Importance of suspended particulate organic matter in the diet of Nephrops norvegicus (Linnaeus, 1758)

The extent to which commercially important Nephrops norvegicus lobsters feed on particulates in the wild is unknown, even though this could be an important way for burrow-dwelling females to avoid starvation during the long breeding season. This was investigated using δ13C and δ15N isotopic signatures in tissues with long and short turnover rates to provide diet discrimination and compare this between males and females. Secondary objectives examined size-related differences and calculated the trophic position based on the new results. Almost half the diet (47%) was made up of suspended particulate organic matter (POMsusp) alone. Fish was another important item in the diet, with plankton and invertebrate sources coming much lower down in dietary importance. Significantly more suspension feeding was observed in small or medium sized individuals than large ones in both sexes. However, there were no sex-related patterns, despite females being restricted to burrows for part of the analysis period. Female diet was almost identical to males and POMsusp comprised a large component of the diet in both sexes. The trophic position was estimated at 2.94 ± 0.16 (mean ± SD), which was at the lower end of the range reported in previous studies (2.60 to 4.32).


Results
importance of suspended particulate organic matter to Nephrops' diet. Consumer isotopic values fell within the range of food source isotopic values for all four time periods (Fig. 1), indicating that all major Nephrops' dietary sources were included in the analysis and no important dietary sources were missing. This arrangement between consumer and source values is a precondition for the SIMMR Bayesian mixing model to work adequately 35 . The SIMMR model output showed that POM susp and fish were the main food sources for all consumer groups and time periods in the analysis. The estimated means of POM susp contribution to the diet ranged between 12.0-47.4% but was generally high, >20% (Supplementary Table S1). Meanwhile, apart from fish, contributions from the other sources (phyto-and zooplankton, filter feeders, polychaetes and crustaceans) were much lower, with means ranging from 2.3-15.6% (Fig. 2). The estimated means for fish contribution to the diet ranged from 18.2-60.9%. Therefore, the main question about the importance of POM susp in Nephrops' diet was accepted to be the case in this study.
After combining sources, the estimated means of 'active feeding' (i.e. filter feeders, polychaete, crustaceans and fish) ranged from 42.5-76.2%, while that of 'suspension feeding' (i.e. POM susp , phytoplankton and zooplankton) ranged from 26.5-57.5%. More details on average contributions from all food sources and probability distributions of 'active feeding' and 'suspension feeding' to Nephrops' diet can be seen respectively in Supplementary  Table S1 and Fig. S1.
Size-related differences in Nephrops' diet. There were some significant Nephrops size-related differences in suspension feeding (i.e. POM susp , phytoplankton and zooplankton). The results showed that suspension feeding took place significantly more in small or medium sized individuals than large ones. Surprisingly, this occurred more frequently in male comparisons than in females. Suspension feeding was significantly higher in the small and medium sized males compared to larger males in the early time period, i.e. 'Spring long': 8 th March-29 th May (Fig. 3). In the late time period, i.e. 'Summer long' (4 th May-25 th July), it was significantly higher only in the small males compared to large males (Fig. 3). For females, the small size class was significantly more likely to suspension feed than larger ones in the 'Spring short' period (10 th -29 th May) (Fig. 3). No significant differences could be detected between the size groups in either sex in the 'Summer short' time period (6 th -25 th July) (Fig. 3).
Sex-related differences in Nephrops' diet. The comparison of suspension feeding between males and females produced no significant results across time periods including: 8 th March-29 th May ('Spring long'), which includes part of the long breeding season of females, and 4 th May-25 th July ('Summer long') when both sexes are non-burrow dwelling and capable of feeding outside of burrows (Fig. 4). The contribution of active feeding and suspension feeding to the diet of males and females was also equivalent, with imperceptible differences observed between the sexes ( Supplementary Fig. S1). Thus, the hypothesis of differences between male and female's diet, related to the reproductive cycle of females is not upheld at Clew Bay. trophic position of Nephrops norvegicus. The overall trophic position of N. norvegicus in Clew Bay, based on isotopic signatures in muscle and hepatopancreas (considered together) was estimated to be 2.94 ± 0.16 (mean ± SD), which represents an average over all time periods. The trophic position across different periods varied from 2.79 ± 0.12 ('Spring short') to 3.06 ± 0.10 ('Spring long'/'Summer long' -with these two periods being identical; Table 1).

Discussion
The primary aim of this research was to investigate feeding on suspended particulate organic matter (POM susp ) in wild Nephrops as this may be a mechanism of avoiding seasonal starvation. Before looking at its contribution to the diet, we could establish that we had captured all the important dietary sources by the arrangement of isotopic values in the δ 13 C/δ 15 N bi-plots (Fig. 1). The position of consumer tissues within the polygon of sampled food sources for each of the four periods indicated that no important food source(s) were missing in the analysis.
At times, almost half (47%) of the diet of Nephrops was made up of POM susp . These lobsters did show variety in their diet, however, and another important item in the diet was fish, while plankton and invertebrate sources came far below these items in dietary importance. Reliance on POM susp and fish, rather than on invertebrates, appears initially surprising, considering predominance of invertebrates in stomach content analysis 8,9 . However, many crustaceans are predatory on fish, which is apparently independent of their size 36 . Capture of flatfish, for example, may present little difficulty to Nephrops, whose diet may also be subsidised from discards arising from inshore fishing activity in Clew Bay. The high level of feeding on particulate matter was more surprising, however another burrowing decapod crustacean, Neotrypaea californiensis, has recently been shown to be primarily reliant on POM susp as a food source 33 .
Due to their burrowing lifestyle and long breeding season in females, much effort has gone into investigating seasonal starvation in Nephrops 10, [37][38][39] . The ability to feed on particulate food sources would help to counteract starvation brought about by these lifestyle restrictions. When Nephrops individuals were maintained in unfiltered seawater in an aquarium, they showed an intermediate nutritional status between control animals with no access to food and those from the wild, which was suggestive of suspension feeding, at least in-extremis with no other food available 13 . The present results add to this by demonstrating the importance of suspension feeding to the diet of wild individuals, showing that they utilised this food source at a significant level. Previous work has theorised that 65-68% of daily energy intake was available for growth from suspension feeding at sufficient particulate densities 8 . Although our study does not address energy intake directly, our estimates of suspension feeding in the diet often www.nature.com/scientificreports www.nature.com/scientificreports/ reached 50% (particularly in short-term tissues). This likely represents a considerable amount of suspension feeding-derived energy that is available for growth (Supplementary Table S1).
In fact, it has long been acknowledged that POM susp is an important seasonal source of food for benthic organisms in Winter 40,41 . Not all suspended food particles were equally important, however, for example phytoand zooplankton were far less important than POM susp in Nephrops ( Fig. 2; Supplementary Table S1). Sediment organic matter (SOM) could be another important food source for benthic organisms like Nephrops. A practical difficulty is distinguishing SOM from POM susp because the latter eventually falls to the seafloor and therefore forms one component of the SOM. Although we cannot discount the possibility that, as well or instead of feeding on POM susp , Nephrops also picks POM up off the sediment while deposit feeding on a mix of SOM/POM, we do not have evidence to support this idea. POM susp and SOM present distinct isotopic signatures 42 . For example, compared with POM susp , SOM was shown to be enriched in δ 13 C and depleted in δ 15 N 42 . Had SOM with this 42 profile been substituted for POM susp in the present study, Nephrops samples would have fallen outside the isotopic polygon. Also, the δ 13 C/δ 15 N bi-plot in our study showed no missing dietary items, which might have been expected had SOM been important in the diet. Other studies have also shown distinct POM and SOM signatures 28,43 . Future studies combining fatty acid analysis and SIA may further disentangle the various sources of organic particulates and their relative importance, including sources found inside lobster burrows (e.g. 33 ).
The hypothesis suggesting a size-related difference in Nephrops diet was accepted for several comparisons. Suspension feeding was higher in smaller compared to larger size classes for males in particular, e.g. during 'Spring long': 8 th March-29 th May (small and medium males compared to large males) and 'Summer long': 4 th May-25 th July (small males compared to large males). Males may suffer more competition for active food items than females 44 which may force smaller individuals to rely on particulate food sources. The same size-related difference, i.e. a higher proportion of suspended food in the diet of small females compared with larger ones, was borne out in only one time period: the 'Spring short' feeding on 10 th -29 th May. This was not seen for the equivalent tissue later in the season 'Summer short' on 6 th -25 th July. Interestingly, the size-related differences we observed did not appear to be related to a limitation on predation capacity in smaller lobsters. Indeed, at times, the contribution to the diet by fish was even higher in smaller individuals than in the larger ones, e.g. 48.1% versus 44.0% and 44.7% versus 28.1% for small versus large males in two of the four periods analysed (Fig. 2, Supplementary  Table S1). However, without further research it is difficult to interpret the reason for this, for example, it is possible that smaller individuals may feed on fisheries discard or on larger individual's leftover prey. www.nature.com/scientificreports www.nature.com/scientificreports/ Although the isotopic signal from long and short-term storage tissues varied substantially, there was no difference in 'Spring' and 'Summer' diets when similar storage tissues were compared ( Supplementary Fig. S1). The difference between long and short-term storage tissues arises because these represent different time intervals, 19 and 81 days respectively. Active feeding was higher in long-term storage tissues, whereas suspension feeding was increased in short-term tissues ( Supplementary Fig. S1). Without further experiments, the reasons for this are unclear, however.
The hypothesis related to sex-specific diets was rejected. Males and females were remarkably similar in diet, even in the Spring period (8 th March-29 th May), even though the tissues sampled from this time represented a period when females were mostly brooding. Larval release at Clew Bay begins around the second week of April 45 , until which point, the females stay inside burrows to brood their developing embryos. Our results demonstrate that this period of burrow dwelling does not prevent females from accessing the same food items as males. It has been suggested that feeding by females during the breeding season may simply take place closer to the burrow mouth 10,46 or females may bury food within or adjacent to burrows 15 . Although the sexes have similar diets, as shown in the present study, the overall opportunity for feeding may be reduced in females. However, starvation and sex-specific reduction in nutritional status has been previously examined and found to be absent 10 , with corroborating evidence from biochemical markers that suggest good nutritional condition throughout the year in females 38 . Although Clew Bay is a particularly shallow site, the same major food groups (i.e. plankton and particulates, macroinvertebrates and fish) are available in deeper habitats [7][8][9][10][11][12] . Nephrops diurnal emergence does vary with depth 47 but we can think of no plausible reason for this to interact with the availability of POM susp or other food groups. Therefore, we believe the results are transferrable to other Nephrops populations, although other locations may have slightly different groups of macroinvertebrates (echinoderms, in particular, were not abundant in the sediments at Clew Bay).
The importance of POM susp as a dietary source in Nephrops has implications for its trophic position. Based on isotopic signatures in the present study, trophic level was calculated to be 2.94 ± 0.16 (mean ± SD) i.e. at the lower end of previous estimates for Nephrops (2.60-4.32 14,15,43 ) that were also derived using SIA. SIA can reveal lower trophic status in consumers compared with stomach contents analysis, because the latter can underestimate soft-bodied prey 25 . In the case of Nephrops, it would be almost impossible to detect from stomach contents, the www.nature.com/scientificreports www.nature.com/scientificreports/ fact that up to half of the diet derived from POM susp . Such disproportionate measurement of prey items acts to artificially inflate trophic position based on stomach contents alone.
As the present study shows, smaller and medium-sized males fed on significantly higher suspended food than larger ones, therefore the potential to suspension feed may be an important mechanism for avoiding aggressive encounters over food between males. This is potentially important because the growth (and hence biomass) of male individuals is strongly density-dependent at Clew Bay 44 (-densities at Clew Bay vary between 0 and 15 individuals per pot fished 45 ). Body size also varies across fishing grounds -smaller Nephrops are found at FUs with higher stock densities, most likely as a result of reduced growth potential due to intraspecific competition 5,44 . We suggest that feeding on POM is an important lifestyle adaptation in both males (counteracting competitive interactions) and females (counteracting burrow-dwelling) but that Nephrops diet is remarkably similar in the sexes. The knowledge that fish is also an important component of the diet in all groups examined at Clew Bay ). An asterix denotes a significant sex-related difference in suspension feeding in the diet i.e. p BIC > 0.95 (provided by the SIMMR package according the Bayesian paradigm), however no such differences were observed.  Table 2) and the 25-July-2014 ('Summer long' or 'Summer short' - Table 2). Nephrops were collected by baited creels on both dates. As Nephrops interacts with benthic communities both on and beneath the sediment, these potential prey items were obtained in two ways: (1) via five 15 min bottom trawl, using a standard 2 m beam trawl with a chain mat, a stretched mesh (bar length: 20 mm) and a codend liner with a knotless mesh (bar length: 4 mm) and (2) using a day grab Van Veen 12.110-250 cm²/3.14 L. A total of 17 different putative prey taxa were sampled from different groups: tunicates, polychaetes, bivalves, gastropods, crustaceans and fish on both dates (see Supplementary Table S2). Zooplankton and phytoplankton were sampled on both days using a 57 cm ring diameter and 250 µm mesh WP-2 plankton net towed behind the boat (15 min tows). The choice of plankton net assumed Nephrops can feed on plankton items larger than 300-500 µm 13 . Assuming that Nephrops consumers fell within the range of diet sources in an isotope bi-plot, we could be satisfied that no important food sources were missing from the analysis (indeed, this was the case -see Results section 2.1). To sample POM susp , water samples were taken via a Niskin bottle triggered at around 1 m above the seafloor. All samples were held on ice during transit and then transferred to a −20 °C freezer until processing for stable isotopes analysis.

Period
Stable isotope sample preparation. Potential Nephrops food 'source' tissues were processed as follows: phytoplankton and zooplankton samples were cleaned under the microscope. POM susp was concentrated by filtering seawater (around 5 L was filtered for each sample) on precombusted glass filters and stored frozen (−20 °C). POM susp samples were acid-washed to remove any carbonates, which consisted of adding 1 ml 0.1 M HCl, following the protocol developed by 50 . All macrofaunal items that were dominant in both abundance and biomass in grabs were sampled for SIA using various tissues, depending on the organism (see Supplementary Table S2 for details). 'Consumer' (Nephrops) tissues were subsampled from the fisheries catch by selecting n = 10 replicate individuals within each of three size classes (small, medium and large) for both sexes (see Supplementary Table S3). After thawing at room temperature, carapace length, weight without chelipeds (to avoid bias due to claw loss) and sex was recorded for all individuals. Nephrops tissue was sampled from muscle (tail) and hepatopancreas for both males and females, with hepatopancreas in this case representing a shorter-term storage tissue and muscle representing a longer-term storage tissue (see below).
All tissues sampled were oven dried in 2 ml tubes at 60 °C for at least 48 h. Each dried sample was then ground with a mortar and pestle to a fine homogenous powder. Varying amounts of lipids amongst species and tissue types can result in errors in δ 13 C isotope values if not removed from the tissue prior to measurement 51 . Therefore, all source and consumer samples underwent lipid correction of three 8 ml washes (or until the supernatant was clear) of 2:1 chloroform:methanol solvent according methodology developed by 52 . Samples were again dried in the oven at 60 °C for 48 h to remove any remaining solvent. Aliquots of lipid extracted tissue of 400-600 μg were weighed into tin capsules for stable isotope analysis.
Stable isotope ratios (δ 13 C and δ 15 N) of all samples were measured at the Stable Isotope Core Laboratory of Washington State University using an elemental analyser (ECS 4010, Costech Analytical, Valencia, CA) connected to a continuous flow isotope ratio mass spectrometer (Delta PlusXP, Thermofinnigan, Bremen) and expressed as parts per thousand (‰) (further details can be found in the Supplementary Methods).  www.nature.com/scientificreports www.nature.com/scientificreports/ data within a Bayesian framework. SIMMR model outputs are posterior probability distributions representing the likelihood of a specific source being part of the diet of the consumer, with their respective credible intervals. SIMMR was run based on the following input data: 13 C and 15 N isotope signatures of consumers, mean 13 C and 15 N isotope signatures of sources i.e. putative prey groups and their standard deviations and estimates for 13 C and 15 N trophic enrichment factors (means and standard deviations -see below).

Data analysis.
For the initial analysis, to show the importance of POM susp in the diet and to ensure that all dietary sources were captured in the analysis, sources were divided into seven taxa/groupings: (i) Crustaceans; (ii) Filter feeders; (iii) Fish; (iv) Phytoplankton; (v) Polychaeta; (vi) POM susp and (vii) Zooplankton. Meanwhile, consumers were grouped in all possible combinations of size (small, medium, large), sex (male and female) for long/ short-term storage tissues (respectively, muscle and hepatopancreas), providing 12 different combinations overall. Comparison of diet between these consumer groups formed the basis of further hypothesis testing, i.e. statistical comparisons of 'active feeding' versus 'suspension feeding' , as described in 'Statistical design' , below. The SIMMR model was run twice based on isotopic signatures (for both consumers and food sources) collected in each of the first and second sampling days. Next, four experimental time 'Periods' were defined based on the combination of the two sampling dates and two different tissues representing a long (muscle) or short (hepatopancreas) residence times (rt) ( Table 2). Residence time for muscle tissue was 81.1 days, obtained from isotopic incorporation rates and discrimination factors in Neogonodactylus bredini (mantis shrimp) 54 , while rt for hepatopancreas was estimated as 19.3 days. This was calculated from the 13 C half-life for hepatopancreas tissues in Callinectes sapidus 55 (further details of these calculations can be found in the Supplementary Methods).
Trophic enrichment (or 'fractionation') factors (TEFs) of 3.0 ± 0.6‰ for δ 13 C and 0.9 ± 0.3‰ for δ 15 N were chosen, based on estimates from mantis shrimp muscle 54 , which is the best taxon-specific information available. These values contrast with widely-used values from previous meta-analysis 56 that present averages from 61 different species of aquatic and terrestrial vertebrates and invertebrates in a variety of taxa: arthropods, molluscs, nematodes, birds, fish and mammals (for information, values in 56 were 0.5 ± 0.13‰ for δ 13 C and 2.3 ± 0.18‰ for δ 15 N). Nevertheless, we chose the mantis shrimp values 54 , firstly, on the basis that these fractionation values were calculated from a decapod crustacean: taxonomic relatedness is important due to evidence that TEFs are taxon-specific due to shared physiological processes at taxon level [57][58][59][60][61] . Secondly, the values in 54 represented lipid-corrected stable isotope ratios for consumers and prey, as also used in our study, and were from a diet shift controlled laboratory experiment.
Statistical design. Each group of consumers subjected to hypothesis testing included 10 replicate consumer samples (n = 10). This sample size seems adequate in bootstrapped simulations, which have shown an absence of large biases in statistical inference of stable isotope data with >8 replicate consumer samples 62 . For hypothesis testing, sources were combined by the function 'combine_sources' of SIMMR package into 'active feeding' (i.e. filter feeders, polychaete, crustaceans and fish) and 'suspension feeding' (i.e. POM susp , phytoplankton and zooplankton). In order to test our hypotheses, posterior distributions of 'suspension feeding' by consumers were compared in several ways. Size-related differences in consumers were examined across all 8 possible 'Sex' x 'Period' combinations (i.e. all combinations of males and females in 4 time periods). Sex-related differences in consumers were examined across 6 combinations of 3 'Size' groups in 2 periods, 'Spring long' and 'Summer long'. These periods represent an equivalent number of feeding days but with the key difference that 'Spring long' included part of the period where females were brooding embryos in Clew Bay, i.e. up until ~10 th April 45 , whereas 'Summer long' was a non-brooding period. Any dietary differences associated with female brooding could be judged against males using this comparison. Please note that, as they had just completed their reproductive cycle and had spawned, none of the females sampled actually contained embryo masses, however we could assume that 84-92% of our sample (n = 60) of females had bred, based on previous work 17, [63][64][65] . The suspension feeding contribution was compared across each of the above groups using the function 'compare_ groups' from the SIMMR package. This function gives the probability p BIC of 'any diet source's proportion in one treatment being greater than the proportion of the same source in another treatment' with p BIC > 0.95 considered to indicate significant differences 66 .
The trophic position of Nephrops was determined based on the isotopic signatures of consumers and prey according to a modified version of the following equation 67 : 15 15 where δ N c 15 is the isotopic signature of the consumer Nephrops, N base is that of the food base (herein phytoplankton), λ is the trophic position of the base (λ = 1 for primary producers) and Δ n is an estimate of the average increase in Δ 15 N per trophic position/level, herein set at 3.4‰ based on estimates for aquatic food webs 67,68 . However, because the TEF for δ 15 N, is significantly lower for decapods 54 than for many other taxa 56,67,68 , we modified the Eq. (1) to incorporate this and prevent an erroneous underestimation of Nephrops' trophic position, as follows: c b ase n 15 15 where 0.9 is the TEF for δ 15 N of mantis shrimp 54 corresponding to the average increase in Δ 15 N per trophic position/level; this was subtracted from the isotopic signatures of the consumers in Eq. (1) to facilitate a more accurate trophic position calculation for a decapod, as is the case in the present study. Because this manipulation of the equation underestimates the trophic position in one level, a correction was required by adding one trophic position/level at the end of the calculation, as seen in Eq. (2). For estimating Nephrops' overall trophic position, the isotopic signatures in the tissues (muscle and hepatopancreas together) of all individuals sampled in both sampling days (n = 120) were used. In this case, the trophic