Introduction

Cephalopods form a diverse and important prey base for many pelagic predators including marine mammals, seabirds, and fishes (West et al. 2017; Furness et al. 1984; Xavier et al. 2011; Santos et al. 2001). All cephalopods have chitinous mandibles, called beaks, which can be used for taxonomic identification and body size estimation. Beaks are more resistant to digestion than soft tissues and are often the only cephalopod remains found in the stomachs of mammals (Santos et al. 2001; West et al. 2017) and birds (Furness et al. 1984; Xavier et al. 2011). While cephalopod remains from both soft tissue (including whole and partially-digested individuals) and beaks are commonly found in fish stomachs, cephalopod identifications from beaks are frequently excluded from diet analyses because they have different residence times than soft tissues and can confound estimates of prey importance (Santos et al. 2001; Xavier et al. 2011). However, if soft tissue digestion rates are correlated with moisture content (e.g., Arai et al. 2003), excluding beaks from diet analyses is likely to disproportionately underrepresent the importance of cephalopods with higher moisture contents.

Quantifying the diet composition of marine predators aids greatly in discerning foraging patterns and associated food web structure. Prey depth distributions and abundances can elucidate vertical and ecological niche partitioning by pelagic predators and food web linkages between shallow and deep habitats (Boyd et al. 2015; Carroll et al. 2017). Many deep-sea organisms have relatively gelatinous tissues with higher moisture contents and reduced swimming abilities compared to their epipelagic counterparts (Childress 1995; Seibel et al. 1997). Although gelatinous taxa (e.g., Chiroteuthidae and Cranchiidae) have energy densities 3–9 times lower than muscular taxa (e.g., Ommastrephidae and Onychoteuthidae; Clarke et al. 1985; Schaafsma et al. 2018), gelatinous cephalopods may be relatively abundant and easy to capture in deeper habitats (Seibel et al. 1997). Thus, excluding beaks from diet analyses may underestimate predator foraging depths and the contributions of gelatinous cephalopods to deep-sea food webs.

The longnose lancetfish (Alepisaurus ferox) is a cosmopolitan predator that consumes a high diversity of fishes, crustaceans, and cephalopods throughout the upper 2000 m of the water column (Moteki et al. 2001). The cephalopod component of lancetfish diets in the central North Pacific Ocean has been reported primarily from soft tissue remains (Choy et al. 2013; Portner et al. 2017). Lancetfish forage across a large depth range on a diversity of cephalopod taxa and are thus useful for examining diet contributions from distinct prey remains (i.e., beaks and soft tissues).

We present a comprehensive analysis of the cephalopod component of lancetfish diets in the central North Pacific. The goals of this study were to (1) quantify cephalopods from soft tissues and beaks, (2) compare depth habitats and energetic contributions of gelatinous and muscular cephalopods, and (3) examine how including cephalopod prey identified from beaks affects our understanding of lancetfish foraging behavior and diet ontogeny. We show that the practice of including cephalopods identified from beaks serves to supplement conventional diet analyses to increase our ecological knowledge of marine food webs.

Materials and methods

Sample and data collection

Lancetfish were sampled from 2009 to 2018 in the central North Pacific Ocean from the Hawaii-based longline fishery, from both shallow-set and deep-set longlines (Figure S1). This fishery operates year-round between 0–35° N and 135–175° W, as described in Woodworth-Jefcoats et al. (2018). Stomachs (n = 2795) were collected by fishery observers of the National Oceanic and Atmospheric Administration (NOAA) Pacific Islands Region Observer Program (described fully in Choy et al. 2013 and Portner et al. 2017). Stomach contents were dissected in the lab where cephalopod remains were identified to the lowest possible taxon, measured, and enumerated (following methods in Choy et al. 2013). For loose beaks not associated with soft tissue remains, we determined the number of cephalopod individuals within each stomach by tallying the highest number of either upper or lower beaks within each taxon. Only lancetfish stomachs containing cephalopods were included in our analyses (n = 1267, representing 45.3% of all stomachs analyzed), and we only consider cephalopod prey items here. More complete diet descriptions including a portion of these stomachs are reported in Choy et al. 2013 and Portner et al. 2017.

Cephalopod size and habitat depth

For cephalopods identified from soft tissue specimens, we used calipers to measure dorsal mantle length (ML; Jereb and Roper 2010) to the nearest millimeter and a calibrated balance to obtain wet mass to the nearest hundredth of a gram. Lower rostral length (LRL) or upper hood length of loose beaks was measured to the nearest tenth of a millimeter for squid and octopod beaks, respectively (Clarke 1986) using calipers or a calibrated ocular micrometer. For cephalopods identified from beaks, we estimated ML and mass using published taxon-specific regressions when available (Table S1). Robust beak to ML and ML to mass regressions were not available for the genus Walvisteuthis, which are abundant lancetfish prey (Portner et al. 2017). We used paired beak–ML data from 4 Walvisteuthis sp. specimens from the Tree of Life Web Project (Bolstad et al. 2015) and direct measurements of 98 intact soft-tissue individuals from this study to generate ordinary least squares regression models relating LRL to ML (equation a, Figure S2a) and ML to body mass (BM, equation b, Figure S2b).

$${\text{ML}}_{{\text{(mm)}}} = 21.646 \, ^{*} \, {\text{LRL}}_{{\text{(mm)}}} + 10.008 \, ({\text{R}}^{2} = 0.93, \, {\text{n}} = 4)$$
(a)
$${\text{BM}}_{{\text{(g)}}} = 0.0006\, ^{*} \,{{\text{ML}}_{{\text{(mm)}}}}^{2.5213} \, ({\text{R}}^{2} = 0.87, \ {\text{n}} = 98)$$
(b)

To characterize cephalopod habitat depth ranges, we relied extensively on the peer-reviewed literature and best-available distribution data from field and taxonomy guides. “Median habitat depth” was quantified for each taxon as the midpoint between the minimum and maximum reported depths. For cephalopods that exhibit ontogenetic descent, median habitat depths were quantified separately for adults and juveniles based on the reported ML at descent, and individuals were assigned to appropriate habitat depths based on their estimated sizes (Table S2). Observations of trawled specimens off Hawaii (e.g., Young 1978) were primarily used to assign depth habitats. However, reported depths were referenced across a taxon’s global distribution so that habitat assignments were unlikely to be limited by sampling depths in a single region (Table S2).

Cephalopod moisture content and energy density

Energy densities (kJ/g wet mass) were available for 8 out of the 66 taxa in our diet dataset. However, moisture content and energy density are inversely related (Ciancio et al. 2007; Schaafsma et al. 2018) and moisture content data were more readily available across the range of cephalopod taxa in our dataset. To estimate the energy density of cephalopod taxa from moisture content, we fit an ordinary least squares regression between energy density (ED) and moisture content (MC, as a percentage of body mass) using paired energy density and moisture content values for whole cephalopods (n = 14 taxa) from published literature (equation c, Figure S3; Clarke et al. 1985; Perez 1994; Schaafsma et al. 2018).

$${\text{ED}}_{{\text{(kJ}}/{\text{g)}}} = - 0.26731 \, {*} \, {\text{MC}} + 25.91423\, ({\text{R}}^{2} = 0.93, \, {\text{n}} = 14)$$
(c)

We measured moisture content for 114 cephalopods (represented as 10 unique taxa from five families) collected in Monterey Bay using samplers outfitted on remotely operated vehicles and midwater trawls. Whole cephalopods were frozen at − 80 °C at sea before being weighed and freeze-dried in the lab (following Hamilton et al. 2021). Percent moisture content was calculated as (1 − (Massdry/Masswet))*100. In the absence of species-specific moisture measurements, we used genus- or family-level averages (Table S2). We also used published moisture content data for six other cephalopod families (Clarke et al. 1985; Perez 1994; Schaafsma et al. 2018). Moisture content data for Argonautidae were not available and were substituted with data from a benthic octopus (Smale and Buchan 1981) as the two taxa have similar morphologies and tissue densities. There was limited variability in moisture content with size across taxa (Figure S4) and an average moisture content was applied to all individuals of a given taxa regardless of size. We assigned energy densities to each prey taxon using equation c and calculated the total energetic value (kJ) of individual prey (n = 1593). Family-level moisture content, mass, and energy densities are summarized in Table 1.

Table 1 Family-level mean moisture content, mass, and energy density for cephalopod prey. Table S2 contains the associated references with cited moisture content values
Full size table

Data treatment and diet analyses

We quantified cephalopod consumption for families contributing greater than 1% mean proportional abundance (\(\overline{p }\)) across all lancetfish sampled. Families contributing less than 1% \(\overline{p }\) were grouped into an “Other” category. This threshold was applied to limit the influence of rare taxa on comparisons among sample groups and data treatments. Cephalopods not identified at least to family were considered “Unidentified”. Lancetfish exhibit ontogenetic variability in diet at approximately 97 cm fork length (FL) (Portner et al. 2017), thus lancetfish were grouped into “small” (< 97 cm FL, n = 339) and “large” size classes (≥ 97 cm FL, n = 930). We summarized \(\overline{p }\) of cephalopod families represented by soft tissues and beaks within each size class and quantified the mean change in proportion when beaks were included in the analyses (\(\overline{p }\) beaks included – \(\overline{p }\) beaks excluded). Unless stated otherwise, all figure panels include data from both soft tissue and beak-identified cephalopods.

The main goal of our analyses was to examine how including and excluding beak-identified cephalopods affects our understanding of the overall importance of cephalopods in lancetfish diets. To visualize the effects of including cephalopod beaks on diet similarity between lancetfish size classes, we performed non-metric multidimensional scaling (nMDS) in the package “vegan” version 2.6-2 (Oksanen et al. 2019) in R version 3.5.3 (R Development Core Team 2019). To minimize the effects of zero-inflated data on similarity analyses between individual lancetfish (Clarke et al. 2006), we calculated family-level \(\overline{p }\) for groups of lancetfish that we would expect to have similar diets based on capture location and season. Both lancetfish size classes were grouped by year into northern and southern regions (at 25ºN) and seasonally (“winter” and “not winter” for small; “winter/spring” and “summer/fall” for large) following Portner et al. 2017. Unique size class groupings with fewer than five stomachs were excluded from similarity analyses to reduce the likelihood of skewed diet comparisons from small sample sizes (n = 37 and 34 stomachs when beaks were included and excluded, respectively). We performed analysis of similarities (ANOSIM) and permutational multivariate analysis of variance (PERMANOVA) in PRIMER-e version 7 (Clarke and Gorley 2015) to quantify diet similarity within and between size classes and determine how diet overlap changes with the inclusion of beak-identified cephalopods. All similarity analyses were performed on pairwise Morisita–Horn similarity indices between size classes with the R package “vegan” and excluded “Unidentified” and “Other” groups.

We also assessed differences in the sizes of cephalopods consumed between lancetfish size classes and whether including beaks influenced cephalopod size structure. For all of these comparisons, data were not normally distributed and sample groups were compared using the nonparametric Wilcoxon Rank-Sum Test in the package “stats” version 3.5.3 in R (R Development Core Team 2019). Cephalopod masses are reported as mean ± SD unless stated otherwise. We did not summarize the effects of including beak-identified cephalopods on small lancetfish diets because so few beaks were found within this lancetfish size class (n = 32 beak-identified cephalopods, Table S3).

Ecological designations for analyses

Cephalopods were categorized as being “gelatinous” or “muscular” based on moisture content and morphology. Cephalopods with greater than 88% moisture content are comparable to the moisture content of other gelatinous taxa (e.g., cnidarians and tunicates) (Doyle et al. 2007; Perrault 2019), and are hereafter considered “gelatinous”. Taxa with less than 88% moisture content are hereafter referred to as “muscular”. The morphology of taxa represented in these moisture content groups also aligns with swimming ability estimates by Murphy et al. 2020, who assigned activity level based on fin-to-ML ratios and mantle musculature. Gelatinous cephalopods in our dataset are considered low activity and muscular cephalopods are high and medium activity level.

Cephalopod taxa with median depths shallower than 500 m are hereafter referred to as “shallow-dwelling”, while cephalopods with median depths deeper than 500 m are “deep-dwelling”. This cut-off was based on the frequency of median depths occupied by all cephalopods consumed, where distinct maxima were observed above and below 500 m.

Results

A total of 2818 cephalopods representing 27 families were collected and identified from lancetfish stomachs (n = 1994 soft tissue and n = 824 beak-identified; Table S4), of which 401 individuals could not be identified at least to family (“Unidentified”, n = 181 soft tissue and n = 220 beak-identified). 22.6% of lancetfish stomachs that included cephalopods (n = 286) only contained beak-identified cephalopods. Length and mass data were available for 88.5% of identified cephalopods (n = 1739). Energy density and moisture content data were either measured or estimated from published regressions for 87.1% of identified cephalopods with mass data (n = 1516).

Small lancetfish mainly consumed muscular squids in the families Onychoteuthidae and Ommastrephidae, and gelatinous cephalopods represented less than 25% of total cephalopod prey by number (%N; Fig. 1a, Fig. 2 and 3). 96% of cephalopods from small lancetfish stomachs were represented as soft tissues (n = 688, Table S3), and the most abundant cephalopod was Hyaloteuthis pelagica (Family Ommastrephidae; n = 96, Fig. 2d). More than half of the cephalopods consumed by large lancetfish were gelatinous (Fig. 1b). 38% of cephalopods consumed by large lancetfish were identified from beaks (n = 792, Table S3), and the number of beak-identified cephalopods per stomach did not increase with FL within larger lancetfish (Figure S5). Amphitretid octopods, including Japetella diaphana (n = 262; Fig. 3c), were the most abundant cephalopod consumed by large lancetfish and were almost exclusively identified from soft tissues. Cranchiid squids in the genus Taonius (Fig. 3b) were the second most abundant prey for large lancetfish (n = 192), 97% of which were identified from beaks. Including beak-identified cephalopods in large lancetfish diets decreased the mean proportions of Amphitretidae and increased the mean proportions of Cranchiidae, Chiroteuthidae, and Histioteuthidae (Fig. 1c).

Fig. 1
figure 1

Mean proportional abundance of cephalopods in the stomachs of a small (< 97 cm FL) and b large lancetfish (≥ 97 cm FL). c Bar graph showing changes in the mean proportional abundance (\(\overline{p }\)) of cephalopods in the diets of large lancetfish when beaks were included versus excluded. Gelatinous families are indicated with bold outlines. Abbreviated family names: Amphitretidae (AMPH), Chiroteuthidae (CHIRO), Cranchiidae (CRAN), Argonautidae (ARGO), Enoploteuthidae (ENOP), Histioteuthidae (HIST), Octopodidae (OCTOD), Octopoteuthidae (OCTOT), Ommastrephidae (OMMA), Onychoteuthidae (ONYC), Tremoctopodidae (TREM), Unidentified (UNID)

Full size image
Fig. 2
figure 2

Representative taxa from muscular cephalopod families commonly consumed by longnose lancetfish. a Argonauta argo (Argonautidae), b Octopoteuthis deletron (Octopoteuthidae), c Onychoteuthis borealijaponica (Onychoteuthidae), d Hyaloteuthis pelagica (Ommastrephidae), e Walvisteuthis youngorum (Onychoteuthidae). Plate by A. Cano-Lasso Carretero

Full size image
Fig. 3
figure 3

Representative taxa from gelatinous cephalopod families commonly consumed by longnose lancetfish. a Grimalditeuthis bonplandi (Chiroteuthidae), b Taonius belone (Cranchiidae), c Japetella diaphana (Amphitretidae), d Galiteuthis phyllura (Cranchiidae), e Chiroteuthis spoeli (Chiroteuthidae). Plate by A. Cano-Lasso Carretero

Full size image

Diet variability between lancetfish size classes

Small and large lancetfish consumed significantly different cephalopod taxa (Fig. 4). When beak-identified cephalopods were included, large lancetfish diets became more different from each other (Table S5, increase in dispersion in Fig. 4b relative to 4a). Small and large lancetfish diets also became more different. Large lancetfish consumed larger cephalopods (52.7 ± 227.5 g) than did small lancetfish (5.2 ± 11.3 g; Wilcoxon rank sum, W = 167,620, p < 0.001; Fig. 5a). Including beak-identified cephalopods significantly increased the mean size of chiroteuthids (Wilcoxon rank sum, W = 6509, p < 0.001) and cranchiids (Wilcoxon rank sum, W = 3096.5, p < 0.001) consumed by large lancetfish (Fig. 5b). The 125 largest chiroteuthids and cranchiids by mass (represented by Chiroteuthis, Megalocranchia, Taonius, and Galiteuthis; Fig. 3) were identified from beaks. There were also clear differences in depth of forage between lancetfish size classes, with larger lancetfish feeding on cephalopods in deeper waters (Fig. 6a, b). Large chiroteuthid and cranchiid squids represented by beaks drastically increased the proportion of deeper-dwelling cephalopods in large lancetfish diets (Fig. 6c).

Fig. 4
figure 4

Non-metric multidimensional scaling (nMDS) plots showing diet overlap between small (< 97 cm FL) and large lancetfish (≥ 97 cm FL). Panels depict changes in overlap when beak-identified cephalopods were a excluded or b included. ANOSIM results (Global R and p-value) are reported for each treatment. Ordination vectors show families that explained the greatest diet variability (p < 0.001), and ellipses represent one standard deviation about the mean. The legend in panel A also applies to panel B

Full size image
Fig. 5
figure 5

a Log-transformed masses of cephalopods consumed by small (< 97 cm FL; n = 496 cephalopods) and large lancetfish (≥ 97 cm FL; n = 1249 cephalopods). b Log-transformed masses of cephalopods in the families Chiroteuthidae and Cranchiidae consumed by large lancetfish when beaks were excluded (n = 190 chiroteuthids and n = 130 cranchiids) versus included (n = 273 chiroteuthids and n = 341 cranchiids). Mean (black diamonds) and median (horizontal lines) masses and total mass distributions (density plots) are indicated on the boxplots for each treatment group

Full size image
Fig. 6
figure 6

Median depth ranges of cephalopods consumed by a small (< 97 cm FL) and b large lancetfish (≥ 97 cm FL). c Bar graph showing changes in the mean proportional abundance (\(\overline{p }\)) of cephalopods by habitat depth in the diets of large lancetfish when beaks were included versus excluded. No cephalopod taxa had median depth ranges between 1500 and 1750 m

Full size image

Depth-specific energetic values of cephalopods consumed by large lancetfish

Large lancetfish mostly consumed small muscular cephalopods at shallower depths (n = 288, median mass = 4.4 g) and large gelatinous cephalopods in deeper habitats (n = 687, median mass = 12.3 g; Fig. 7a, b). The mean mass and energetic value of gelatinous cephalopods increased with depth but did not change for muscular cephalopods (Fig. 7a, c). Deeper-dwelling gelatinous and muscular cephalopods had similar energetic values (95.3 ± 125.8 kJ and 120.2 ± 169.4 kJ, respectively; Fig. 7c), but gelatinous cephalopods were more abundant (79.6% of cephalopods with median depths > 500 m; Fig. 7b).

Fig. 7
figure 7

Masses and energetic contents of muscular (dark red) and gelatinous (light red) cephalopods consumed by large lancetfish (≥ 97 cm FL). All mass values are log-transformed. a Density plot showing the relative frequency of cephalopod mass (log-transformed) within 250 m median habitat depth bins. The Y-axis labels represent the maximum depth of each bin. The vertical lines indicate the mean mass per group. No taxa had median depth ranges between 1500 and 1750 m. b Density plot showing relative frequency of median depths for cephalopods consumed. c Mean (± SD) relationships between cephalopod mass and energetic content (both log-transformed) with respect to habitat depth

Full size image

Discussion

Cephalopod beaks are commonly excluded from stomach content analyses to conservatively describe fish diets. However, excluding prey items may obscure important energetic pathways in pelagic food webs, especially for deep-sea predators whose feeding observations are limited. We demonstrate significant changes in the taxonomic composition and size of cephalopods abundantly consumed by lancetfish when beak-identified prey are integrated into diet analyses, and present new evidence on the importance of gelatinous cephalopods in deep-sea food webs.

Beak-identified cephalopods reveal niche partitioning among lancetfish size classes

Including beaks revealed increased use of mesopelagic resources by large lancetfish, providing insights into the drivers of ontogenetic diet shifts. Many pelagic cephalopods exhibit ontogenetic descent (Roper and Young 1975), so larger cephalopods may be more abundant at depth. While the size of muscular cephalopods did not change significantly with foraging depth, large lancetfish did feed on larger gelatinous cephalopods with increasing depth when beak-identified prey were considered. Ontogeny in lancetfish diets could be explained by changes in predator–prey habitat overlap and increased accessibility to larger cephalopods in deeper waters.

Large, gelatinous cephalopods provide calorie-rich meals at depth

The energy density of gelatinous cephalopods per gram of tissue is 3–16 times lower than muscular taxa (Table 1, Figure S3). However, the large gelatinous cephalopods consumed by large lancetfish provided similar individual energetic values as smaller muscular cephalopods (Fig. 7c). Cranchiid and chiroteuthid squids had much greater masses on average than any other prey regularly consumed by lancetfish (Choy et al. 2013; Portner et al. 2017), suggesting they may be easier to capture than large muscular prey. Thus, predators may have access to a larger size range of prey in the mesopelagic and some of these taxa may represent calorie-rich options for other mesopelagic predators.

Gelatinous cephalopods in pelagic food webs

Gelatinous squids are important prey for several seabirds and deep-diving mammals, and these associations have primarily been made using the remains of beaks in stomachs. Cranchiids (e.g., Taonius pavo and Teuthowenia sp.) are abundant prey of beaked whales, which can forage at depths over 1500 m (Tyack et al. 2006) (e.g., Cuvier’s beaked whales (West et al. 2017), strap-toothed whales (Sekiguchi et al. 1996), and bottlenose whales (Fernández et al. 2014)). Cranchiid (e.g., Galiteuthis glacialis) and chiroteuthid (e.g., Chiroteuthis sp.) squids are also abundant in seabird diets (e.g., albatross (Cooper and Klages 1995; Xavier et al. 2003), petrel (Bester et al. 2011), and shearwater (Neves et al. 2012)). Gelatinous octopods (especially the amphitretid Japetella sp.) have been described in the diets of several fishes (Galván-Magaña et al. 2006; Ménard et al. 2013) but are low-calorie prey compared to the gelatinous squids consumed by large lancetfish. If gelatinous cephalopods are better represented as beaks across predator diets, more thoughtful consideration of beaks in fish diet analyses is likely to change our understanding of foraging depths, prey size spectra, and resource partitioning among pelagic predators.

Cephalopods as prey in a changing ocean

Cephalopods have relatively short lifespans and respond quickly to changes in environmental conditions (O’Dor and Webber 1986). This plasticity in life history may help explain the proliferation of muscular cephalopod species in the global ocean from the 1950s–2010s (Doubleday et al. 2016). Muscular, epipelagic cephalopods in the families Ommastrephidae (e.g., Dosidicus gigas and Sthenoteuthis oualaniensis) and Argonautidae (Argonauta sp.) are abundant in the diets of many commercially targeted fishes (e.g., swordfish and yellowfin tuna; Olson et al. 2014; Trujillo-Olvera et al. 2018). Climate-induced changes in oceanographic conditions have been linked to an increased reliance on ommastrephid squids by yellowfin tuna from the eastern tropical Pacific (Olson et al. 2014). This diet shift may reflect bottom-up changes in prey availability for yellowfin tuna that results in increased diet overlap with small lancetfish. However, lancetfish switch to feeding on large gelatinous cephalopods at depth with increasing size, reducing resource overlap with commercially targeted fishes. Predators that forage more broadly across the water column may thus have a competitive advantage when responding to shifts in prey availability under changing environmental conditions.

Caveats and future directions

Interpretations from our findings were limited by scarce biological information for many of the cephalopods consumed. Median habitat depths used in this study were based on the reported range of a given taxa and may not reflect the most used depths. Relationships between beak morphologies and body size for many deep pelagic cephalopods are not well understood (Xavier et al. 2007) as beak identification requires extensive taxonomic expertise. Beak-identified cephalopods may also be skewed toward larger individuals if smaller beaks have lower residence times in stomachs. This bias does not detract from our findings that gelatinous cephalopods are an important part of lancetfish diets and are largely overlooked without beak identifications. Although disentangling the relationship between beak size and retention rate was beyond the scope of this study, simply excluding beaks will bias overall diet composition. Increased sampling of deeper-dwelling cephalopods and more regular examinations of beaks in predator diets would greatly improve our understanding of deep pelagic food web structure.