52] (seabird index = cholesterol/(cholesterol + sitosterol)). Given the large quantity of stigmastanol and the absence of sitosterol in many of our sample ponds, we included this plant stanol within the ratio. As cholesterol and sitosterol can be microbially reduced in aerobic, natural environments to 5α-cholestanol and stigmastanol, respectively (through 5α-stanone intermediates) [16,28], we also included the aerobic degradation product of cholesterol–5α-cholestanol, in this ratio. Therefore, we propose an expanded bird sterol index, where the

3.1

The per cent abundances of all nine sterols and stanols from the top and bottom samples of eight sediment cores were depicted with a stratigraphical approach using the C2 software. These samples are arranged in order of increasing bird sterol index (equation (3.1)), so that the sediment samples with the greatest bird impact plot at the base of the stratigraphy. Pearson correlation analyses, which included all the top and bottom samples plus the bird guano samples, were used to examine the relationship between the sterols and bird ratios to δ¹⁵N. All surface (or ‘top’) samples plus the mean sterol composition for guano from each of the three waterbird species were summarized through principal component analysis (PCA) in Canoco v. 5.0 [53]. Samples were converted to relative abundances of the total concentration of sterols quantified in each sample, and square-root transformed prior to analyses to equalize the variance among sterols. This unconstrained ordination procedure was used to examine how the lake and guano samples varied with respect to sterol and stanol composition. Pre-disturbance (or ‘bottom’) samples and sediment samples within cores (i.e. not 0 cm) were run passively in the ordination, and all modern and guano samples were run actively. Plotting the modern and pre-disturbance assemblages in the same ordination space enabled a comparison of the relative changes in sterol and stanol composition through time, highlighting the similarities and differences among sites in terms of both sterol composition and magnitude of change (i.e. trajectory through time).

4. Results and discussion

Sterol and stanol concentrations measured in the guano samples of waterbird species were highly variable both within a species and among the three species examined. Typically, cholesterol was the main sterol found in the guano of all three taxa with concentrations of 205 ± 144 µg g⁻¹ (n = 3) in cormorants, 173 ± 188 µg g⁻¹ (n = 3) in ring-billed gulls and 483 ± 501 µg g⁻¹ (n = 3) in herring gulls, comprising 72%, 36% and 86%, respectively, of the total sterols. In the same order, sitosterol concentrations were 48 ± 23 µg g⁻¹, 289 ± 347 µg g⁻¹ and 50 ± 63 µg g⁻¹. Of the total plant sterols (combining sitosterol and stigmastanol), ring-billed gulls had the greatest proportion and concentration within their faeces. Similar to previous work [29,52,54], these results suggest that cholesterol is the dominant sterol in bird faeces; however, the diet of a single bird living in a dynamic landscape and close to a city centre may vary widely. For example, herring gulls are generalist predators on fishes, insects, small mammals and birds, and the eggs and young of co-occurring species, as well as opportunistic scavengers of dead animals or garbage from landfills [36,55]. Thus, the resultant large deviation for the sterol concentrations found in the gull samples may result from a wide foraging range and diet. Ring-billed gulls especially are known scavengers and dump feeders, whereas cormorants are strictly piscivorous, consuming primarily alewife (Alosa pseudoharengus) historically and now more commonly round goby (Neogobius melanostomus) in Lake Ontario [56]. Ring-billed gull numbers have rebounded rapidly in part owing to an opportunistic, omnivorous diet and their behavioural adaptability to nest on urban lands close to anthropogenic food sources such as landfill dumps [36], which are reflected in the variable sterol concentrations in guano and large phytosterol component. Additionally, isotopic tracers suggest a dietary shift in herring gulls from aquatic to terrestrial food sources since the early 1980s [57], with similar diet shifts known for ring-billed gulls [36]. Gulls inhabiting temperate regions have highly varied diets, including individual specialization, and the recent exploitation of foods from anthropogenic sources could lead to markedly different δ¹⁵N values and faecal sterol and stanol composition between individuals. For example, ring-billed gulls from Gull Island are known to exploit grains from nearby agricultural fields, especially during harvests [58]. This adaptability to terrestrial diet sources when aquatic prey is less abundant would yield highly variable concentrations of faecal sterols, making gulls more challenging birds to track through time with these markers.

The high proportions of plant sterols in the guano of the study species were corroborated with the δ¹⁵N signatures of guano from the same birds. Elevated plant consumption or omnivory in the ring-billed gull diet, initially suggested from the high proportion of phytosterols in their guano, was also indicated by the relatively low δ¹⁵N value of their guano. Ring-billed gull guano had the lowest mean δ¹⁵N of 3.58‰ (n = 3), second to cormorant guano with a mean δ¹⁵N of 6.28‰ (n = 3), and herring gull guano had the highest mean δ¹⁵N of 8.60‰ (n = 3). The low δ¹⁵N of ring-billed gull guano in this study is consistent with previous work that showed when these birds forage in predominantly anthropogenic habitats they exhibit significantly lower ¹⁵N in blood cells, characteristic of the anthropogenic food items composing their diet (french fries, bread, composite meat, and various corn-based products depleted in ¹⁵N by approx. 3.1–5.2‰ relative to arthropods) [7]. However, it is important to note that we examined the sterols and δ¹⁵N in a total of three guano samples from each gull species, collected at a similar time of year. Given the large range in gull diet [36] and potential to exploit agricultural fields and grains, the faecal sterols and δ¹⁵N probably fluctuate substantially by region and time-of-year sampled. At the time of sampling, ring-billed gulls were largely consuming grains from agricultural fields (electronic supplementary material, figure S3). Typically, for this region, ring-billed gull and herring gull diets should not differ substantially; however, our data demonstrate the specificity that faecal sterols can provide about an animal's diet.

In addition, ponds surrounded by only gulls or a mixture of gulls and cormorants had the greatest δ¹⁵N values measured in surficial sediments, followed by ponds surrounded solely by cormorants, followed by the reference sites (figure 2). This elevated δ¹⁵N corresponded to the numbers of birds on each island with gulls representing the largest colonies followed by mixed-species colonies and smaller cormorant colonies. However, the δ¹⁵N of the sediments of all ponds receiving bird inputs was pointedly higher than the δ¹⁵N measured in bird guano (figure 2). Although the δ¹⁵N of bird guano may fluctuate depending on seasonal diet, we suggest that in the presence of excess N and anoxia, sediment denitrification processes have most likely increased the δ¹⁵N signatures of the sediment. Denitrification, the anaerobic bacterial reduction of NO₃ to N₂ and N₂O, has been identified as driving additional fractionation in NO₃, increasing δ¹⁵N of NO₃. Denitrification can have a high enrichment factor (+5‰ to 40‰) [11,59] by anaerobic bacteria living in sustained low oxygen conditions, and generally denitrification shows increasing rates with increasing temperature [12]. Important factors that influence denitrification in aquatic systems, such as temperature, the supply of nitrate and organic matter, and oxygen concentration [12], would be at play in our temperate, nutrient-rich study ponds. Dissolved oxygen and temperature data loggers that collected hourly measurements over several weeks in East Brother Pond have captured nightly periods of anoxia and corresponding peak temperatures up to approximately 20°C [60]. Additionally, in combination with in-pond anaerobic denitrification, the volatilization of ammonia from guano deposited on land would enrich the δ¹⁵N signature of guano that can wash into ponds as runoff. The probability of ammonia volatilization of bird guano is emphasized as surface soils dry and air temperatures rise beginning in the spring and continuing through into the summer [61], and is certainly a process occurring on our study islands with high nesting gull and cormorant numbers.

Figure 2.

Sterol and stanol per cent relative abundances for the top and bottom sediment samples from eight ponds arranged in order from lowest expanded bird sterol index at the top (equation (3.1)) to highest at bottom. The bird index published in Cheng et al. [52] is also shown for comparison, as well as δ¹⁵N and % organic matter. GI, Gull Island; MD2, Main Duck; EB, East Brother; LG, Little Galloo; HB, High Bluff; PGN, Pigeon; FDI, False Duck Island.

Cholesterol (7.1 ± 7.5 µg g⁻¹, n = 8) and sitosterol (2.0 ± 1.2 µg g⁻¹, n = 8) were present in all Lake Ontario pond surface sediments, suggesting widespread inputs from animal- and plant-derived material to these ponds (figure 2). Indeed, ponds and guano with higher δ¹⁵N had significantly higher cholesterol (Pearson correlation, R = 0.70, p < 0.001, d.f. = 17, n = 19), but did not have higher sitosterol (R = −0.14, p = 0.57, d.f. = 17, n = 19). Hence, ornithogenic input appears to be the dominant source of cholesterol in many of these ponds. The ratio that incorporates both sitosterol and cholesterol, as well as their naturally reduced 5α-stanols, cholestanol and stigmastanol, demonstrated the strongest correlation to δ¹⁵N (R = 0.70, p < 0.001, d.f. = 17, n = 19) compared to the previous index published in Cheng et al. [52] (R = 0.61, p < 0.01, d.f. = 17, n = 19). In many of the non-impacted or reference samples, stigmastanol was the dominant phytosterol and, in fact, sitosterol was absent in the bottom sediments of Little Galloo. Without the incorporation of stigmastanol into the ratio, the Cheng et al. [52] index would infer that the Little Galloo bottom sediments had the greatest bird influence. In aerobic, natural environments, sitosterol can be microbially degraded to stigmastanol, through 5α-stanone intermediates [16,29]. Thus, stigmastanol is probably present at greater concentrations in the bottom sediments of cores prior to both anthropogenic eutrophication and the presence of large bird colonies that deliver excess nutrients to the ponds and can cause anoxia in the sediments. This may explain the high relative abundances of both stigmastanol and cholestanol in the bottom sediments of Main Duck, Calf, and East Brother, all which also have very low δ¹⁵N signatures, suggesting no bird inputs and reduced productivity relative to the present-day ponds. The sterol and stanol stratigraphy highlights that the surface sediments of the reference ponds, Main Duck and Calf, have the highest percentages of cholestanol (figure 2). Cholestanol can be formed naturally in the environment by bacteria [6] or, as mentioned, sometimes from the reduction of cholesterol in the environment [16] and, therefore, typically occurs in unpolluted environments. Thus, higher percentages of this stanol can be a marker for true reference sites, receiving little-to-no bird inputs.

Overall, the total sterol content of guano and sediments can be highly variable, and it is often best to look at relative abundances of the sterols in relation to the total sterol content [52]. Calf and Main Duck surface samples, our true reference sites, had 27% and 35% cholesterol, respectively, and bird ratios of 0.45 and 0.40, respectively. False Duck with approximately 100 nesting cormorants had 52% cholesterol and a bird ratio of 0.57 in the top sediments. All the remaining sites, with approximately 1300–53 000 nesting waterbirds, had cholesterol ranging between 62 and 72%, and ratios between 0.67 and 0.77. Gull Island and Little Galloo, with by far the greatest number of nesting waterbirds between 30 000 (Little Galloo) and 53 000 (Gull Island), had the highest bird ratios, approximately 0.76. Therefore, based on both per cent cholesterol and the expanded bird index, we can further split our impact sites into two groups: low and high impact. Additionally, the two impact sites with known vegetation die-offs owing to the direct toxicity of cormorant guano and its alterations to soil chemistry [62], Pigeon and East Brother, had the lowest relative proportions of plant sterols in their surface sediments. Across all impact sites, East Brother had the highest relative proportion of plant sterols in the bottom sediments (89%), representing the greatest change in plant sterols and overall sterol and stanol composition through time (greater than 100 years) [45] (figures 2 and 3a). Based on our preliminary observations, this may be a reliable marker for distinguishing heavy cormorant influence on islands from the influence of nesting gulls or terns.

Figure 3.

(a) Principle component analysis of the relative abundance of sterols and stanols quantified in the surface sediments from the eight study ponds and the mean sterol composition of guano collected from double-crested cormorants (DCCO), herring gulls (HG) and ring-billed gulls (RBG) depicted with stars. The bottom samples (hollow circles) from each sediment core are plotted passively and connected to the surface samples (solid circles) with a dashed arrow. Principle component axis 1 (PC1) and axis 2 (PC2) explain a cumulative 88.2% of the variation in the sterols across sites, with the axis eigenvalues displayed in brackets; (b) the sterol and stanol composition (% of the total sterol concentration measured) from the three sediment cores—High Bluff, Calf and Little Galloo, passively plotted in the ordination from (a) with the sedimentary core depth intervals labelled next to each point. GI, Gull Island; MD2, Main Duck; EB, East Brother; LG, Little Galloo; HB, High Bluff; PGN, Pigeon; FDI, False Duck Island. (Online version in colour.)

PCA can be used as a monitoring tool for tracking the spatial differences in modern sites and visually tracking the arrival and influence of birds on sites over time. The ‘top-bottom’ sediment analysis shows that several of the sites (e.g. Gull, Pigeon) have had high proportions of cholesterol relative to all other sterols delivered to the lake's older sediments (figures 2 and 3a). This indicates that either birds have been present at these sites for the length of the sediment cores, greater than 200 years at Pigeon [43], or that the mixing of cholesterol-rich surficial sediments contaminated the bottom layer of the sediment core. The latter may potentially be the case for Gull Island, as sedimentary ²¹⁰Pb activity was relatively steady until the 14 cm level, at which point there was a rapid decay until 16 cm (electronic supplementary material, figure S1). However, for Pigeon Pond, the presence of diatoms beginning at approximately 1850 (8 cm) and increases in sedimentary chlorophyll a and δ¹⁵N suggest that the stratigraphic integrity of the sediment record had not been compromised [43]. Elevated δ¹⁵N to the base of the record (approx. 15‰) [43] corroborated the sterol abundances and bird ratio, suggesting that waterbirds have been nesting on this Lake Ontario island for several hundred years.

The sterol composition of the High Bluff sediment core transitioned from a dominance of sitosterol (24 cm) to shared abundances of sitosterol and stigmastanol (16 and 20 cm), followed by an increase towards a greater representation of cholesterol at 12 cm (figure 3b). In concordance, δ¹⁵N values at the bottom of the core (approx. 6–7‰) were close to values for Lake Ontario sediments (approx. 4‰) [63], as well as the non-impacted sites of Stewart et al. [43], and increased from approximately 6 to 13‰ (from 20 to 7 cm with a notable increase at 12 cm [60]). Both these proxies indicate inputs from animals with cholesterol as a dominant component of their faeces as well as input from animals with a high trophic position within the Lake Ontario aquatic food web (as there is an approx. 3‰ difference per trophic level). These are consistent indicators associated with the influence of a large bird colony nesting around a pond, as has been found on other islands in this region [43], and in the High Arctic [38,52].

Within the PCA ordination, Little Galloo bottom sediments (18 cm) were more distinct than any other site with a greater composition of coprostanol, epicoprostanol and coprostanone. The presence of a historically manned lighthouse on Galloo Island (established in 1820), a close neighbour of Little Galloo Island, may explain the higher abundance of these largely human-associated sterols at the base of the core. Coprostanol can be up to 10 times more abundant in human faeces than most other animals and the biota in the digestive tracts of most birds appears largely incapable of biohydrogenating precursor sterols to 5β-stanols [29]. From 5 to 15 cm, the samples were clustered with the plant sterols and then from 5 to 0 cm there was movement towards the cholesterol quadrant of the PCA (figure 3b). δ¹⁵N in the bottom of the Little Galloo core was already elevated (approx. 15‰), increased to 17.5‰ around 14.25 cm (approx. 1930s), and then remained stable until the surface of the core [60]. The timing of this increase in δ¹⁵N pre-dated the estimated arrival of ring-billed gulls to Little Galloo Island around 1940 [40]. However, little is known about the ring-billed gull colony on Little Galloo before it was first documented in 1945 [40], and thus our sediment record may indicate an earlier arrival date of the ring-billed gull colony in contrast to documented nesting. Cormorants began nesting on Little Galloo Island in approximately 1969 [42]. As cormorants are more likely than ring-billed gulls to deposit guano in their nesting areas (D. Tyerman 2016, personal communication), the onset of cormorant nesting would explain the greater representation of cholesterol in the top 5 cm of the core.

In the Little Galloo Pond core, the unsupported ²¹⁰Pb activity displayed a plateau with very little decay until the second last interval, between 16 and 18 cm. In conjunction with this plateau, unsupported ²¹⁰Pb activity never decreased below background, and the total Pb concentrations decreased through the core to the surface, not aligning with established Pb profiles in Lake Ontario sediments (electronic supplementary material, figure S1) [49]. Therefore, even given the large changes in sterol abundances in the Little Galloo sediment core, the dating proxies suggested sediment mixing, making reliable inferences on the timing of bird occupation on this island not possible.

A passive comparison of the sterol and stanol composition of Calf Pond downcore sediments relative to its surface suggested that the sediments remained dominated by plant sterols. The Calf Pond dating profile indicated that unsupported ²¹⁰Pb activity reached supported ²¹⁰Pb values (background) at 40 cm depth (electronic supplementary material, figure S1) and was thus interpreted as being approximately 150 years old. In conjunction with the decay and total Pb profiles that increased over time to the surface, aligning well with Graney et al. [49], we used the CRS-dates for this core. For the duration of the sediment record, δ¹⁵N was low and stable [60], and phytosterols dominated. Together, these proxies indicated that this site has been relatively unaffected by large nesting colonies of waterbirds. Algal proxies suggested Calf Pond is a naturally productive site [60] with inferences that the site became more nutrient-rich around the 1990s. These more productive pond conditions may reflect the movement to modern sediments with greater proportions of sitosterol, and less stigmastanol, which we suggest may be present in more pristine pond conditions. Calf Pond also experiences diurnal anoxia with several nights recorded to have zero oxygen [60], which could inhibit the conversion of sitosterol to stigmastanol. Alternatively, within PCA-space, the Calf pond sediments may be interpreted as becoming more similar to several of the low-bird impact sites (e.g. EB TOP, FDI TOP). This may coincide with known, brief, unsuccessful nesting attempts of cormorants on Calf Island in the 1990s, shortly followed by the brief success of a colony of 170 cormorant nests in 2008 and the 24 nests in 2009 (I. Mazzocchi 2016–2017, personal communication).

5. Conclusion

Our results clearly demonstrate the value of using sterol compounds and their ratios measured in lake sediments to track the activity of different waterbird species and colony nesting sizes on temperate ponds. These variations were driven, in part, by species-specific differences in sterol and stanol signatures, including the presence of greater phytosterols in omnivorous gull species presumably linked to their scavenging of grain products. This research highlights that we should continue developing and testing taxon-specific markers in sediments, as even across birds nesting in a similar region, differences in diet and digestion pathways allow enhanced marker specificity that traditional proxies alone, e.g. δ¹⁵N, cannot provide. However, birds with more stable, natural diets may be more reliably tracked with sterols and stanols, and we suggest that, if omnivorous birds are studied, more than three guano samples should be collected that broadly span the breeding and nesting season of a species. Furthermore, we recommend that this approach could potentially be insightful in deeper ponds and lakes with high-resolution sedimentary archives, less affected by bioturbation and wind mixing. The high proportions of faecal sterols, predominantly cholesterol, throughout several of the sediment cores suggest that birds have been nesting on these Lake Ontario islands for several decades. Integrating sterols and stanols into future biovector studies can help pinpoint the identity of the vector of nutrient and contaminant subsidies, which may allow us to better understand and predict ecosystem-wide consequences in lakes through time.

Data accessibility

Data associated with this manuscript are available in the electronic supplementary material.

Authors' contributions

J.P.S., C.G., N.M., E.M.S., K.E.H. and J.M.B. conceived of the study, coordinated and participated in the design of the study; K.E.H. and E.M.S. carried out the laboratory work, participated in data and statistical analysis, and drafted the manuscript; C.G., N.M. and E.M.S. collected field data; J.M.B. provided conceptual and analytical development for the sterol/stanol method application; M.M. contributed to study design and helped draft the manuscript; L.K. coordinated laboratory analysis and contributed to drafting the manuscript. All authors gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

An N.S.E.R.C. Discovery Grant funded to J.M.B., and the N.S.E.R.C. Brockhouse Canada Prize awarded to J.M.B. and J.P.S. A W. Garfield Weston Postdoctoral Fellowship supported KEH during the time of data collection and analysis.