Keywords: habitat degradation, invertebrates, niche overlap, optimal foraging, population dynamics, seagrass, stable isotopes, trophic niche

2.1. Study area

The study area was located near the central Tyrrhenian coast of Italy, within an area whereP. oceanica meadows extend discontinuously for 40 km along the coastline. In this area, the conservation status of theP. oceanica meadows is lower at the north end (Paticchio,2013). Intense illegal trawling during the past century increased coastal urbanization and water turbidity due to the presence of stream mouths are considered the main causes of meadow degradation (Paticchio,2013). Water turbidity can increase during autumn‐winter, when rainfall peaks. Within this area, we selected the meadow characterized by the highest conservation status, just off the “Salt marsh Nature Reserve” of Tarquinia (Vt) (42°12′N 11°42′E) (Paticchio,2013). The presence of the Reserve has protected this area from intense urbanization and the intensification of controls since the 1990s has reduced illegal fishing in the area. Although the current general conditions do not differ substantially from those of the 1990s, the meadow is characterized by varying degrees of coverage.

Samplings were carried out at a fixed depth of 6 m in three locations differing in the degree ofP. oceanica coverage. Coverage was estimated in circular areas of 60 m in diameter (2,826 m² surface area) by two scuba divers and by photographic analysis. Each diver operated independently during a preliminary survey, providing two independent visual estimates per location. The degree of coverage varied between locations but was highly homogeneous within each location. Accordingly, we defined a high‐coverage location (“H,” coverage: 92.5 ± 2.5%), an intermediate‐coverage location (“I,” coverage: 70.0 ± 5.0%), and a low‐coverage location (“L,” coverage: 50.0 ± 5.0%) (Kruskal–Wallis test and Mann–Whitney pairwise comparisons: Hc = 9.85,p < .01). Locations H and L were 1,850 m apart, with location I being positioned half‐way between the two. Macroinvertebrates were sampled using litterbags (2‐cm mesh size) anchored to the meadow bed, each containing 20‐g dry weight ofP. oceanica leaf litter (Costantini, Rossi, Fazi, & Rossi,2009). Two sampling sites per location were chosen. Six litterbags were placed randomly at each sampling site (making a total of 12 litterbags per sampling location), as far as possible from the edge of the seagrass patch in order to represent environmental conditions characterizingP. oceanica patches across three levels of seagrass coverage. The minimum distance between litterbags within each site was 10 m. Sediments, attachedP. oceanica leaves (both with and without evident epiphyte colonization) and leaf litter, representing the trophic sources at the base of the macroinvertebrate food web, were harvested together with macroinvertebrates at each location (Table S1). This study relies on census and isotopic data presented in Calizza, Costantini, et al. (2013), where a detailed description of invertebrate and resource samplings, processing and species’ isotopic data can be found. Specifically, a total of 39 benthic macroinvertebrate species were found and the community composition differed between locations. Macroinvertebrate density decreased with meadow degradation (one‐way ANOVA and Tukey's pairwise comparisons,p < .01), that is, from location H (34.4 ± 6.0 individuals per litterbag) to location L (16 ± 5.2 individuals per litterbag), and species richness was higher in H (26 species) and lower in I and L (both 19 species). In addition, the percentage of both coarse (>1 mm) and ultrafine (<0.56 mm) sediments accounted for by organic matter decreased from H to L. The Percent Ash Free Dry Mass (AFDM%) of coarse sediment was 6.9 ± 0.1% in H and 3.2 ± 0.5% in L (one‐way ANOVA and Tukey's test,p < .05), while the AFDM% of ultrafine sediment (the richest fraction for organic matter content) was 8.1 ± 0.2% in H, with ultrafine sediment being absent from locations I and L. Both these aspects suggest that locations I and L were characterized by limiting conditions as a consequence of habitat degradation, as reported for other seagrass habitats (Bell, Brooks, Robbins, Fonseca, & Hall,2001; Boström, Jackson, & Simenstad,2006; Bowden, Rowden, & Attrill,2001; Orth et al.,2006; Van der Heide et al.,2007). In the text, we will refer to “location scale” to indicate the sampling location spatial scale (i.e., accounting for differences between H, I, and L) and to “meadow scale” to indicate results describing the density and trophic organization of species in the study area as a whole (i.e., when not accounting for differences between sampling locations, see below).

2.2. Target species

We focused on the three most abundant species:Microdeutopus obtusatus Myers (Amphipoda),Athanas nitescens Leach (Decapoda), andCymodoce truncata Leach (Isopoda) as Target Species (TS) (sensu Lawlor,1979) undergoing competition with each other and with the remaining nontarget species (NTS) in the community. These three species were chosen as they (1) accounted for 70% of macroinvertebrate organisms in H, 84% in I and 76% in L, as well as 76% of the overall invertebrate community at the meadow scale; (2) were present at all sampling locations; (3) were characterized by similar habitat use and foraging behavior, all having the potential for active swimming if necessary, but preferring to search for shelter and remain inconspicuous, and (4) displayed similar body size (specimens considered in this study were between 11 and 14 mm in length forM. obtusatus, between 12 and 16 mm forA. nitescens, and between 11 and 15 mm forC. truncata), which can be an important determinant of metabolic rates and interspecific competition outcomes (Basset,1995; Brown, Gillooly, Allen, Savage, & West,2004; di Lascio, Rossi, & Costantini,2011). In addition, from points (3) and (4), it may be supposed that these three species are subject to similar predatory pressure, the effect of which is not explicitly accounted for in this study (Calizza, Rossi, & Costantini,2013; Lima,2002; Mancinelli, Costantini, & Rossi,2007).

2.3. δ¹³C and δ¹⁵N distribution and mixing models

For each target species at each location, population‐wide niche metrics were applied to invertebrate isotopic data in accordance with Jackson et al. (2011) and Layman, Arrington, Montaña, and Post (2007), usingstable isotope Bayesian ellipses in R (SIBER) as part of the R statistical computing package (R Development Core Team2012). The range (i.e., the difference between maximum and minimum values, CR) and variance (σ²) of δ¹³C were considered to be measures of the range of exploited resources and monodimensional niche width, respectively (Layman, Arrington, et al.,2007; Layman, Qattrocchi, et al.,2007; Rossi et al.,2015; Sanders, Vogel, & Knop,2015). Isotopic niche space was calculated as the Standard Ellipse Area corrected for degrees of freedom (SEAc) encompassing the core (i.e., around 40%) of isotopic observations for each population, along with the isotopic total area (TA) occupied by specimens. Distribution of SEAc values obtained for each species was pairwise‐compared between locations by means of Welch'st‐test, which made it possible to account for unequal variance in SEAc distributions. The overlap between species pairs’ TA and SEAc at both the location and meadow scale was also calculated (SIBER analysis, Jackson et al.,2011). Overlaps between species pairs are expressed as the percentage of the TA or SEAc of each species.

At the location scale, intraspecific isotopic dissimilarity at the individual level (ND_ii) was quantified as the mean isotopic (i.e., Euclidean) distance between each specimen and all remaining conspecifics within each sampling location. The mean intraspecific isotopic dissimilarity for each population (MND_ii) was then obtained as the mean ND_ii value of all specimens in that population. The higher the MND_ii, the higher the mean isotopic dissimilarity between conspecifics at each location. Similarly, for each TS specimen, the mean isotopic distance from each nonconspecific specimen (NDij) belonging to any other species in the community (both target and nontarget species) was also calculated, obtaining a measure of mean interspecific isotopic dissimilarity between species pairs (MNDij).

The δ¹³C and δ¹⁵N values of resources did not vary significantly across locations (PERMANOVA,p > .05;Table S1). However, the isotopic distribution of each species (both TS and NTS) at the meadow scale was obtained from the individual δ¹³C and δ¹⁵N values observed at any given location, standardized with respect to the centroid of resources in that location by subtracting the resource centroids from the individual isotopic values. All the standardized values were then considered within a single isotopic niche space and isotopic metrics were calculated in order to describe niche metrics and species’ overlap at the meadow scale.

A Bayesian isotopic mixing model available as an open source R package (SIAR, Stable Isotope Analysis in R) was used to assess the relative contributions to consumers’ diets of attachedP. oceanica leaves, fresh (“Green”) and decomposed (“Brown”)P. oceanica leaf litter, epiphytes and sediment organic matter (SOM) (Careddu et al.,2015; Rossi et al.,2015). In accordance with McCutchan, Lewis, Kendall, and McGrath (2003), the isotopic shifts between consumers and resources for δ¹³C and δ¹⁵N were assumed to be 0.4‰ and 2.3‰, respectively. Metabolic isotopic fractionation by consumers is prohibitively difficult to measure in the field, which would need dedicated feeding experiments (e.g., Madeira, di Lascio, Carlino, Costantini, & Pons,2013; Rossi, Costantini, & Brilli,2007). Nevertheless, fractionation has been shown to be predictably influenced by consumer and diet type, consumers’ metabolic pathways, and environmental conditions (Henschke et al.,2015; McCutchan et al.,2003). Thus, considering a set of species relying on a shared pool of potential resources on limited spatial and temporal scales and of similar body size, it can be assumed that differences in specimens’ isotopic signatures reflect differences in their food use (Bašić & Britton,2016; Fry,2006; Jackson et al.,2011; Layman, Qattrocchi, et al.,2007; Rossi et al.,2015; Sanders et al.,2015; Swanson et al.,2015; Yao et al.,2016).

At the meadow scale, the proportional contribution of each resource item to the diet of each target species (P_Xmeadow) was calculated as follows:

whereP_X is the proportional contribution of resourceX to the diet; H, I, and L are the sampling locations, andN is the population density. This calculation provides a simple weighted measure of the contribution of each resource to the diet of consumers at the meadow scale. It takes account of the spatial (i.e., interlocation) variability of both species density and resource consumption by each species and is not affected by potential isotopic variations in resource items between locations.

Based on diet composition, trophic niche width was measured as the diversity (i.e., Shannon diversity index, Hs) of resources in the diet. Diet similarity between species pairs was calculated by means of the Bray–Curtis similarity index, based on the identity and proportion of resources consumed. As complementary information, the variability among TSs in the consumption of each resource at each location was obtained as the coefficient of variation (C.V.) of the contribution of any given resource to their diet.

2.4. Competition strength and carrying capacity

The strength of competition between species pairs was calculated with reference to:

1. The overlap in resource use (Abrams,1977; Levins,1968; Pianka,1969). In this case, competition strength is indicated by β_ij, that is, the effect of speciesj on speciesi (sensu Levins,1968), withi ≠ j, in accordance with the formula:

   $${\mathrm{\beta }}_{ij}=\frac{{\sum }_h\mathrm{p}ih\ast \mathrm{p}jh}{{\sum }_h{(\mathrm{p}ih)}^2}$$

   (2)

   where pih and pjh are the proportional consumptions of any given resourceh by speciesi and speciesj, respectively, obtained as outputs of Bayesian mixing models;
2. The ratio of intraspecific to interspecific isotopic dissimilarity for each species pair, based on mean individual isotopic distances. In this case, competition strength is indicated by α_ij, that is, the effect of speciesj on speciesi, withi ≠ j, in accordance with the formula:

   $${\mathrm{\alpha }}_{ij}=\frac{1}{n}\sum _{i=1}^n\frac{{\text{ND}}_{ii}}{{\text{ND}}_{ij}}$$

   (3)

   where ND_ii is the intraspecific isotopic dissimilarity and ND_ij the interspecific isotopic dissimilarity of any given specimen of speciesi, andn is the number of specimens ini. This measure provides a mean value of interaction strength and associated standard error which is dependent on differences in ND_ii and ND_ij between specimens. At the species assemblage level (i.e., considering the three TSs and the effect on them of each remaining NTS), the carrying capacity (K) for each TS at each sampling location was calculated with reference to Lotka–Volterra competition models:

   $$\frac{\mathrm{d}{N}_i}{\mathrm{d}t}=r_i{N}_i\left(\frac{K_i-N_i{\sum }_{i\ne j}^n{\mathrm{\alpha }}_{ij}{N}_j}{K_i}\right)$$

   (4)

   and direct competition coefficients coupled with competitors’ densities (Levins,1968; Pianka,1974), as follows:

   $$K_i=(N_i+{\mathrm{\alpha }}_{i2}{N}_2+{\mathrm{\alpha }}_{i3}{N}_3+\dots +{\mathit{\alpha }}_{ij}{N}_j)$$

   (5)

   whereK_i is the carrying capacity for species_i; α_i2, α_i3, …, α_ij are the effects on speciesi of species 2, 3, …,j, respectively, andN_i,N₂,N_3, …,N_j are the densities of speciesi, 2, 3, …,j, respectively. In the text, we will refer to the product of α_ij andN_j as the limiting effect of speciesj on speciesi.

2.5. Indirect competition and species assemblage stability

The stability of the species assemblage (i.e., the local Lyapunov stability) was investigated with reference to (1) the community matrix (M), accounting for the pairwise competition coefficients (either β_ij or α_ij), and (2) the resulting classical Jacobian matrix (J) (May,1974a; Montoya, Woodward, Emmerson, & Solé,2009; Whittaker & Levin,1975). The stability of any given n‐species matrix can be inferred from its eigenvalues, with stability being expected for Jacobian matrices having only negative eigenvalues (in their real part) (Allesina & Tang,2012; May,1974b).

In order to account for the effects of both direct and indirect competition, the inverse of the Jacobian matrix (J⁻¹) was considered. Each element of J⁻¹ describes the net effect of speciesj on speciesi, taking into account all indirect pathways that link speciesi andj via intermediate competitors (Montoya et al.,2009; Wootton,2002). To account for the uncertainty in matrix elements, which is not directly accounted for in inverse Jacobian matrix calculation, we created modified inverse Jacobian matrices to simulate under‐ and overestimates of matrix elements by forcing a substantial (20%) decrease (matrix denoted as ↓J⁻¹) or increase (matrix denoted as ↑J⁻¹) in each matrix element. In addition, in order to test the effect on species assemblage stability of the distribution of interspecific interaction strengths, a further set of 10 random matrices (${\mathrm{J}}_{\mathrm{r}}^{-1}$) was created by randomly rearranging original off‐diagonal matrix elements. Here, we did not randomize the interspecific interaction strengths (or the isotopic distance and diet composition underlying competition strengths), as this would obviously produce a change in the eigenvalues of the matrix, which would not be informative in this case. On the other hand, randomization of the position of matrix elements made it possible to verify whether alternative stable configurations could exist given the observed magnitude of interaction strengths while ignoring their distribution between species pairs.

3.1. Isotopic distribution, trophic niche, and competition strength

Although TS density decreased from H to L, it was not significantly affected by meadow degradation (Kruskal–Wallis test,M. obtusatus: Hc = 3.24p = .20;A. nitescens: Hc = 3.21p = .20;C. truncata: Hc = 2.12p = .35) (Table 1). In contrast, NTS density decreased significantly from H to L (n° of individuals per litterbag: H = 10.7 ± 2.9, I = 4.5 ± 1.0, L = 3.8 ± 0.8; One‐way ANOVA,F = 4.35p < .05; evenness of nontarget specimens across litterbags in H = 0.81, in I = 0.88, in L = 0.92). Moving from H to L, the isotopic distribution of each TS varied (PERMANOVA,p < .01) (Figure 1), and both TA and SEAc increased (Table 1). As neither the mean nor the variance (σ²) of δ¹³C values differed between sampling sites within each location, for all TSs (Table S2), the values of each TS were pooled between sites and analyzed at the location scale for subsequent comparisons. For all TSs, σ² varied across locations and at the meadow scale (Levene's test for homogeneity of variances,p always <.05), whereas the σ² of resources did not vary significantly (Table S3). Specifically, from H to L, σ² increased by 68% forM. obtusatus, 89% forA. nitescens, and 84% forC. truncata, whereas the σ² of resources increased by just 22%, being lowest in I and highest in L. In addition, the variance of the mean δ¹³C of each TS between locations was higher than that of any resource item (Table S3). Similarly, across locations, CR increased by 25% forM. obtusatus, 40% forA. nitescens, and 57% forC. truncata, being lowest in H and highest in L, while the CR of resources increased by just 19%, being lowest in I and highest in L (Table S3). Neither the σ² nor the CR of TSs was affected by the number of isotopic observations (i.e., number of individuals; σ²:R² < .01,p = .99; CR:R² = .19,p = .15).

The MND_ii differed between TSs, but increased from H to L for all of them (two‐way ANOVA and post hoc comparison,p < .05), and was not related to their density (r² = .17,p = .18) (Table 1). At the meadow scale, MND_ii was higher than that of H and I but was similar to that of L (Kruskal–Wallis and Mann–Whitney comparisons,p < .05 for all TSs). Resource use varied both between species and between meadow locations (Figure 2). On average, trophic generalism in the use of resources was higher forC. truncata (Hs = 1.48 ± 0.10), intermediate forM. obtusatus (Hs = 0.90 ± 0.15) and lower inA. nitescens (Hs = 0.77 ± 0.13). Mean diet similarity between TS pairs was lowest in H and highest in I and L and at the meadow scale (Figure 2; one‐way ANOVA for repeated‐measures and Tukey's pairwise comparisons,p < .05). Accordingly, the variability (i.e., C.V.) in the consumption of each resource among TSs was highest in H and lowest in L and at the meadow scale (Figure 2).

Direct interspecific competition between TSs increased in strength from H to L, the only exception being the effect ofA. nitescens onM. obtusatus, which weakened with fallingP. oceanica coverage (Figure 3). The values of α_ij and β_ij were strongly related (Figure 3), displaying the same pattern of variation in competition strength along the meadow coverage gradient (r² = .88,p < .0001, slope = 0.91, 95% confidence interval on slope: 0.71, 1.10;p slope equal to 1 = 0.41; intercept = 0.12, 95% confidence interval on intercept: −0.02, 0.20). On average (i.e., across locations),A. nitescens had the strongest (α_ij = 0.84 ± 0.10),M. obtusatus the intermediate (α_ij = 0.78 ± 0.14), andC. truncata the weakest (α_ij = 0.59 ± 0.10) competitive effect on the two remaining TSs (Kruskal–Wallis and Mann–Whitney pairwise comparisons,p < .05). The mean strength of competition with NTSs did not vary with meadow degradation forM. obtusatus, whereas it increased significantly from H to L for bothA. nitescens andC. truncata (Table 1). Considering the mean effect of each NTS on the three TSs, the α_ij values decreased with increasing NTS density (Fig. S2). This pattern was also observed when considering the effect of NTSs on bothA. nitescens andM. obtusatus in isolation, whereas it was less evident forC. truncata (Fig. S2).

Competition betweenM. obtusatus (M) andA. nitescens (A) was always asymmetrical, regardless of the degree of meadow coverage, with α_MA > α_AM in H and I (t‐test,p < .01), and the opposite in L (t‐test,p < .01). In contrast, competition betweenC. truncata andA. nitescens, and betweenM. obtusatus andC. truncata, was asymmetrical only in H and I, respectively, withC. truncata being the subordinate competitor in both cases (t‐test,p < .01). Higher α_ij values between TS pairs were reflected in higher overlaps between isotopic TAs (r² = .36,p < .01), but not in higher overlaps between SEAcs (r² = .10,p > .05) (Table S4). Accordingly, the overlap between TAs was positively correlated with interpopulation diet similarity (r² = .59,p < .01) and inversely correlated with variability in the consumption of each resource at each location (r² = .97,p < .01). The α_ij measured at the meadow scale was significantly higher than the α_ij measured at the location scale for four of six species–pair interactions (Fig. S3, pairedt‐test,p always <.05). Accordingly, the overlap between TAs was higher at the meadow scale than at the location scale (pairedt‐test,t = 6.4,p < .01) (Table S4).

3.2. Carrying capacity and species assemblage stability

Carrying capacity (K) decreased from H to L for all TSs (Figure 4). The mean population density across meadow locations was 40.4 ± 7.2% of K forM. obtusatus, 32.5 ± 4.2% forA. nitescens, and 17.2 ± 1.8% forC. truncata. The limiting effect on each TS of the other two TSs increased with its MND_ii (r² = .62,p < .01) and was not related to population density (r² = .004,p = .87). On the other hand, the limiting effect of NTSs was related neither to TS MND_ii (r² = .20,p = .22) nor to TS density (r² = .02,p = .71), but increased with the evenness of the target specimens’ spatial distribution (i.e., across litterbags) at the location scale (r² = .46,p < .05). For all possible pairwise interactions between TSs, stable equilibria (as inferred from two‐species competition models, i.e., whenK_i/α_ij > K_j andK_j/α_ji > K_i) were expected in H and I, but not in L or at the meadow scale (Fig. S4).

The mean competition strength in the community matrix (M) with reference to both β_ij and α_ij pairwise competition coefficients was lowest in H and highest in L (Table 2). At the meadow scale, mean competition strength was higher than in H and was similar to competition strength in I (β_ij) and L (α_ij). Considering both β_ij and α_ij, the leading eigenvalue (λ) of the Jacobian matrix (J) increased from H to L and was >0 at the meadow scale (Table 2). Net competition effects increased species assemblage stability at the location scale (in H, I and L λ⁻¹ was always <λ), but not when considering the entire meadow (Table 2). The analysis of both ↓J⁻¹ and ↑J⁻¹ gave similar results to those obtained with J⁻¹ (Table 2). In contrast, the analysis of${\mathrm{J}}_{\mathrm{r}}^{-1}$ failed to replicate the results obtained with J⁻¹. Indeed,${\mathrm{\lambda }}_{\mathrm{r}}^{-1}$ was <0 in H (${\mathrm{\lambda }}_{\mathrm{r}}^{-1}$ < 0 for six and five random matrices considering β_ij and α_ij, respectively) and I (${\mathrm{\lambda }}_{\mathrm{r}}^{-1}$ < 0 for three and four random matrices considering β_ij and α_ij, respectively), whereas${\mathrm{\lambda }}_{\mathrm{r}}^{-1}$ was always >0 in L (Table 2 andTable S5).

4.1. Stable isotopes and competition strength

The overlap in resource use (β_ij) and the isotopic similarity of populations based on intra‐ and interspecific individual isotopic distances (α_ij) gave equivalent results in terms of the strength and pattern of competitive interactions along the habitat loss gradient. This was also reflected in the direct and indirect effects of competition on species assemblage stability, obtained with direct and inverse Jacobian matrices, respectively. Bayesian mixing models take account of uncertainty in both consumer and resource isotopic signatures, as well as in isotopic fractionation occurring between them. While in the computation of α_ij, it is not possible to account for potential interindividual variability in isotopic fractionation for each population, such potential variability should be considered similar across locations given the spatial scale investigated, and thus does not represent an obstacle to the comparison of populations’ isotopic data across locations (Araújo et al.,2007). In addition, the consistent information obtained by means of Euclidean distances and Bayesian isotopic mixing models can be considered an indirect test of the robustness of our interpretation of results based on isotopic differences between specimens.

α_ij proposed here and β_ij as originally proposed by Levins share the same theoretical basis and characteristics. Indeed, (1) α_ij increases as intraspecific dissimilarity increases with respect to interspecific dissimilarity; (2) MND_ij and MND_ji are equivalent, whereas differences in MND_ii and MND_jj make α_ij and α_ji asymmetric; and (3) α_ij approaches 1 only when intra‐ and interspecific dissimilarity is equivalent. In turn, (4) α_ii (i.e., the term on the diagonal in the community matrix) is always 1, regardless of niche width and the number of resources consumed by speciesi. Unlike β_ij, the calculation of α_ij does not require the isotopic characterization of potential food sources, nor does it require knowledge of the proportional contribution of each food item to the diet of each consumer. Indeed, such information is carried within the isotopic signature of each specimen, which is the result of the relative contribution of all trophic pathways converging in that organism (Bentivoglio et al.,2015; Careddu et al.,2015; Post,2002; Rossi et al.,2015), thereby explaining the positive isometric correlation between β_ij and α_ij.

Notably, the use of α_ij makes it possible to infer competition strength from specimens’ isotopic signatures when: (1) not all potential food sources are known, (2) all potential food sources are known but they are missing from the dataset or cannot be sampled, such as resources obtained by consumers in deep or extreme habitats; (3) medium‐ to long‐term description of the diet based on stomach content analysis or direct observation would be prohibitively difficult, broadening the range of animal groups for which field‐based competition coefficients can be obtained. In addition, calculation of α_ij makes it possible to quantify both asymmetry in species–pair interactions and the effect of competition on the carrying capacity for each population, based on a measure of isotopic similarity which yields time‐integrated information on foraging behavior at the organism level.

4.2.Posidonia oceanica habitat loss, niche overlap, and interspecific competition

Considering the three most abundant species in the invertebrate community (i.e., the TSs), the proportional overlap between their isotopic total areas reflected diet similarity and competition strength, which increased with habitat degradation. This was not observed when considering the overlap between standard ellipse areas encompassing the core of the isotopic niche, in accordance with classical niche theory, which does not expect high overlap on the central part of a given resource axis under limiting conditions (Whittaker & Levin,1975). Asymmetric competition characterized five of nine target species–pair interactions, withP. oceanica coverage reduction even inverting the outcome of competition betweenA. nitescens andM. obtusatus. Real food webs are expected to exhibit asymmetry in interspecific interactions, potentially stabilizing ecological communities (Rooney et al.,2006). In this case, TS trophic‐functional traits (e.g., intraspecific isotopic similarity), but not demographic ones (e.g., density), were related to differences in competition strength between populations, representing a key aspect linking habitat degradation and species assemblage organization along the disturbance gradient (Boström‐Einarsson et al.,2014; Valiente‐Banuet et al.,2015).

As with competition between TSs, the mean strength of competition between TSs and less abundant (i.e., nontarget) species increased withP. oceanica coverage reduction, suggesting that meadow degradation had a widespread effect within the invertebrate community. At the location scale, the limiting effect of NTSs increased with the evenness of the spatial distribution of specimens, which increased with decliningP. oceanica coverage. This is consistent with the expectation of increased home‐range exploration by consumers in conditions of low‐quality resource patches (Calizza et al.,2012; Pyke et al.,1977). A more even distribution of specimens within their foraging home range as a result of habitat degradation and resource depletion implies a higher probability of species co‐occurrence on each given resource patch. This leads to isotopically close nonconspecifics as a result of increased home range exploration and overlap of resources encountered during harvesting (Basset,1995; Pyke et al.,1977; Rossi et al.,2015).

4.3. Competition, stability, and spatial scale of interactions

Competition between TSs was stronger when measured at the meadow scale than at the location scale, due to increased isotopic and diet similarity when not accounting for the trophic segregation of populations between locations. Spatial patterns and compartmentalization in species interactions have been shown to promote species coexistence, in conjunction with habitat complexity and differences between competitors in home‐range harvesting behavior (Basset,1995; Durrett & Levin,1998). At the species assemblage level, stability decreased in the low‐coverage location, due to stronger mean competition and lower carrying capacity. These results imply that (1) habitat degradation lowered the species assemblage's potential to “absorb” environmental changes without destabilization and that (2) although related toP. oceanica coverage, stability was low (i.e., λ was negative but close to 0) when considering the direct outcomes of competition alone. At the location scale, the net effect of competition (Basset & Rossi,1990; Lawlor,1979) was to increase species assemblage stability, mitigating the negative effect of habitat degradation. Indirect responses mediated by species interactions have the potential to exceed the direct effect of environmental changes at both population and community levels, leading to disturbance propagation or attenuation within the community (Bewick et al.,2014; Montoya et al.,2009). Here, the net effect of species interactions was to reverse the outcomes of competition on population dynamics and stabilize communities (Lawlor,1979; Montoya et al.,2009). The fact that these results were not affected by potential under‐ or overestimation of matrix elements is indicated by the similar results obtained from the analysis of ↓J⁻¹ and ↑J⁻¹. On the other hand, the stabilizing inverse relationship between direct and indirect competition became weaker with meadow degradation and was not observed at the meadow scale (seeTables S6 and S7 in the online supplementary material).

Lastly, the random rearrangement of original matrix elements in${\mathrm{J}}_{\mathrm{r}}^{-1}$ provided alternative stable configurations of the species assemblage at high and intermediateP. oceanica coverage, but not at lowP. oceanica coverage, where none of the random configurations satisfied the criteria for stability (i.e., in L,${\mathrm{\lambda }}_{\mathrm{r}}^{-1}$ was always >0). The distribution of interaction strengths between species has been shown to play a major role in community persistence (Montoya et al.,2009; Tang, Pawar, & Allesina,2014). Our results suggest that under lowP. oceanica coverage, the species assemblage may have a low probability of recovering a stable configuration following changes in interspecific interactions, such as those potentially associated with species invasion or local extinction. This may have important implications for biodiversity organization and persistence in degraded seagrass meadows, where both habitat degradation and species invasion are expected to increase in the near future (Hemminga & Duarte,2000; Orth et al.,2006).

1. Abrams, P. A. (1977). Density‐independent mortality and interspecific competition: A test of Pianka's niche overlap hypothesis. The American Naturalist, 111, 539–552. [Google Scholar]
2. Allesina, M. S., & Tang, S. (2012). Stability criteria for complex ecosystems. Nature, 483, 205–208. [DOI] [PubMed] [Google Scholar]
3. Araújo, M. S., Bolnick, D. I., Machado, G., Giaretta, A. A., & dos Reis, S. F. (2007). Using δ¹³C stable isotopes to quantify individual‐level diet variation. Oecologia, 152, 643–654. [DOI] [PubMed] [Google Scholar]
4. Araújo, M. S., Guimarães, P. R.Jr, Svanbäck, R., Pinheiro, A., Guimarães, P., dos Reis, S. F., & Bolnick, D. I. (2008). Network analysis reveals contrasting effects of intraspecific competition on individual vs. population diets. Ecology, 89, 1981–1993. [DOI] [PubMed] [Google Scholar]
5. Araújo, M. S., Langerhans, R. B., Giery, S. T., & Layman, C. A. (2014). Ecosystem fragmentation drives increased diet variation in an endemic live bearing fish of the Bahamas. Ecology and Evolution, 4, 3298–3308. [DOI] [PMC free article] [PubMed] [Google Scholar]
6. Auer, S. K., & Martin, T. E. (2013). Climate change has indirect effects on resource use and overlap among coexisting bird species with negative consequences for their reproductive success. Global Change Biology, 19, 411–419. [DOI] [PubMed] [Google Scholar]
7. Bašić, T., & Britton, R. (2016). Characterizing the trophic niches of stocked and resident cyprinid fishes: Consistency in partitioning over time, space and body sizes. Ecology and Evolution, 6, 5093–5104. [DOI] [PMC free article] [PubMed] [Google Scholar]
8. Basset, A. (1995). Body size‐related coexistence: An approach through allometric constraints on home range use. Ecology, 76, 1027–1035. [Google Scholar]
9. Basset, A., & Rossi, L. (1990). Competitive trophic niche modifications in three populations of detritivores. Functional Ecology, 4, 685–669. [Google Scholar]
10. Beckerman, A. P., Petchey, O. L., & Warren, P. H. (2006). Foraging biology predicts food web complexity. Proceedings of the National Academy of Sciences USA, 103, 13745–13749. [DOI] [PMC free article] [PubMed] [Google Scholar]
11. Bell, S. S., Brooks, R. A., Robbins, B. D., Fonseca, M. S., & Hall, M. O. (2001). Faunal response to fragmentation in seagrass habitats: Implication for seagrass conservation. Biological Conservation, 10, 115–123. [Google Scholar]
12. Bentivoglio, F., Calizza, E., Rossi, D., Carlino, P., Careddu, G., Rossi, L., & Costantini, M. L. (2015). Site‐scale isotopic variations along a river course help localize drainage basin influence on river food webs. Hydrobiologia, 770, 257–272. [Google Scholar]
13. Bewick, S., Stuble, K. L., Lessard, J. P., Dunn, R. R., Adler, F. R., & Sanders, N. L. (2014). Predicting future coexistence in a North American ant community. Ecology and Evolution, 4, 1804–1819. [DOI] [PMC free article] [PubMed] [Google Scholar]
14. Bolnick, D. I. (2001). Intraspecific competition favours niche width expansion in Drosophila melanogaster. Nature, 410, 463–466. [DOI] [PubMed] [Google Scholar]
15. Borrelli, J. J., Allesina, S., Amarasekare, P., Arditi, R., Chase, I., Damuth, J., … Ginzburg, L. R. (2015). Selection on stability across ecological scales. Trends in Ecology and Evolution, 30, 417–425. [DOI] [PubMed] [Google Scholar]
16. Boström, C., Jackson, E. L., & Simenstad, C. A. (2006). Seagrass landscapes and their effects on associated fauna: A review. Estuarine, Coastal and Shelf Science, 68, 383–403. [Google Scholar]
17. Boström‐Einarsson, L., Bonin, M. C., Munday, P. L., & Jones, J. P. (2014). Habitat degradation modifies the strength of interspecific competition in coral dwelling damselfishes. Ecology, 95, 3056–3067. [Google Scholar]
18. Bowden, D. A., Rowden, A. A., & Attrill, M. J. (2001). Effect of patch size and in‐patch location on the infaunal macroinvertebrate assemblages ofZostera marina seagrass beds. Journal of Experimental Marine Biology and Ecology, 259, 133–154. [DOI] [PubMed] [Google Scholar]
19. Brown, J. H., Gillooly, J. F., Allen, A. P., Savage, V. M., & West, G. B. (2004). Toward a metabolic theory of ecology. Ecology, 85, 1771–1789. [Google Scholar]
20. Calizza, E., Costantini, M. L., Carlino, P., Bentivoglio, F., Orlandi, L., & Rossi, L. (2013).Posidonia oceanica habitat loss and changes in litter‐associated biodiversity organization: A stable isotope‐based preliminary study. Estuarine Coastal and Shelf Science, 135, 137–145. [Google Scholar]
21. Calizza, E., Costantini, M. L., & Rossi, L. (2015). Effect of multiple disturbances on food web vulnerability to biodiversity loss in detritus‐based systems. Ecopshere, 6, 1–20. [Google Scholar]
22. Calizza, E., Costantini, M. L., Rossi, D., Carlino, P., & Rossi, L. (2012). Effects of disturbance on an urban river food web. Freshwater Biology, 57, 2613–2628. [Google Scholar]
23. Calizza, E., Rossi, L., & Costantini, M. L. (2013). Predators and resources influence phosphorus transfer along an invertebrate food web through changes in prey behaviour. PLoS ONE, 8, e65186. doi:10.1371/journal.pone.0065186 [DOI] [PMC free article] [PubMed] [Google Scholar]
24. Careddu, G., Costantini, M. L., Calizza, E., Carlino, P., Bentivoglio, F., Orlandi, L., & Rossi, L. (2015). Effects of terrestrial input on macrobenthic food webs of coastal sea are detected by stable isotope analysis in Gaeta Gulf. Estuarine Coastal and Shelf Science, 154, 158–168. [Google Scholar]
25. Corman, A. M., Mendel, B., Voigt, C. C., & Garthe, S. (2015). Varying foraging patter ns in response to competition? A multicolony approach in a generalist seabird. Ecology and Evolution, 6, 974–986. [DOI] [PMC free article] [PubMed] [Google Scholar]
26. Costantini, M. L., Rossi, L., Fazi, S., & Rossi, D. (2009). Detritus accumulation and decomposition in a coastal lake: Acquatina‐southern Italy. Aquatic Conservation: Marine and Freshwater Ecosystems, 19, 566–574. [Google Scholar]
27. Costantini, M. L., Zaccarelli, N., Mandrone, S., Rossi, D., Calizza, E., & Rossi, L. (2012). NDVI spatial pattern and the potential fragility of mixed forested areas in volcanic lake watersheds. Forest Ecology and Management, 285, 133–141. [Google Scholar]
28. Durrett, R., & Levin, S. (1998). Spatial aspects of interspecific competition. Theoretical Population Biology, 53, 30–43. [DOI] [PubMed] [Google Scholar]
29. Flaherty, E. A., & Ben‐David, M. (2010). Overlap and partitioning of the ecological and isotopic niches. Oecologia, 119, 1409–1416. [Google Scholar]
30. Fry, B. (2006). Stable isotope ecology. Berlin, Germany: Springer. [Google Scholar]
31. Heck, K. L., & Orth, R. J. (2006). Predation in seagrass beds In Larkum A. W. D., Orth R. J. & Duarte C. M. (Eds.), Seagrasses: Biology, ecology, and conservation. Dordrecht, the Netherlands: Springer. [Google Scholar]
32. Hemminga, M. A., & Duarte, C. M. (2000). Seagrass ecology. Cambridge, UK: Cambridge University Press. [Google Scholar]
33. Henschke, N., Everett, J. D., Suthers, I. M., Smith, J. A., Hunt, B. P. V., Doblin, M. A., & Taylor, M. D. (2015). Zooplankton trophic niches respond to different water types of the western Tasman Sea: A stable isotope analysis. Deep‐Sea Research I, 104, 1–8. [Google Scholar]
34. Holdridge, E. M., Cuellar‐Gempeler, C., & terHost, C. P. (2016). A shift from exploitation to interference competition with increasing density affects population and community dynamics. Ecology and Evolution, 6, 5333–5341. [DOI] [PMC free article] [PubMed] [Google Scholar]
35. Jackson, A. L., Inger, R., Parnell, A. C., & Bearhop, S. (2011). Comparing isotopic niche widths among and within communities: SIBER – Stable Isotope Bayesian Ellipses in R. Journal of Animal Ecology, 80, 595–602. [DOI] [PubMed] [Google Scholar]
36. Jona‐Lasinio, G., Costantini, M. L., Calizza, E., Pollice, A., Bentivoglio, F., Orlandi, L., … Rossi, L. (2015). Stable isotope‐based statistical tools as ecological indicator of pollution sources in Mediterranean transitional water ecosystems. Ecological Indicators, 55, 23–31. [Google Scholar]
37. Kondoh, M. (2003). Foraging adaptation and the relationship between food‐web complexity and stability. Science, 299, 1388–1381. [DOI] [PubMed] [Google Scholar]
38. Krauss, J., Bommarco, R., Guardiola, M., Heikkinen, R. K., Helm, A., Kuussaari, M., … Steffan‐Dewenter, I. (2010). Habitat fragmentation causes immediate and time‐delayed biodiversity loss at different trophic levels. Ecology Letters, 13, 507–605. [DOI] [PMC free article] [PubMed] [Google Scholar]
39. di Lascio, A., Rossi, L., & Costantini, M. L. (2011). Different temperature tolerance of northern and southern European populations of a freshwater Isopod Crustacean species (Asellus aquaticus L.). Fundamental and Applied Limnology, 179, 193–201. [Google Scholar]
40. Lawlor, L. R. (1979). Direct and indirect effect of n‐species competition. Oecologia, 43, 355–364. [DOI] [PubMed] [Google Scholar]
41. Layman, C. A., Arrington, D. A., Montaña, C. G., & Post, D. M. (2007). Can stable isotope ratios provide for community‐wide measures of trophic structure?Ecology, 88, 42–48. [DOI] [PubMed] [Google Scholar]
42. Layman, C. A., Qattrocchi, J. P., Peyer, C. M., & Allgeier, J. E. (2007). Niche width collapse in a resilient top predators following ecosystem fragmentation. Ecology Letters, 10, 937–944. [DOI] [PMC free article] [PubMed] [Google Scholar]
43. Levins, R. (1968). Evolution in changing environments: Some theoretical explorations. Princeton, NJ: Princeton University Press. [Google Scholar]
44. Lima, S. L. (2002). Putting predators back into behavioural predator‐prey interactions. Trends in Ecology and Evolution, 17, 70–75. [Google Scholar]
45. Madeira, F., di Lascio, A., Carlino, P., Costantini, M. L., & Pons, X. (2013). Change in carbon stable isotope ratios of the predatory bugOrius majusculus after dietary shifts. Entomologia Experimentalis et Applicata, 148, 287–296. [Google Scholar]
46. Mancinelli, G., Costantini, M. L., & Rossi, L. (2007). Top‐down control of reed detritus processing in a lake littoral zone: Experimental evidence of a seasonal compensation between fish and invertebrate predation. International Review of Hydrobiology, 92, 117–134. [Google Scholar]
47. Mancinelli, G., & Vizzini, S. (2015). Assessing anthropogenic pressures on coastal marine ecosystems using stable CNS isotopes: State of the art, knowledge gaps, and community‐scale perspectives. Estuarine Coastal and Shelf Science, 156, 195–204. [Google Scholar]
48. May, R. M. (1974a). On the theory of niche overlap. Theoretical Population Biology, 5, 297–332. [DOI] [PubMed] [Google Scholar]
49. May, R. M. (1974b). Stability and complexity in model ecosystems. Princeton, NJ: Princeton University Press. [Google Scholar]
50. McCutchan, J. H., Lewis, W. M., Kendall, C., & McGrath, C. C. (2003). Variation in trophic shift for stable isotope ratios of carbon, nitrogen, and sulfur. Oikos, 102, 378–390. [Google Scholar]
51. Montoya, J. M., Woodward, G., Emmerson, M. E., & Solé, R. V. (2009). Press perturbations and indirect effects in real food webs. Ecology, 90, 2426–2433. [DOI] [PubMed] [Google Scholar]
52. Munday, P. L. (2004). Habitat loss, resource specialization, and extinction on coral reefs. Global Change Biology, 10, 1642–1647. [Google Scholar]
53. O'Gorman, E. J., Fitche, J. E., & Crowe, T. P. (2012). Multiple anthropogenic stressors and the structural properties of food webs. Ecology, 93, 441–448. [DOI] [PubMed] [Google Scholar]
54. Orlandi, L., Bentivoglio, F., Carlino, P., Calizza, E., Rossi, D., Costantini, M. L., & Rossi, L. (2014). δ¹⁵N variation inUlva lactuca as a proxy for anthropogenic nitrogen inputs in coastal areas of Gulf of Gaeta (Mediterranean Sea). Marine Pollution Bulletin, 84, 76–82. [DOI] [PubMed] [Google Scholar]
55. Orth, R. G., Carruthers, T. J. B., Denninson, W. C., Duarte, C. M., Fourqueran, J. V., Heck, K. L.Jr, … Williams, S. (2006). A global crisis for seagrass ecosystems. BioSciences, 56, 987–996. [Google Scholar]
56. Parnell, A. C., Phillips, D. L., Bearhop, S., Semmens, B. X., Ward, E. J., Moore, J. W., … Inger, R. (2013). ayesian stable isotope mixing models. Environmetrics, 24, 387e399. doi:10.1371/journal.pone.0009672 [Google Scholar]
57. Paticchio, N. (2013). Meccanismi di regressione e frammentazione dell'habitat nelle praterie di Posidonia oceanica. PhD dissertation,Sapienza University of Rome, Rome, 259 pp.
58. Pianka, E. R. (1969). Sympatry of desert lizards (Ctenotus) in Western Australia. Ecology, 50, 1012–1030. [Google Scholar]
59. Pianka, E. R. (1974). Niche overlap and diffuse competition. Proceedings of the National academy of Sciences USA, 71(2141), 2145. [DOI] [PMC free article] [PubMed] [Google Scholar]
60. Post, D. M. (2002). Using stable isotopes to estimate trophic position: Models, methods, and assumptions. Ecology, 83, 703–718. [Google Scholar]
61. Pyke, G. H., Pulliam, H. R., & Charnov, E. L. (1977). Optimal foraging: A selective review of theory and tests. Quaternary Review of Biology, 52, 137–154. [Google Scholar]
62. R Development Core Team(2012) R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. [Google Scholar]
63. Rooney, N., McCann, K., Gellner, G., & Moore, J. C. (2006). Structural asymmetry and the stability of diverse food webs. Nature, 442, 265–269. [DOI] [PubMed] [Google Scholar]
64. Rossi, L., Costantini, M. L., & Brilli, M. (2007). Does stable isotope analysis separate transgenic and traditional corn (Zea mays L.) detritus and their consumers?Applied Soil Ecology, 35, 449–453. [Google Scholar]
65. Rossi, L., Costantini, M. L., Carlino, P., di Lascio, A., & Rossi, D. (2010). Autochthonous and allochthonous plant contributions to coastal benthic detritus deposits: A dual‐stable isotope study in a volcanic lake. Aquatic Sciences, 72, 227–236. [Google Scholar]
66. Rossi, L., di Lascio, A., Carlino, P., Calizza, E., & Costantini, M. L. (2015). Predator and detritivore niche width helps to explain biocomplexity of experimental detritus‐based food webs in four aquatic and terrestrial ecosystems. Ecological Complexity, 23, 14–24. [Google Scholar]
67. Sanders, D., Vogel, E., & Knop, E. (2015). Individual and species‐specific traits explain niche size and functional role in spiders as generalist predators. Journal of Animal Ecology, 84, 134–142. [DOI] [PubMed] [Google Scholar]
68. SCBD‐Secretariat of the Convention on Biological Diversity(2010) Global biodiversity outlook 3. Montréal: SCBD‐Secretariat of the Convention on Biological Diversity. [Google Scholar]
69. Shea, K., & Chesson, P. (2002). Community ecology theory as a framework for biological invasions. Trends in Ecology and Evolution, 17, 170–176. [Google Scholar]
70. Svanbäck, R., & Bolnick, D. I. (2007). Intraspecific competition drives increased resource use diversity within a natural population. Proceedings of the Royal Society B: Biological Sciences, 274, 839–844. [DOI] [PMC free article] [PubMed] [Google Scholar]
71. Swanson, H. K., Lysy, M., Power, M., Stasko, A. D., Johnson, J. D., & Reist, J. D. (2015). A new probabilistic method for quantifying n‐dimensional ecological niches and niche overlap. Ecology, 96, 318–324. [DOI] [PubMed] [Google Scholar]
72. Tang, S., Pawar, S., & Allesina, S. (2014). Correlation between interaction strengths drives stability in large ecological networks. Ecology Letters, 17, 1094–1100. [DOI] [PubMed] [Google Scholar]
73. Valiente‐Banuet, A., Aizen, M. A., Alcántara, J. M., Arroyo, J., Cocucci, A., Galetti, M., … Zamora, R. (2015). Beyond species loss: The extinction of ecological interactions in a changing world. Functional Ecology, 29, 299–307. [Google Scholar]
74. Valladares, G., Cagnolo, L., & Salvo, A. (2012). Forest fragmentation leads to food web contraction. Oikos, 121, 299–305. [Google Scholar]
75. Van der Heide, T., van Nes, E. H., Geerling, G. W., Smolders, A. J. P., Bouma, T. J., & van Katwijk, M. M. (2007). Positive feedbacks in seagrass ecosystems – Implications for success in conservation and restoration. Ecosystems, 10, 1311–1322. [Google Scholar]
76. Vizzini, S., Sarà, G., Michener, R. H., & Mazzola, A. (2002). The role and contribution of the seagrassPosidonia oceanica L. Delile organic matter for secondary consumers as revealed by carbon and nitrogen stable isotope analysis. Acta Oecologia, 23, 277–285. [Google Scholar]
77. Whittaker, R., & Levin, S. (1975). Niche: Theory and application. Stroudsbourg, PA: Dowden, Hutchison and Ross, Inc.. [Google Scholar]
78. Wootton, J. T. (2002). Indirect effects in complex ecosystems: Recent progress and future challenges. Journal of Sea Research, 48, 157–172. [Google Scholar]
79. Yao, X., Huang, G., Xie, P., & Xu, J. (2016) Trophic niche differences between coexisting omnivores silver carp and bighead carp in a pelagic food web. Ecological Research, 31, 831–839. [Google Scholar]

Supplementary Materials