Do deposit-feeders compete? Isotopic niche analysis of an invasion in a species-poor system

Successful establishment of invasive species is often related to the existence of vacant niches. Competition occurs when invaders use the same limiting resources as members of the recipient community, which will be reflected in some overlap of their trophic niches. The concept of isotopic niche has been used to study trophic niche partitioning among species. Here, we present a two-year field study comparing isotopic niches of the deposit-feeding community in a naturally species-poor system. The isotopic niche analyses showed no overlap between a recent polychaete invader and any of the native species suggesting that it has occupied a vacant niche. Its narrow isotopic niche suggests specialized feeding, however, the high δ15N values compared to natives are most likely due to isotope fractionation effects related to nitrogen recycling and a mismatch between biological stoichiometry of the polychaete and the sediment nitrogen content. Notably, highly overlapping isotopic niches were inferred for the native species, which is surprising in a food-limited system. Therefore, our results demonstrate that invaders may broaden the community trophic diversity and enhance resource utilization, but also raise questions about the congruence between trophic and isotopic niche concepts and call for careful examination of assumptions underlying isotopic niche interpretation.

Successful establishment of invasive species is often related to the existence of vacant niches. Competition occurs when invaders use the same limiting resources as members of the recipient community, which will be reflected in some overlap of their trophic niches. The concept of isotopic niche has been used to study trophic niche partitioning among species. Here, we present a two-year field study comparing isotopic niches of the deposit-feeding community in a naturally species-poor system. The isotopic niche analyses showed no overlap between a recent polychaete invader and any of the native species suggesting that it has occupied a vacant niche. Its narrow isotopic niche suggests specialized feeding, however, the high d 15 N values compared to natives are most likely due to isotope fractionation effects related to nitrogen recycling and a mismatch between biological stoichiometry of the polychaete and the sediment nitrogen content. Notably, highly overlapping isotopic niches were inferred for the native species, which is surprising in a food-limited system. Therefore, our results demonstrate that invaders may broaden the community trophic diversity and enhance resource utilization, but also raise questions about the congruence between trophic and isotopic niche concepts and call for careful examination of assumptions underlying isotopic niche interpretation.
C lassic theory suggests that no two species can have the same niche; the less effective competitor will be excluded from an area 1 . The basis for coexistence is niche differentiation, often obtained through resource partitioning 2 . Invasive species are often generalists 3 that can take advantage of formerly unexploited resources, thereby occupying a distinct niche space relative to the native community. Hence, invasion success is often explained by occupation of a previously vacant niche 3,4 . Alternatively, the invader might share trophic niche space with some members of the native community but be competitively superior, for example, by having higher feeding rate or food conversion efficiency and hence higher production 3,5 . Finally, if food resources are not limiting, co-existence may result from interference competition for space 6 . The ecological strategies of invasive species can cause food web changes and influence ecosystem functioning [7][8][9] . However, there also examples of invasive species replacing natives with similar trophic niches, which results in little change in overall ecosystem functioning 10 .
Stable isotopes have become powerful tools for studying food webs and trophic relationships as they provide time-integrated information on diet and habitat use 11 . For nitrogen, ratios of 15 N to 14 N (expressed as d 15 N) exhibit stepwise enrichment with trophic transfer, while ratios of carbon isotopes (d 13 C) vary substantially among primary producers and therefore mainly reflect primary carbon sources. The distribution of individuals in isotopic space is often used to quantify niche size 12 . Several aspects of trophic diversity and dietary divergence between individuals/species can be quantified and compared 13 . As we often lack the data to determine the true trophic niche of consumers in field studies, particularly for omnivores and deposit-feeders, use of the isotopic niche as a proxy is becoming common in ecological and evolutionary studies 14 . The isotope niche of a population integrates temporal (seasonal to multi-year) andat least to some extentspatial (population habitat) variations 14 . This quantitative approach to describe trophic space by stable isotopes has been used to test ecological theories 15 including trophic ecology of invasive species in their novel environments 16,17 , but has also been criticized 18 . To evaluate how well the isotopic niche aligns with the trophic niche, more studies are needed that give the additional information required to infer ecological mechanisms from stable isotope data.
The Baltic Sea is a good model system for testing hypotheses on community structure and function as its low number of species makes it easier to elucidate trophic relationships 19,20 and food-web effects of invasive species 8,21 . A mere three macrofaunal species constitute the native deposit-feeding community in sub-thermocline soft bottoms in the northern Baltic Sea. During the last two decades, three deposit-feeding species of the polychaete genus Marenzelleria spp. have invaded the Baltic Sea 22 , with mainly Marenzelleria arctia cooccurring with the native species below the summer thermocline at 15-20 m depth 23 . Today, M. arctia has become one of the most abundant species in many Baltic Sea areas covered by the Swedish National Monitoring Program (data available at www.smhi.se).
In temperate waters, deposit-feeders depend crucially on the sedimentation of the spring phytoplankton bloom as a source of food 21,24,25 . This is especially true for the amphipod Monoporeia affinis in the Baltic Sea, and its population decline over the last two decades has been linked to a reduced spring-bloom input 24 . A recent experimental study using isotope enriched algae found little overlap in use of freshly deposited spring bloom material between M. arctia and the native species, suggesting that M. arctia is not a strong competitor for the nutritious diatom input during the experimental duration of three weeks 21 . However, due to efficient utilization of old organic matter by M. arctia, competition can be expected to intensify once the spring bloom input to the sediment is exhausted and food limitation sets in 21 . The well-studied trophic ecology of the native species, especially the habitat and resource partitioning between the amphipods M. affinis and Pontoporeia femorata 20,26,27 provides a solid background for applications of isotope niche in ecological studies.
Here, we compare the isotopic niches of the three native species and the non-indigenous M. arctia that together constitute the deposit-feeding community in a coastal area of the northern Baltic proper, where the invader has co-existed with the native species for about 20 years. To explain the invasion success of M. arctia in the Baltic Sea, we tested two alternative hypotheses: 1) There is little or no isotopic niche overlap between the co-existing species; this would indicate that M. arctia has occupied a vacant niche. 2) The invasive polychaete has a larger isotopic niche than the native species; this would suggest that M. arctia is a generalist feeder potentially outcompeting native species.
We also tested relationships between isotopic niche area and body condition to provide a better understanding of ecological implications of the observed variation in niche space and to test whether a large isotopic niche may confer growth benefits, as an expected outcome of generalist feeding 28 . Finally, C:N stoichiometry of the deposit-feeders and a relationship between sediment-consumer d 15 N enrichment and food quality 29 , as measured by nitrogen content in sediment, were evaluated to understand differences in d 15 N of the consumers.

Results
Species composition and sediment properties for each station and year. The total and relative abundance of species varied among stations, with densities of the invasive species Marenzelleria arctia ranging from 23 to 677 individuals m 22 (Fig. 1). As a group, depositfeeders about doubled in abundance from 2009 to 2010 at all stations. Sediment carbon and nitrogen content were 5.0% and 0.7% (Håldämman), 5.2% and 0.7% (Uttervik) and 3.5% and 0.5% (Mörkö), respectively (n 5 6-8, analytical precision within 0.1%).
Isotopic niche overlaps. The isotopic niche, measured as the standard ellipse area (SEAc), of the invasive species M. arctia differed clearly from those of the native species in all datasets (Fig. 2). By contrast, SEAc values for native species overlapped on five of six occasions (Table 1). In 2010 at stn Mörkö, the niche area overlap among the three native species was nearly complete (Fig. 2), with Monoporeia affinis overlapping by 98% with Pontoporeia femorata. The highest median niche area overlap was found between P. femorata (58% of its niche area) and Macoma balthica (51%; n 5 4).
Species-specific differences in niche indices and relation to body condition. M. affinis or P. femorata had the largest and M. arctia the smallest SEA B (SEA estimated via Bayesian interference) of all species (Fig. 3). A similar pattern was found for the other niche indices (Fig. 4). M. arctia had significantly smaller d 13 C-range (indicative of the range of basal resources utilized) range than M. affinis and P. femorata (F 3, 18  Sediment-consumer enrichment in d 15 N and nitrogen content in sediment. There was a significant decrease in the enrichment values in consumers with increasing %N in sediment. Moreover, M. arctia, which had the lowest C:N ratio among the species tested ( Fig. 4), also had significantly higher d 15 N values than the three native species, which may reflect greater limitation by low protein quality in sediments ( Table 2, Fig. 5).

Discussion
The isotopic niche concept is increasingly used by ecologists but there are still too few studies that evaluate how well the isotopic niche aligns with the trophic niche. The well-studied trophic ecology of our deposit-feeding community allows us to discuss whether the assumptions underlying isotopic niche interpretations may have The isotopic niche of the non-native Marenzelleria arctia was clearly distinct from those of all native species in the deposit-feeding guild, thus supporting Hypothesis 1. The clear separation between M. arctia and the native species, regardless of M. arctia population density and organic content of the sediment, suggests limited competition for resources and provides evidence that this invasive species has occupied a vacant niche in the Baltic Sea soft-bottom habitat. The field data presented here strongly support experimental findings showing efficient resource partitioning between M. arctia and the native species 21 . The species poverty of Baltic sediments, common to other brackish environments where osmoregulation capacity limits species distribution 30 , has likely left more niches vacant than in more species-rich systems. Such vacant niches canat least in partexplain the high invasibility of brackish waters 31 . The consequences of invasion-induced species richness for ecosystem productivity are complex even in a species-poor system like the Baltic Sea 21 , but the well-known trophic ecology of the studied macrofaunal species gives insight into the mechanisms behind the observed patterns. Below we discuss them in detail.
The isotopic niche of M. arctia differs from those of the native species in three main ways: more enriched d 15 N, more depleted d 13 C and, compared to the amphipods, a smaller niche area as well as lower values for the other niche indices representing basal resources and dietary divergence between individuals. The relatively narrow isotopic niche of M. arctia and the absence of a positive relationship  Table 1. Sediment isotope value is denoted 1, early spring bloom isotope value is denoted r and late spring bloom isotope value is denoted &. Seston (stn Mörkö only) is denoted m. Resource data are mean 6 standard deviation. between its body condition and niche size suggests specialized feeding, in contrast to our Hypothesis 2. It is, thus, likely that M. arctia occupies a previously vacant niche. The resemblance of its carbon isotope value to both the sediment and diatom carbon isotope would indicate a strong dependence on the diatom input. M. arctia do feed on diatoms during experimental conditions but grow nearly as well when offered aged sediment only 21 , whereas native species are known to critically depend on the diatoms for growth and survival 21,24,25 . After lipid correction 32 , the carbon isotope signature of the native species became however clearly heavier than both the sediment and diatom signatures (Fig. 2), which is most likely related to amphipod' higher requirements for carbon to build lipid reserves 20,21,33 and, therefore, less discrimination against the heavier isotope. Furthermore, the dissolved inorganic carbon pool in the water becomes more enriched as the season progresses 34 , resulting in more enriched carbon values in phytoplankton later in the season, as evident from the late spring bloom isotope data that better resemble the carbon signal of the native species (Fig. 2). Although dinoflagellates, which dominate the late spring bloom, are a carbon rich food source, they settle mostly as cysts which are not efficiently digested by macrofauna 35 , and their value as food for benthos may therefore be questioned. It is, however, important to remember that the phytoplankton or seston isotope signatures do not necessarily reflect the isotope signal of decomposing organic material which has undergone a microbial conditioning in the sediment that may enrich isotope values 36 . High contribution of microbially processed organic matter might explain the higher d 15 N of M. arctia. An alternative explanation for greater d 15 N enrichment of M. arctia could be a higher isotopic fractionation by the polychaete, due to a mismatch between the nitrogen requirements and its availability in the sediments. It has been hypothesized that nitrogen isotopic discrimination should decrease as dietary protein quality increases 29 , and the significant negative relationship between sediment-consumer enrichment in d 15 N and sediment %N observed in our study (Fig. 5) supports this hypothesis. Moreover, as M. arctia has relatively high nitrogen requirements 21 due to its nitrogen content of about 10% compared to , 8% for the native species 37 , the higher enrichment observed for the polychaete (Fig. 5) is not surprising. Indeed, a higher flux of amino acids through various metabolic pathways, where transamination and deamination take place, should result in elevated 15 N fractionation, particularly when food is suboptimal in terms of the protein quality and similarity in the essential amino acid composition 38 . Finally, the high d 15 N signal could indicate that M. arctia has a higher trophic position by feeding also on meiofauna or on dead macrofauna. This, however, seems unlikely as Marenzelleria spp. are classified as facultative suspension-(surface) deposit-feeders 39 and have been shown to efficiently incorporate labelled phytodetritus 21 . Furthermore, Viitasalo 40 and Urban-Malinga et al. 41 found no evidence of predation by Marenzelleria spp. on cladoceran resting eggs or meiofauna in experimental studies. We therefore suggest that isotope enrichment originating from recycling of N in the benthic food web and greater organismal fractionation by the polychaete are the main causes of higher d 15 N in M. arctia than in the native species. We note that assigning the high d 15 N to predatory feeding behaviour in this species in the analysis of the benthic food web structure would lead to erroneous interpretations of the trophic positions of higher level consumers, such as benthivorous fish, when this signal is translated higher up in the food web.
Isotopic niche assessment relies on several assumptions, namely: (i) there is no isotopic overlap in food sources; (ii) isotopically identical food sources should yield identical delta values, i.e. variability in diet-consumer fractionation is negligible; (iii) temporal integration is similar between the species, i.e. the isotopic signatures reflect the diets assimilated over similar periods; and (iv) all species are in isotopic equilibrium with their diets. Hence, it is important to critically evaluate how well the isotopic niche aligns with the trophic niche. Assumption (i) is violated since both diatoms and sediment have very similar isotope values, although none can be directly considered equivalent of the isotope signal of the assimilated diet. As discussed above, due to differential elemental requirements and biological stoichiometry, species-or stage-specific fractionation of nitrogen and carbon isotopes would likely violate assumption (ii). Indeed, variation in growth and metabolic rates (e.g., M. affinis has faster growth, respiration rate and lipid accumulation than P. femorata 33 ) is likely to result in differential fractionation 42 , violating assumptions (ii), but also (iii) and (iv) and, hence, affecting the interpretation of isotopic niche indices and niche overlap for these species. Differences in growth rate between long-lived bivalves and semelparous amphipods should then be even more pronounced, with similar implications as above. Moreover, asynchronous moulting in crustaceans as a function of individual growth rate would further increase d 15 N variability among individuals in a mixed population 43 compared to a population with strong cohort structure. Therefore,  isotopic niche area comparisons should be interpreted with caution and high isotopic niche overlap observed between the native species (up to 100%), should not be considered as a proof of equally extensive trophic overlap. Indeed, the amphipod species with largely overlapping trophic niches partition resources by depth in sediment when found together, with P. femorata feeding deeper in the sediment 20,21,26,27 . Such context-dependent small-scale partitioning is not likely to be reflected in isotopic niche overlap, hence violating assumption (i). In contrast, the overlap between M. affinis and M. balthica isotopic niches likely reflects at least some food competition, since both species are known to feed in the top centimetre of the sediment 21 . Moreover, asymmetrical competition for phytodetritus between these species was suggested in the latter study, with incorporation rates in M. affinis negatively influenced by the presence of M. balthica. This, and predation by M. affinis on larval M. balthica 44 , could explain why these species seldom co-occur in high abundances 45 . Increased competition for resources could result in trophic niche widening, if the consumers are forced to broaden their diets and shift to suboptimal resources in order to meet energy requirements 46 . However, the common view is that it is a lack of competition rather than increased competition that allows consumers to extend their realised niche 47 . In support of this, we found a positive correlation between isotopic niche size and body condition (reflected by C:N ratio) for P. femorata and M. balthica, suggesting that supplemental feeding is physiologically advantageous 28 .
Studies of invasive species often examine their deleterious effects on native species, community composition, food-web functioning and ecosystem services. Previous research focused on competitive interactions between invasive Marenzelleria spp. and native deposit-feeders 48,49 since its invasion in the northern Baltic Sea coincided with population crash in M. affinis. Recent evidence indicates, however, that the observed crash in amphipod abundance was an effect of climate-induced food shortage, rather than the invasion 24 . Moreover, it was later found that there are three species of the genus Marenzelleria in the Baltic, which differ in habitat requirements 23 and sediment reworking 50 , making it difficult to interpret some of the early interspecific competition experiments 48,49 and to extrapolate results from shallow 51 to deeper waters. Furthermore, the capacity of Marenzelleria spp. to bury deeper in the sediment than the native species is feared to facilitate remobilization of old contaminants to the overlying water 52 . On the other hand, several recent studies have suggested positive effects of the M. cf. arctia invasion in deeper areas of the Baltic Sea, including enhanced long-term retention of phosphorus in sediments through increased oxygenation 53 , burial of fresh detritus deeper in the sediment where mineralization is lower 21,54 and suppressed cyanobacterial recruitment 55,56 , counteracting eutrophication symptoms. Finally, as discussed above, the results of our field study and the earlier experiment with M. arctia 21 indicate that this invader and the native species show resource partitioning that may enhance resource usage and energy transfer efficiency through broadening the community's trophic niche. That M. arctia can grow rapidly on sediment lacking recent bloom input 21 means that detritus that would otherwise have been lost to bacterial mineralization is now efficiently converted to secondary production and ultimately to fish production. It is not known to what extent M. arctia can replace declining M. affinis populations as fish food, but Marenzelleria spp. is eaten by sand gobies, which spend winter in deep water 57 , indicating that it can function as food for other benthivorous fish as well.
In conclusion, there is no isotopic niche overlap between M. arctia and the native deposit-feeding community during the growth season. This strongly indicates resource partitioning between the invader and the native species. The native species, on the other hand, often exhibit almost completely overlapping isotopic niches, although it is important to remember that this does not rule out resource partitioning in space (e.g., by sediment depth), variability in fractionation, or use of different foods with similar isotopic values, resulting in similar isotopic niches. Hence, caution in interpretation of the isotopic niche is necessary as it may not fully represent the trophic niche, and studies testing correspondence between trophic and isotopic niches in ecologically relevant settings are warranted. Although the wider food-web consequences of the M. arctia invasion are difficult to assess, as a food web component, M. arctia has likely enhanced resource utilization in the Baltic Sea, with potentially positive effects further up the food chain.

Methods
Sampling procedures and isotope analyses. We sampled sediment and depositfeeding macrofauna during May to September in 2009 (4 sampling events) and 2010 (2-3 sampling events) at three coastal stations in the north-western Baltic proper; stn Håldämman (30 m depth, 58u499 N, 17u349 E), stn Uttervik (20 m depth, 58u509 N, 17u329 E), and stn Mörkö (23 m depth, 58u549 N, 17u429 E). We used a benthic sled, set to collect the top 1-2 cm sediment, which was then sieved through a 1 mm sieve to retain fauna. The two most abundant species, the bivalve Macoma balthica and the non-indigenous polychaete Marenzelleria arctia 23 were found in high numbers at all stations on all sampling occasions. Benthic community composition data for May 2009 and 2010 were obtained from Swedish National Monitoring Programme (SNMP, data available at www.smhi.se; stations 6001 and 6010 near stn Uttervik and stn Håldämman, respectively). Benthic community data from stn Mörkö in October in 2008 and 2011 were obtained from the SYVAB's marine monitoring program in Himmerfjärden Bay (Himmerfjärden Eutrophication Study; www2.ecology.su.se). At stn Mörkö, the amphipod Monoporeia affinis was less abundant in May and absent in July 2009. The amphipod Pontoporeia femorata was never found at stn Uttervik and absent from stn Håldämman in May and July 2009 (Fig. 1). On each sampling occasion, we selected up to 10 individuals of similar size (2 mg dry weight, shell-free for M. balthica) of each species from each station for isotope analysis. In total, 655 individuals were analysed, about 200 each for M. affinis, M. balthica and M. arctia and 81 for P. femorata. Animals and bulk sediment (upper 1-2 cm, collected by the sled on each sampling occasion were oven-dried (60uC), packed individually into tin capsules and analysed for elemental and stable isotope content (carbon and nitrogen) at the UC Davis Stable Isotope Facility, USA. The C and N isotope ratios are expressed in the d notation, using equation (1): where R is the ratio between the heavy and light isotopes ( 13 C: 12 C or 15 N: 14 N). The stable isotope ratio, d, is defined as the deviation from an international reference standard (Vienna PeeDee Belemnite for C, and atmospheric nitrogen gas for N), given in % since it is a small figure. Higher d indicates a higher proportion of the heavy isotope. Samples were run in continuous flow with a standard deviation of ,0.2% among replicate standard samples both for C and N. Before calculations and statistical analyses, all d 13 C-values were corrected for lipid content using the C:N ratio 32 . Data on spring bloom isotopic signal originate from different sources (Table 3). Stable isotope data for early (dominated by the diatom Thalassiosira sp., n55) and late (dominated by Dinophysis sp. n55) spring bloom (March and April, respectively) for stations Uttervik and Håldämman were obtained from Rolff 58 . Spring bloom material from stn Mörkö was sampled in April (n53, sample dominated by Thalassiosira sp. but also containing ciliates and rotifers). Seston from mid-May was sampled at 20 m depth using water bottle sampler (22.7 L). The entire volume was filtered with 90 mm sieve and particular matter of the filtrate was collected on GFF (n53). All samples were oven-dried (60uC), packed into tin capsules and analysed for elemental and stable isotope content as described above.
Data analyses and statistics. Isotopic diversity indices 13 were calculated for each species, station and year using the SIBER (Stable Isotope Bayesian Ellipses in R) package for R v.2.10.1 59 . The isotope niche of a population integrates temporal (seasonal to multi-year) andat least to some extentspatial (population habitat) variations. To focus on our hypotheses, we pooled population data within a season as advocated by Antonio and Richoux 14 . Moreover, deposit-and suspension-feeders are commonly used in food web analysis to represent the baseline signal smoothing-out fluctuations in stable isotope signatures of primary producers that would obscure trophic position estimates in a food web 60 . Therefore, pooling population data within a season would incorporate variations related to both source signature and individual feeding preferences consistent with trophic niche definition. Three of the indices developed by Layman et al. 13 were used here; trophic diversity (dN, d 15 N-range), multiple basal resources and a potential for niche diversification (dC, d 13 C-range) and dietary divergence between individuals (mean nearest neighbour distance; MNND), which describes how individuals are distributed relative to one another within a population's dietary niche space. Instead of the convex hull area 13 , which is the total area encompassed by all points on a d 13 C2d 15 N bi-plot, we calculated the standard ellipse area, SEA 59 (comparable to SD in univariate cases). SEA C (c denotes that SEA was corrected for small sample size) provides information about the core aspects of a population's niche and is less sensitive to outliers and small samples sizes (n , 30 according to Syväranta et al. 61 ). Overlap in SEA C between species was calculated for each species combination. Differences in SEA C between species were estimated via Bayesian interference (SEA B ), according to Jackson et al. 59 . Differences in the other diversity indices (dN, dC and MNND) between species were tested using one-way ANOVA for each index separately, followed by Tukey's HSD. Residuals were inspected for normality and MNND data were log-transformed to improve distribution of the residuals. We also performed Pearson product moment correlations to explore possible relationships between (i) Marenzelleria arctia density and niche size (SEA C ) of native species and (ii) SEA C and the C:N ratio (high values indicate better body condition with larger lipid reserves) for each species. The effect of %N in sediment on sediment-consumer enrichment in d 15 N was tested with general linear model (GLM) with sediment %N and C:N ratio of deposit-feeders as independent variables. ANOVA, GLM and Pearson correlation tests were performed in STATISTICA 12 (StatSoft Inc.). Raw data on isotope composition and C/N ratio are available from the Dryad Digital Repository: http://datadryad.org.