Assessing the Contribution of Posidonia oceanica to Mediterranean Secondary Production Through Stable Isotope Analysis

Abstract

The role of seagrasses in providing a complex habitat for marine organisms is globally documented; however, few studies have investigated the trophic incorporation of endemic Mediterranean Posidonia oceanica into marine food webs. Meadows of P. oceanica are declining due to climate change and anthropogenic pressures, emphasising the need to determine its contribution in local trophic dynamics. We investigated whether benthic marine invertebrate (BMI) and fish consumers assimilate carbon directly from P. oceanica seagrass or other sources along the seagrass meadow margins in Malta. We sampled and analysed the δ¹³C and δ¹⁵N isotope values of P. oceanica, particulate organic matter (POM), macroalgae, 14 invertebrate taxa, and 10 fishes at three locations marginal to P. oceanica seagrass meadows. Stable isotope ratios were significantly different between all taxa (F₂₆ = 17.37, R² = 0.68, p < 0.01) and locations (F₂ = 34.22, R² = 0.10, p < 0.01). The source, invertebrate, and fishes were enriched in both ¹³C and ¹⁵N at Baħar iċ-Ċagħaq relative to the other locations, L’Aħrax and Golden Bay, likely due to the increased effluent. Stable isotope mixing models were somewhat confounded as POM and macroalgae had similar δ¹³C and δ¹⁵N values at each site, hampering efforts to define the resource use of the sampled taxa. However, Posidonia oceanica made the lowest contribution for both consumer groups at all locations, consistent with the results of other Mediterranean studies, suggesting that P. oceanica does not contribute significantly to the diet of consumers at seagrass meadow margins within Maltese waters.

1. Introduction

Seagrasses are vascular marine plants which are classed as keystone species within temperate and tropical regions around the world by virtue of the diverse array of marine ecosystem services they provide [1,2]. Seagrass meadows are ecologically important, providing a habitat, food supplies, and a refuge to marine species [3], and also serve an important socio-economic role by supporting the provision of ecosystem services in coastal waters such as carbon sequestration [4], water column oxygenation [5], the stabilisation of the sediment [6], coastal protection [7], the improvement of water quality, and potential sanitisation against disease [8]. Despite their critical ecological importance, seagrass meadows are undergoing a global decline in distribution as result of climate change, ocean warming and marine heatwaves [9,10,11,12,13], non-native species invasions [14,15], and increased habitat destruction, as well as pollution [16,17,18]. Climate change is also inducing range shifts of [9,19] and metabolic [20] changes in seagrasses. This highlights the need to further characterise their ecological role as well as the mechanisms for anticipating and mitigating impacts on seagrass meadows, whilst attempting to minimise the loss of ecosystem services.

In the Mediterranean Sea, Posidonia oceanica (L.) Delile is an endemic seagrass species usually occurring in a depth range of 0–40 m [21], depending on the in situ water turbidity, where it supports highly productive marine ecosystems [10,21,22]. Campagne et al. [23] estimate that the diverse ecosystem services provided by P. oceanica equate to a value of €284–514 per hectare per year. The documented decline in Mediterranean Sea biodiversity [9,11,12] further emphasises the importance of characterising a meadow’s consumer ecology so as to be able to predict how associated communities may respond to habitat loss or to a direct or indirect decrement in the carbon source.

The role of P. oceanica meadows in providing a habitat and substrate for marine organisms and as a sink of ‘blue’ carbon (and, thus, as possible climate change mitigation) is well-documented [21,24,25,26,27]; however, only a few studies have investigated their role within marine food webs, especially in the central Mediterranean [22,28,29]. Seagrasses in tropical areas are known to be eaten by mega grazers, including dugongs and sea turtles that feed directly on the leaves [30]. Temperate and Mediterranean seagrass meadows lack large grazers but harbour smaller invertebrates (borers) [31] and fishes, with members of the Sparidae family, including Sarpa salpa, as the most relevant [20,28]. Previous studies based on the stable isotope analysis (SIA) of Amphibolis griffithii and Posidonia sinuosa meadow communities in Western Australia and of Posidonia oceanica meadow communities in Italy in the central Mediterranean found that grazers predominantly assimilate carbon from sources other than the seagrasses themselves, including epiphytic algae and/or detritus containing particulate organic matter from pelagic sources [22,28,29]. Moreover, there is evidence that P. oceanica may act as an indirect food source through the detrital pathway [22] and provide dissolved organic carbon for nearby primary macroalgal production [32]. The present study contributes to addressing existing knowledge gaps revolving around the trophic contribution to multiple food web levels made by P. oceanica at sites where it is prevalent.

The present study aims to achieve this by determining whether invertebrate and fish consumers along the P. oceanica seagrass meadow margins in Malta are assimilating carbon directly from the seagrass itself or from other sources in and around these habitats. The SIA of tracers is a useful tool that has been widely used to resolve the food web structure and to determine the trophic contribution of sources to a mixture [22,33,34]. Our objectives were (i) to quantify and analyse ¹³C and ¹⁵N isotope signatures of P. oceanica and organisms associated with meadows at three different locations around Malta, and (ii) to quantify the trophic contribution of various potential sources, including P. oceanica seagrass, to consumers sampled at the same locations. It was hypothesised that P. oceanica does not provide a direct trophic resource to higher food web levels [22,28,29] and that other sources, such as macroalgae and pelagic carbon, make the highest contribution to consumer diets [22,35].

2. Materials and Methods

2.1. Study Area and Sample Collection

Sampling events took place during August 2018 along Posidonia oceanica meadow margins at depths of ca. 10 m at three locations around Malta: L’Aħrax (35.9980° N, 14.3667° W), Baħar iċ-Ċagħaq (35.9483° N, 14.4533° E) and Golden Bay (35.9381° N, 14.3310° E) (Figure 1). All three sampling locations were characterised by continuous and reticulate P. oceanica meadows settled on sand and on matte, with surface water temperatures recorded during the sampling expeditions ranging between 27 and 28 degrees Centigrade.

Benthic marine invertebrates (BMIs) were sampled directly by hand with the use of macrofauna hand nets on SCUBA and fishes were targeted using baited traps. Taxa associated with seagrass communities were identified prior to sampling events. Individuals of selected taxa were sampled randomly at each site and subsequently identified to species level where possible, otherwise to the highest taxonomic grouping (e.g., genus, family, etc.). For example, several polychaete worms were recorded using the paraphyletic group term ‘Polychaete’. Three Posidonia oceanica shoots were sampled through SCUBA diving at each location, along with a minimum of 50 g of vegetal material for macroalgal species. Due to the lack of plankton samples, Mytilus spp. filter feeders were used as a proxy for a pelagic source of carbon, referred to, here, from now on, as particulate organic matter (POM) [36]. Macroalgae and P. oceanica were scrubbed and rinsed in distilled water to remove any epiphytic growth. P. oceanica samples were subdivided into leaf, rhizome, and root sections to test for tissue related variability across all samples. Soft muscle tissue was dissected from all mollusc and crustacean samples and used for SIA. All producer and consumer samples were stored at −20 °C before being dried at 60 °C for 48 h.

2.2. Stable Isotope Analysis (IRMS)

Carbon (δ¹³C) and nitrogen (δ¹⁵N) stable isotope analysis took place at the Stable Isotopes in Nature Laboratory, University of New Brunswick, Canada. All samples were homogenised to a fine powder and subsamples (1 ± 0.1 mg for animal tissue, 3 ± 0.1 mg for plant and macroalgal samples) were placed in aluminium foil containers for analysis. Samples were flash-combusted at 900 °C in an ECS 4010 elemental analyser (Costech Analytical Technologies, Valencia, CA, USA). Carbon dioxide and nitrogen gases were separated on a gas chromatography column at 50 °C and the resulting gas was passed through a Delta Plus XP continuous-flow isotope ratio mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) for measurement of δ¹³C and δ¹⁵N values. All isotope data were reported as δ notation (‰) relative to Vienna Pee Dee Belemnite and atmospheric air for C and N, respectively. The instrumentation was calibrated to international standards of Vienna Pee Dee Belemnite carbonate and atmospheric nitrogen using International Atomic Energy Authority (IAEA, Vienna, Austria) certified materials CH-7 and N2 for carbon and nitrogen measurements, respectively. Duplicates of certain samples were run every 18 samples to ensure continuous accuracy of the instrumentation throughout the runs. The analytical error (mean ± SD: 0.1 ± 0.1‰) was calculated for both carbon and nitrogen based on repeat analyses of in-house standards: nicotinamide (δ¹³C and δ¹⁵N mean ± SD:  32.3 ± 0.1‰ and 2.0 ± 0.1‰), bovine liver (δ¹³C and δ¹⁵N:  18.8 ± 0.1‰ and 7.2 ± 0.1‰), muskellunge muscle (δ¹³C and δ¹⁵N:  22.3 ± 0.1‰ and 14.0 ± 0.1‰), and USGS 61—USGA, Reston, VA, USA (δ¹³C and δ¹⁵N:  34.8 ± 0.1‰ and  2.9 ± 0.1‰).

2.3. Statistical Analyses

The mean and standard deviation δ¹³C and δ¹⁵N of sampled producer and consumer taxa were calculated and visualised across sites using biplots. The mean and standard deviation δ¹³C and δ¹⁵N of P. oceanica rhizome, leaf, and root samples were also calculated and plotted for each study site to investigate potential variability in the isotopic signatures between tissues; PERMANOVA, a multivariate permutation-based analogue of Analysis of Variance [37], was used to assess variation in stable isotope ratios across two factors and their interaction term: ‘Taxon’—a fixed effect with three levels (‘Plant’—i.e., macroalgae, P. oceanica, and POM -, BMI, and fish), and ‘Location’—a random effect with three levels (L’Aħrax, Baħar iċ-Ċagħaq, and Golden Bay). PERMANOVAs were performed on Bray–Curtis similarity matrices derived from δ¹³C and δ¹⁵N stable isotope ratios of relevant organisms; δ¹³C values were multiplied by −1 and +5 was added to δ¹⁵N values, as Bray–Curtis matrix requires positive values. Analysis was carried out using the vegan package in R-4.4.2 [38].

Stable isotope mixing models were used to estimate the contribution of potential sources to BMI and fish consumers at each site [32,39]. Following the idiom ‘you are what you eat’, stable isotope mixing models provide a tool to quantify the relative contribution of multiple putative pre-organisms to a consumer [39]. To quantify trophic contribution of each source (macroalgae, P. oceanica seagrass, and POM) across the three sites, and to discern whether P. oceanica seagrass makes a notable contribution to seagrass community consumers, two three-source mixing models were carried out in R using the MixSIAR package [33]. To obtain POM SI values, we subtracted the fractionation values, +2.0 and +3.6 for ¹³C and ¹⁵N, respectively [36], from the Mytilus spp. values. Each model contained site species values for each of the three prey sources; trophic fractionation was assumed to be 1.3 (±0.4)‰ and 4.4 (±1)‰ for δ¹³C and δ¹⁵N, respectively (post 2002). The first model quantified the trophic contribution of sources to BMI across the three sample sites, and the second quantified trophic contribution of sources to fishes. Fishes were only sampled at Baħar iċ-Ċagħaq and Golden Bay; therefore, the model only applies to these two locations, and trophic fractionation values and associated uncertainties were doubled when estimating fish resource use to account for additional trophic steps.

3. Results

The sampled taxa included 3 carbon sources, 14 benthic marine invertebrate (BMI) consumers, and 10 fish consumers. A total of 153 samples were collected and processed for stable isotope analysis. These included potential carbon sources, macroalgae (n = 45), and Posidonia oceanica seagrass (n = 36), and BMI consumers (n = 38) including Mytilus spp., as well as fish consumers (n = 31) across two of the three sample sites (

Table 1. Mean (±standard deviation) and number of observations (n) of δ¹³C and δ¹⁵N values of sampled taxa from Posidonia oceanica margin communities at three sampling locations around Malta, L’Aħrax, Baħar iċ-Ċagħaq, and Golden Bay.

Taxon	L’Aħrax			Baħar			Golden Bay
	n	δ13C	δ15N	n	δ13C	δ15N	n	δ13C	δ15N
Carbon sources
Macroalgae	12	−18.3 (0.3)	1.3 (0.7)	7	−18.6 (2.2)	4 (0.9)	26	−21.9 (3.3)	0.8 (1)
Marine POM *	1	−20.3	2.1	1	−19.9	4.6	1	−19.1	1.5
Posidonia oceanica	15	−15.6 (0.8)	0.5 (0.5)	7	−14.4 (0.6)	6 (1.5)	14	−15.8 (1.1)	1.0 (0.6)
Invertebrates
Amphipod	1	−18.6	2.7				1	−20.6	2.3
Decapod	4	−16.7 (1.4)	3.6 (0.8)	1	−16.9	4.1	1	−16.2	3.9
Echinoderm	1	−16.7	4.7
Halochynthis papillosa	1	−21.3	2.7
Isopod	1	−18.2	4.4
Polychaete	2	−21.2 (1.5)	3.5 (1.8)	1	−17.6	8.9	5	−19.5 (1.2)	5.6 (1.6)
Porifera	1	−19.1	4.3				5	−18.4 (1.1)	4.3 (0.9)
Sipunculid	1	−16.9	4.8				2	−19.6 (0.4)	3.6 (0.6)
Cirriped				1	−17.2	8.1	1	−18.6	6
Hermodice carunculata				2	−18.2 (1.6)	9.1 (0.3)	2	−18.8 (0.7)	8.8 (2.2)
Hexaplex sp.				1	−18.1	10.2
Mollusc							1	−17.8	2.4
Pagurid							1	−17.2	3
Spirastrella sp.							1	−20.2	4.2
Fishes
Chromis chromis				4	−18.8 (0.2)	7.8 (0.4)	5	−19.5 (0.2)	6.7 (0.2)
Coris julis				4	−15.9 (0.2)	12.9 (0.8)
Diplodus vulgaris				2	−16.2 ± (1)	13.9 (1.4)	2	−17.2 (0.1)	9.2 (0.8)
Oblada melanura				3	−17.6 ± (0.2)	10.2 (0.5)
Labrus viridis							1	−17.2	10.6
Sciana umbra							1	−17.2	7.9
Scorpaena sp.							1	−18.1	8.1
Serranus cabrilla							3	−18.6 (0.3)	8.1 (0.3)
Sparisoma cretense							4	−19.9 (0.5)	5.5 (0.2)
Uranoscopus scaber							1	−18.5	8

* Derived from Mytilus spp. values.

). The P. oceanica tissue sampled at each of the three sites consisted of leaves, rhizomes, and roots, with the quantities for each site ranging as follows: Ahrax (Leaf n = 5, Rhizome n = 5, and Root n = 5), Baħar iċ-Ċagħaq (Leaf n = 3, Rhizome n = 2, and Root n = 2), and Golden Bay (Leaf n = 4, Rhizome n = 5, and Root n = 5).

3.1. Stable Isotope Values

The carbon isotope ratios of producers and consumers typically ranged between −23‰ and −15‰ across all sites, although they were slightly enriched in Baħar iċ-Ċagħaq (range: −19.9 to −14.4‰) relative to the other two sites, L’Aħrax (range: −20.3 to −15.6‰) and Golden Bay (range: −21.9 to −15.8‰) (Table 1; Figure 2). Posidonia oceanica was enriched in ¹³C relative to macroalgae and POM (Table 1, Figure 2). We observed some tissue-related variability in P. oceanica values (Figure 3), although this was not standard across all sites. Carbon (δ¹³C) values spanned 2–3‰ at each site, with rhizome values typically being intermediate between leaf and root values, although the variation in δ¹³C among tissues was below statistically identifiable levels (ANOVA: F_2,31 = 2.3, p = 0.1). The variation in nitrogen was similarly negligible (ANOVA: F_2,31 = 2.2, p = 0.13), although Baħar iċ-Ċagħaq was something of an outlier as δ¹⁵N values spanned 5‰. All P. oceanica tissue values were, therefore, combined prior to subsequent mixing models. Macroalgae displayed highly variable δ¹³C values at each site, which, in all cases, encompassed the estimated marine POM δ¹³C value (Figure 2).

We observed considerable variation in stable isotope values among taxa (PERMANOVA: PseudoF_2,145 = 73.8, p < 0.001), locations (PseudoF_2,145 = 30.8, p < 0.001), and their interaction term (PseudoF_3,145 = 5.9, p < 0.001). Across all sites, benthic invertebrates were enriched in ¹⁵N relative to primary producers, with most fishes further enriched in δ¹⁵N. At Baħar iċ-Ċagħaq, the δ¹⁵N values of sources and most fish were ¹⁵N-enriched (range: 7.8–13.9‰) relative to Golden Bay (range: 5.5–10.6‰) (Table 1; Figure 2). The δ¹⁵N SI values of BMI consumers were also enriched at Baħar iċ-Ċagħaq (range: 4.1–10.2‰) compared to L’Aħrax (range: 2.7–4.8‰) and Golden Bay (range: 2.3–8.8‰) (Table 1; Figure 2) While this pattern was evident in most samples, decapods and Chromis chromis were notable exceptions (Figure 2).

3.2. Stable Isotope Mixing Model

Stable isotope mixing models performed reasonably well. In the BMI model, all 123 variables had an acceptable Gelman Rubig diagnostic of <1.05. For the fish model, the Gelman Rubig diagnostics of all 94 variables were also <1.05. However, some posterior density estimates for macroalgae and POM were not reliable; for instance, at the L’Aħrax location, the estimated contribution of macroalgae to the BMI diet showed a bimodal distribution, likely due to a high variability in macroalgae SI values and to a considerable overlap between macroalgae and POM values. As such, while we present the estimated contributions of all three sources to all consumer groups, the specific estimates for macroalgae and marine POM should be interpreted with some caution. However, the estimates for P. oceanica, the principal interest of this study, are robust in all cases.

The models revealed that consumers were assimilating carbon primarily from sources other than P. oceanica seagrass, although, for the reasons outlined above, it is difficult to determine whether consumers are reliant on POM, macroalgae, or a combination of both (

Table 2. Summary of two three-source stable isotope mixing models showing estimated mode (± 95% credibility interval) contributions of macroalgae, particulate organic matter (POM), and Posidonia oceanica seagrass to invertebrate (BMI) and fish consumers among P. oceanica margin communities. General estimated contributions across all locations and location specific estimates are provided.

Model	Location	Macroalgae	POM	P. oceanica
BMI	All	34 (5–74)	45 (8–82)	12 (1–60)
	L’Aħrax	66 (6–100)	25 (0–85)	3 (0–24)
	Baħar iċ-Ċagħaq	11 (0–86)	79 (7–100)	3 (0–21)
	Golden Bay	18 (0–52)	80 (37–100)	1 (0–15)
Fish	All	26 (2–75)	37 (5–84)	25 (2–74)
	Baħar iċ-Ċagħaq	0 (0–20)	1 (77–100)	0 (0–8)
	Golden Bay	0 (0–100)	100 (0–100)	0 (0–1)

; Figure 4). This is evident in the extremely large 95% credibility intervals associated with the estimated contribution of macroalgae and POM to all BMI and Fish populations (Table 2). However, the estimated contributions of P. oceanica are minor, and appear robust. Amongst the pooled BMI, the model contribution of P. oceanica was not above 3% at any location, with 95% credibility intervals between 0 and 24%. This pattern was further emphasised in fishes, where the model contribution was zero in both locations with upper 95% credibility intervals below 10% (Table 2).

4. Discussion

Our study was designed to assess the degree to which P. oceanica is consumed by benthic macroinvertebrates and fishes, which use Posidonia meadows as their principal habitat. Despite the challenges in interpreting the stable isotope mixing model data, we have demonstrated that the P. oceanica-associated communities we have sampled are assimilating their carbon from sources other than P oceanica. Furthermore, we accounted for potential intra-plant variability by sampling three tissue types (leaf, rhizome, and root) and ascertaining that they had similar SI compositions. Below, we place these findings within the appropriate regional and international context, highlighting the limitations associated with his current study, and how these may be overcome in future investigations.

4.1. δ¹³C and δ¹⁵N Isotopes of Sources and Consumers

This study is the first to provide baseline tracer values (δ¹³C and δ¹⁵N) for Posidonia oceanica seagrasses and organisms associated with their meadows around Malta, in the central Mediterranean (δ¹³C = 15.5 ± 1.0‰, δ¹⁵N = 1.7 ± 2.3‰; Table 2). Vizzini et al. [22] reported mean isotope values of δ¹³C −11.3 ± 0.3‰ and δ¹⁵N 2.8 ± 0.4‰ for P. oceanica, which were enriched relative to the values recorded in the present study. This difference might be explained in terms of the site-specific differences in water quality and nutrient enrichment levels. For instance, a number of sites sampled within the Vizzini et al. [22] study, notably the Stagnone of Marsala, are prone to nutrient enrichment by virtue of the semi-enclosed and shallow nature of the same sites, whilst all the coastal sites sampled within Maltese waters were exposed to hydrodynamic elements.

The PERMANOVA analysis indicated that the SI values obtained for Plants, i.e., sources, for BMI, and for Fish consumers were significantly dissimilar to each other across sites, reinforcing their distinct trophic positions and highlighting the effectiveness of stable isotopes as a tool to infer trophic relationships in this region. Isotope values were notably distinct between the three sources of macroalgae, particulate organic matter (POM) and P. oceanica, except at Baħar iċ-Ċagħaq, where POM and macroalgae were similar (Figure 2), due to the highly variability values observed in macroalgae. This could be linked to the overall isotopic enrichment of taxa at this location, which could be the result of the increased nutrient loading from effluents in this area linked to anthropogenic activities. This is confirmed by the fact that Malta’s third River Basin Management Plan (RBMP), submitted to the EU Commission as part of Malta’s Water Framework Directive (WFD) reporting obligations, assigns higher nutrient levels to the coastal water body (MTMTC104) housing the Baħar iċ-Ċagħaq site when compared to the two coastal water bodies (MTMTC 103 and MTMTC 109) housing the L’Aħrax and Golden Bay sampling locations [40]. In the present study, macroalgae showed marked variability for both isotopic values at all three locations. This may have been the result of not targeting specific alga species that are known to be grazed upon by benthic marine invertebrate (BMI) and fish consumers, resulting in a considerable amount of noise in the data. Macroalgae were also not sampled consistently in terms of the species, size, and substrate which may also have had an effect on SI composition. This is consistent with the research by Vizzini et al. [22], in which macroalgal SI variability was attributed to species-specific physiological factors, including enzymatic discrimination during photosynthesis which allow macroalgae to use different carbon and nitrogen sources (specifically HCO₃-/dissolved CO₂ and NO₃-/NH₄), as well as differences in size which can affect the metabolism, and carbon and nitrogen assimilation rates, as well as isotopic composition. These results appear to be in contrast with studies carried out in Corsica [30] where P. oceanica organic matter might represent an important carbon source for the specific taxa of macroalgae. POM stable isotope values averaged δ¹³C −19.8 ± 0.6‰ and δ¹⁵N 2.7 ± 1.6‰ across sites. To our knowledge, there are no published POM stable isotope values for Maltese waters against which to compare these values; however, studies in other regions of the Mediterranean produced similar δ¹³C values [22,41], which suggests seagrass detritus contributes very little to POM [22]. It is, therefore, unlikely that seagrass detritus contributes indirectly to consumers via detrital pathways based on the sources studied here.

4.2. Trophic Contribution of Sources to Consumers

The mixing model results support our hypothesis that consumers are primarily consuming particulate organic matter (POM) or macroalgae rather than Posidonia oceanica, which makes very little trophic contribution to benthic marine invertebrate (BMI) and fish consumers within this study, which is in line with other studies in the Mediterranean and further afield [22,28]. Macroalgae were the dominant food source among invertebrates at L’Aħrax, contributing up to 66% to its diet, whereas, at Baħar iċ-Ċagħaq and Golden Bay, POM was dominant, contributing up to 79% and 80%, respectively (Figure 3). POM also made up the entirety of the studied fish diet, contributing 100% at Baħar iċ-Ċagħaq and Golden Bay (Figure 3). Nevertheless, the mixing model results regarding macroalgal and POM contributions need to be interpreted with caution due to the similar isotope ratios of macroalgae and POM. Further investigation, possibly using δ¹³C stable isotope ratios of amino acids, may be insightful in this regard.

4.3. Study Limitations

Due to the opportunistic nature of the sampling events, taxa were not consistently sampled across sites in terms of taxonomy and quantity. Moving forward, fish should be sampled at all sites where possible, and the source and BMI taxa need to be sampled consistently across sites and identified to the species level where possible. This is especially important for the macroalgae, considering the noise in the data. Analysis could not be carried out using specific genera to increase the power of the models, due to the lack of samples and consistency across sites. For such a reason, we could not detect a specific relationship between the carbon composition of P. oceanica and some specific taxa of macroalgae as pointed out in other studies [32]. Further investigation could also sample and analyse the isotopic signatures of POM directly by taking water samples of phytoplankton and zooplankton using a Niskin bottle deployed by boat, for instance. Due to the absence of plankton samples, Mytilus spp. were used as a proxy for POM [35]. However, since only very few Mytilus spp. were sampled at each location (Mytilus spp. individuals are infrequent within Maltese coastal waters), the degree of variation in stable isotope ratios could not be estimated, which had to be arbitrarily set for analysis, which again reduced the robustness of the current study.

Sampling limitations restricted the number and diversity of the species which were collected at each site. Consequently, we assessed the contribution of P. oceanica to broad classifications of BMI and fish rather than at a taxon-specific level. While this allowed us to address concerns regarding low sample sizes, it masks the likely variation in research used at the species or individual level. For example, fish species which are known to feed directly on the P. oceanica biomass were not sampled, such as Sarpa salpa, which is known to be one of the main Mediterranean P. oceanica grazers [42], besides other known non-fish P. oceanica grazers (e.g., Paracentrotus lividus, and a number of decapod species) [43,44]. This shortcoming, however, does not appear to be making a significant contribution to the wider community. Despite not necessarily relying on the seagrass itself, some organisms like P. lividus may be feeding directly on seagrass to graze epiphytic diatoms [43]. Further study could aim to include these species amongst targeted consumers, and to quantify the isotope ratios of P. oceanica-associated epiphytic diatoms in order to assess their contribution to the wider food web within Posidonia meadows.

The utility of two tracer stable isotopes to resolve trophic dynamics was limited in this instance due to the similarity of isotope ratios between putative prey sources. Mixing models depend on several assumptions [33]. For instance, they assume that all sources contributing to a mixture are known and quantified. This is not necessarily the case in this study. We chose three sources to best represent different carbon sources along the margins of P. oceanica margins in coastal marine settings; however, there are likely to be other sources, for example, sedimentary organic matter [22]. Carbon isotope values of consumers fall within the range of sources sampled at all sites, suggesting that our study encompassed all significant food web sources. Other approaches could reveal a higher resolution of what is grazed upon (e.g., which species of macroalgae) and may reduce noise in the data, for instance, through the use of stomach content analysis and compound specific fatty acid stable isotope analysis, or the inclusion of additional isotopic markers such as sulphur (δ³⁴S) [22,45,46]. Alternatively, a DNA barcoding approach could be used to determine whether P. oceanica is present in the digestive system of Posidonia consumers; this approach may also discern if P. oceanica is ingested but not assimilated by consumers targeting epiphytic diatoms.

5. Conclusions

We sampled and analysed δ¹³C and δ¹⁵N in potential carbon sources, macroalgae, POM, and P. oceanica seagrass, to quantify the trophic contribution of these sources to BMI and fish consumers at three locations along the P. oceanica seagrass meadow margins in Maltese coastal waters. Isotopic signatures varied significantly between taxa (source, BMI, and fishes) and across locations. Stable isotope values of sources and consumers were enriched at Baħar iċ-Ċagħaq, probably due to the lower water quality arising from in situ anthropogenic activities, including intensive coastal urbanisation, desalination plant discharges, bunkering, yacht marina facilities, and intensive boating activity. Posidonia oceanica contributed the least to BMI at all three sampling locations (L’Aħrax, Baħar iċ-Ċagħaq, and Golden Bay) and to fish at two of the three sampling locations (Baħar iċ-Ċagħaq and Golden Bay). Based on the mixing model outputs, we can be confident that P. oceanica does not contribute significantly to consumer diets at the sampled seagrass meadow margins in coastal waters around Malta, presumably due to its largely indigestible nature. This is consistent with findings from studies elsewhere in the Mediterranean and from further afield [22,28,29]. We cannot conclude, however, that BMI and fishes are predominantly consuming POM or macroalgae. Further investigation into the matter is needed and future studies could adopt a multi-faceted approach, integrating stomach content analysis with the compound-specific stable isotope analysis of fatty acids [22,46].