Cryptic diets of forage fish: jellyfish consumption observed in the Celtic Sea and western English Channel

To establish if fishes’ consumption of jellyfish changes through the year, we conducted a molecular gut‐content assessment on opportunistically sampled species from the Celtic Sea in October and compared these with samples previously collected in February and March from the Irish Sea. Mackerel Scomber scombrus were found to feed on hydrozoan jellyfish relatively frequently in autumn, with rare consumption also detected in sardine Sardina pilchardus and sprat Sprattus sprattus. By October, moon jellyfish Aurelia aurita appeared to have escaped predation, potentially through somatic growth and the development of stinging tentacles. This is in contrast with sampling in February and March where A. aurita ephyrae were heavily preyed upon. No significant change in predation rate was observed in S. sprattus, but jellyfish predation by S. scombrus feeding in autumn was significantly higher than that seen during winter. This increase in consumption appears to be driven by the consumption of different, smaller jellyfish species than were targeted during the winter.

To establish if fishes' consumption of jellyfish changes through the year, we conducted a molecular gut-content assessment on opportunistically sampled species from the Celtic Sea in October and compared these with samples previously collected in February and March from the Irish Sea. Mackerel Scomber scombrus were found to feed on hydrozoan jellyfish relatively frequently in autumn, with rare consumption also detected in sardine Sardina pilchardus and sprat Sprattus sprattus. By October, moon jellyfish Aurelia aurita appeared to have escaped predation, potentially through somatic growth and the development of stinging tentacles. This is in contrast with sampling in February and March where A. aurita ephyrae were heavily preyed upon. No significant change in predation rate was observed in S. sprattus, but jellyfish predation by S. scombrus feeding in autumn was significantly higher than that seen during winter. This increase in consumption appears to be driven by the consumption of different, smaller jellyfish species than were targeted during the winter.

K E Y W O R D S
16s mtDNA, Celtic Sea, diet, English Channel, gelatinous zooplankton, molecular gut-content analysis 1 | INTRODUCTION Fisheries in the Irish Sea are important for the regional economy: in 2016 the UK-based fleet landed 36,600 t worth £57.8 million (Richardson et al., 2017), while the Irish fleet caught a further 11,253 t (CSO, 2018). However, Irish Sea fisheries are facing challenges from increasingly abundant scyphomedusae jellyfish (hereafter referred to as jellyfish, unless stated otherwise; Lynam et al., 2011). Jellyfish blooms (mass aggregations of jellyfish in a localised area) in other regions have caused economic losses to fisheries by bursting fishing nets, contaminating catches, reducing the abundance of fish by competing for the same resources and killing fish through irritation of gills with their stinging tentacles (Richardson et al., 2009). Fish farms can also suffer damage and mass mortality from jellyfish blooms (Doyle et al., 2008). Preventing jellyfish blooms from affecting human enterprise has been difficult and many direct interventions have been ineffective (Richardson et al., 2009).
Commercially-important fish species such as herring Clupea harengus L. 1758 and whiting Merlangius merlangus L. 1758 were shown to consume jellyfish in the Irish Sea using a molecular assay (Lamb et al., 2017). However, the observed scyphomedusae consumption occurred when jellyfish in the Irish Sea were juvenile and lacked the size or defensive structures to deter predation; it remains unknown if they are consumed throughout the year or used as a seasonal resource.
Complex and dynamic interspecific relationships are common in marine ecosystems: assuming unchanging predation throughout the year is likely to misrepresent a species' trophic role. For example, C. harengus are known to limit cod Gadus morhua L. 1758 recruitment by feeding on juvenile G. morhua when they are part of the ichthyoplankton (Koster & Mollmann, 1996). However, upon maturation, G. morhua feed on small C. harengus (Bailey & Batty, 1984), reversing the interspecific relationship. A dynamic relationship like this may be present in jellyfish as they have a complex life cycle featuring multiple, functionally different life stages (Lucas, 2001). During February and March, consumption of jellyfish was probably targeting ephyrae (a juvenile form of jellyfish) (Lamb et al., 2017). Ephyrae are just a few millimetres in diameter and often lack the stinging tentacles seen in mature jellyfish as these can take several weeks to develop once they join the plankton community (Holst, 2012).
Although all Irish Sea and Celtic Sea jellyfish ephyrae measure a few millimetres in diameter, there is considerable variation in size and stinging ability by maturation (Holst, 2012). Mauve stinger jellyfish Pelagia noctiluca (Forsskål, 1775) remain small, with a mean ± SD diameter of 4.5 (± 1.2) cm, although large individuals can reach 12 cm . Other common species are known to grow larger: Aurelia aurita L. 1758 can reach 25 cm diameter (Omori et al., 1995), while barrel jellyfish bells Rhizostoma pulmo (Macri, 1778) are known to approach 1 m in diameter (Russell, 1970). Large predators such as leatherback turtles Dermochelys coriacea (Vandelli, 1761) feed on whole medusa (Heaslip et al., 2012;Houghton et al., 2006), but it remains to be seen if the pelagic fish species identified previously as jellyfish consumers in the Irish Sea (Lamb et al., 2017) (Table 1), measured, weighed, and had their stomachs removed and frozen on-board. Scalpels and gloves were changed and cutting boards cleaned using fresh water between species dissection. If jellyfish were found in the haul, they were identified to species level and a small sample of bell tissue was preserved in 100% ethanol.
Additional jellyfish samples were obtained from plankton sampling at night when the ship was stationary at designated samples points (ICES, 2016) using ring-nets equipped with a General Oceanics

| PCR and sequencing
The protocol developed previously by Lamb et al. (2017)

| Statistical analysis
A Fisher's exact test was performed to determine if differences in predation could be observed between seasons (February and March compared with October). Since multiple hypotheses (different species) were tested, a one-stage false detection rate correction (Pike, 2011) was applied (reported as q-values) to avoid the chance of a type-2 error. All statistical analyses were performed using R (www.r-project.org).

| The pelagic food-web
The most frequent consumer of jellyfish in the early season Consequently, we were unable to capture the late-season benthic component of the food web. Furthermore, the sample size was limited by a single-individual processing these samples.
Although these factors limit our ability to assess temporal variation for the predatory fish species, these data still refine our understanding of jellyfish predation in the food-web. Small jellyfish were a food source for S. scombrus (Figure 2): this has been recorded previously when S. scombrus switched from filter feeding to a biting in order to consume the small hydrozoan Aglantha digitale (O.F. Müller, 1776) (10-40 mm bell height; Runge et al., 1987). However, in contrast to the widespread predation by fish during February and March (Lamb et al., 2017) very little predation on jellyfish was found across the pelagic community during October. A possible explanation of diet shifts may be related to the relative abundance of other prey items.
For example, S. sprattus switch to preying on fish eggs in the winter when other zooplankton levels are depressed (Pliru et al., 2012). It is possible that widespread predation of scyphomeduase jellyfish ephyrae in the February and March is in response to poor availability of other zooplankton; greater zooplankton availability in October may result in a switch away from jellyfish and result in the observed predation rates.

| Escaping predation?
Although no difference in seasonal predation was detected in S. sprattus, statistical analysis demonstrated S. scombrus fed on jellyfish more frequently in the samples collected in October than those in February and March. This was unexpected, as we anticipated the consumption of larger jellyfish to be more difficult and that rates of predation would therefore decline later in the year. Upon closer inspection however, the results do not contradict this hypothesis: L. tetraphylla has a bell diameter of 1-3 cm (Russell, 1953) (Lucas, 2001), although overwintering populations have recently been recorded in other ecosystems (Ceh et al., 2015;Purcell et al., 2018). Another explanation is that larger species of jellyfish, particularly A. aurita, which were frequently preyed upon early in the season (which were likely ephyrae, although it should be noted this is inferred through phenological trends as the molecular techniques lack the ability to reveal this), may have escaped predation through somatic growth, leaving only small species like L. tetraphylla vulnerable to predation (Figure 3). Finally, it should be acknowledged that unknown sea-specific phenomena may be driving the observed differences.
The data presented here show that, in contrast to early-season sampling, late-season predation is limited: S. scombrus were the only species to feed frequently on jellyfish, although some predation was also detected in S. pilchardus and S. sprattus. The type of jellyfish consumed also changed: the small hydrozoan species L. tetraphylla was the preferred prey item in October, accounting for 80.7% predation across all species. The shift from widespread predation of juvenile jellyfish to rare predation of adults suggests energy flows from jellyfish Jellyfish predation as a function of bell area (bar height). Predation of ephyrae is inferred using data from 2008-2009(Lamb et al., 2017, medusae predation was inferred using 2015 data presented in this paper. n.b. Bar graphs are overlaid, not stacked. Detection of predation ( ) and non-detection ( ) are shown for medusae ( , and ) and ephyrae ( and ). Mean Lirope tetrayphylla size and Aurelia aurita were taken from literature (Russell (1953) and Bastian et al. (2011), respectively). Pelagia noctiluca ephyrae size is taken from Sandrini & Avian (1983). All other mean medusae, and ephyrae, bell areas are reported in Holst (2012). The mouth gape of Scomber scombrus was calculated using mean S. scombrus size in this study and a S. scombrus-specific allometric scaling function Karachle & Stergiou (2011)