Lipid load triggers migration to diapause in Arctic Calanus copepods—insights from underwater imaging

Lipid load triggers migration to diapause in Arctic Calanus copepods—insights from underwater... Abstract Copepod lipids fuel the Arctic marine ecosystem, but information on the fine-scale distribution of copepods and lipids is nonexistent. This study investigated the fine-scale (1 m) vertical distribution of the copepods Calanus hyperboreus, Calanus glacialis and Metridia longa during a Lagrangian drift in the North Water Polynya using the Lightframe On-sight Keyspecies Investigation (LOKI) imaging system. A copepod species- and stage-specific automatic identification model based on machine learning, a subcategory of artificial intelligence, was used to identify images taken by LOKI. Lipids were measured from images of copepods taken over the whole water column (1 m resolution). Diel vertical migration (DVM) in all three species was detected. In C. hyperboreus and C. glacialis C4-females as well as M. longa C5-females lipid load of deep copepod individuals was significantly higher than that of shallower individuals. Vertical distribution profiles and individual lipid loads suggested that individuals with lower lipid load continued DVM, while others with high lipid load ceased migrating, remaining at depth. Calanus hyperboreus individuals seemed to migrate to diapause at lower lipid fullness (50%) than C. glacialis (60%). A bioenergetics model showed that Calanus females had enough lipids to diapause for over a year, highlighting the significant lipid overhead they use for capital breeding. INTRODUCTION In the Arctic Ocean strong seasonality results in a pulsed availability of resources with peak phytoplankton biomass occurring during late spring and summer (Søreide et al., 2010). Zooplankton graze this overabundance of phytoplankton and can channel about a third of that biomass in the form of highly energetic lipids to higher trophic levels such as forage fish, whales and seabirds (Welch et al., 1992; Søreide et al., 2008). Copepods constitute up to 80% of this zooplankton biomass in the Arctic and many are efficient grazers whose behavior and life-cycle strategies are finely tuned to the transient dynamics of primary producers (Auel and Hagen, 2002). However, recent and rapid environmental changes in the Arctic Ocean have already shown impacts on zooplankton distributions, survival and phenology, with the potential to disrupt this sensitive yet critical ecological balance (Falk-Petersen et al., 2007; Richardson, 2008; Søreide et al., 2010; Doney et al., 2012). One strategy copepods display to adjust to environmental variability is extensive diel vertical migration (DVM) during which they move up and down in the water column on a daily cycle, even in the context of Arctic midnight sun during summer (Rabindranath et al., 2010; Darnis et al., 2017). During classic synchronized DVM individuals of a population move together to shallow waters at night to feed, and back to deeper waters during the day to take shelter from visual predators, mainly fish (Darnis et al., 2017), but other types of DVM exist. Such strategies allow copepods to feed efficiently and to rapidly accumulate the mass they need to develop and thrive in the extreme Arctic environment. Seasonal dynamics of lipid accumulation in Calanus hyperboreus and Calanus glacialis are well documented (Ashjian et al., 1995; Lee et al., 2006; Freese et al., 2017). Lipids accumulated by Arctic copepods contain a high proportion of energy-rich wax esters that propagate into the pelagic food web (Lee et al., 2006). This accumulation of large reserves is critical for another ubiquitous strategy displayed by the small number of large Calanus species that overwhelmingly dominate the biomass of Arctic mesozooplankton, the diapause (Thibault et al., 1999). This diapause strategy is characterized by the building of large reserves ahead of a long period of reduced metabolism that coincides with an extensive seasonal vertical migration towards the relative security of the depths (Hirche, 1997; Darnis and Fortier, 2014). This phenomenon is not limited to the interior of the deep oceanic basins, but rather occurs all along the North Atlantic and Arctic Ocean productive shelves as long as regional bathymetry reaches ca. 150 m and below. While diapause dynamics (e.g. timing of descent to diapause, diapause depth and duration) display considerable intra- and interspecies differences (Johnson et al., 2008), it remains essentially dependent on the processes of lipid accumulation and mobilization (Johnson et al., 2008; Falk-Petersen et al., 2009; Maps et al., 2010, 2014; Tarrant et al., 2008, 2016; Baumgartner and Tarrant, 2017). The lipid accumulation window (LAW) hypothesis, for example, proposes that individual copepods initiate diapause only when they have managed to accumulate enough lipid stores to support the (reduced) metabolic demand during dormancy and the maturation process in pre-adult stages (Johnson et al., 2008; Maps et al., 2010; Baumgartner and Tarrant, 2017). Since diapause involves an ontogenetic migration towards the relative security of deeper waters, it implies that a vertical gradient in lipid content should be observed among individuals. Testing this hypothesis in situ has been challenging, owing to the coarse spatio-temporal resolution that net-based sampling methods can provide. Until very recently, sampling for copepod lipids has been done in layers of tens of meters, essentially through multiple closing/opening nets systems (ARIES, Jónasdóttir, 1999; BIONESS, Davies et al., 2015; Hydro-Bios, Laakmann et al., 2009; Pepin and Head, 2009; MOCNESS, Saumweber and Durbin, 2006). Actually, even in studies where sampling was stratified, lipid contents were provided as pooled measurements (mean ± standard deviation), sometimes split between late copepodite and adult stages. But the relative lipid content between a “surface” and a “deep” layer was reported rather anecdotally rather than as the focus of the study (e.g. Lischka and Hagen, 2007; but see Pepin and Head, 2009). As a result, there is currently no information available on the vertical distribution of total lipids (mg) or lipid fullness (LF) (%) at both the individual scale and at a fine vertical resolution of a few meters. However, such information is relevant to many biological processes such as feeding, mate finding, trophic interactions, etc. (Hays et al., 2001; Pond and Tarling, 2011; Zarubin et al., 2014). This potentially limits our detailed understanding of the ecology and life history strategies of these key copepod taxa. Underwater imaging systems that take photographs of individual zooplankton in situ are able to overcome the low vertical resolution of nets (Hirche et al., 2014; Luo et al., 2014), and are therefore useful for studying copepod lipids. This study uses the Lightframe On-sight Keyspecies Investigation (LOKI) system (Schulz et al., 2010; Schmid et al., 2016). Images were identified using a copepod stage-specific automatic zooplankton identification model (Schmid et al., 2016) based on machine learning. The appearance of copepod lipid sacs was investigated in hundreds of LOKI images of individuals of three copepod species that dominate the biomass of mesozooplankton in the Arctic Ocean: the diapausing species C. hyperboreus and C. glacialis, and the non-diapausing Metridia longa. Individuals were automatically identified to the development stage, and their total lipid amount (mg) and LF (%) were estimated. These data of unprecedented level of resolution were used in a bioenergetics model in order to estimate the potential diapause duration of deep-dwelling individuals for the two Calanus congeners (Maps et al., 2014). The objective of this study was to use new imaging technology and its high resolution to explore the relationship between individual lipid content and the vertical migration behavior and diapause strategy of Arctic copepod species. METHOD Study area We sampled zooplankton during the 2013 ArcticNet expedition onboard the CCGS Amundsen. Sampling followed a Lagrangian surface drifter on August 15 and 16. Vertical tows of LOKI were performed in the vicinity of Ellesmere Island, NU, in the North Water Polynya (NOW; Supplementary Fig. S1). The NOW, situated in northwestern Baffin Bay between Greenland and Ellesmere Island, is one of the largest recurrent polynyas in the Arctic with a size of 5–8 × 104 km2 and is considered one of the most productive areas north of the Arctic Circle (Deming et al., 2002). Ice-free waters occur there as early as the beginning of April compared to mid-July in adjacent regions of the Canadian Archipelago (Fortier et al., 2002). LOKI system The LOKI is a mesozooplankton imaging system. A plankton concentration net, with a mouth opening of 0.28 m2 and a mesh size of 200 μm is attached on top of the inflow of the camera (Supplementary Fig. S2). The LOKI camera-imaging channel has a length of 31.3 mm, width of 20.75 mm and height of 4 mm (i.e. a volume of 2.6 cm3). A cod-end collected all zooplankton after imaging, and samples were preserved in 4% formaldehyde–seawater solution buffered with sodium borate for further analyses in the laboratory. At each time step (Table I), LOKI was deployed to 10 m above the seafloor prior to being hauled back to the surface at a constant speed of 24 m min−1. Images of mesozooplankton were collected and stored on an internal Solid State Drive in 8-bit lossless greyscale format. Environmental conditions (temperature, oxygen, conductivity, fluorescence and pressure) were recorded during each LOKI deployment using onboard sensors. For photosynthetically active radiation (EPAR), we used data collected by the ArcticNet main CTD (Biospherical QCP-2300). Detailed information on LOKI design and setup are presented in Schulz et al. (2010) and Schmid et al. (2016). Table I: LOKI deployments carried out during the Lagrangian drift Date Time Latitude °N Longitude °W Bottom depth (m) 15 August 2013 6:40 76° 22′ 53′′ 77° 21′ 53′′ 354 15 August 2013 10:30 76° 21′ 3′′ 77° 29′ 4′′ 381 15 August 2013 18:45 76° 20′ 57′′ 77° 35′ 19′′ 340 15 August 2013 23:20 76° 19′ 55′′ 77° 39′ 51′′ 273 15 August 2013 2:40 76° 17′ 37′′ 77° 45′ 30′′ 275 Date Time Latitude °N Longitude °W Bottom depth (m) 15 August 2013 6:40 76° 22′ 53′′ 77° 21′ 53′′ 354 15 August 2013 10:30 76° 21′ 3′′ 77° 29′ 4′′ 381 15 August 2013 18:45 76° 20′ 57′′ 77° 35′ 19′′ 340 15 August 2013 23:20 76° 19′ 55′′ 77° 39′ 51′′ 273 15 August 2013 2:40 76° 17′ 37′′ 77° 45′ 30′′ 275 Table I: LOKI deployments carried out during the Lagrangian drift Date Time Latitude °N Longitude °W Bottom depth (m) 15 August 2013 6:40 76° 22′ 53′′ 77° 21′ 53′′ 354 15 August 2013 10:30 76° 21′ 3′′ 77° 29′ 4′′ 381 15 August 2013 18:45 76° 20′ 57′′ 77° 35′ 19′′ 340 15 August 2013 23:20 76° 19′ 55′′ 77° 39′ 51′′ 273 15 August 2013 2:40 76° 17′ 37′′ 77° 45′ 30′′ 275 Date Time Latitude °N Longitude °W Bottom depth (m) 15 August 2013 6:40 76° 22′ 53′′ 77° 21′ 53′′ 354 15 August 2013 10:30 76° 21′ 3′′ 77° 29′ 4′′ 381 15 August 2013 18:45 76° 20′ 57′′ 77° 35′ 19′′ 340 15 August 2013 23:20 76° 19′ 55′′ 77° 39′ 51′′ 273 15 August 2013 2:40 76° 17′ 37′′ 77° 45′ 30′′ 275 Image data preparation and automatic identification of copepod taxa using machine learning Every LOKI image was linked to the corresponding data from the environmental sensors based on a timestamp and calculated depth. Fifty-two image parameters including circularity, Hu moments, Fourier descriptors as well as several values describing geometry and connectivity within objects (from Morphological Spatial Pattern Analysis) were measured on all collected images using LOKI software and GUIDOS toolbox (Schulz et al., 2010; see Schmid et al., 2016 for a complete list of image parameters). Representativeness of LOKI image data was ensured by applying the ZOOMIE v 1.0 (Zooplankton Multiple Image Exclusion) software to all images (Schmid et al., 2015). ZOOMIE filters out redundant images of unique individuals. The copepod stage-specific automatic zooplankton identification model developed in Schmid et al. (2016) was applied to all images, using Salford Systems’ predictive modeler (version 7.0) implementation of the random forests (RF) algorithm (Breiman, 2001). RF is a part of machine learning, a subcategory of artificial intelligence. Schmid’s model is highly efficient at identifying images of C. hyperboreus, C. glacialis and M. longa, when presented with many zooplankton types. Accuracy was tested using model internal testing with withheld test data, and external testing in which automatic identification of images from deployments unknown to the model was manually verified. Externally tested accuracies (true positive samples over total manual classification) for C. hyperboreus, C. glacialis and M. longa were 86%, 82% and 83%, respectively. Externally tested specificities (true positive samples over total classifier prediction) for C. hyperboreus, C. glacialis and M. longa were 87%, 81% and 84%, respectively (Schmid et al., 2016). All images belonging to the different developmental stages of C. hyperboreus, C. glacialis and M. longa (Supplementary Fig. S3) were selected for further analysis. Because of morphological similarities and size overlap, the pairs of C. glacialis C2 and C. hyperboreus C1 as well as C. glacialis C3 and C. hyperboreus C2 were treated as single groups by the automatic identification model (Schmid et al., 2016). The automatic identification model does not distinguish Calanus finmarchicus from other congeners. Investigation of cod-end samples revealed that C. finmarchicus abundance was 174 ± 29 ind. m−2 (mean ± 1 SD) during the drift, compared to 12 483 ± 213 ind. m−2 of C. glacialis, as which C. finmarchicus would most likely be identified as by the automatic identification model. Image observations of individuals in each taxon were binned for each 1 m of the water column. Counts of detections of each taxon in 1 m bins were corrected based on the classification probabilities (Solow et al., 2001) of all taxa from the automatic identification model (Schmid et al., 2016). The applied classification probabilities were the averages of the two confusion matrices, from external testing, developed in Schmid et al. (2016). A total of 6983 automatically identified images had been algorithmically and randomly selected over all taxa, and then manually validated, in order to generate these matrices (Schmid et al. 2016). The corrected number of observations was used to estimate taxon abundances (m−3) per 1 m of the water column, based on the volume filtered. See Schmid et al. (2016) for additional details on each step of image processing. Lipid analysis We used two complementary measures of individual lipid content: total lipids (TL, in mg) and LF (in % of body volume). Since inter- and intraspecific variability in individual size can be very large (Forster and Hirst, 2012), total lipid measures were not directly comparable among stages and species, whereas LF could provide a more general picture of the individuals’ condition and lipid metabolism among copepods. Total lipids and LF were estimated for a total of 822 LOKI images of the development stages known to accumulate lipids in a visible lipid sac in their prosome. Images of individual C. hyperboreus stages C3 to adult female (F), C. glacialis C4 to F and M. longa stages C5 to F were sourced from two of the five deployments: one during daytime (18:45 h DST, UTC −4; 380 μE m−2 s−1 incident EPAR), and one at night (2:40 h, 18 μE m−2 s−1 incident EPAR). For C. hyperboreus C3–C5, C. glacialis C4/C5 and M. longa C5/F, 50 individuals were analyzed from each of the two deployments. For. C. glacialis and C. hyperboreus F, a total of 42 and 80 individuals respectively were analyzed, due to their different relative availabilities. Copepod images for lipid analysis were sampled randomly from the whole water column, but images where the copepod was not in a lateral position had to be discarded due to the methodology used (Vogedes et al. 2010), and a new image was drawn randomly. For all individuals, the area of the lipid sac (A, in mm2) was measured (Fig. 1), and used to estimate total lipids according to Vogedes et al. (2010): TL=0.197A1.38. (1) Fig. 1. View largeDownload slide Image of a Calanus hyperboreus female with its lipid sac highlighted in red. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 1. View largeDownload slide Image of a Calanus hyperboreus female with its lipid sac highlighted in red. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) LF was simply defined as the area of the lipid sac divided by the area of the prosome times 100. Potential diapause duration estimation Potential diapause duration was computed for the stages and species known to perform extensive diapause in the area, i.e. C. hyperboreus C3 to F and C. glacialis C4 to F, following Maps et al. (2014): D=4−M(Cout14−Cin14) (2) where M (μg C1/4 s−1) is the metabolic constant for dormant individuals at 0°C. We did not include any effect of temperature since the observed temperature at the expected depth of diapause (>150 m) was for the largest part constant at 0°C (Supplementary Fig. S4). Cin and Cout are the individual body mass (μg carbon) at the onset and at the end of dormancy, respectively (see Equations (3) and (4) below). To estimate these critical values, we made the assumption that the maximum potential diapause duration would be reached upon lipid stores depletion. Total lipid values estimated from the images were converted into carbon using a carbon to lipid conversion factor λ = 0.81, based on the average lipid composition of calanoid copepods (Kattner and Graeve, 1991). Total body carbon of each individual was calculated from the lipid carbon and LF: Cin=λTLLF. (3) Then the difference between Cout and the lipid carbon gave the minimum (structural) carbon content Cout: Cout=Cin−λTL. (4) We likely overestimated the minimal carbon content at the end of diapause since the structural component of the calanoid copepods’ body contains much less carbon per unit of dry weight than its lipid stores, as revealed by comparing the elemental C to N ratios of lipid full vs. lipid poor individuals (e.g. Pepin and Head, 2009). As a result, we consider our estimate of potential maximum diapause duration to be conservative. Data and R-code (R Core Team, 2017) to reproduce these results can be found at https://zenodo.org/record/1203741. RESULTS Environmental setting A dicothermal layer prevailed during the drift study with relatively warm (4.3°C) surface waters sitting on a glacial (−1.5 °C) layer centered around 100 m over warmer deep waters (Supplementary Fig. S4). A strong halocline with salinity 31.4 at the surface and 34.1 at depth provided vertical stability. Fluorescence in surface waters reached 1.4 mg m−3 by the end of the drift (Supplementary Fig. S4). Daylight prevailed from 7:00 to 23:00 with maximum EPAR in the water column reaching 239 μE m−2 s−1 (Supplementary Fig. S4). A constant T-S signature among profiles (not shown) confirmed that the same vertical water mass structure was sampled during the 24 h drift over a 15 km distance. Vertical distributions of copepods The mean abundances (±1 SD) of C. hyperboreus, C. glacialis and M. longa during the drift, as estimated from automatically identified underwater imagery, were 11 465 ± 320, 12 563 ± 265 and 15 668 ± 378 ind. m−2, respectively. The abundance of each species varied little among profiles. While the combined young Calanus copepodite stages C2/C1 distributed vertically according to the fluorescence profile, especially at night (2:40 h profile, Fig. 2), the older C3/C2 showed tendencies of avoiding the immediate surface layer (<10 m) during daytime, before congregating in or slightly under the subsurface chlorophyll maximum (SCM) during night. Older C. hyperboreus development stages presented more distinct vertical distribution patterns. C3 congregated in the SCM at night (2:40 h profile), moved between 100 and 200 m well below the SCM during the day and then moved back towards it at dusk (Fig. 3). Similarly, the bulk of the abundant C4 stage was distributed around 200 m in daytime, moving towards the SCM at dusk and converging into it at night. C5 were distributed all over the water column with greater abundance at depth near the bottom. Females were distributed in the deeper half of the water column (Fig. 3). Calanus glacialis C1 distribution matched chlorophyll a fluorescence at all times within the 0–120 m layer and displayed little vertical movement throughout the diel cycle (Fig. 4). The abundant C4 and C5 were distributed over the entire water column with a noticeable increase in density immediately above and below the SCM at night. The rare female C. glacialis were distributed close to the seafloor at all times (Fig. 4). Vertical distributions of M. longa C1–C4 remained similar throughout the drift (Fig. 5; Supplementary Fig. S5). C4, C5 and females were found throughout the water column but avoided the immediate surface layer (<10 m) in daytime and concentrated in the SCM at night. Males were rare, and always situated deep in the water column (Supplementary Fig. S5). Fig. 2. View largeDownload slide Vertical distribution of the unresolved Calanus glacialis C2 / C. hyperboreus C1 and C. glacialis C3 / C. hyperboreus C2 complexes (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 2. View largeDownload slide Vertical distribution of the unresolved Calanus glacialis C2 / C. hyperboreus C1 and C. glacialis C3 / C. hyperboreus C2 complexes (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 3. View largeDownload slide Vertical distribution of Calanus hyperboreus C3 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 3. View largeDownload slide Vertical distribution of Calanus hyperboreus C3 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 4. View largeDownload slide Vertical distribution of Calanus glacialis C1 and C4 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 4. View largeDownload slide Vertical distribution of Calanus glacialis C1 and C4 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 5. View largeDownload slide Vertical distribution of Metridia longa C4 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. The vertical distributions of the remaining M. longa stages are given in Supplementary Fig. S5. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 5. View largeDownload slide Vertical distribution of Metridia longa C4 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. The vertical distributions of the remaining M. longa stages are given in Supplementary Fig. S5. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Copepod lipids In C. hyperboreus, we observed maximum mean total lipids of 2.45 mg and maximum mean lipid fullness of 79% in females, while minimum mean total lipids of 0.09 mg and minimum mean lipid fullness of 39% were found in C3 (Fig. 6; see Supplementary Table S1 for all means and standard errors). Calanus glacialis female individuals contained maximum mean total lipids of 0.52 mg and maximum mean lipid fullness of 80%, while minimum mean total lipids of 0.17 mg and minimum mean lipid fullness of 43% were found in C4 (Fig. 6; Supplementary Table S1). Finally, M. longa females contained maximum mean total lipids of 0.1 mg and maximum mean lipid fullness of 36%, while minimum mean total lipids of 0.01 mg and minimum mean lipid fullness of 12% were found in C5 (Fig. 6; Supplementary Table S1). While in most taxa total lipids and lipid fullness followed normal distributions, C. hyperboreus C3 and female lipid fullness showed a tendency towards a bimodal distribution (Supplementary Fig. S6). Fig. 6. View largeDownload slide Total lipids (a) and lipid fullness (b) for different copepodite stages and female (F) Calanus hyperboreus, C. glacialis and Metridia longa. Fig. 6. View largeDownload slide Total lipids (a) and lipid fullness (b) for different copepodite stages and female (F) Calanus hyperboreus, C. glacialis and Metridia longa. Vertical distribution of lipids In C. hyperboreus C4, C5 and F, total lipids and lipid fullness significantly increased with depth during both day and night (Fig. 7; see Supplementary Table S2 for a linear regression summary). Vertical profiles of total lipids and lipid fullness differed significantly between day and night sampling only in C. hyperboreus C4 (Fig. 7; see Supplementary Table S3 for an analysis of covariance (ANCOVA) summary). Total lipids and lipid fullness of C. glacialis from C4 to female stages were significantly higher in deeper water compared to shallower water during both day and night samplings (Fig. 8; see Supplementary Table S2). Vertical profiles of lipid fullness in individuals differed significantly between day and night samplings in C. glacialis C4 and C5 (Fig. 8; Supplementary Table S3), while for total lipids this was only the case in C. glacialis C4. Total lipids and lipid fullness of M. longa C5 and females were significantly higher in deeper water compared to shallower water during night sampling (Fig. 9; Supplementary Table S2), but not during the daytime. Fig. 7. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Calanus hyperboreus individuals located at different depths. C3–C5, copepodite stages 3, 4 and 5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 7. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Calanus hyperboreus individuals located at different depths. C3–C5, copepodite stages 3, 4 and 5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 8. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Calanus glacialis individuals located at different depths. C4 and C5, copepodite stages C4 and C5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 8. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Calanus glacialis individuals located at different depths. C4 and C5, copepodite stages C4 and C5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 9. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Metridia longa individuals located at different depths. C5, copepodite stage C5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 9. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Metridia longa individuals located at different depths. C5, copepodite stage C5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Potential diapause duration Potential maximum diapause duration (D) estimated for all the developmental stages of the Calanus congeners increased almost linearly with their size (Fig. 10). The youngest developmental stages thought to be able to undergo diapause, i.e. C. hyperboreus C3 and C. glacialis C4, could on average be diapausing during 78 and 112 days, respectively. Subsequent stages could all exceed an arbitrary duration threshold of six months. It also appeared that C. glacialis females could remain dormant for almost a year, whereas C. hyperboreus females potential maximum diapause duration could well exceed this duration. Fig. 10. View largeDownload slide Estimated maximum potential diapause duration for observed development stages of Calanus hyperboreus and Calanus glacialis. Dotted line: 6-month threshold; dashed line: 1-year threshold. Fig. 10. View largeDownload slide Estimated maximum potential diapause duration for observed development stages of Calanus hyperboreus and Calanus glacialis. Dotted line: 6-month threshold; dashed line: 1-year threshold. DISCUSSION Diel vertical migration In this study, underwater images of zooplankton species and stages taken by the LOKI system were automatically, and with high accuracy, identified using a previously developed machine learning model (Schmid et al., 2016). The resulting fine-scale vertical distribution data (1 m resolution) showed clear differences in the DVM of the different developmental stages. The observed DVM in C. hyperboreus, C. glacialis and M. longa corresponded to the classic patterns reported in previous Arctic studies (Kosobokova, 1978; Fortier et al., 2001; Falk-Petersen et al., 2008; Daase et al., 2008, 2015). Our observations are consistent with studies showing the clearest signs of classic DVM in copepods during late Arctic summer and autumn (Kosobokova, 1978; Cottier et al., 2006; Wallace et al., 2010). The ultimate cause for DVM is thought to be predator avoidance, by reducing detection from visual predators such as Themisto sp. and polar cod in surface layers (Hays, 2003), while the proximate cause for DVM is thought to be changes in light intensity (Cohen and Forward, 2009). Local hydrography has also been shown to affect DVM patterns, for instance through thermal optima at depth (Berge et al., 2014), water mass differences (Lawson et al., 2004) and thermoclines (Williams, 1985). However, the observations of individual copepods by means of underwater imaging and at multiple time steps in this study showed that also the small/young copepodite stages (see for instance the C. glacialis C3 / C. hyperboreus C2 complex, Fig. 2) readily migrated from immediate surface waters to deeper waters, and back again, crossing thermo-, and halocline. This suggests that hydrography, and especially clines, were not the most important drivers, and not inhibiting, in this study. Since smaller C. glacialis C3 and C. hyperboreus C2 stages swim at lower Reynolds numbers than larger stages, an effect of clines would be visible first in these weaker swimmers. The youngest and smallest copepodite stages (i.e. C. glacialis C2 / C. hyperboreus C1 complex and M. longa C1) showed no signs of DVM, which can be explained by a reduced visual predation pressure in their size range (Brooks and Dodson, 1965; Fortier et al., 2001; Daase et al., 2015). Instead, these young stages were vertically distributed in proportion to the abundance of their algal food throughout the drift. Especially during the night-time deployment at 2:40, individuals of all copepodite stages except Calanus females accumulated within the SCM (Rabindranath et al., 2010), most likely in order to feed (Falk-Petersen et al., 2008; Daase et al., 2008, 2015). Our results suggest that inter- and intraspecific differences observed in the DVM patterns are linked to the condition of individual copepods (Ohman, 1990; Hays et al., 2001), specifically their lipid fullness. Furthermore, the distribution profiles shown in this study demonstrate that the majority of a given copepod population (e.g. C. glacialis C5, Fig 4.) was constantly vertically migrating, rather than being at a certain “resting depth”, a term which is often used in DVM studies. The role of lipids in vertical migration and diapause In addition to the diapause requirements, C. hyperboreus depends on its lipid reserves for egg spawning in winter (Conover, 1967; Hirche, 1991), while C. glacialis utilizes stored lipids but also grazes on ice algae to spawn during early spring (Hirche and Kattner, 1993; Søreide et al., 2010). As an interesting counterpoint, M. longa is a smaller but abundant omnivorous species that differs from Calanus spp. in that it feeds and reproduces year-round. Since diapause is not a feature of its life cycle (Darnis and Fortier, 2014), it does not accumulate lipid reserves to the level found in Calanus spp. (Ashjian et al., 1995). As a result, we expected very different vertical gradients in lipid content between these species and among their developmental stages. Our individual-based observations obtained on finely resolved vertical and temporal scales showed that lipid-rich C. hyperboreus C5-F, as well as C. glacialis F, were found almost exclusively below 150 m. The lack of any significant increase in numbers within the surface layer at night strongly suggests that these individuals were already diapausing. On the contrary, the changes in lipid fullness observed in younger C. glacialis C4 and C5 stages as well as C. hyperboreus C4 suggest that some individuals with lower lipid fullness were still migrating from their deep refuge during the day towards the surface layer during the night, most likely for feeding (Figs 7–9; Supplementary Table S3), while individuals from the same stages with higher lipid fullness remained deep and did not migrate back up to surface waters. These results support the LAW hypothesis and may even allow estimation of the threshold beyond which active DVM would stop for a given developmental stage. Lipid fullness observations of C. hyperboreus C4 and C. glacialis C5 suggested that C. hyperboreus C4 individuals were able to start diapausing at lower lipid fullness (∼50%) than C. glacialis C5 (∼60%). Observations of lipid-filled Calanus sp. at depth are consistent with the understanding that Calanus sp. are strong seasonal migrators (Darnis and Fortier, 2014) that can diapause as soon as stage C3 for C. hyperboreus (Hirche, 1997) or C4 for C. glacialis (Melle and Skjoldal, 1998; Falk-Petersen et al., 2009). While total lipids reported in this study for C. hyperboreus C4–C5, C. glacialis C4–C5 and M. longa females are within the ranges reported in the literature (Hirche and Kattner, 1993; Ashjian et al., 1995; Scott et al., 2000; Auel et al., 2003), no comparable study has ever shown total lipids and lipid fullness for C. hyperboreus C3 and M. longa C5. The average lipid fullness in C. hyperboreus and C. glacialis females was around 80%, which is higher than what was reported by previous studies elsewhere around the Arctic (Hirche and Kattner, 1993; Ashjian et al., 1995; Scott et al., 2000; Auel et al., 2003). Total lipids and lipid fullness differences between studies can be explained by the individual feeding success (Miller et al., 1998; Baumgartner et al., 2011) as well as seasonal and geographical variability in environmental conditions. High amounts of total lipids and replete oil sacs of Calanus sp. found in our study suggest that these copepods found plenty of suitable phytoplankton food during the productive spring/summer of 2013. Such individual conditions led to high potential diapause duration estimates for Calanus spp. copepodite stages. Substantial lipid reserves of C. hyperboreus C4-F and C. glacialis C5-F were sufficient for a potential diapause from late August until next spring, when the first production peak of ice algae starts (late March–early April). Calanus hyperboreus C5 and both congener’s females even had a sizable lipid surplus, enabling potential diapause of over a year in females. Obviously, females of either species do not remain in diapause for over a year. However, both are known to produce eggs from their lipid reserves, and while C. glacialis reaches its maximum egg production during the spring-summer phytoplankton bloom, C. hyperboreus is a strict capital breeder (Hirche, 1991). This lipid overhead shows that all female individuals of the two Calanus congeners would be able to safely undergo diapause and survive through the long Arctic winter as well as successfully reproduce. While M. longa does not go into diapause, the substantial lipid fullness of its females reinforces the hypothesis that M. longa relies on a mixed strategy of active feeding and accumulating lipids to overcome the harsh winter conditions of the Arctic marine ecosystem, and even produce some eggs (Conover and Huntley, 1991). Higher temperatures of ambient water can substantially increase the metabolic demand of diapausing copepods, leading to a shortening of potential diapause length (Maps et al., 2014). However, temperature of deeper Arctic waters (>200 m) is not expected to increase to a level that would negatively affect diapause length in the near future (Seidov et al., 2015). Climate change-induced mismatch scenarios earlier in the season (Falk-Petersen et al., 2007), however, have the potential to disrupt lipid accumulation and could lead to starvation of copepods during diapause (Auel et al., 2003). This would also negatively affect the capital breeding strategy of Arctic-adapted Calanus species. In such a scenario, M. longa would likely fare better than Calanus spp. owing to its omnivorous, year-round feeding strategy, buffered by some lipid-accumulating capabilities. CONCLUSION This study used a machine learning—artificial intelligence approach to automatically identify underwater imagery, leading to fine-scale (1 m) resolution data on the distribution of individual copepods and species life stages as well as environmental parameters. Data derived from LOKI can be used in studies that aim at integrating the physiology and ecology of organisms, as was demonstrated here with the in situ lipid data of copepods. The present study shows that for Calanus individuals within a given developmental stage, individuals that accumulated high lipid fullness had entered diapause and ceased DVM, while individuals with lower lipid fullness still performed DVM and had not yet entered diapause. The lipid fullness threshold for migration to diapause varied between Calanus species and stages. Potential diapause durations of over a year were estimated in Calanus females, highlighting the large amounts of lipids going into egg production. It is hoped that the fine scale insights on copepod abundance, total lipids and lipid fullness provided in this study will be useful for a range of studies, especially for modeling experiments (e.g. Dufour et al., 2016). SUPPLEMENTARY DATA Supplementary data can be found online at Journal of Plankton Research online. ACKNOWLEDGEMENTS We would like to thank the officers and crew of the CCGS Amundsen for their dedication and professionalism at sea. We are grateful to Jordan Grigor and Cyril Aubry for support throughout the project and reviewing of manuscripts. Thanks to Kevin Gonthier for his help with processing of LOKI data as well as to Catherine Boudreau for her help with the lipid analysis. The help of Maxime Geoffroy, Marianne Falardeau and Alexis Burt with LOKI deployments is much appreciated. Thanks to Andrew Solow and Heidi Sosik for providing MATLAB code for the method described in Solow et al. (2001). We thank three anonymous reviewers for providing helpful comments on earlier versions of the manuscript. FUNDING The Canada Foundation for Innovation (CFI) (grant no. CFI224324) and the Network of Centres of Excellence ArcticNet (grant no. 12000). M.S. received postgraduate scholarships from the Canada Excellence Research Chair (CERC) in Remote Sensing of Canada’s New Arctic Frontier (Takuvik) and stipends from Québec-Océan. This is a joint contribution to Québec-Océan, ArcticNet and the Canada Research Chair on the response of marine arctic ecosystems to climate warming. REFERENCES Ashjian , C. J. , Smith , S. L. and Lane , P. V. Z. ( 1995 ) The Northeast Water Polynya during summer 1992: distribution and aspects of secondary production of copepods . J. Geophys. Res. , 100 , 4371 – 4388 . Google Scholar CrossRef Search ADS Auel , H. and Hagen , W. ( 2002 ) Mesozooplankton community structure, abundance and biomass in the central Arctic Ocean . Mar. Biol. , 140 , 1013 – 1021 . Google Scholar CrossRef Search ADS Auel , H. , Klages , M. and Werner , I. ( 2003 ) Respiration and lipid content of the Arctic copepod Calanus hyperboreus overwintering 1 m above the seafloor at 2,300 m water depth in the Fram Strait . Mar. Biol. , 143 , 275 – 282 . Google Scholar CrossRef Search ADS Baumgartner , M. F. , Lysiak , N. , Schuman , C. , Urban-Rich , J. and Wenzel , F. W. ( 2011 ) Diel vertical migration behavior of Calanus finmarchicus and its influence on right and sei whale occurrence . Mar. Ecol. Prog. Ser. , 423 , 167 – 184 . Google Scholar CrossRef Search ADS Baumgartner , M. F. and Tarrant , A. M. ( 2017 ) The physiology and ecology of diapause in marine copepods . Annu. Rev. Mar. Sci. , 9 , 387 – 411 . Google Scholar CrossRef Search ADS Berge , J. , Cottier , F. , Varpe , Ø. , Renaud , P. E. , Falk-Petersen , S. , Kwasniewski , S. , Griffiths , C. , Søreide , J. E. et al. ( 2014 ) Arctic complexity: a case study on diel vertical migration of zooplankton . J. Plankton Res. , 36 , 1279 – 1297 . Google Scholar CrossRef Search ADS PubMed Breiman , L. ( 2001 ) Random forests . Mach. Learn. , 45 , 5 – 32 . Google Scholar CrossRef Search ADS Brooks , J. L. and Dodson , S. I. ( 1965 ) Predation, body size, and composition of plankton . Science , 150 , 28 – 35 . Google Scholar CrossRef Search ADS PubMed Cohen , J. H. and Forward , R. B. J. ( 2009 ) Zooplankton diel vertical migration—a review of proximate control . Oceanogr. Mar. Biol. Annu. Rev. , 47 , 77 – 109 . Conover , R. J. ( 1967 ) Reproductive cycle, early development, and fecundity in laboratory populations of the copepod Calanus hyperboreus . Crustaceana , 13 , 61 – 72 . Google Scholar CrossRef Search ADS Conover , R. J. and Huntley , M. ( 1991 ) Copepods in ice-covered seas—distribution, adaptations to seasonally limited food, metabolism, growth patterns and life cycle strategies in polar seas . J. Mar. Syst. , 2 , 1 – 41 . Google Scholar CrossRef Search ADS Cottier , F. R. , Tarling , G. A. , Wold , A. and Falk-Petersen , S. ( 2006 ) Unsynchronised and synchronised vertical migration of zooplankton in a high Arctic fjord . Limnol. Oceanogr. , 51 , 2586 – 2599 . Google Scholar CrossRef Search ADS Daase , M. , Eiane , K. , Aksnes , D. L. and Vogedes , D. ( 2008 ) Vertical distribution of Calanus spp. and Metridia longa at four Arctic locations . Mar. Biol. Res. , 4 , 193 – 207 . Google Scholar CrossRef Search ADS Daase , M. , Hop , H. and Falk-Petersen , S. ( 2015 ) Small-scale diel vertical migration of zooplankton in the High Arctic . Polar Biol. , 39 , 1213 – 1223 . Google Scholar CrossRef Search ADS Darnis , G. and Fortier , L. ( 2014 ) Temperature, food and the seasonal vertical migration of key arctic copepods in the thermally stratified Amundsen Gulf (Beaufort Sea, Arctic Ocean) . J. Plankton Res. , 36 , 1092 – 1108 . Google Scholar CrossRef Search ADS Darnis , G. , Hobbs , L. , Geoffroy , M. , Grenvald , J. C. , Renaud , P. E. , Berge , J. , Cottier , F. , Kristiansen , S. et al. ( 2017 ) From polar night to midnight sun: diel vertical migration, metabolism and biogeochemical role of zooplankton in a high Arctic fjord (Kongsfjorden, Svalbard) . Limnol. Oceanogr. , 62 , 1586 – 1605 . Google Scholar CrossRef Search ADS Davies , K. T. A. , Taggart , C. T. and Smedbol , R. K. ( 2015 ) Interannual variation in diapausing copepods and associated water masses in a continental shelf basin, and implications for copepod buoyancy . J. Mar. Syst. , 151 , 35 – 46 . Google Scholar CrossRef Search ADS Deming , J. W. , Fortier , L. and Fukuchi , M. ( 2002 ) The International North Water Polynya Study (NOW): a brief overview . Deep Sea Res. Part II Top. Stud. Oceanogr. , 49 , 4887 – 4892 . Google Scholar CrossRef Search ADS Doney , S. C. , Ruckelshaus , M. , Duffy , J. E. , Barry , J. P. , Chan , F. , English , C. A. , Galindo , H. M. , Grebmeier , J. M. et al. ( 2012 ) Climate change impacts on marine ecosystems . Annu. Rev. Mar. Sci. , 4 , 11 – 37 . Google Scholar CrossRef Search ADS Dufour , K. , Maps , F. , Plourde , S. , Joly , P. and Cyr , F. ( 2016 ) Impacts of intraguild predation on Arctic copepod communities . Front. Mar. Sci. , 3 , 185 . Google Scholar CrossRef Search ADS Falk-Petersen , S. , Leu , E. , Berge , J. , Kwasniewski , S. , Nygård , H. , Røstad , A. , Keskinen , E. , Thormar , J. et al. ( 2008 ) Vertical migration in high Arctic waters during autumn 2004 . Deep Sea Res. Part II Top. Stud. Oceanogr. , 55 , 2275 – 2284 . Google Scholar CrossRef Search ADS Falk-Petersen , S. , Mayzaud , P. , Kattner , G. and Sargent , J. R. ( 2009 ) Lipids and life strategy of Arctic Calanus . Mar. Biol. Res. , 5 , 18 – 39 . Google Scholar CrossRef Search ADS Falk-Petersen , S. , Pavlov , V. , Timofeev , S. and Sargent , J. R. ( 2007 ) Climate variability and possible effects on arctic food chains: the role of Calanus. In Ørbæk , J. B. , Kallenborn , R. , Tombre , I. , Hegseth , E. N. , Falk-Petersen , S. and Hoel , A. H. (eds) , Arctic Alpine Ecosystems and People in a Changing Environment . Springer , Berlin , pp. 147 – 166 . Google Scholar CrossRef Search ADS Forster , J. and Hirst , A. G. ( 2012 ) The temperature-size rule emerges from ontogenetic differences between growth and development rates . Funct. Ecol. , 26 , 483 – 492 . Google Scholar CrossRef Search ADS Fortier , M. , Fortier , L. , Hiroshi , H. , Hiroaki , S. and Legendre , L. ( 2001 ) Visual predators and the diel vertical migration of copepods under Arctic sea ice during the midnight sun . J. Plankton Res. , 23 , 1263 – 1278 . Google Scholar CrossRef Search ADS Fortier , M. , Fortier , L. , Michel , C. and Legendre , L. ( 2002 ) Climatic and biological forcing of the vertical flux of biogenic particles under seasonal Arctic sea ice . Mar. Ecol. Prog. Ser. , 225 , 1 – 16 . Google Scholar CrossRef Search ADS Freese , D. , Søreide , J. E. , Graeve , M. and Niehoff , B. ( 2017 ) A year-round study on metabolic enzymes and body composition of the Arctic copepod Calanus glacialis: implications for the timing and intensity of diapause . Mar. Biol. , 164 , 3 . Google Scholar CrossRef Search ADS Hays , G. C. ( 2003 ) A review of the adaptive significance and ecosystem consequences of zooplankton diel vertical migrations . Hydrobiologia , 503 , 163 – 170 . Google Scholar CrossRef Search ADS Hays , G. C. , Kennedy , H. and Frost , B. W. ( 2001 ) Individual variability in diel vertical migration of a marine copepod: why some individuals remain at depth when others migrate . Limnol. Oceanogr. , 46 , 2050 – 2054 . Google Scholar CrossRef Search ADS Hirche , H. J. ( 1991 ) Distribution of dominant calanoid copepod species in the Greenland sea during late fall . Polar Biol. , 11 , 351 – 362 . Google Scholar CrossRef Search ADS Hirche , H. J. ( 1997 ) Life cycle of the copepod Calanus hyperboreus in the Greenland Sea . Mar. Biol. , 128 , 607 – 618 . Google Scholar CrossRef Search ADS Hirche , H. J. , Barz , K. , Ayon , P. and Schulz , J. ( 2014 ) High resolution vertical distribution of the copepod Calanus chilensis in relation to the shallow oxygen minimum zone off northern Peru using LOKI, a new plankton imaging system . Deep Sea Res. Part II Top. Stud. Oceanogr. , 88 , 63 – 73 . Google Scholar CrossRef Search ADS Hirche , H. J. and Kattner , G. ( 1993 ) Egg production and lipid content of Calanus glacialis in spring: indication of a food-dependent and food-independent reproductive mode . Mar. Biol. , 117 , 615 – 622 . Google Scholar CrossRef Search ADS Johnson , C. L. , Leising , A. W. , Runge , J. A. , Head , E. J. H. , Pepin , P. , Plourde , S. and Durbin , E. G. ( 2008 ) Characteristics of Calanus finmarchicus dormancy patterns in the Northwest Atlantic . ICES J. Mar. Sci. , 65 , 339 – 350 . Google Scholar CrossRef Search ADS Jónasdóttir , S. H. ( 1999 ) Lipid content of Calanus finmarchicus during overwintering in the Faroe-Shetland Channel . Fish. Oceanogr. , 8 , 61 – 72 . Google Scholar CrossRef Search ADS Kattner , G. and Graeve , M. ( 1991 ) Wax ester composition of the dominant calanoid copepods of the Greenland Sea/Fram Strait region . Polar Res. , 10 , 479 – 485 . Google Scholar CrossRef Search ADS Kosobokova , K. N. ( 1978 ) Diurnal vertical distribution of Calanus hyperboreus Kroyer and Calanus glacialis Jaschnov in the central Polar Basin . Oceanology , 18 , 476 – 480 . Laakmann , S. , Kochzius , M. and Auel , H. ( 2009 ) Ecological niches of Arctic deep-sea copepods: vertical partitioning, dietary preferences and different trophic levels minimize inter-specific competition . Deep Sea Res. Part I Oceanogr. Res. Pap. , 56 , 741 – 756 . Google Scholar CrossRef Search ADS Lawson , G. L. , Wiebe , P. H. , Ashjian , C. J. , Gallager , S. M. , Davis , C. S. and Warren , J. D. ( 2004 ) Acoustically-inferred zooplankton distribution in relation to hydrography west of the Antarctic Peninsula . Deep Sea Res. Part II Top. Stud. Oceanogr. , 51 , 2041 – 2072 . Google Scholar CrossRef Search ADS Lee , R. F. , Hagen , W. and Kattner , G. ( 2006 ) Lipid storage in marine zooplankton . Mar. Ecol. Prog. Ser. , 307 , 273 – 306 . Google Scholar CrossRef Search ADS Lischka , S. and Hagen , W. ( 2007 ) Seasonal lipid dynamics of the copepods Pseudocalanus minutus (Calanoida) and Oithona similis (Cyclopoida) in the Arctic Kongsfjorden (Svalbard) . Mar. Biol. , 150 , 443 – 454 . Google Scholar CrossRef Search ADS Longhurst , A. R. ( 1976 ) Vertical migration. In Walsh , J. J. and Cushing , D. H. (eds) , The Ecology of the Seas . Blackwell , Philadelphia , pp. 116 – 137 . Luo , J. Y. , Grassian , B. , Tang , D. , Irisson , J.-O. , Greer , A. T. , Guigand , C. M. , McClatchie , S. and Cowen , R. K. ( 2014 ) Environmental drivers of the fine-scale distribution of a gelatinous zooplankton community across amesoscale front . Mar. Ecol. Prog. Ser. , 510 , 129 – 149 . Google Scholar CrossRef Search ADS Maps , F. , Plourde , S. and Zakardjian , B. ( 2010 ) Control of dormancy by lipid metabolism in Calanus finmarchicus: a population model test . Mar. Ecol. Prog. Ser. , 403 , 165 – 180 . Google Scholar CrossRef Search ADS Maps , F. , Record , N. R. and Pershing , A. J. ( 2014 ) A metabolic approach to dormancy in pelagic cope- pods helps explaining inter- and intra-specific variability in life-history strategies . J. Plankton Res. , 36 , 18 – 30 . Google Scholar CrossRef Search ADS Melle , W. and Skjoldal , H. R. ( 1998 ) Reproduction and development of Calanus finmarchicus, C. glacialis and C. hyperboreus in the Barents Sea . Mar. Ecol. Prog. Ser. , 169 , 211 – 228 . Google Scholar CrossRef Search ADS Miller , C. B. , Morgan , C. A. , Prahl , F. G. and Sparrow , M. A. ( 1998 ) Storage lipids of the copepod Calanus finmarchicus from Georges Bank and the Gulf of Maine . Limnol. Oceanogr. , 43 , 488 – 497 . Google Scholar CrossRef Search ADS Ohman , M. D. ( 1990 ) The demographic benefits of diel vertical migration by zooplankton . Ecol. Monogr. , 60 , 257 – 281 . Google Scholar CrossRef Search ADS Pepin , P. and Head , E. J. H. ( 2009 ) Seasonal and depth-dependent variations in the size and lipid contents of stage 5 copepodites of Calanus finmarchicus in the waters of the Newfoundland Shelf and the Labrador Sea . Deep-Sea Res. Part I Oceanogr. Res. Pap. , 56 , 989 – 1002 . Google Scholar CrossRef Search ADS Pond , D. W. and Tarling , G. A. ( 2011 ) Phase transitions of wax esters adjust buoyancy in diapausing Calanoides acutus . Limnol. Oceanogr. , 56 , 1310 – 1318 . Google Scholar CrossRef Search ADS R Core Team ( 2017 ) A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. www.r-project.org. Rabindranath , A. , Daase , M. , Falk-Petersen , S. , Wold , A. , Wallace , M. I. , Berge , J. and Brierley , A. S. ( 2010 ) Seasonal and diel vertical migration of zooplankton in the High Arctic during the autumn midnight sun of 2008 . Mar. Biodiversity , 41 , 365 – 382 . Google Scholar CrossRef Search ADS Richardson , A. J. ( 2008 ) In hot water: zooplankton and climate change . ICES J. Mar. Sci. , 65 , 279 – 295 . Google Scholar CrossRef Search ADS Saumweber , W. J. and Durbin , E. G. ( 2006 ) Estimating potential diapause duration in Calanus finmarchicus . Deep Sea Res. Part II Top. Stud. Oceanogr. , 53 , 2597 – 2617 . Google Scholar CrossRef Search ADS Schmid , M. S. , Aubry , C. , Grigor , J. and Fortier , L. ( 2015 ) ZOOMIE v1. 0 (Zooplankton Multiple Image Exclusion). https://zenodo.org/record/17928. Schmid , M. S. , Aubry , C. , Grigor , J. and Fortier , L. ( 2016 ) The LOKI underwater imaging system and an automatic identification model for the detection of zooplankton taxa in the Arctic Ocean . Methods Oceanogr. , 15–16 , 129 – 160 . Google Scholar CrossRef Search ADS Schulz , J. , Barz , K. , Ayon , P. , Lüdtke , A. , Zielinski , O. , Mengedoht , D. and Hirche , H. J. ( 2010 ) Imaging of plankton specimens with the lightframe on-sight keyspecies investigation (LOKI) system . J. Eur. Opt. Soc. Rapid , 5 , 10017s . Google Scholar CrossRef Search ADS Scott , C. L. , Kwasniewski , S. , Falk-Petersen , S. and Sargent , J. R. ( 2000 ) Lipids and life strategies of Calanus finmarchicus, Calanus glacialis and Calanus hyperboreus in late autumn, Kongs- fjorden, Svalbard . Polar Biol. , 23 , 510 – 516 . Google Scholar CrossRef Search ADS Seidov , D. , Antonov , J. I. , Arzayus , K. M. , Baranova , O. K. , Biddle , M. , Boyer , T. P. , Johnson , D. R. , Mishonov , A.V. et al. ( 2015 ) Oceanography north of 60°N from World Ocean Database . Prog. Oceanogr. , 132 , 153 – 173 . Google Scholar CrossRef Search ADS Solow , A. , Davis , C. and Hu , Q. ( 2001 ) Estimating the taxonomic composition of a sample when individuals are classified with error . Mar. Ecol. Prog. Ser. , 216 , 309 – 311 . Google Scholar CrossRef Search ADS Søreide , J. E. , Falk-Petersen , S. , Hegseth , E. N. , Hop , H. , Carroll , M. L. , Hobson , K. A. and Blachowiak- Samolyk , K. ( 2008 ) Seasonal feeding strategies of Calanus in the high-Arctic Svalbard region . Deep Sea Res. Part II Top. Stud. Oceanogr. , 55 , 2225 – 2244 . Google Scholar CrossRef Search ADS Søreide , J. E. , Leu , E. , Berge , J. , Graeve , M. and Falk-Petersen , S. ( 2010 ) Timing of blooms, algal food quality and Calanus glacialis reproduction and growth in a changing Arctic . Glob. Chang. Biol. , 16 , 3154 – 3163 . Tarrant , A. M. , Baumgartner , M. F. , Lysiak , N. S. J. , Altin , D. , Storseth , T. R. and Hansen , B. H. ( 2016 ) Transcriptional profiling of metabolic transitions during development and diapause preparation in the copepod Calanus finmarchicus . Integr. Comp. Biol. , 56 , 1157 – 1169 . Google Scholar CrossRef Search ADS PubMed Tarrant , A. M. , Baumgartner , M. F. , Verslycke , T. and Johnson , C. L. ( 2008 ) Differential gene expression in diapausing and active Calanus finmarchicus (Copepoda) . Mar. Ecol. Prog. Ser. , 355 , 193 – 207 . Google Scholar CrossRef Search ADS Thibault , D. , Head , E. J. H. and Wheeler , P. A. ( 1999 ) Mesozooplankton in the Arctic Ocean in summer . Deep Sea Res. Part I Oceanogr. Res. Pap. , 46 , 1391 – 1415 . Google Scholar CrossRef Search ADS Vogedes , D. , Varpe , O. , Søreide , J. E. , Graeve , M. , Berge , J. and Falk-Petersen , S. ( 2010 ) Lipid sac area as a proxy for individual lipid content of arctic calanoid copepods . J. Plankton Res. , 32 , 1471 – 1477 . Google Scholar CrossRef Search ADS PubMed Wallace , M. I. , Cottier , F. R. , Berge , J. , Tarling , G. A. , Griffiths , C. and Brierley , A. S. ( 2010 ) Comparison of zooplankton vertical migration in an ice-free and a seasonally ice-covered Arctic fjord: an insight into the influence of sea ice cover on zooplankton behavior . Limnol. Oceanogr. , 55 , 831 – 845 . Google Scholar CrossRef Search ADS Welch , H. E. , Bergmann , M. A. , Siferd , T. D. , Martin , K. A. , Curtis , M. F. , Crawford , R. E. , Conover , R. J. and Hop , H. ( 1992 ) Energy flow through the marine ecosystem of the Lancaster Sound Region, Arctic Canada . Arctic. , 45 , 343 – 357 . Google Scholar CrossRef Search ADS Williams , R. ( 1985 ) Vertical distribution of Calanus finmarchicus and C. helgolandicus in relation to the development of the seasonal thermocline in the Celtic Sea . Mar. Biol. , 86 , 145 – 149 . Google Scholar CrossRef Search ADS Zarubin , M. , Farstey , V. , Wold , A. , Falk-Petersen , S. and Genin , A. ( 2014 ) Intraspecific differences in lipid content of calanoid copepods across fine-scale depth ranges within the photic layer . PLoS One , 9 , e92935 . Google Scholar CrossRef Search ADS PubMed Author notes Corresponding editor: Roger Harris © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Plankton Research Oxford University Press

Lipid load triggers migration to diapause in Arctic Calanus copepods—insights from underwater imaging

Loading next page...
 
/lp/ou_press/lipid-load-triggers-migration-to-diapause-in-arctic-calanus-copepods-0xqSvi910S
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0142-7873
eISSN
1464-3774
D.O.I.
10.1093/plankt/fby012
Publisher site
See Article on Publisher Site

Abstract

Abstract Copepod lipids fuel the Arctic marine ecosystem, but information on the fine-scale distribution of copepods and lipids is nonexistent. This study investigated the fine-scale (1 m) vertical distribution of the copepods Calanus hyperboreus, Calanus glacialis and Metridia longa during a Lagrangian drift in the North Water Polynya using the Lightframe On-sight Keyspecies Investigation (LOKI) imaging system. A copepod species- and stage-specific automatic identification model based on machine learning, a subcategory of artificial intelligence, was used to identify images taken by LOKI. Lipids were measured from images of copepods taken over the whole water column (1 m resolution). Diel vertical migration (DVM) in all three species was detected. In C. hyperboreus and C. glacialis C4-females as well as M. longa C5-females lipid load of deep copepod individuals was significantly higher than that of shallower individuals. Vertical distribution profiles and individual lipid loads suggested that individuals with lower lipid load continued DVM, while others with high lipid load ceased migrating, remaining at depth. Calanus hyperboreus individuals seemed to migrate to diapause at lower lipid fullness (50%) than C. glacialis (60%). A bioenergetics model showed that Calanus females had enough lipids to diapause for over a year, highlighting the significant lipid overhead they use for capital breeding. INTRODUCTION In the Arctic Ocean strong seasonality results in a pulsed availability of resources with peak phytoplankton biomass occurring during late spring and summer (Søreide et al., 2010). Zooplankton graze this overabundance of phytoplankton and can channel about a third of that biomass in the form of highly energetic lipids to higher trophic levels such as forage fish, whales and seabirds (Welch et al., 1992; Søreide et al., 2008). Copepods constitute up to 80% of this zooplankton biomass in the Arctic and many are efficient grazers whose behavior and life-cycle strategies are finely tuned to the transient dynamics of primary producers (Auel and Hagen, 2002). However, recent and rapid environmental changes in the Arctic Ocean have already shown impacts on zooplankton distributions, survival and phenology, with the potential to disrupt this sensitive yet critical ecological balance (Falk-Petersen et al., 2007; Richardson, 2008; Søreide et al., 2010; Doney et al., 2012). One strategy copepods display to adjust to environmental variability is extensive diel vertical migration (DVM) during which they move up and down in the water column on a daily cycle, even in the context of Arctic midnight sun during summer (Rabindranath et al., 2010; Darnis et al., 2017). During classic synchronized DVM individuals of a population move together to shallow waters at night to feed, and back to deeper waters during the day to take shelter from visual predators, mainly fish (Darnis et al., 2017), but other types of DVM exist. Such strategies allow copepods to feed efficiently and to rapidly accumulate the mass they need to develop and thrive in the extreme Arctic environment. Seasonal dynamics of lipid accumulation in Calanus hyperboreus and Calanus glacialis are well documented (Ashjian et al., 1995; Lee et al., 2006; Freese et al., 2017). Lipids accumulated by Arctic copepods contain a high proportion of energy-rich wax esters that propagate into the pelagic food web (Lee et al., 2006). This accumulation of large reserves is critical for another ubiquitous strategy displayed by the small number of large Calanus species that overwhelmingly dominate the biomass of Arctic mesozooplankton, the diapause (Thibault et al., 1999). This diapause strategy is characterized by the building of large reserves ahead of a long period of reduced metabolism that coincides with an extensive seasonal vertical migration towards the relative security of the depths (Hirche, 1997; Darnis and Fortier, 2014). This phenomenon is not limited to the interior of the deep oceanic basins, but rather occurs all along the North Atlantic and Arctic Ocean productive shelves as long as regional bathymetry reaches ca. 150 m and below. While diapause dynamics (e.g. timing of descent to diapause, diapause depth and duration) display considerable intra- and interspecies differences (Johnson et al., 2008), it remains essentially dependent on the processes of lipid accumulation and mobilization (Johnson et al., 2008; Falk-Petersen et al., 2009; Maps et al., 2010, 2014; Tarrant et al., 2008, 2016; Baumgartner and Tarrant, 2017). The lipid accumulation window (LAW) hypothesis, for example, proposes that individual copepods initiate diapause only when they have managed to accumulate enough lipid stores to support the (reduced) metabolic demand during dormancy and the maturation process in pre-adult stages (Johnson et al., 2008; Maps et al., 2010; Baumgartner and Tarrant, 2017). Since diapause involves an ontogenetic migration towards the relative security of deeper waters, it implies that a vertical gradient in lipid content should be observed among individuals. Testing this hypothesis in situ has been challenging, owing to the coarse spatio-temporal resolution that net-based sampling methods can provide. Until very recently, sampling for copepod lipids has been done in layers of tens of meters, essentially through multiple closing/opening nets systems (ARIES, Jónasdóttir, 1999; BIONESS, Davies et al., 2015; Hydro-Bios, Laakmann et al., 2009; Pepin and Head, 2009; MOCNESS, Saumweber and Durbin, 2006). Actually, even in studies where sampling was stratified, lipid contents were provided as pooled measurements (mean ± standard deviation), sometimes split between late copepodite and adult stages. But the relative lipid content between a “surface” and a “deep” layer was reported rather anecdotally rather than as the focus of the study (e.g. Lischka and Hagen, 2007; but see Pepin and Head, 2009). As a result, there is currently no information available on the vertical distribution of total lipids (mg) or lipid fullness (LF) (%) at both the individual scale and at a fine vertical resolution of a few meters. However, such information is relevant to many biological processes such as feeding, mate finding, trophic interactions, etc. (Hays et al., 2001; Pond and Tarling, 2011; Zarubin et al., 2014). This potentially limits our detailed understanding of the ecology and life history strategies of these key copepod taxa. Underwater imaging systems that take photographs of individual zooplankton in situ are able to overcome the low vertical resolution of nets (Hirche et al., 2014; Luo et al., 2014), and are therefore useful for studying copepod lipids. This study uses the Lightframe On-sight Keyspecies Investigation (LOKI) system (Schulz et al., 2010; Schmid et al., 2016). Images were identified using a copepod stage-specific automatic zooplankton identification model (Schmid et al., 2016) based on machine learning. The appearance of copepod lipid sacs was investigated in hundreds of LOKI images of individuals of three copepod species that dominate the biomass of mesozooplankton in the Arctic Ocean: the diapausing species C. hyperboreus and C. glacialis, and the non-diapausing Metridia longa. Individuals were automatically identified to the development stage, and their total lipid amount (mg) and LF (%) were estimated. These data of unprecedented level of resolution were used in a bioenergetics model in order to estimate the potential diapause duration of deep-dwelling individuals for the two Calanus congeners (Maps et al., 2014). The objective of this study was to use new imaging technology and its high resolution to explore the relationship between individual lipid content and the vertical migration behavior and diapause strategy of Arctic copepod species. METHOD Study area We sampled zooplankton during the 2013 ArcticNet expedition onboard the CCGS Amundsen. Sampling followed a Lagrangian surface drifter on August 15 and 16. Vertical tows of LOKI were performed in the vicinity of Ellesmere Island, NU, in the North Water Polynya (NOW; Supplementary Fig. S1). The NOW, situated in northwestern Baffin Bay between Greenland and Ellesmere Island, is one of the largest recurrent polynyas in the Arctic with a size of 5–8 × 104 km2 and is considered one of the most productive areas north of the Arctic Circle (Deming et al., 2002). Ice-free waters occur there as early as the beginning of April compared to mid-July in adjacent regions of the Canadian Archipelago (Fortier et al., 2002). LOKI system The LOKI is a mesozooplankton imaging system. A plankton concentration net, with a mouth opening of 0.28 m2 and a mesh size of 200 μm is attached on top of the inflow of the camera (Supplementary Fig. S2). The LOKI camera-imaging channel has a length of 31.3 mm, width of 20.75 mm and height of 4 mm (i.e. a volume of 2.6 cm3). A cod-end collected all zooplankton after imaging, and samples were preserved in 4% formaldehyde–seawater solution buffered with sodium borate for further analyses in the laboratory. At each time step (Table I), LOKI was deployed to 10 m above the seafloor prior to being hauled back to the surface at a constant speed of 24 m min−1. Images of mesozooplankton were collected and stored on an internal Solid State Drive in 8-bit lossless greyscale format. Environmental conditions (temperature, oxygen, conductivity, fluorescence and pressure) were recorded during each LOKI deployment using onboard sensors. For photosynthetically active radiation (EPAR), we used data collected by the ArcticNet main CTD (Biospherical QCP-2300). Detailed information on LOKI design and setup are presented in Schulz et al. (2010) and Schmid et al. (2016). Table I: LOKI deployments carried out during the Lagrangian drift Date Time Latitude °N Longitude °W Bottom depth (m) 15 August 2013 6:40 76° 22′ 53′′ 77° 21′ 53′′ 354 15 August 2013 10:30 76° 21′ 3′′ 77° 29′ 4′′ 381 15 August 2013 18:45 76° 20′ 57′′ 77° 35′ 19′′ 340 15 August 2013 23:20 76° 19′ 55′′ 77° 39′ 51′′ 273 15 August 2013 2:40 76° 17′ 37′′ 77° 45′ 30′′ 275 Date Time Latitude °N Longitude °W Bottom depth (m) 15 August 2013 6:40 76° 22′ 53′′ 77° 21′ 53′′ 354 15 August 2013 10:30 76° 21′ 3′′ 77° 29′ 4′′ 381 15 August 2013 18:45 76° 20′ 57′′ 77° 35′ 19′′ 340 15 August 2013 23:20 76° 19′ 55′′ 77° 39′ 51′′ 273 15 August 2013 2:40 76° 17′ 37′′ 77° 45′ 30′′ 275 Table I: LOKI deployments carried out during the Lagrangian drift Date Time Latitude °N Longitude °W Bottom depth (m) 15 August 2013 6:40 76° 22′ 53′′ 77° 21′ 53′′ 354 15 August 2013 10:30 76° 21′ 3′′ 77° 29′ 4′′ 381 15 August 2013 18:45 76° 20′ 57′′ 77° 35′ 19′′ 340 15 August 2013 23:20 76° 19′ 55′′ 77° 39′ 51′′ 273 15 August 2013 2:40 76° 17′ 37′′ 77° 45′ 30′′ 275 Date Time Latitude °N Longitude °W Bottom depth (m) 15 August 2013 6:40 76° 22′ 53′′ 77° 21′ 53′′ 354 15 August 2013 10:30 76° 21′ 3′′ 77° 29′ 4′′ 381 15 August 2013 18:45 76° 20′ 57′′ 77° 35′ 19′′ 340 15 August 2013 23:20 76° 19′ 55′′ 77° 39′ 51′′ 273 15 August 2013 2:40 76° 17′ 37′′ 77° 45′ 30′′ 275 Image data preparation and automatic identification of copepod taxa using machine learning Every LOKI image was linked to the corresponding data from the environmental sensors based on a timestamp and calculated depth. Fifty-two image parameters including circularity, Hu moments, Fourier descriptors as well as several values describing geometry and connectivity within objects (from Morphological Spatial Pattern Analysis) were measured on all collected images using LOKI software and GUIDOS toolbox (Schulz et al., 2010; see Schmid et al., 2016 for a complete list of image parameters). Representativeness of LOKI image data was ensured by applying the ZOOMIE v 1.0 (Zooplankton Multiple Image Exclusion) software to all images (Schmid et al., 2015). ZOOMIE filters out redundant images of unique individuals. The copepod stage-specific automatic zooplankton identification model developed in Schmid et al. (2016) was applied to all images, using Salford Systems’ predictive modeler (version 7.0) implementation of the random forests (RF) algorithm (Breiman, 2001). RF is a part of machine learning, a subcategory of artificial intelligence. Schmid’s model is highly efficient at identifying images of C. hyperboreus, C. glacialis and M. longa, when presented with many zooplankton types. Accuracy was tested using model internal testing with withheld test data, and external testing in which automatic identification of images from deployments unknown to the model was manually verified. Externally tested accuracies (true positive samples over total manual classification) for C. hyperboreus, C. glacialis and M. longa were 86%, 82% and 83%, respectively. Externally tested specificities (true positive samples over total classifier prediction) for C. hyperboreus, C. glacialis and M. longa were 87%, 81% and 84%, respectively (Schmid et al., 2016). All images belonging to the different developmental stages of C. hyperboreus, C. glacialis and M. longa (Supplementary Fig. S3) were selected for further analysis. Because of morphological similarities and size overlap, the pairs of C. glacialis C2 and C. hyperboreus C1 as well as C. glacialis C3 and C. hyperboreus C2 were treated as single groups by the automatic identification model (Schmid et al., 2016). The automatic identification model does not distinguish Calanus finmarchicus from other congeners. Investigation of cod-end samples revealed that C. finmarchicus abundance was 174 ± 29 ind. m−2 (mean ± 1 SD) during the drift, compared to 12 483 ± 213 ind. m−2 of C. glacialis, as which C. finmarchicus would most likely be identified as by the automatic identification model. Image observations of individuals in each taxon were binned for each 1 m of the water column. Counts of detections of each taxon in 1 m bins were corrected based on the classification probabilities (Solow et al., 2001) of all taxa from the automatic identification model (Schmid et al., 2016). The applied classification probabilities were the averages of the two confusion matrices, from external testing, developed in Schmid et al. (2016). A total of 6983 automatically identified images had been algorithmically and randomly selected over all taxa, and then manually validated, in order to generate these matrices (Schmid et al. 2016). The corrected number of observations was used to estimate taxon abundances (m−3) per 1 m of the water column, based on the volume filtered. See Schmid et al. (2016) for additional details on each step of image processing. Lipid analysis We used two complementary measures of individual lipid content: total lipids (TL, in mg) and LF (in % of body volume). Since inter- and intraspecific variability in individual size can be very large (Forster and Hirst, 2012), total lipid measures were not directly comparable among stages and species, whereas LF could provide a more general picture of the individuals’ condition and lipid metabolism among copepods. Total lipids and LF were estimated for a total of 822 LOKI images of the development stages known to accumulate lipids in a visible lipid sac in their prosome. Images of individual C. hyperboreus stages C3 to adult female (F), C. glacialis C4 to F and M. longa stages C5 to F were sourced from two of the five deployments: one during daytime (18:45 h DST, UTC −4; 380 μE m−2 s−1 incident EPAR), and one at night (2:40 h, 18 μE m−2 s−1 incident EPAR). For C. hyperboreus C3–C5, C. glacialis C4/C5 and M. longa C5/F, 50 individuals were analyzed from each of the two deployments. For. C. glacialis and C. hyperboreus F, a total of 42 and 80 individuals respectively were analyzed, due to their different relative availabilities. Copepod images for lipid analysis were sampled randomly from the whole water column, but images where the copepod was not in a lateral position had to be discarded due to the methodology used (Vogedes et al. 2010), and a new image was drawn randomly. For all individuals, the area of the lipid sac (A, in mm2) was measured (Fig. 1), and used to estimate total lipids according to Vogedes et al. (2010): TL=0.197A1.38. (1) Fig. 1. View largeDownload slide Image of a Calanus hyperboreus female with its lipid sac highlighted in red. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 1. View largeDownload slide Image of a Calanus hyperboreus female with its lipid sac highlighted in red. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) LF was simply defined as the area of the lipid sac divided by the area of the prosome times 100. Potential diapause duration estimation Potential diapause duration was computed for the stages and species known to perform extensive diapause in the area, i.e. C. hyperboreus C3 to F and C. glacialis C4 to F, following Maps et al. (2014): D=4−M(Cout14−Cin14) (2) where M (μg C1/4 s−1) is the metabolic constant for dormant individuals at 0°C. We did not include any effect of temperature since the observed temperature at the expected depth of diapause (>150 m) was for the largest part constant at 0°C (Supplementary Fig. S4). Cin and Cout are the individual body mass (μg carbon) at the onset and at the end of dormancy, respectively (see Equations (3) and (4) below). To estimate these critical values, we made the assumption that the maximum potential diapause duration would be reached upon lipid stores depletion. Total lipid values estimated from the images were converted into carbon using a carbon to lipid conversion factor λ = 0.81, based on the average lipid composition of calanoid copepods (Kattner and Graeve, 1991). Total body carbon of each individual was calculated from the lipid carbon and LF: Cin=λTLLF. (3) Then the difference between Cout and the lipid carbon gave the minimum (structural) carbon content Cout: Cout=Cin−λTL. (4) We likely overestimated the minimal carbon content at the end of diapause since the structural component of the calanoid copepods’ body contains much less carbon per unit of dry weight than its lipid stores, as revealed by comparing the elemental C to N ratios of lipid full vs. lipid poor individuals (e.g. Pepin and Head, 2009). As a result, we consider our estimate of potential maximum diapause duration to be conservative. Data and R-code (R Core Team, 2017) to reproduce these results can be found at https://zenodo.org/record/1203741. RESULTS Environmental setting A dicothermal layer prevailed during the drift study with relatively warm (4.3°C) surface waters sitting on a glacial (−1.5 °C) layer centered around 100 m over warmer deep waters (Supplementary Fig. S4). A strong halocline with salinity 31.4 at the surface and 34.1 at depth provided vertical stability. Fluorescence in surface waters reached 1.4 mg m−3 by the end of the drift (Supplementary Fig. S4). Daylight prevailed from 7:00 to 23:00 with maximum EPAR in the water column reaching 239 μE m−2 s−1 (Supplementary Fig. S4). A constant T-S signature among profiles (not shown) confirmed that the same vertical water mass structure was sampled during the 24 h drift over a 15 km distance. Vertical distributions of copepods The mean abundances (±1 SD) of C. hyperboreus, C. glacialis and M. longa during the drift, as estimated from automatically identified underwater imagery, were 11 465 ± 320, 12 563 ± 265 and 15 668 ± 378 ind. m−2, respectively. The abundance of each species varied little among profiles. While the combined young Calanus copepodite stages C2/C1 distributed vertically according to the fluorescence profile, especially at night (2:40 h profile, Fig. 2), the older C3/C2 showed tendencies of avoiding the immediate surface layer (<10 m) during daytime, before congregating in or slightly under the subsurface chlorophyll maximum (SCM) during night. Older C. hyperboreus development stages presented more distinct vertical distribution patterns. C3 congregated in the SCM at night (2:40 h profile), moved between 100 and 200 m well below the SCM during the day and then moved back towards it at dusk (Fig. 3). Similarly, the bulk of the abundant C4 stage was distributed around 200 m in daytime, moving towards the SCM at dusk and converging into it at night. C5 were distributed all over the water column with greater abundance at depth near the bottom. Females were distributed in the deeper half of the water column (Fig. 3). Calanus glacialis C1 distribution matched chlorophyll a fluorescence at all times within the 0–120 m layer and displayed little vertical movement throughout the diel cycle (Fig. 4). The abundant C4 and C5 were distributed over the entire water column with a noticeable increase in density immediately above and below the SCM at night. The rare female C. glacialis were distributed close to the seafloor at all times (Fig. 4). Vertical distributions of M. longa C1–C4 remained similar throughout the drift (Fig. 5; Supplementary Fig. S5). C4, C5 and females were found throughout the water column but avoided the immediate surface layer (<10 m) in daytime and concentrated in the SCM at night. Males were rare, and always situated deep in the water column (Supplementary Fig. S5). Fig. 2. View largeDownload slide Vertical distribution of the unresolved Calanus glacialis C2 / C. hyperboreus C1 and C. glacialis C3 / C. hyperboreus C2 complexes (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 2. View largeDownload slide Vertical distribution of the unresolved Calanus glacialis C2 / C. hyperboreus C1 and C. glacialis C3 / C. hyperboreus C2 complexes (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 3. View largeDownload slide Vertical distribution of Calanus hyperboreus C3 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 3. View largeDownload slide Vertical distribution of Calanus hyperboreus C3 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 4. View largeDownload slide Vertical distribution of Calanus glacialis C1 and C4 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 4. View largeDownload slide Vertical distribution of Calanus glacialis C1 and C4 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 5. View largeDownload slide Vertical distribution of Metridia longa C4 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. The vertical distributions of the remaining M. longa stages are given in Supplementary Fig. S5. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 5. View largeDownload slide Vertical distribution of Metridia longa C4 to female (F) copepodite stages (black bars) and phytoplankton fluorescence (green line) for five different profiles over the 24-h duration of a drift study in the North Water. Sampling time and incident EPAR (μE m−2 s−1) at the beginning of a profile are given above each panel. Note the differences in maximum depth among profiles. The vertical distributions of the remaining M. longa stages are given in Supplementary Fig. S5. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Copepod lipids In C. hyperboreus, we observed maximum mean total lipids of 2.45 mg and maximum mean lipid fullness of 79% in females, while minimum mean total lipids of 0.09 mg and minimum mean lipid fullness of 39% were found in C3 (Fig. 6; see Supplementary Table S1 for all means and standard errors). Calanus glacialis female individuals contained maximum mean total lipids of 0.52 mg and maximum mean lipid fullness of 80%, while minimum mean total lipids of 0.17 mg and minimum mean lipid fullness of 43% were found in C4 (Fig. 6; Supplementary Table S1). Finally, M. longa females contained maximum mean total lipids of 0.1 mg and maximum mean lipid fullness of 36%, while minimum mean total lipids of 0.01 mg and minimum mean lipid fullness of 12% were found in C5 (Fig. 6; Supplementary Table S1). While in most taxa total lipids and lipid fullness followed normal distributions, C. hyperboreus C3 and female lipid fullness showed a tendency towards a bimodal distribution (Supplementary Fig. S6). Fig. 6. View largeDownload slide Total lipids (a) and lipid fullness (b) for different copepodite stages and female (F) Calanus hyperboreus, C. glacialis and Metridia longa. Fig. 6. View largeDownload slide Total lipids (a) and lipid fullness (b) for different copepodite stages and female (F) Calanus hyperboreus, C. glacialis and Metridia longa. Vertical distribution of lipids In C. hyperboreus C4, C5 and F, total lipids and lipid fullness significantly increased with depth during both day and night (Fig. 7; see Supplementary Table S2 for a linear regression summary). Vertical profiles of total lipids and lipid fullness differed significantly between day and night sampling only in C. hyperboreus C4 (Fig. 7; see Supplementary Table S3 for an analysis of covariance (ANCOVA) summary). Total lipids and lipid fullness of C. glacialis from C4 to female stages were significantly higher in deeper water compared to shallower water during both day and night samplings (Fig. 8; see Supplementary Table S2). Vertical profiles of lipid fullness in individuals differed significantly between day and night samplings in C. glacialis C4 and C5 (Fig. 8; Supplementary Table S3), while for total lipids this was only the case in C. glacialis C4. Total lipids and lipid fullness of M. longa C5 and females were significantly higher in deeper water compared to shallower water during night sampling (Fig. 9; Supplementary Table S2), but not during the daytime. Fig. 7. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Calanus hyperboreus individuals located at different depths. C3–C5, copepodite stages 3, 4 and 5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 7. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Calanus hyperboreus individuals located at different depths. C3–C5, copepodite stages 3, 4 and 5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 8. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Calanus glacialis individuals located at different depths. C4 and C5, copepodite stages C4 and C5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 8. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Calanus glacialis individuals located at different depths. C4 and C5, copepodite stages C4 and C5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 9. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Metridia longa individuals located at different depths. C5, copepodite stage C5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 9. View largeDownload slide Total lipid amounts (left panels) and lipid fullness (right panels) in Metridia longa individuals located at different depths. C5, copepodite stage C5; F, females. In each panel, the light blue crosses show daytime (18:45 h) results. The dark blue crosses show night-time (2:40 h) results. Regression lines for the two datasets (day and night) are shown as light blue and dark blue lines with the 95% confidence intervals shown as gray areas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Potential diapause duration Potential maximum diapause duration (D) estimated for all the developmental stages of the Calanus congeners increased almost linearly with their size (Fig. 10). The youngest developmental stages thought to be able to undergo diapause, i.e. C. hyperboreus C3 and C. glacialis C4, could on average be diapausing during 78 and 112 days, respectively. Subsequent stages could all exceed an arbitrary duration threshold of six months. It also appeared that C. glacialis females could remain dormant for almost a year, whereas C. hyperboreus females potential maximum diapause duration could well exceed this duration. Fig. 10. View largeDownload slide Estimated maximum potential diapause duration for observed development stages of Calanus hyperboreus and Calanus glacialis. Dotted line: 6-month threshold; dashed line: 1-year threshold. Fig. 10. View largeDownload slide Estimated maximum potential diapause duration for observed development stages of Calanus hyperboreus and Calanus glacialis. Dotted line: 6-month threshold; dashed line: 1-year threshold. DISCUSSION Diel vertical migration In this study, underwater images of zooplankton species and stages taken by the LOKI system were automatically, and with high accuracy, identified using a previously developed machine learning model (Schmid et al., 2016). The resulting fine-scale vertical distribution data (1 m resolution) showed clear differences in the DVM of the different developmental stages. The observed DVM in C. hyperboreus, C. glacialis and M. longa corresponded to the classic patterns reported in previous Arctic studies (Kosobokova, 1978; Fortier et al., 2001; Falk-Petersen et al., 2008; Daase et al., 2008, 2015). Our observations are consistent with studies showing the clearest signs of classic DVM in copepods during late Arctic summer and autumn (Kosobokova, 1978; Cottier et al., 2006; Wallace et al., 2010). The ultimate cause for DVM is thought to be predator avoidance, by reducing detection from visual predators such as Themisto sp. and polar cod in surface layers (Hays, 2003), while the proximate cause for DVM is thought to be changes in light intensity (Cohen and Forward, 2009). Local hydrography has also been shown to affect DVM patterns, for instance through thermal optima at depth (Berge et al., 2014), water mass differences (Lawson et al., 2004) and thermoclines (Williams, 1985). However, the observations of individual copepods by means of underwater imaging and at multiple time steps in this study showed that also the small/young copepodite stages (see for instance the C. glacialis C3 / C. hyperboreus C2 complex, Fig. 2) readily migrated from immediate surface waters to deeper waters, and back again, crossing thermo-, and halocline. This suggests that hydrography, and especially clines, were not the most important drivers, and not inhibiting, in this study. Since smaller C. glacialis C3 and C. hyperboreus C2 stages swim at lower Reynolds numbers than larger stages, an effect of clines would be visible first in these weaker swimmers. The youngest and smallest copepodite stages (i.e. C. glacialis C2 / C. hyperboreus C1 complex and M. longa C1) showed no signs of DVM, which can be explained by a reduced visual predation pressure in their size range (Brooks and Dodson, 1965; Fortier et al., 2001; Daase et al., 2015). Instead, these young stages were vertically distributed in proportion to the abundance of their algal food throughout the drift. Especially during the night-time deployment at 2:40, individuals of all copepodite stages except Calanus females accumulated within the SCM (Rabindranath et al., 2010), most likely in order to feed (Falk-Petersen et al., 2008; Daase et al., 2008, 2015). Our results suggest that inter- and intraspecific differences observed in the DVM patterns are linked to the condition of individual copepods (Ohman, 1990; Hays et al., 2001), specifically their lipid fullness. Furthermore, the distribution profiles shown in this study demonstrate that the majority of a given copepod population (e.g. C. glacialis C5, Fig 4.) was constantly vertically migrating, rather than being at a certain “resting depth”, a term which is often used in DVM studies. The role of lipids in vertical migration and diapause In addition to the diapause requirements, C. hyperboreus depends on its lipid reserves for egg spawning in winter (Conover, 1967; Hirche, 1991), while C. glacialis utilizes stored lipids but also grazes on ice algae to spawn during early spring (Hirche and Kattner, 1993; Søreide et al., 2010). As an interesting counterpoint, M. longa is a smaller but abundant omnivorous species that differs from Calanus spp. in that it feeds and reproduces year-round. Since diapause is not a feature of its life cycle (Darnis and Fortier, 2014), it does not accumulate lipid reserves to the level found in Calanus spp. (Ashjian et al., 1995). As a result, we expected very different vertical gradients in lipid content between these species and among their developmental stages. Our individual-based observations obtained on finely resolved vertical and temporal scales showed that lipid-rich C. hyperboreus C5-F, as well as C. glacialis F, were found almost exclusively below 150 m. The lack of any significant increase in numbers within the surface layer at night strongly suggests that these individuals were already diapausing. On the contrary, the changes in lipid fullness observed in younger C. glacialis C4 and C5 stages as well as C. hyperboreus C4 suggest that some individuals with lower lipid fullness were still migrating from their deep refuge during the day towards the surface layer during the night, most likely for feeding (Figs 7–9; Supplementary Table S3), while individuals from the same stages with higher lipid fullness remained deep and did not migrate back up to surface waters. These results support the LAW hypothesis and may even allow estimation of the threshold beyond which active DVM would stop for a given developmental stage. Lipid fullness observations of C. hyperboreus C4 and C. glacialis C5 suggested that C. hyperboreus C4 individuals were able to start diapausing at lower lipid fullness (∼50%) than C. glacialis C5 (∼60%). Observations of lipid-filled Calanus sp. at depth are consistent with the understanding that Calanus sp. are strong seasonal migrators (Darnis and Fortier, 2014) that can diapause as soon as stage C3 for C. hyperboreus (Hirche, 1997) or C4 for C. glacialis (Melle and Skjoldal, 1998; Falk-Petersen et al., 2009). While total lipids reported in this study for C. hyperboreus C4–C5, C. glacialis C4–C5 and M. longa females are within the ranges reported in the literature (Hirche and Kattner, 1993; Ashjian et al., 1995; Scott et al., 2000; Auel et al., 2003), no comparable study has ever shown total lipids and lipid fullness for C. hyperboreus C3 and M. longa C5. The average lipid fullness in C. hyperboreus and C. glacialis females was around 80%, which is higher than what was reported by previous studies elsewhere around the Arctic (Hirche and Kattner, 1993; Ashjian et al., 1995; Scott et al., 2000; Auel et al., 2003). Total lipids and lipid fullness differences between studies can be explained by the individual feeding success (Miller et al., 1998; Baumgartner et al., 2011) as well as seasonal and geographical variability in environmental conditions. High amounts of total lipids and replete oil sacs of Calanus sp. found in our study suggest that these copepods found plenty of suitable phytoplankton food during the productive spring/summer of 2013. Such individual conditions led to high potential diapause duration estimates for Calanus spp. copepodite stages. Substantial lipid reserves of C. hyperboreus C4-F and C. glacialis C5-F were sufficient for a potential diapause from late August until next spring, when the first production peak of ice algae starts (late March–early April). Calanus hyperboreus C5 and both congener’s females even had a sizable lipid surplus, enabling potential diapause of over a year in females. Obviously, females of either species do not remain in diapause for over a year. However, both are known to produce eggs from their lipid reserves, and while C. glacialis reaches its maximum egg production during the spring-summer phytoplankton bloom, C. hyperboreus is a strict capital breeder (Hirche, 1991). This lipid overhead shows that all female individuals of the two Calanus congeners would be able to safely undergo diapause and survive through the long Arctic winter as well as successfully reproduce. While M. longa does not go into diapause, the substantial lipid fullness of its females reinforces the hypothesis that M. longa relies on a mixed strategy of active feeding and accumulating lipids to overcome the harsh winter conditions of the Arctic marine ecosystem, and even produce some eggs (Conover and Huntley, 1991). Higher temperatures of ambient water can substantially increase the metabolic demand of diapausing copepods, leading to a shortening of potential diapause length (Maps et al., 2014). However, temperature of deeper Arctic waters (>200 m) is not expected to increase to a level that would negatively affect diapause length in the near future (Seidov et al., 2015). Climate change-induced mismatch scenarios earlier in the season (Falk-Petersen et al., 2007), however, have the potential to disrupt lipid accumulation and could lead to starvation of copepods during diapause (Auel et al., 2003). This would also negatively affect the capital breeding strategy of Arctic-adapted Calanus species. In such a scenario, M. longa would likely fare better than Calanus spp. owing to its omnivorous, year-round feeding strategy, buffered by some lipid-accumulating capabilities. CONCLUSION This study used a machine learning—artificial intelligence approach to automatically identify underwater imagery, leading to fine-scale (1 m) resolution data on the distribution of individual copepods and species life stages as well as environmental parameters. Data derived from LOKI can be used in studies that aim at integrating the physiology and ecology of organisms, as was demonstrated here with the in situ lipid data of copepods. The present study shows that for Calanus individuals within a given developmental stage, individuals that accumulated high lipid fullness had entered diapause and ceased DVM, while individuals with lower lipid fullness still performed DVM and had not yet entered diapause. The lipid fullness threshold for migration to diapause varied between Calanus species and stages. Potential diapause durations of over a year were estimated in Calanus females, highlighting the large amounts of lipids going into egg production. It is hoped that the fine scale insights on copepod abundance, total lipids and lipid fullness provided in this study will be useful for a range of studies, especially for modeling experiments (e.g. Dufour et al., 2016). SUPPLEMENTARY DATA Supplementary data can be found online at Journal of Plankton Research online. ACKNOWLEDGEMENTS We would like to thank the officers and crew of the CCGS Amundsen for their dedication and professionalism at sea. We are grateful to Jordan Grigor and Cyril Aubry for support throughout the project and reviewing of manuscripts. Thanks to Kevin Gonthier for his help with processing of LOKI data as well as to Catherine Boudreau for her help with the lipid analysis. The help of Maxime Geoffroy, Marianne Falardeau and Alexis Burt with LOKI deployments is much appreciated. Thanks to Andrew Solow and Heidi Sosik for providing MATLAB code for the method described in Solow et al. (2001). We thank three anonymous reviewers for providing helpful comments on earlier versions of the manuscript. FUNDING The Canada Foundation for Innovation (CFI) (grant no. CFI224324) and the Network of Centres of Excellence ArcticNet (grant no. 12000). M.S. received postgraduate scholarships from the Canada Excellence Research Chair (CERC) in Remote Sensing of Canada’s New Arctic Frontier (Takuvik) and stipends from Québec-Océan. This is a joint contribution to Québec-Océan, ArcticNet and the Canada Research Chair on the response of marine arctic ecosystems to climate warming. REFERENCES Ashjian , C. J. , Smith , S. L. and Lane , P. V. Z. ( 1995 ) The Northeast Water Polynya during summer 1992: distribution and aspects of secondary production of copepods . J. Geophys. Res. , 100 , 4371 – 4388 . Google Scholar CrossRef Search ADS Auel , H. and Hagen , W. ( 2002 ) Mesozooplankton community structure, abundance and biomass in the central Arctic Ocean . Mar. Biol. , 140 , 1013 – 1021 . Google Scholar CrossRef Search ADS Auel , H. , Klages , M. and Werner , I. ( 2003 ) Respiration and lipid content of the Arctic copepod Calanus hyperboreus overwintering 1 m above the seafloor at 2,300 m water depth in the Fram Strait . Mar. Biol. , 143 , 275 – 282 . Google Scholar CrossRef Search ADS Baumgartner , M. F. , Lysiak , N. , Schuman , C. , Urban-Rich , J. and Wenzel , F. W. ( 2011 ) Diel vertical migration behavior of Calanus finmarchicus and its influence on right and sei whale occurrence . Mar. Ecol. Prog. Ser. , 423 , 167 – 184 . Google Scholar CrossRef Search ADS Baumgartner , M. F. and Tarrant , A. M. ( 2017 ) The physiology and ecology of diapause in marine copepods . Annu. Rev. Mar. Sci. , 9 , 387 – 411 . Google Scholar CrossRef Search ADS Berge , J. , Cottier , F. , Varpe , Ø. , Renaud , P. E. , Falk-Petersen , S. , Kwasniewski , S. , Griffiths , C. , Søreide , J. E. et al. ( 2014 ) Arctic complexity: a case study on diel vertical migration of zooplankton . J. Plankton Res. , 36 , 1279 – 1297 . Google Scholar CrossRef Search ADS PubMed Breiman , L. ( 2001 ) Random forests . Mach. Learn. , 45 , 5 – 32 . Google Scholar CrossRef Search ADS Brooks , J. L. and Dodson , S. I. ( 1965 ) Predation, body size, and composition of plankton . Science , 150 , 28 – 35 . Google Scholar CrossRef Search ADS PubMed Cohen , J. H. and Forward , R. B. J. ( 2009 ) Zooplankton diel vertical migration—a review of proximate control . Oceanogr. Mar. Biol. Annu. Rev. , 47 , 77 – 109 . Conover , R. J. ( 1967 ) Reproductive cycle, early development, and fecundity in laboratory populations of the copepod Calanus hyperboreus . Crustaceana , 13 , 61 – 72 . Google Scholar CrossRef Search ADS Conover , R. J. and Huntley , M. ( 1991 ) Copepods in ice-covered seas—distribution, adaptations to seasonally limited food, metabolism, growth patterns and life cycle strategies in polar seas . J. Mar. Syst. , 2 , 1 – 41 . Google Scholar CrossRef Search ADS Cottier , F. R. , Tarling , G. A. , Wold , A. and Falk-Petersen , S. ( 2006 ) Unsynchronised and synchronised vertical migration of zooplankton in a high Arctic fjord . Limnol. Oceanogr. , 51 , 2586 – 2599 . Google Scholar CrossRef Search ADS Daase , M. , Eiane , K. , Aksnes , D. L. and Vogedes , D. ( 2008 ) Vertical distribution of Calanus spp. and Metridia longa at four Arctic locations . Mar. Biol. Res. , 4 , 193 – 207 . Google Scholar CrossRef Search ADS Daase , M. , Hop , H. and Falk-Petersen , S. ( 2015 ) Small-scale diel vertical migration of zooplankton in the High Arctic . Polar Biol. , 39 , 1213 – 1223 . Google Scholar CrossRef Search ADS Darnis , G. and Fortier , L. ( 2014 ) Temperature, food and the seasonal vertical migration of key arctic copepods in the thermally stratified Amundsen Gulf (Beaufort Sea, Arctic Ocean) . J. Plankton Res. , 36 , 1092 – 1108 . Google Scholar CrossRef Search ADS Darnis , G. , Hobbs , L. , Geoffroy , M. , Grenvald , J. C. , Renaud , P. E. , Berge , J. , Cottier , F. , Kristiansen , S. et al. ( 2017 ) From polar night to midnight sun: diel vertical migration, metabolism and biogeochemical role of zooplankton in a high Arctic fjord (Kongsfjorden, Svalbard) . Limnol. Oceanogr. , 62 , 1586 – 1605 . Google Scholar CrossRef Search ADS Davies , K. T. A. , Taggart , C. T. and Smedbol , R. K. ( 2015 ) Interannual variation in diapausing copepods and associated water masses in a continental shelf basin, and implications for copepod buoyancy . J. Mar. Syst. , 151 , 35 – 46 . Google Scholar CrossRef Search ADS Deming , J. W. , Fortier , L. and Fukuchi , M. ( 2002 ) The International North Water Polynya Study (NOW): a brief overview . Deep Sea Res. Part II Top. Stud. Oceanogr. , 49 , 4887 – 4892 . Google Scholar CrossRef Search ADS Doney , S. C. , Ruckelshaus , M. , Duffy , J. E. , Barry , J. P. , Chan , F. , English , C. A. , Galindo , H. M. , Grebmeier , J. M. et al. ( 2012 ) Climate change impacts on marine ecosystems . Annu. Rev. Mar. Sci. , 4 , 11 – 37 . Google Scholar CrossRef Search ADS Dufour , K. , Maps , F. , Plourde , S. , Joly , P. and Cyr , F. ( 2016 ) Impacts of intraguild predation on Arctic copepod communities . Front. Mar. Sci. , 3 , 185 . Google Scholar CrossRef Search ADS Falk-Petersen , S. , Leu , E. , Berge , J. , Kwasniewski , S. , Nygård , H. , Røstad , A. , Keskinen , E. , Thormar , J. et al. ( 2008 ) Vertical migration in high Arctic waters during autumn 2004 . Deep Sea Res. Part II Top. Stud. Oceanogr. , 55 , 2275 – 2284 . Google Scholar CrossRef Search ADS Falk-Petersen , S. , Mayzaud , P. , Kattner , G. and Sargent , J. R. ( 2009 ) Lipids and life strategy of Arctic Calanus . Mar. Biol. Res. , 5 , 18 – 39 . Google Scholar CrossRef Search ADS Falk-Petersen , S. , Pavlov , V. , Timofeev , S. and Sargent , J. R. ( 2007 ) Climate variability and possible effects on arctic food chains: the role of Calanus. In Ørbæk , J. B. , Kallenborn , R. , Tombre , I. , Hegseth , E. N. , Falk-Petersen , S. and Hoel , A. H. (eds) , Arctic Alpine Ecosystems and People in a Changing Environment . Springer , Berlin , pp. 147 – 166 . Google Scholar CrossRef Search ADS Forster , J. and Hirst , A. G. ( 2012 ) The temperature-size rule emerges from ontogenetic differences between growth and development rates . Funct. Ecol. , 26 , 483 – 492 . Google Scholar CrossRef Search ADS Fortier , M. , Fortier , L. , Hiroshi , H. , Hiroaki , S. and Legendre , L. ( 2001 ) Visual predators and the diel vertical migration of copepods under Arctic sea ice during the midnight sun . J. Plankton Res. , 23 , 1263 – 1278 . Google Scholar CrossRef Search ADS Fortier , M. , Fortier , L. , Michel , C. and Legendre , L. ( 2002 ) Climatic and biological forcing of the vertical flux of biogenic particles under seasonal Arctic sea ice . Mar. Ecol. Prog. Ser. , 225 , 1 – 16 . Google Scholar CrossRef Search ADS Freese , D. , Søreide , J. E. , Graeve , M. and Niehoff , B. ( 2017 ) A year-round study on metabolic enzymes and body composition of the Arctic copepod Calanus glacialis: implications for the timing and intensity of diapause . Mar. Biol. , 164 , 3 . Google Scholar CrossRef Search ADS Hays , G. C. ( 2003 ) A review of the adaptive significance and ecosystem consequences of zooplankton diel vertical migrations . Hydrobiologia , 503 , 163 – 170 . Google Scholar CrossRef Search ADS Hays , G. C. , Kennedy , H. and Frost , B. W. ( 2001 ) Individual variability in diel vertical migration of a marine copepod: why some individuals remain at depth when others migrate . Limnol. Oceanogr. , 46 , 2050 – 2054 . Google Scholar CrossRef Search ADS Hirche , H. J. ( 1991 ) Distribution of dominant calanoid copepod species in the Greenland sea during late fall . Polar Biol. , 11 , 351 – 362 . Google Scholar CrossRef Search ADS Hirche , H. J. ( 1997 ) Life cycle of the copepod Calanus hyperboreus in the Greenland Sea . Mar. Biol. , 128 , 607 – 618 . Google Scholar CrossRef Search ADS Hirche , H. J. , Barz , K. , Ayon , P. and Schulz , J. ( 2014 ) High resolution vertical distribution of the copepod Calanus chilensis in relation to the shallow oxygen minimum zone off northern Peru using LOKI, a new plankton imaging system . Deep Sea Res. Part II Top. Stud. Oceanogr. , 88 , 63 – 73 . Google Scholar CrossRef Search ADS Hirche , H. J. and Kattner , G. ( 1993 ) Egg production and lipid content of Calanus glacialis in spring: indication of a food-dependent and food-independent reproductive mode . Mar. Biol. , 117 , 615 – 622 . Google Scholar CrossRef Search ADS Johnson , C. L. , Leising , A. W. , Runge , J. A. , Head , E. J. H. , Pepin , P. , Plourde , S. and Durbin , E. G. ( 2008 ) Characteristics of Calanus finmarchicus dormancy patterns in the Northwest Atlantic . ICES J. Mar. Sci. , 65 , 339 – 350 . Google Scholar CrossRef Search ADS Jónasdóttir , S. H. ( 1999 ) Lipid content of Calanus finmarchicus during overwintering in the Faroe-Shetland Channel . Fish. Oceanogr. , 8 , 61 – 72 . Google Scholar CrossRef Search ADS Kattner , G. and Graeve , M. ( 1991 ) Wax ester composition of the dominant calanoid copepods of the Greenland Sea/Fram Strait region . Polar Res. , 10 , 479 – 485 . Google Scholar CrossRef Search ADS Kosobokova , K. N. ( 1978 ) Diurnal vertical distribution of Calanus hyperboreus Kroyer and Calanus glacialis Jaschnov in the central Polar Basin . Oceanology , 18 , 476 – 480 . Laakmann , S. , Kochzius , M. and Auel , H. ( 2009 ) Ecological niches of Arctic deep-sea copepods: vertical partitioning, dietary preferences and different trophic levels minimize inter-specific competition . Deep Sea Res. Part I Oceanogr. Res. Pap. , 56 , 741 – 756 . Google Scholar CrossRef Search ADS Lawson , G. L. , Wiebe , P. H. , Ashjian , C. J. , Gallager , S. M. , Davis , C. S. and Warren , J. D. ( 2004 ) Acoustically-inferred zooplankton distribution in relation to hydrography west of the Antarctic Peninsula . Deep Sea Res. Part II Top. Stud. Oceanogr. , 51 , 2041 – 2072 . Google Scholar CrossRef Search ADS Lee , R. F. , Hagen , W. and Kattner , G. ( 2006 ) Lipid storage in marine zooplankton . Mar. Ecol. Prog. Ser. , 307 , 273 – 306 . Google Scholar CrossRef Search ADS Lischka , S. and Hagen , W. ( 2007 ) Seasonal lipid dynamics of the copepods Pseudocalanus minutus (Calanoida) and Oithona similis (Cyclopoida) in the Arctic Kongsfjorden (Svalbard) . Mar. Biol. , 150 , 443 – 454 . Google Scholar CrossRef Search ADS Longhurst , A. R. ( 1976 ) Vertical migration. In Walsh , J. J. and Cushing , D. H. (eds) , The Ecology of the Seas . Blackwell , Philadelphia , pp. 116 – 137 . Luo , J. Y. , Grassian , B. , Tang , D. , Irisson , J.-O. , Greer , A. T. , Guigand , C. M. , McClatchie , S. and Cowen , R. K. ( 2014 ) Environmental drivers of the fine-scale distribution of a gelatinous zooplankton community across amesoscale front . Mar. Ecol. Prog. Ser. , 510 , 129 – 149 . Google Scholar CrossRef Search ADS Maps , F. , Plourde , S. and Zakardjian , B. ( 2010 ) Control of dormancy by lipid metabolism in Calanus finmarchicus: a population model test . Mar. Ecol. Prog. Ser. , 403 , 165 – 180 . Google Scholar CrossRef Search ADS Maps , F. , Record , N. R. and Pershing , A. J. ( 2014 ) A metabolic approach to dormancy in pelagic cope- pods helps explaining inter- and intra-specific variability in life-history strategies . J. Plankton Res. , 36 , 18 – 30 . Google Scholar CrossRef Search ADS Melle , W. and Skjoldal , H. R. ( 1998 ) Reproduction and development of Calanus finmarchicus, C. glacialis and C. hyperboreus in the Barents Sea . Mar. Ecol. Prog. Ser. , 169 , 211 – 228 . Google Scholar CrossRef Search ADS Miller , C. B. , Morgan , C. A. , Prahl , F. G. and Sparrow , M. A. ( 1998 ) Storage lipids of the copepod Calanus finmarchicus from Georges Bank and the Gulf of Maine . Limnol. Oceanogr. , 43 , 488 – 497 . Google Scholar CrossRef Search ADS Ohman , M. D. ( 1990 ) The demographic benefits of diel vertical migration by zooplankton . Ecol. Monogr. , 60 , 257 – 281 . Google Scholar CrossRef Search ADS Pepin , P. and Head , E. J. H. ( 2009 ) Seasonal and depth-dependent variations in the size and lipid contents of stage 5 copepodites of Calanus finmarchicus in the waters of the Newfoundland Shelf and the Labrador Sea . Deep-Sea Res. Part I Oceanogr. Res. Pap. , 56 , 989 – 1002 . Google Scholar CrossRef Search ADS Pond , D. W. and Tarling , G. A. ( 2011 ) Phase transitions of wax esters adjust buoyancy in diapausing Calanoides acutus . Limnol. Oceanogr. , 56 , 1310 – 1318 . Google Scholar CrossRef Search ADS R Core Team ( 2017 ) A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. www.r-project.org. Rabindranath , A. , Daase , M. , Falk-Petersen , S. , Wold , A. , Wallace , M. I. , Berge , J. and Brierley , A. S. ( 2010 ) Seasonal and diel vertical migration of zooplankton in the High Arctic during the autumn midnight sun of 2008 . Mar. Biodiversity , 41 , 365 – 382 . Google Scholar CrossRef Search ADS Richardson , A. J. ( 2008 ) In hot water: zooplankton and climate change . ICES J. Mar. Sci. , 65 , 279 – 295 . Google Scholar CrossRef Search ADS Saumweber , W. J. and Durbin , E. G. ( 2006 ) Estimating potential diapause duration in Calanus finmarchicus . Deep Sea Res. Part II Top. Stud. Oceanogr. , 53 , 2597 – 2617 . Google Scholar CrossRef Search ADS Schmid , M. S. , Aubry , C. , Grigor , J. and Fortier , L. ( 2015 ) ZOOMIE v1. 0 (Zooplankton Multiple Image Exclusion). https://zenodo.org/record/17928. Schmid , M. S. , Aubry , C. , Grigor , J. and Fortier , L. ( 2016 ) The LOKI underwater imaging system and an automatic identification model for the detection of zooplankton taxa in the Arctic Ocean . Methods Oceanogr. , 15–16 , 129 – 160 . Google Scholar CrossRef Search ADS Schulz , J. , Barz , K. , Ayon , P. , Lüdtke , A. , Zielinski , O. , Mengedoht , D. and Hirche , H. J. ( 2010 ) Imaging of plankton specimens with the lightframe on-sight keyspecies investigation (LOKI) system . J. Eur. Opt. Soc. Rapid , 5 , 10017s . Google Scholar CrossRef Search ADS Scott , C. L. , Kwasniewski , S. , Falk-Petersen , S. and Sargent , J. R. ( 2000 ) Lipids and life strategies of Calanus finmarchicus, Calanus glacialis and Calanus hyperboreus in late autumn, Kongs- fjorden, Svalbard . Polar Biol. , 23 , 510 – 516 . Google Scholar CrossRef Search ADS Seidov , D. , Antonov , J. I. , Arzayus , K. M. , Baranova , O. K. , Biddle , M. , Boyer , T. P. , Johnson , D. R. , Mishonov , A.V. et al. ( 2015 ) Oceanography north of 60°N from World Ocean Database . Prog. Oceanogr. , 132 , 153 – 173 . Google Scholar CrossRef Search ADS Solow , A. , Davis , C. and Hu , Q. ( 2001 ) Estimating the taxonomic composition of a sample when individuals are classified with error . Mar. Ecol. Prog. Ser. , 216 , 309 – 311 . Google Scholar CrossRef Search ADS Søreide , J. E. , Falk-Petersen , S. , Hegseth , E. N. , Hop , H. , Carroll , M. L. , Hobson , K. A. and Blachowiak- Samolyk , K. ( 2008 ) Seasonal feeding strategies of Calanus in the high-Arctic Svalbard region . Deep Sea Res. Part II Top. Stud. Oceanogr. , 55 , 2225 – 2244 . Google Scholar CrossRef Search ADS Søreide , J. E. , Leu , E. , Berge , J. , Graeve , M. and Falk-Petersen , S. ( 2010 ) Timing of blooms, algal food quality and Calanus glacialis reproduction and growth in a changing Arctic . Glob. Chang. Biol. , 16 , 3154 – 3163 . Tarrant , A. M. , Baumgartner , M. F. , Lysiak , N. S. J. , Altin , D. , Storseth , T. R. and Hansen , B. H. ( 2016 ) Transcriptional profiling of metabolic transitions during development and diapause preparation in the copepod Calanus finmarchicus . Integr. Comp. Biol. , 56 , 1157 – 1169 . Google Scholar CrossRef Search ADS PubMed Tarrant , A. M. , Baumgartner , M. F. , Verslycke , T. and Johnson , C. L. ( 2008 ) Differential gene expression in diapausing and active Calanus finmarchicus (Copepoda) . Mar. Ecol. Prog. Ser. , 355 , 193 – 207 . Google Scholar CrossRef Search ADS Thibault , D. , Head , E. J. H. and Wheeler , P. A. ( 1999 ) Mesozooplankton in the Arctic Ocean in summer . Deep Sea Res. Part I Oceanogr. Res. Pap. , 46 , 1391 – 1415 . Google Scholar CrossRef Search ADS Vogedes , D. , Varpe , O. , Søreide , J. E. , Graeve , M. , Berge , J. and Falk-Petersen , S. ( 2010 ) Lipid sac area as a proxy for individual lipid content of arctic calanoid copepods . J. Plankton Res. , 32 , 1471 – 1477 . Google Scholar CrossRef Search ADS PubMed Wallace , M. I. , Cottier , F. R. , Berge , J. , Tarling , G. A. , Griffiths , C. and Brierley , A. S. ( 2010 ) Comparison of zooplankton vertical migration in an ice-free and a seasonally ice-covered Arctic fjord: an insight into the influence of sea ice cover on zooplankton behavior . Limnol. Oceanogr. , 55 , 831 – 845 . Google Scholar CrossRef Search ADS Welch , H. E. , Bergmann , M. A. , Siferd , T. D. , Martin , K. A. , Curtis , M. F. , Crawford , R. E. , Conover , R. J. and Hop , H. ( 1992 ) Energy flow through the marine ecosystem of the Lancaster Sound Region, Arctic Canada . Arctic. , 45 , 343 – 357 . Google Scholar CrossRef Search ADS Williams , R. ( 1985 ) Vertical distribution of Calanus finmarchicus and C. helgolandicus in relation to the development of the seasonal thermocline in the Celtic Sea . Mar. Biol. , 86 , 145 – 149 . Google Scholar CrossRef Search ADS Zarubin , M. , Farstey , V. , Wold , A. , Falk-Petersen , S. and Genin , A. ( 2014 ) Intraspecific differences in lipid content of calanoid copepods across fine-scale depth ranges within the photic layer . PLoS One , 9 , e92935 . Google Scholar CrossRef Search ADS PubMed Author notes Corresponding editor: Roger Harris © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Journal of Plankton ResearchOxford University Press

Published: Apr 16, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off