Journal list menu

Volume 63, Issue 3 p. 1168-1180
Free Access

Stable isotope analysis of micronekton around Hawaii reveals suspended particles are an important nutritional source in the lower mesopelagic and upper bathypelagic zones

Kristen Gloeckler

Kristen Gloeckler

Department of Oceanography, University of Hawaii Manoa, Honolulu, Hawaii

Search for more papers by this author
C. Anela Choy

C. Anela Choy

Monterey Bay Aquarium Research Institute, Moss Landing, California

Search for more papers by this author
Cecelia C. S. Hannides

Cecelia C. S. Hannides

Department of Oceanography, University of Hawaii Manoa, Honolulu, Hawaii

Search for more papers by this author
Hilary G. Close

Hilary G. Close

Department of Geology and Geophysics, University of Hawaii, Honolulu, Hawaii

Search for more papers by this author
Erica Goetze

Erica Goetze

Department of Oceanography, University of Hawaii Manoa, Honolulu, Hawaii

Search for more papers by this author
Brian N. Popp

Brian N. Popp

Department of Geology and Geophysics, University of Hawaii, Honolulu, Hawaii

Search for more papers by this author
Jeffrey C. Drazen

Corresponding Author

Jeffrey C. Drazen

Department of Oceanography, University of Hawaii Manoa, Honolulu, Hawaii

Correspondence: [email protected]Search for more papers by this author
First published: 18 December 2017
Citations: 48


Several studies have found that the respiratory demand for carbon by the mesopelagic community exceeds carbon supply through the particulate sinking flux by up to two to three orders of magnitude, suggesting that mesopelagic communities rely on additional overlooked carbon sources. Suspended particles (defined as 0.7–53 μm) have been suggested as one of these sources but few studies have evaluated their contribution to the mesopelagic food web. We use amino acid compound-specific nitrogen isotope analysis to investigate whether suspended particles are an important nutritional source to fish, cephalopod, and crustacean micronekton species in the central North Pacific. Our results suggest that micronekton feed from food webs fueled by a variety of nutritional sources including surface dwelling phytoplankton and bacteria, sinking particles, and suspended particles, with micronekton becoming more reliant on suspended particles with increasing habitat depth. Several species were identified as feeding from a primarily suspended particle-based food web including the fishes Cyema atrum, Cyclothone pallida, Melanocetus johnsonii, Serrivomer sector, and the pelagic octopod Japetella diaphana. We also found that micronekton species feeding from a suspended particle food web cannot be identified using bulk tissue δ15N values. Our data support the notion that suspended particles are an important nutritional source in the lower mesopelagic and upper bathypelagic and they should be accounted for when estimating carbon supply for these communities.

The mesopelagic (∼ 200–1000 m) and bathypelagic (1000–3000 m) zones of the ocean comprise the largest habitat on the planet (Robison 2004). However, due to the remoteness of this habitat and difficulty in sampling its inhabitants, very little is known about the function and structure of the pelagic food web and the corresponding ecosystem. Many studies have examined the controls on primary production as well as the distribution and feeding ecology of top predators in the open ocean, but few studies have been done on the mesopelagic zooplankton and micronekton which are the primary trophic link between primary producers and top predators (Brodeur and Yamamura 2005).

The mesopelagic food web plays an important role in fisheries and the sequestration of carbon through the ocean's biological pump. For example, ecosystem-based fisheries management, which links exploited commercial species to their prey and competitors (Grumbine 1994; Choy et al. 2016), relies on an understanding of the entire food web including mesopelagic zooplankton and micronekton. Some zooplankton and micronekton migrate daily to surface waters at night to feed under the cover of darkness and then return to depth during the day to hide from predators; this diel vertical migration carries organic carbon to mesopelagic depths (Longhurst et al. 1990; Steinberg et al. 2000; Al-Mutairi and Landry 2001; Steinberg et al. 2008; Hannides et al. 2009aa).

Several studies have attempted to estimate the supply of carbon transported to depth via the biological carbon pump as well as the demand for carbon in the mesopelagic, and have found that the respiratory demand for carbon exceeds the amount of carbon being delivered there by sinking particulate organic carbon (POC) fluxes by as much as two to three orders of magnitude (Boyd et al. 1999; Steinberg et al. 2008; Baltar et al. 2009; Burd et al. 2010). Thus, either carbon demand in the mesopelagic zone is overestimated or carbon supply to the mesopelagic is underestimated. Many of the studies that estimated carbon supply used sediment traps to measure the downward flux of carbon from the surface. Some common sediment trap designs can lead to inaccurate estimates of sinking POC flux because free-swimming organisms enter the sediment traps, and particles can be swept out by flow within the sediment traps (Buesseler 2007). Organisms may also sustain themselves through diel vertical migration or by feeding from a pool of smaller particles that are not effectively captured in sediment traps (Steinberg et al. 2000; Al-Mutairi and Landry 2001; Steinberg et al. 2008; Baltar et al. 2009; Giering et al. 2014; Mayor et al. 2014). In the North Atlantic, mesopelagic carbon budgets were balanced by using a very shallow and dynamic upper boundary to the zone and by excluding the respiratory demand of vertical migrators from the budgets, a mechanism to account for their feeding in surface waters (Giering et al. 2014).

Few studies have estimated the relative contribution of small particles to downward carbon flux. Particles are frequently sampled operationally by size into pools that are thought to be primarily “suspended” (small, here 0.7–53 μm) or primarily “sinking” (large, here > 53 μm) in studies of vertical carbon flux (e.g., Bishop and Edmond 1976; Lam et al. 2011; Puigcorbè et al. 2015) and organic geochemical studies (e.g., Abramson et al. 2010). The standing stock concentration of organic carbon in small particles in the mesopelagic is quantitatively larger than that of large particles (Verdugo et al. 2004; Baltar et al. 2009). In some places such as the Eastern Tropical North Pacific, suspended particles have been shown to be major contributors to carbon flux (Puigcorbè et al. 2015) and therefore, if consumed, could make up some of the deficit in the carbon supply at mesopelagic depths and below. Baltar et al. (2009) found a strong positive relationship between suspended particle concentrations and bacterial activity in the subtropical North Atlantic, suggesting that suspended particles are an important source of carbon to deep ocean microbial communities. However, Baltar et al. (2009) did not investigate how the suspended particles might supply carbon to higher trophic levels.

Stable nitrogen isotope analysis is a commonly used tool for studying food webs. The nitrogen isotope composition (δ15N values) has often been used as an indicator of relative trophic position (TP) because it increases ∼ 2–3‰ with respect to the food source between predator and prey (Heikoop et al. 1998). However, the nitrogen isotope composition of an organism will also vary with the δ15N value of the base of the food web. Amino acid compound-specific isotope analysis (AA-CSIA) is a tool which can be used to distinguish between changes in nitrogen isotope composition driven by differences at the base of the food web from changes in TP. “Trophic” amino acids (e.g., alanine, aspartic acid, glutamic acid, isoleucine, proline, valine), fractionate in a predictable way (∼ 4–7‰) with increasing TP, while “source” amino acids (e.g., glycine, lysine, methionine, phenylalanine, serine, tyrosine) have δ15N values that change by < 1‰ with increasing trophic level (McClelland and Montoya 2002; Chikaraishi et al. 2009). Therefore, source amino acids reflect the nitrogen isotopic compositions of different food sources at the base of the food web. Additionally, this technique allows one to estimate the TP of an organism from a single sample using the difference between the δ15N values of “trophic” and “source” amino acids (Chikaraishi et al. 2009; Bradley et al. 2015). Choy et al. (2012) used AA-CSIA to confirm that the lanternfishes and dragonfishes from five global ocean regions had the same TPs, and further that the large differences observed in bulk tissue δ15N values of each species were driven by differences in the δ15N values of primary producers in the different regions as recorded in the δ15NsourceAA values in these fishes.

Sinking particles and suspended particles have distinct δ15NsourceAA values in the mesopelagic zone at Sta. ALOHA (Hannides et al. 2016; see results below), and these differences can be used to trace the relative contributions of these two nutritional sources into the diets of organisms at higher trophic levels. Hannides et al. (2013) measured the δ15NsourceAA values of suspended particles and zooplankton, finding that δ15NsourceAA values of suspended particles increased by up to 14‰ between the surface and mesopelagic depths. A similar increasing trend was present at smaller magnitudes in zooplankton, suggesting that zooplankton in the mesopelagic relied on suspended particles by as much as 38%. Choy et al. (2015), found that the δ15N values of three source amino acids (phenylalanine, glycine, and serine) also increased with depth in a variety of micronektonic fishes with at least one fish, Cyema atrum, appearing to rely on nitrogen largely from a suspended particle based food web.

We build upon the findings of Hannides et al. (2013) and Choy et al. (2015), who demonstrated evidence for zooplankton and micronekton feeding from a suspended particle based food web in the mesopelagic, by characterizing both suspended and sinking particles and extending animal types and depths of occurrence. In this study, we further investigate whether suspended particles are an important carbon source for a diversity of mesopelagic micronekton around Hawaii from the surface to depths of 1500 m. We use δ15NsourceAA values in micronekton and compare these values to the δ15NsourceAA values of sinking and suspended particles from the same ecosystem. We find that suspended particles are an important nutritional source to animals living in the lower mesopelagic and upper bathypelagic zones. Additionally, we investigated whether micronekton species that feed from a suspended particle food web can be identified using bulk tissue δ15N analysis because this method is faster and more affordable than AA-CSIA. We found that micronekton species feeding from a suspended particle food web cannot be identified using bulk tissue δ15N values alone and therefore AA-CSIA is necessary to identify nutritional links to particle food resources.


Sample collection

Micronekton were collected using a 10 m2 multiple opening closing net and environmental sensing system (MOCNESS) at Sta. ALOHA (22.45°N, 158°W) in March and August of 2011 and in February and September of 2014 with a few samples from other locations around Oahu in 2011 (Choy et al. 2015). Micronekton were collected over five depth zones between the surface and 1500 m: 0–100 m, 100–500 m, 500–700 m, 700–1000 m, and 1000–1500 m. At sea, micronekton were sorted and identified to the most specific taxonomic level, then measured and photographed. Standard length measurements were taken for fish, carapace length and total length were taken for crustaceans and both mantle length and total length were taken for cephalopods. For most fishes, white muscle tissue was removed and frozen in a cryovial in liquid nitrogen. Small fishes, crustaceans, and gelatinous organisms were frozen whole or individuals were pooled for sufficient tissue required for stable isotope analysis. Specimens were transferred to a −80°C freezer until the samples could be prepared for stable isotope analysis.

Bulk tissue nitrogen isotope analysis

Eighty-three samples (individual specimens or small groups of conspecifics) were selected for stable isotope analysis. Samples selected for stable isotope analysis represented different combinations of trophic strategies (suspension feeding, zooplanktivores, micronektonivores), depth guilds (epipelagic, mesopelagic, bathypelagic) and migrating behaviors based on available ecological information (e.g., Clarke 1973; Maynard 1982). Each sample was freeze dried and ground using a ceramic mortar and pestle. For bulk tissue carbon and nitrogen isotope analysis, approximately 0.5 mg of each sample was weighed and placed into a tin boat. Carbon and nitrogen isotopic compositions were determined using an isotope ratio mass spectrometer (DeltaPlusXP) coupled to an elemental analyzer (Costech Model 4010). Isotopic ratios are given in δ-notation relative to the international standards VPDB and atmospheric N2. Accuracy and precision were < 0.2‰ based on glycine and homogenized fish tissue reference materials analyzed every 10 samples. The isotopic compositions of the reference materials have been extensively characterized using NIST certified reference materials in the UH laboratory and verified independently in other isotope laboratories.

Amino acid compound specific stable isotope analysis

For AA-CSIA, approximately 15 mg (dry weight) of each sample underwent acid hydrolysis and derivatization yielding trifluoroacetic (TFA) amino acid esters following the methods of Popp et al. (2007) and Hannides et al. (2009bb). The nitrogen isotope composition of the TFA amino acid esters were determined using an isotope ratio mass spectrometer (Thermo Scientific Delta V Plus or Thermo Scientific MAT 253 IRMS) interfaced with a Thermo Finnigan GC-C III. Samples were injected onto a BPx5 forte capillary column (60 m × 0.32 mm × 1.0 μm film thickness) at an injector temperature of 180°C with a constant helium flow rate of 1.4 mL min−1. The column was initially held at 50°C for 2 min and then increased at a rate of 15°C min−1 to 120°C. Temperature was then increased at a rate of 4°C min−1 to 195°C, then to 255°C at a rate of 5°C min−1 and finally to 300°C at a rate of 15°C min−1, holding at the final temperature for 8 min. Each sample was analyzed in triplicate and co-injected with the reference compounds norleucine (Nor) and aminoadipic acid (AAA) of known isotopic composition. A suite of pure amino acids of known nitrogen isotopic composition (Ala, Thr, Ile, Pro, Glu, and Phe) was also injected every three runs as an extra measure of accuracy for the instrument. Reference compounds Nor and AAA as well as the suite of amino acids were used to normalize the measured isotope values. Standard deviation for all amino acids averaged ± 0.4‰ (range 0.0–3.1‰).

DNA barcoding Cyclothone individuals

Cyclothone is a cosmopolitan genus of meso/bathypelagic fishes within the family Gonostomatidae that consists of the most abundant micronekton in Hawaiian waters (Maynard et al. 1975) and most other regions of the oceans (e.g., Sutton et al. 2013) and have been considered the most abundant vertebrates on the planet (Herring 2002; Nelson 2006). Members of the genus Cyclothone are small, can be difficult to identify morphologically, and have known cryptic species (Miya and Nishida 1997). Individuals of Cyclothone pallida, exhibited variability in isotopic values; to help explain some of this variability DNA barcoding was employed to test for the presence of cryptic species. DNA was extracted using the DNeasy Blood & Tissue kit (Qiagen) following the manufacturer's recommended protocols, using freeze-dried and ground tissue. A 624-bp fragment of mitochondrial cytochrome c oxidase subunit I (mtCOI) was amplified in polymerase chain reaction (PCR) using primers described in Ward et al. (2005). Reactions were run in 20 μL volumes with 10 μL MangoMix (Bioline), 0.15 μL bovine serum albumin (NEB Biolabs), and 2 μL of template DNA. PCR cycle conditions were as follows: 95°C for 2 min, followed by 40 cycles of 95°C for 30 s, 45°C for 30 s, 72°C for 1 min, with a final extension at 72°C for 4 min. PCR reactions were purified using shrimp alkaline phosphatase and exonuclease I, and Sanger sequences were obtained from both strands on an ABI 3730XL. Sequences were edited, aligned, and trimmed using Genious v7.1.8 (Biomatters), with a final alignment using MUSCLE (Edgar 2004) and including sequences from NCBI of Cyclothone microdon, Cyclothone atraria, Cyclothone pacifica, Cyclothone acclinidens, Cyclothone parapallida, Cyclothone pseudopallida, and Cyclothone braueri (EU148134–EU148139, FJ164515–FJ164523, GQ860355–GQ860358, GU071722, GU071728, GU071729, GU071740, GU071741, GU440298, HQ010054, JN640877–JN640879, KF929802, Steinke et al. 2009; Bucklin et al. 2010; Kenaley et al. 2014; Harada et al. 2015). Kimura 2-parameter (K2P) genetic distances were calculated in MEGA v6.06 (Tamura et al. 2013).

Data analysis

For each sample analyzed, the value of δ15NsourceAA represents a weighted mean of the four “source” amino acids glycine, serine, phenylalanine, and lysine, weighted by each compound's standard deviation. TP was calculated using the difference between the trophic amino acids alanine, leucine and glutamic acid, and the source amino acids glycine, lysine and phenylalanine, a β value of 3.6‰, and a trophic discrimination factor (TDFAA) of 5.7‰ (after Bradley et al. 2015). For statistical comparison to micronekton, weighted average values of δ15NsourceAA for suspended and sinking particles were calculated from compound-specific amino acid data, using the same amino acids as for micronekton. Briefly, particles were collected on each cruise using in situ filtration (WTS-LV, McLane Research Laboratories), employing in situ size fractionation by filtering sequentially onto 53 μm nylon (Nitex) mesh and two, stacked 0.7 μm precombusted glass microfiber filters (GF/F) mounted on a two-tiered filter holder. Filters were freeze-dried and analyzed using AA-CSIA using the methods described above. Particle data from eight discrete depths (25–1205 m) were averaged across two seasons (February 2014 and August 2014) to match the seasonal pooling of the micronekton data (see below), and data from sampling replicates (depths closer than 11 m) were averaged.

Among possible predictors of isotopic variability in micronekton, we first evaluated the effects of seasons. This effect was best assessed separately for the four fish species (Bolinichthys distofax, C. pallida, Chauliodus sloani, Melanocetus johnsonii) in which samples were collected in both summer and winter. No significant differences were found (Wilcoxon matched pairs test; p = 0.47) which was expected given their multiyear lifespan. Subsequently, a statistical model was created to test the hypothesis that variability in δ15NsourceAA values of micronekton are due to differences in species, length, depth, region, and year (R version 3.1.2). As a preliminary assessment, each of the predictors for the model was plotted against δ15NsourceAA values of the micronekton. Not all variables were found to have linear relationships with the δ15NsourceAA values of the micronekton so a general additive model (R-package “mgcv”) was used to best explain the variability in the data. Finally, to explore intraspecific variation, linear regressions with depth and length were used to investigate variability in δ15NsourceAA values in C. pallida, the only species with a large enough sample size to perform these analyses.


Summary of data collected

The δ15N values of amino acids were measured for 83 individuals representing 25 different species of micronekton (Table 1), a fraction of which were previously reported in Choy et al. (2015). Crustaceans ranged in size from 21 mm total length in Euphausia sp. to 102.5 mm total length in Sergia sp. Cephalopods ranged in size from 31.1 mm total length in Histioteuthis sp. to 290 mm mantle length in Sthenoteuthis oualaniensis. Fish ranged in size from 22.9 mm standard length in Hygophum proximum to 507 mm total length in Serrivomer sector. TPs overlapped between the three types of feeding guilds with micronektonivores ranging from a TP of 3.1 to 4.5, zooplanktivores ranging from a TP of 2.6 to 4.4, and a TP of 3.8 for the suspension feeder in this study, Euphausia sp. Bulk tissue δ15N values ranged from 5.4‰ in juvenile Thunnus albacares to 13‰ in M. johnsonii.

Table 1. Summary of data collected for each species in this study. The symbol (J) after a species name indicates the individuals were juveniles of the species. N is the sample size. For animal type column, Ceph, Cephalopod; Crust, Crustacean. For feeding guild column, M, micronektonivore; S, suspension feeder; Z, zooplanktivore. For size type column, SL, standard length; FL, fork length; TL, total length; CL, carapace length; ML, mantle length. Source AA values (mean ± standard deviation) were calculated using a weighted mean of glycine, serine, phenylalanine, and lysine δ15N values. TP was calculated using the difference between the trophic amino acids alanine, leucine, and glutamic acid, and the source amino acids glycine, lysine, and phenylalanine, a β value of 3.6‰, and a TDFAA of 5.7‰ (Bradley et al. 2015).
No. Species N Animal type Day depth range (m) Night depth range (m) Depth Ref. Feeding guild Size (mm) Size type δ15N (‰) source AA δ15N (‰) δ13C (‰) Average TP
1 Histioteuthis sp. 2 Ceph 375–850 100–500 13 M 31.1 TL 2 ± 1.3 10.4 ± 2.1 −20.1 ± 1.1 4.0 ± 1.3
2 Japetella diaphana 2 Ceph 725–1065 725–1065 13 Z 48.4–53.2 ML 4.1 ± 0.6 7.6 −20 2.9 ± 0.1
3 Sthenoteuthis oauliensis 3 Ceph 650 0 16 M 245–490 ML −1.2 ± 1.1 8.7 ± 0.7 −18.2 ± 0.1 4.5 ± 0.5
4 Euphausia sp. 3 Crust 175 84 9 S 21–40.1 TL 0.5 ± 1.3 6.8 ± 0.8 −19.7 ± 0.4 3.8 ± 0.6
5 Gennadas sp. 3 Crust 600–1100 100–1000 12 Z 10.8–11.9 CL 1.3 ± 2 6 ± 0.2 −19.6 ± 0.5 3.8 ± 0.1
6 Oplophorus gracilirostris 2 Crust 490–650 60–750 15 Z 11.8–12.2 CL 0.3 ± 0.7 5.4 ± 0.9 −20.8 ± 1 3.3 ± 0.1
7 Sergia sp. 3 Crust 800 150 15 Z 22.5–27.2 CL 1.7 ± 2.6 7.2 ± 1.2 −19 ± 1.4 3.8 ± 0.2
8 Stomatopod (J, unidentified) 2 Crust 0–100 0–100 7 Z 12.2–33 TL 0.5 ± 0.9 4.7 ± 0.4 −19.7 ± 0.1 2.8 ± 0.4
9 A. cornuta 2 Fish 750–1150 275–980 1 M 88.6 SL −0.5 ± 3.7 7 ± 0 −18.7 ± 0.1 3.5 ± 0.7
10 B. distofax 4 Fish 490–690 490–690 2 Z 48.6–87.7 SL 0.2 ± 1.1 8.6 ± 0.9 −18.6 ± 0.7 3.8 ± 0.2
11 B. longipes 2 Fish 525–725 50–150 2 Z 32–46 SL 0.1 ± 0.6 6 ± 0.4 −19 ± 1 2.9
12 C. hippurus (J) 1 Fish 0–100 0–100 3, 4 Z 180 FL −0.2 6.2 17.3 3.3
13 C. sloani 5 Fish 480–825 45–225 5 M 52.3–174 SL 0.2 ± 1.4 9.7 ± 2.5 −18 ± 0.8 3.4 ± 0.5
14 Cyclothone alba 4 Fish 425–465 425–465 6 Z 24.5–26.5 SL 0.9 ± 1.7 6.4 ± 1.1 −19.9 ± 0.4 2.7 ± 0.1
15 C. pallida 12 Fish 600–1500 600–1500 6, 7 Z 24–58.5 SL 4.3 ± 2.7 11 ± 2.6 −19.5 ± 1.9 3.1 ± 0.4
16 C. braueri 1 Fish 525–750 525–750 6 Z 38.1 SL 1.7 8.2 −22.3 2.7
17 C. atrum 3 Fish 1200–1400 1200–1400 8 Z 103–139 SL 6.9 ± 2.9 7 ± 2 −18.3 ± 1.7 2.9 ± 0.1
18 Exocoetus sp. 2 Fish 0–10 0–10 10, 11 Z 75–160 SL −0.4 ± 3.8 8.6 ± 0.2 −17.7 ± 1.3 3.5 ± 0
19 H. proximum 4 Fish 500–700 0–300 5 Z 22.9–42.2 SL 1.2 ± 3.6 6.8 ± 0.5 −18.7 ± 1.2 2.9 ± 0.1
20 M. johnsonii 3 Fish 500–1500 500–1500 14 M 21–98 SL 5.5 ± 2 12.7 ± 1.5 −19 ± 0.9 3.9 ± 0
21 Myctophum lynchnobium 2 Fish 600–800 0–15 2 Z 24.6–106 SL −1.4 ± 0.4 8.6 ± 2.3 −18.7 ± 0.8 3.5 ± 0.5
22 O. soleatus 3 Fish 450–550 450–550 8 Z 67–85.1 SL 1.9 ± 0.8 9.8 ± 0.6 −20 ± 1.3 4.4 ± 0.6
23 Scopelogadus sp. 3 Fish 600–1000 150–250 1 Z 29–37.4 TL 0 ± 1.2 5.6 ± 0.4 −20.4 ± 0.3 2.6 ± 0.1
24 S. sector 3 Fish 550–1500 550–1500 8 M 356–507 TL 4.5 ± 2.9 10.4 ± 2.6 −18.3 ± 1 3.1 ± 0.1
25 T. albacares (J) 2 Fish 0–110 0–110 17 Z 139–191 FL −0.4 ± 0.1 5.4 ± 0.1 −17.4 ± 0.4 3.2 ± 0.4
  • 1 Clarke, T. A., and P. J. Wagner. 1976. Vertical distribution and other aspects of the ecology of certain mesopelagic fishes taken near Hawaii. Fish. Bull. 74: 635–645.
  • 2 Clarke, T. A. 1973. Some aspects of ecology of lanternfishes (Myctophidae) in the Pacific Ocean near Hawaii. Fish. Bull. 71: 401–434.
  • 3 Palko, B. J., G. L. Beardsly, and W. J. Richards. 1982. Synopsis of the biological data on dolphinfishes, Coryphaena hippurus Linnaeus and Coryphaena equiselis Linnaeus. FAO Fish. Synop. 130: 28.
  • 4 Lasso, J., and L. Zapata. 1999. Fisheries and biology of Coryphaena hippurus (Pisces: Coryphaenidae) in the Pacific coast of Colombia and Panama. Sci. Mar. 63: 387–399.
  • 5 Clarke, T. A. 1974. Some aspects of ecology of stomiatoid fishes in Pacific Ocean near Hawaii. Fish. Bull. 72: 337–351.
  • 6 Maynard, S. D. 1982. Aspects of the biology of the mesopelagic fishes of the genus Cyclothone (Pisces: Gonostomatidae) in Hawaiian waters. Ph.D. thesis. Univ. of Hawaii.
  • 7 Gloeckler and Drazen unpubl. data.
  • 8 Amesbury, S. S. 1975. The vertical structure of midwater fish community off leeward Oahu, Hawaii. Ph.D. thesis. Univ. of Hawaii.
  • 9 Steinberg, D. K., J. S. Cope, S. E. Wilson, and T. Kobari. 2008. A comparison of mesopelagic mesozooplankton community structure in the subtropical and subarctic North Pacific Ocean. Deep-Sea Res. II 55: 1615–1635.
  • 10 Gorelova, T. A. 1980. The feeding of young flyingfishes of the family Exocoetidae and of the smallwing flyingfish, Oxyporhamhus micropterus, of the family Hermirhamphidae. J. Ichthyol. 20: 60–71.
  • 11 Lipskaya, N. 1987. Feeding of flyingfish (Exocoetidae) larvae and fingerlings in the region of the Peruvian upwelling. J. Ichthyol. 27: 108–116.
  • 12 Riggs, F. 1977. Vertical distribution and migration patterns of the Hawaiian mid-water Penaeidae. M.S. thesis. Univ. of Hawaii.
  • 13 Young, R. E. 1978. Vertical distribution and photosensitive vesicles of pelagic cephalopods from Hawaiian waters. Fish. Bull. 76: 583–615.
  • 14 Pietsch, T. W. 2009. Oceanic angler fishes: Extraordinary diversity in the deep-sea. Univ. of California Press.
  • 15 Cowles, D. L., J. J. Childress, and M. E. Wells. 1991. Metabolic rates of midwater crustaceans as a function of depth of occurrence off the Hawaiian Islands: Food availability as a selective factor? Mar. Biol. 110: 75–83.
  • 16 Parry, M. P. 2003. The trophic ecology of two ommastrephid squid species, Ommastrephes bartramii and Sthenoteuthis oaulaniensis, in the North Pacific sub-tropical gyre. Ph.D. thesis. Univ. of Hawaii.
  • 17 Graham, B. S. 2007. Trophic dynamics and movements of tuna in the tropical Pacific Ocean inferred from stable isotope analysis. Ph.D. thesis. Univ. of Hawaii.

Bulk tissue δ15N vs. δ15NsourceAA values and TP

A positive linear correlation was found when comparing bulk tissue δ15N values and δ15NsourceAA values (r2 = 0.26 for all data; Fig. 1). Micronekton were divided into three roughly even groups that each spanned about 0.6 TPs with close to the same number of species in each group. Correlations between δ15N values and δ15NsourceAA values were not found to be significant for any TP group (TP 2.5–3.1, p = 0.11; TP 3.1–3.7, p = 0.22, driven by a single sample; TP 3.7–4.5, p = 0.06). For each TP group, bulk tissue δ15N values increased linearly with increasing δ15NsourceAA values (Fig. 1) and had similar slopes but low r2 values. Y-intercepts increased from the group of the lowest TPs to the group of the highest TPs.

Details are in the caption following the image

Bulk tissue δ15N values of micronekton as a function of δ15NsourceAA values. δ15NsourceAA values have been averaged for each species. Micronekton have been grouped by TP shown by different colors

These relationships were also evaluated intraspecifically for C. pallida (Fig. 2). A strong, positive correlation was found between bulk tissue δ15N values and δ15NsourceAA values (analysis of variance (ANOVA) p < 0.05, df = 12). Bulk tissue δ15N values and TP were also found to have a positive but weaker and nonsignificant correlation (p = 0.07).

Details are in the caption following the image

Bulk tissue δ15N values of C. pallida (fish) as a function of δ15NsourceAA values (a, p < 0.05) and TP (b, p = 0.07).

Amino acid-compound specific stable isotope analysis

There are large differences between δ15NsourceAA values of small and large particles in the mesopelagic: between 250 m and 1200 m, small particles had an average δ15NsourceAA value of 6.3 ± 1.3‰, while large particles had an average δ15NsourceAA value of 2.4 ± 2.2‰. In the upper water column (25–150 m), δ15NsourceAA values small and large particles were similar to one another, averaging −1.1 ± 2.6‰ and −0.1 ± 1.6‰, respectively, but significantly lower than either mesopelagic pool (averages of values shown in Fig. 3).

Details are in the caption following the image

δ15NsourceAA values (weighted mean and standard deviation) of several micronekton species plotted against their median daytime depth of occurrence (a) and their median nighttime depth of occurrence (b). For both figures, the blue shaded area represents the standard deviation of δ15NsourceAA values of small particles (0.7–53 μm) and the red shaded area represents standard deviation of the δ15NsourceAA values of large particles (> 53 μm; Supporting Information Table S2). Where error bars do not appear only data for a single season was available and the width of the shaded area represents a conservative width of 1.0. Circular points represent nonmigrating organisms and triangular points represent migrating organisms. Numbers correspond to taxa in Table 1.

The δ15NsourceAA values of micronekton increased with depth, approaching the δ15NsourceAA values of suspended particles in the lower mesopelagic/upper bathypelagic, particularly when mean nighttime depth of occurrence was considered (Fig. 3). Depth was found to be the only significant predictor of δ15NsourceAA (daytime gam, p < 0.001; nighttime gam, p < 0.001) with daytime depth having less explanatory power (41% of variability in δ15NsourceAA values) compared to the nighttime depths (50% of the variability in δ15NsourceAA values explained). Other predictors including season (assessed separately for four species), species, total length, region of collection, or year of collection were not significant. In the epipelagic, δ15NsourceAA values of micronekton, such as juvenile T. albacares, and larval stomatopods, were similar to the δ15NsourceAA values of small and large particles from the surface to 150 m and this similarity was found to be statistically significant (Fig. 4; SIMPROF p < 0.05, df = 55). In the mesopelagic, the δ15NsourceAA values of migrating micronekton, including myctophids and sergestid shrimp, were also similar to small and large particles in the epipelagic (surface to 150 m; SIMPROF p < 0.05, df = 55). However, the migrators Gennadas sp., Anoplogaster cornuta, and Histioteuthis sp. were more similar to mesopelagic nonmigrating species. Mesopelagic, nonmigrating micronekton had δ15NsourceAA values that fell between those of large and small particles from the epipelagic and deep large particles (250–1205 m), except Opisthoproctus soleatus and Histioteuthis sp. which had statistically significant similarity to deep large particles. Bathypelagic, nonmigrating micronekton had the highest δ15NsourceAA values in between the values for deep large particles and deep small particles (250–1205 m). δ15NsourceAA values for the fish M. johnsonii were statistically similar to those of deep small particles (SIMPROF p < 0.05, df = 55).

Details are in the caption following the image

Dendrogram representing cluster analysis on four source amino acids (Gly, Ser, Phe, and Lys) using group average linkage and Euclidean distance. Red lines indicate statistically significant cluster groups (SIMPROF; p < 0.05). Samples within a group are more similar to each other. Red diamonds = bathypelagic nonmigrators, blue triangles = mesopelagic nonmigrators, green triangles = mesopelagic migrators, blue squares = epipelagic nonmigrators. Symbols are numbered based on Table 1. Labels without symbols indicate different particle samples. Small (0.7–53 μm) = suspended particles; large (> 53 μm) = sinking particles; numbers indicate depth (m) the particle sample was collected.

Variability of δ15NsourceAA values in C. pallida

Twelve C. pallida were analyzed to examine intraspecific variability in isotopic values of amino acids in a more detailed manner (Fig. 5). δ15NsourceAA values increased with depth (p < 0.01) as well as with length (p < 0.05) for this species. The δ15NsourceAA values of 12 individuals of C. pallida were found to be highly variable. The effects of season, year, or capture location could not be fully explored because 10 of the 12 individuals were captured at Sta. ALOHA in the summer of 2014. The variability found in C. pallida δ15NsourceAA values was not unusual with four other species of micronekton having standard deviations higher than that of C. pallida including C. atrum, H. proximum, Exocoetus sp., and S. sector (Table 1; Fig. 3).

Details are in the caption following the image

δ15NsourceAA of 12 individuals of C. pallida plotted against depth (a) (ANOVA p < 0.1, df = 12) and length (b) (ANOVA p < 0.05, df = 12). The x and y axis of the top and bottom figure have been reversed to be consistent with Fig. 3.

DNA barcoding of Cyclothone individuals

Two haplotypes that differed by a single substitution were observed in our C. pallida specimens, with one of these haplotypes a 100% sequence match to a C. pallida sequence reported from an animal collected in the California Current [GU440298]. K2P genetic distances between C. pallida and other species in the genus Cyclothone ranged from d = 0.168–0.299, and between C. pallida populations in the Atlantic and Pacific Oceans of d = 0.004–0.008 (11–14 substitutions), suggesting sufficient polymorphism in this mtCOI fragment to detect species-level differentiation within the genus. Thus, we confirmed the identity of our C. pallida specimens.


We found that lower mesopelagic and upper bathypelagic micronekton from around Hawaii feed from a suspended particle based food web. We could trace this food source into higher trophic levels because small particles have significantly higher δ15N values in their source amino acids than large sinking particles or particles at the surface of the ocean, likely a result of fractionation from microbial degradation (Hannides et al. 2013; Ohkouchi et al. 2017; Yamaguchi et al. 2017).

The δ15NsourceAA values of micronekton increased from the epipelagic to the upper bathypelagic suggesting a depth related shift in nutritional sources from surface derived material to sinking particles and finally toward a suspended particle based food web. This depth related pattern generally followed patterns found in particles (Fig. 3) and zooplankton (Hannides et al. 2016) collected concurrently and from earlier studies (Hannides et al. 2013; Choy et al. 2015). The depth pattern of δ15NsourceAA values can be evaluated by depth horizons and functional groups of micronekton. The δ15NsourceAA values of surface dwelling micronekton (e.g., flying fish) were similar to the δ15NsourceAA values of both small and large particles in surface waters (Fig. 4), indicating that these species feed on a food web based on fresh surface produced material likely comprised of phytoplankton, bacteria, and small protists. Epipelagic zooplankton (0–200 m) have similar δ15NsourceAA values (−1‰ to 0‰) to epipelagic micronekton and they are the likely intermediary trophic step (Hannides et al. 2016).

Mesopelagic, migrating species have similar δ15NsourceAA values to epipelagic species suggesting that mesopelagic migrators also feed from a food web based on surface derived material at night in surface waters as has been suggested in past diet studies (Clarke 1978; Hopkins et al. 1996). For example, Hopkins et al. (1996) reported that migrating myctophids fed primarily on the copepod Pleuromamma xiphias. In this study, δ15NsourceAA values of myctophids Bolinichthys longipes (0.1 ± 0.6‰) and H. proximum (1.2 ± 3.6‰) had δ15NsourceAA values similar to P. xiphias (0.5–2‰, Hannides et al. 2016).

Most lower mesopelagic and upper bathypelagic nonmigrators (800–1300 m; Fig. 3b) had δ15NsourceAA values consistent with a food web based on both in situ sinking and suspended particles (Fig. 4). One exception was the myctophid fish B. distofax, which had δ15NsourceAA values which were similar to surface derived material. Clarke (1973) reported that no B. distofax (formerly B. superlateralis) larger than 53 mm were collected above the day depth range of the species indicating that large individuals of this species did not migrate. The B. distofax collected in this study ranged in size from 48.6 mm to 87.7 mm and for this reason, we classified this species as a nonmigrator. However, we have captured this species between the surface and 100 m at night (including one specimen analyzed for isotopic composition) and suggest it is indeed a vertical migrator.

Mesopelagic nonmigrating species with median depths between 500 m and 700 m had δ15NsourceAA values similar to sinking particles suggesting that these species feed from a sinking particle based food web (Fig. 4). At these depths zooplankton, potential food for many of the micronekton, also have δ15NsourceAA values similar to sinking particles or intermediate between sinking and suspended particles (∼ 1–5‰, Hannides et al. 2016). Deeper living micronekton species (800–1300 m) had δ15NsourceAA values that were more similar to suspended particles (Fig. 4). These species include a piscivorous anglerfish (TP = 3.9, Table 1) suggesting that suspended particles supply enough energy at these depths to be fuel for higher trophic levels. The increased reliance on a suspended particle food web at greater depths makes sense given the availability of other food sources. Large particle concentrations and flux attenuate rapidly with depth (Buesseler et al. 2008) with up to 90% of the flux being attenuated in the first 1000 m (Boyd et al. 2008). Further, daytime sound scattering layers are densest at ∼ 400–700 m in the central North Pacific (Abecassis et al. 2015). Thus, at greater depths, alternative food resources such as small particles appear to become increasingly important to the midwater food web. It has even been hypothesized that smaller particles with greater surface area can be more easily colonized by heterotrophic microbes which could metabolize more refractory material and convert it to accessible amino acids and other compounds for grazers (Mayor et al. 2014).

It is clear that energy in small particles is transferred to some deeper-living micronekton; it is therefore worth evaluating whether this energy is also transferred to higher trophic levels. Choy et al. (2015) measured δ15NsourceAA values in large predatory fishes around Hawaii and using their data we find that δ15NsourceAA values range from 0.04‰ to 3.04‰, which are more similar δ15NsourceAA values of the surface or larger sinking particles. This is likely due to the fact that very few species dive deep enough to access the micronekton that feed largely from the suspended particle food web. Additionally, the effect of consumption of some very deep prey on the isotopic composition of large predatory fishes is likely moderated by more frequent consumption of shallower living prey with lower δ15NsourceAA values as suggested by a number of diet studies (Jackson et al. 2000; Olson and Galván-Magaña 2002; Choy et al. 2013). One large predatory fish species, Smith's escolar (Lepidocybium flavobrunneum), has been documented to have foraging depths of 1000 m, deep enough to access the micronekton which feed to a great extent from the suspended particle based food web (Kerstetter et al. 2008). However, Smith's escolar was found to have a δ15NsourceAA value of 1.4‰, similar to sinking particles in the mesopelagic (Choy et al. 2015). Finally, the high mobility of large pelagic predators may confound our interpretations because isotopic values are known to vary regionally (e.g., Choy et al. 2012) such that a large predator's isotopic values would be an integration of foraging across both regions and depths. Suspended particles are clearly an important nutritional source for lower mesopelagic and upper bathypelagic micronekton; this source is distinguishable from the fresh, surface-derived material utilized by epipelagic and upper mesopelagic micronekton. However, based on the results above, it is unclear if the suspended particle based food web influences populations of commercially important, deep foraging fishes.

δ15NsourceAA values were highly variable within several species of micronekton. C. pallida exhibited particularly high variability in δ15NsourceAA values. There was no evidence of cryptic species (e.g., Miya and Nishida 1997) among individuals of C. pallida which otherwise could have resulted in high isotopic variability. Almost half of the variability could be explained by the length of the individuals (r2 = 0.44; Fig. 5b). Additionally, more than a quarter of the variability in the δ15NsourceAA values could be explained by the median depth of capture of the individuals (r2 = 0.29; Fig. 5a). Maynard (1982) observed that larger C. pallida tend to occur deeper implying that length and depth are two closely related variables. Additionally, the median depth of capture represents a large depth range for a given C. pallida individual. For example, an individual with a median depth of capture of 1250 m could have been caught anywhere between 1500 m and 1000 m. It is possible that these nonmigratory fish that have very low metabolic rates (Smith and Laver 1981) inhabit narrower depth ranges; by sampling more discrete depth ranges, we would find that depth of capture explains more variability in the δ15NsourceAA values for this species. Another possibility for the variability in the δ15NsourceAA values of C. pallida is that these fishes do not live at discrete depths but undergo periodic vertical migrations within their depth range (600–1500 m). For example, Sutton et al. (2008) found the North Atlantic species C. microdon, which were a dominant species between 1500 m and 2300 m, were also found frequently above 750 m. However, if all individuals of C. pallida fed throughout their depth range, i.e., where the sinking particle food web is dominant (500–800 m at night) as well as where the suspended particle food web is dominant (800–1300 m at night), we would expect the δ15NsourceAA values from these two food webs to mix leading to more similar δ15NsourceAA values among individuals. While the observed intraspecific variability points to directions for future study, the interspecific patterns observed with depth are robust. It is clear that micronekton increasingly are supported by a suspended particle food web at greater depths.

An additional goal of this study was to determine if bulk isotope data, rather than AA-CSIA, could be used to identify organisms that fed from a suspended particle food web. Bulk isotope analysis is less time consuming and considerably less expensive than AA-CSIA. We found several species that had very high bulk tissue δ15N values including the mesopelagic, nonmigrating O. soleatus (9.8‰), the bathypelagic, nonmigrating M. johnsonii (13‰), and the mesopelagic, migrating species Histioteuthis sp. (10‰) and C. sloani (9.3‰). Several of these micronekton species had TPs within the range of large pelagic fishes from the north Pacific (TP = 4.3–5.0; Choy et al. 2015). For example, the spookfish O. soleatus had a TP of 4.4. O. soleatus is a small mouthed, sluggish fish which probably has a high TP because it feeds on predatory siphonophores (Mauchline and Gordon 1983). However, other micronekton species with high δ15N values, such as S. sector (TP = 3.1), had lower TPs but high δ15NsourceAA values likely being driven by feeding from a suspended particle food web. Thus, high bulk δ15N values could be interpreted in multiple ways while AA-CSIA can distinguish between TP and dietary sources.

Indeed, we found that δ15NsourceAA values and TP explained variability in bulk tissue δ15N across the suite of micronekton but low r2 values indicate that there may be other factors influencing bulk tissue δ15N values in micronekton (Fig. 1). For example, differences in amino acid composition between species would not affect the δ15NsourceAA values of individual amino acids but could lead to differences in an integrated bulk tissue δ15N value. Further study would be required to evaluate these potential effects. Intraspecifically such factors appear to have much less influence. δ15NsourceAA values were able to explain 62% of the variability in the bulk tissue δ15N values of C. pallida (Fig. 2) implying that within a species the isotopic baseline has a stronger influence on bulk tissue δ15N values than TP, a conclusion similar to that found by others using AA-CSIA to study pelagic marine food webs (e.g., Lorrain et al. 2014). In conclusion, it is difficult to identify species feeding from a suspended particle food web based on bulk isotope data alone. It is reasonable to identify shifting isotopic baselines within a species and across species if TPs are known from diet studies or other work, though uncertainties would remain. Despite costs and analysis times, we recommend the use of AA-CSIA to identify taxa feeding from a suspended particle based food web.

In the open ocean, ultimately the food web is supported by epipelagic phytoplankton but these resources undergo transformations through microbial activity that lead to basal sources of nutrition that are distinct and complicate our understanding of mesopelagic food webs. The discrepancy between the carbon supply and the carbon demand in the mesopelagic and the debate about these budgets (Giering et al. 2014) illustrates our incomplete understanding. Sediment traps are valuable tools for measuring POC flux but can include free-swimming organisms and they can miss smaller particles that are swept out by flow within the sediment traps (Buesseler 2007). In addition to the gravitational flux, some organisms sustain themselves through diel vertical migration. For example, Steinberg et al. (2008) estimated that many zooplankton from Sta. ALOHA were able to transport organic material through their vertical migrations, via feeding at the surface and defecating at depth, corresponding to up to 15% of the carbon demand of the mesopelagic around Hawaii. Further, one study in the northeast Pacific estimated that micronekton could transport up to 23 mg C m−2 d−1, which is 26% of the carbon demand of the mesopelagic zone around Hawaii (Davison et al. 2013), though we would expect a lower abundance of micronekton in the oligotrophic waters around Hawaii to contribute less to carbon transport. The vertical migration of zooplankton and micronekton does not account entirely for the deficit in carbon supply to the mesopelagic. We suggest that suspended particles could make up a portion of the “carbon deficit” found for mesopelagic and bathypelagic fauna. We do not attempt to close the carbon budget of this ecosystem as most of the mesopelagic carbon demand is microbial (Burd et al. 2010); however, suspended particles clearly provide a source of food that is of importance to higher trophic levels. Further work should focus on evaluating the specific consumers of the suspended particles and the diets of deep mesopelagic and bathypelagic micronekton in order to close the loop on this food web pathway.


We thank Natalie Wallsgrove, Cassie Lyons, and Whitney Ko for laboratory assistance and isotope measurements. We thank Astrid Leitner, Lauren Van Woudenberg, Whitney Ko, Liz Hetherington, Kazia Mermel, and Justin Miyano for helping to collect and process micronekton samples at sea. We thank Anna Neuheimer for helping to build and implement the statistical models used to analyze variability in the isotope data. We also thank the captain and crew of R/V Kilo Moana. This study was carried out in accordance with the animal use protocols of the University of Hawai‘i (protocol 10-984). This is SOEST contribution number 10233. Funding for this project was from NSF-OCE 1333734 and 1041329 and partially supported by the Pelagic Fisheries Research Program via Cooperative Agreement NA09OAR4320075 between the Joint Institute for Marine and Atmospheric Research and the National Oceanic and Atmospheric Administration.

    Conflict of Interest

    None declared.