Feeding of capelin (Mallotus villosus) in Newfoundland waters Richard L. O’Driscoll, Morag J.D. Parsons & George A. Rose SARSIA O’Driscoll RL, Parsons MJD, Rose GA. 2001. Feeding of capelin (Mallotus villosus) in Newfound- land waters. Sarsia 86:165-176. The diet of capelin (Mallotus villosus Müller) from six areas off the Newfoundland and Labrador coast was compared over three seasons (January, May-June, August-September) in 1999. A total of 1247 stomachs were examined. Of these, 837 (67 %) contained food. The proportion of empty stom- achs was higher in winter (55 %) than in spring (28 %) or autumn (20 %). Copepods were the major prey over all areas and seasons, occurring in 90 % of non-empty stomachs. Hyperiid amphipods, euphausiids, larvaceans and chaetognaths were also important, occurring in 30 %, 11 %, 9 % and 7 % of non-empty stomachs respectively. The importance of these other prey groups increased with in- creasing capelin size. Larger capelin contained larger prey. There were also spatial and temporal differences in diet. Capelin from Placentia Bay, southeastern Newfoundland, consumed smaller copepods and a higher proportion of amphipods than capelin from other areas. Diet composition, particularly the incidence of lipid-rich Calanus species, may influence capelin growth. Richard L. O’Driscoll*, Morag J.D. Parsons, & George A. Rose, Fisheries Conservation, Fisheries and Marine Institute, Memorial University of Newfoundland, P.O. Box 4920, St. John’s NF, Canada A1C 5R3. *Present address: National Institute of Water and Atmospheric Research (NIWA), PO Box 14-901, Kilbirnie, Wellington, New Zealand. E-mail: firstname.lastname@example.org Keywords: Capelin; Mallotus villosus; diet; feeding; growth; Newfoundland. INTRODUCTION sons. This is the most extensive quantitative account of the diet of capelin from Newfoundland waters to date. Capelin (Mallotus villosus Müller) are an important Such information is important for understanding the role component of many northern marine ecosystems. In of capelin in the Northwest Atlantic ecosystem and al- waters off Newfoundland and Labrador capelin are a lows comparison with recent feeding studies from other key forage species for many species of marine mam- northern regions (e.g. Iceland, Astthorsson & Gislason mals (e.g. humpback whale Megaptera novaeangliae 1997; Barents Sea, Ajiad & Pushchaeva 1992; Bering Borowski, harp seal Phoca groenlandica Fabricius), Sea, Naumenko 1984). seabirds (e.g. murres Uria spp., Atlantic puffin Frater- cula arctica L.) and fish (e.g. Atlantic cod Gadus MATERIAL AND METHODS morhua L., Greenland halibut Reinhardtius hippo- glossoides Walbaum), as well as supporting a moder- SAMPLE COLLECTION ate (quota ~30 000 tons in 1999) commercial fishery. Capelin were sampled using either a Campelen 1800 Despite their central role as a link between zoo- bottom-trawl or an International Young Gadoids Pelagic plankton production and higher trophic levels (Bundy Trawl (IYGPT) mid-water-trawl during acoustic-trawl & al. 2000), relatively little is known about feeding of research surveys in winter (January), spring (May-June) capelin in the Northwest Atlantic. Several studies and autumn (August-September) 1999. Both trawls (Templeman 1948; Kovalyov & Kudrin 1973; Chan & were fitted with a cod-end liner with a mesh size of 13 Carscadden 1976; Marchand & al. 1999) describe mm. Fishing times were between 15-30 min. Tow depth capelin diet in general terms, but only Vesin & al. (1981) ranged from near surface (~20 m) to 300 m. and Gerasimova (1994) provide quantitative accounts Samples were collected at sites across the Northeast of prey composition across a range of capelin sizes. Newfoundland and Labrador Shelf and within major These two studies were restricted spatially (western Gulf bays on the northeast and southeast coast (Fig. 1). Six of St. Lawrence, Vesin & al. 1981; Grand Banks, key areas were identified corresponding to the location Gerasimova 1994) and the work of Gerasimova (1994) of acoustically detected concentrations of capelin (Fig. was only in the spring (April-May). 1). Two areas (Northeast Grand Bank and Labrador) In this paper we present information on capelin diet were sampled in all three seasons, two areas (Trinity over a broad geographical area and across three sea- Bay and Placentia Bay) were sampled in winter and 166 Sarsia 86:165-176 – 2001 Stomach contents from formaldehyde-preserved sam- ples were washed and then examined under a dissect- ing microscope. Contents were sorted into major taxo- nomic groups (usually class or order, e.g. copepods, amphipods, euphausiids, larvaceans, pteropods, chaeto- gnaths, fish larvae, etc) and counted. The prey group that contributed the greatest volume of the diet (meas- ured by displacement) of each individual was noted. Weights of stomach contents were not recorded because a suitably sensitive balance was not available. Intact prey items were measured to the nearest 0.1 mm. Total lengths were measured for all groups except copepods where prosome length was measured. It was impossi- ble to obtain sizes for some groups (e.g. larvaceans, pteropods) because of digestion and distortion. When more than 50 intact copepods were present in the stom- Fig. 1. Location of trawl stations where capelin samples were collected in winter (squares), spring (circles) and autumn ach a random sample of 50 was selected and measured. (stars) 1999. Six key areas are indicated. Approximately 10 % of stomach samples were se- lected based on the prey composition and degree of di- gestion (only relatively undigested samples were se- spring only, and the remaining two areas (Avalon Pe- lected) for more detailed identification. In these sam- ninsula and Funk Island Bank) were only sampled in ples, prey items were identified to the lowest possible one season (spring and autumn respectively). Samples taxonomic level (usually genera). The range of sizes of from four other sites outside these key areas were also each taxon present in the stomach contents was re- examined (Fig. 1). corded. A random sample of 200 capelin was measured from STATISTICAL ANALYSIS each trawl. From this sample a subsample of up to five fish per 10-mm length class was selected for stomach It was difficult to formally compare capelin diet between analysis. Because of spatial differences in length com- seasons and locations because of differences in the size position of catches, it was seldom possible to obtain of fish from different areas. Differences in capelin size five fish from each available length class in every catch. will confound seasonal and spatial patterns because of In winter and autumn, the abdominal wall of each length related selectivity of prey (see Results). capelin selected for stomach analysis was slit and then We used diet information from all samples combined the whole fish was preserved in 4 % formaldehyde- and observed length frequency distributions to calcu- seawater solution for later dissection and analysis. In late the expected proportion of stomachs from each area/ spring, stomachs of selected fish were dissected at sea season containing a prey group. For example, over all and preserved in 4 % formaldehyde-seawater solution. areas 30 % of non-empty stomachs from capelin 130- A further subsample of up to five fish per 10-mm length 139 mm in length contained hyperiid amphipods. Four- class from each trawl was frozen for subsequent otolith teen capelin with non-empty stomachs from Trinity Bay removal and ageing. spring samples were in this length range. If diet was similar over all areas and locations we would expect SAMPLE ANALYSIS 0.3(14) = 4 capelin in the 130-139 mm length class from Total length, weight, sex, maturity and stomach full- Trinity Bay spring samples to contain hyperiids. Math- ness (on a scale of 0-4 where 0 is empty and 4 is full, ematically, expected occurrence of prey group f in area Chan & Carscadden 1976) were assessed for all sub- a and season s, Efsa was calculated as: sampled capelin (i.e. up to ten fish per 10-mm length class). Where appropriate total length (from tip of the x mandible to the end of the ventral lobe of the tail) was E fsa = p fi nasi corrected for shrinkage during preservation or freez- i =1 ing. Thawed lengths of frozen fish were converted to fresh lengths by multiplying by a factor of 1.03 (Win- Where pfi was the overall proportion of stomachs in 10 ters 1982). Lengths of whole fish preserved in formalin mm length class i (i = 1,2,3 … x) that contain prey group where multiplied by a factor of 1.05 (our unpublished f and nasi was the number of capelin samples from fish data). in length class i from area a and season s. The pfi was O’Driscoll – Capelin feeding 167 expressed as a proportion of non-empty stomachs ex- (Table 3). Hyperiid amphipods (Themisto spp.), cept for empty stomachs (pempty,i) where it was the pro- euphausiids (Thysanoessa spp.), larvaceans (Oikopleura portion of total stomachs in length class i. sp.) and chaetognaths (Sagitta sp.) were also important Expected values of prey occurrence Efsa were com- in capelin diet, occurring in 30 %, 11 %, 9 % and 7 % pared to observed values using a two-sided binomial of non-empty stomachs respectively (Table 2). Several test of proportions (Freund 1988). The null hypothesis other groups were present in a small proportion (< 5 %) that diet was similar over all areas and seasons was re- of stomachs (Tables 2-3). jected if a prey group was observed significantly more LENGTH RELATED SELECTIVITY OF PREY or less frequently than expected. The test assumes that individual capelin stomachs were statistically independ- Capelin lengths ranged from 72-193 mm. There was ent samples. This assumption may be violated where strong evidence for differences in selectivity of prey capelin were taken from the same trawl catch, so p- related to capelin length (Fig. 2). Larger prey items such values should be treated with caution. as hyperiid amphipods, mysids, larvaceans, pteropods To examine seasonal and spatial variation in the size and fish larvae occurred more frequently in the stom- composition of copepods in capelin diet, copepod sizes achs of larger capelin. Smaller prey like molluscan lar- were first standardised by capelin length. A linear re- vae, fish eggs and diatoms were eaten by smaller gression was fitted to a plot of log-transformed copepod capelin. Copepods and euphausiids were consumed by prosome lengths as a function of log-transformed all lengths of capelin (Fig. 2). Both copepods and capelin length. Residuals about this fitted regression euphausiids occurred in a wide range of sizes in the provided a standardised measure of copepod size inde- diet (Table 3) due to the presence of different species of pendent of capelin length. Non-parametric Mann copepods and different developmental stages of cope- Whitney and Kruskal Wallis tests (Freund 1988) were pods and euphausiids. There was also evidence for se- then used to compare standardised copepod sizes be- lection within prey groups, with larger capelin eating tween areas and seasons. Kruskal Wallis tests were also larger individuals. This selectivity was particularly clear used to compare the size (length-at-age) of capelin be- for copepods (Fig. 3). tween areas. SEASONAL AND SPATIAL DIFFERENCES IN DIET There were spatial and seasonal differences in capelin RESULTS diet (Fig. 4). In general the proportion of empty stom- DIET COMPOSITION achs was lower in spring (28 %) and autumn (20 %) than in winter (55 %), but in some areas (Labrador and A total of 1247 stomachs from 48 trawl samples were examined (Table 1). Of these 837 (67 %) contained food. Copepods were the dominant prey over all capelin Table 2. Frequency of occurrence and importance of prey lengths, sites and seasons, occurring in 90 % of non- groups from the stomachs of capelin caught in Newfoundland empty stomachs and being the major prey by volume in waters in winter, spring and autumn 1999. Occurrence is the 69 % of non-empty stomachs (Table 2). The most com- number of non-empty stomachs that contained a prey group. mon genera of copepods identified from a subset of Dominance is the number of stomachs in which the prey group capelin stomachs were Calanus, Metridia, and Temora contributed the highest volume of stomach contents. Total number of stomachs examined (Table 1) was 1247. Prey Group Occurrence Dominance Table 1. Summary of samples used to examine seasonal and Copepods 753 578 spatial differences in capelin diet in Newfoundland waters. Hyperiid Amphipods 248 145 First number in each cell is the number of individual stom- Euphausiids 95 42 achs analysed. Second number (in parentheses) is the number Larvaceans 79 31 of trawl sets from which stomach samples were taken. Area Chaetognaths 61 9 locations are shown in Fig. 1. Molluscan larvae 42 0 Pteropods 33 7 Winter Spring Autumn Fish larvae 21 14 Labrador 70 (9) 102 (3) 90 (2) Cirripedian larvae 19 0 Funk Island Bank 0 0 90 (3) Mysids 15 5 Trinity Bay 31 (1) 126 (4) 0 Brachyuran larvae 8 0 Avalon Peninsula 0 55 (2) 0 Cladocerans 7 2 Northeast Grand Banks 63 (2) 90 (3) 29 (1) Fish eggs 7 0 Placentia Bay 107 (3) 298 (11) 0 Diatoms 6 0 Other 37 (1) 0 59 (3) Empty 410 410 168 Sarsia 86:165-176 – 2001 Northeast Grand Banks) the level of feeding remained tion of independent samples (see Material and Meth- low (> 50 % of stomachs empty) in spring. Copepods ods), Table 4 provides useful information about spatial were an important prey in all areas and all seasons, but and seasonal variation in diet taking into account dif- other groups contributed a higher volume of the diet at ferences in capelin size. For example, capelin from specific times and locations. For example larvaceans Labrador in winter had a relatively high proportion of were the dominant prey in 49 % of capelin stomachs empty stomachs (Table 4). In non-empty stomachs the from the Avalon Peninsula in spring (Fig. 4). occurrence of copepods and larvaceans in Labrador Table 4 compares expected values of prey occurrence winter samples was lower than expected, while the pro- (Efsa) to observed values. Although significance levels portion containing euphausiids was higher than expected should be treated with caution because of the assump- (Table 4). In contrast, capelin from the Avalon Penin- sula in spring had lower proportions of empty stom- achs and stomachs containing euphausiids than ex- pected, but relatively high occurrence of copepods, Table 3. Taxa identified from each of the prey groupings in larvaceans and pteropods (Table 4). Table 2. The season (W = winter, S = spring, A = autumn) in which each genera was recorded and range of sizes found in There was also seasonal and spatial variation in the capelin stomachs are also given. size composition of copepods in capelin stomachs. Fig- ure 5 shows the linear regression between log-trans- Prey group Season Size range (mm) formed copepod prosome lengths and log-transformed Copepods capelin lengths used to standardise copepod size and Calanus spp. W,S,A 1.5-7.2* account for length related selectivity (Fig. 3). Residuals Centropages sp. A 0.8-1.0* about this fitted regression were used to examine sea- Euchaeta sp. S - Metridia sp. W,S,A 1.0-2.7* sonal and spatial patterns in copepod size independent Microcalanus sp. S 0.5-0.9* of differences in capelin length (Fig. 6). Paracalanus sp. S 0.9-1.2* In the two areas which were sampled in all three sea- Pseudocalanus sp. S,A 0.6-2.0* sons (Labrador and Northeast Grand Banks) there were Temora spp. W,S,A 0.5-1.1* significant seasonal differences in the size of copepods Hyperiid Amphipods in capelin stomachs (Kruskal Wallis tests: Labrador, χ2 = Themisto spp. W,S,A 1.5-18.2 9.2, d.f = 2, p = 0.01; Northeast Grand Banks, χ2 = 14.0, Euphausiids d.f = 2, p = 0.001). The decrease in the size of copepods Thysanoessa spp. W,S,A 1.7-27.3 Cladocerans in capelin stomachs between winter and spring in these Podon sp. A 0.8-1.1 two areas and also in Trinity Bay (Fig. 6) was largely Mysids due to transition in the diet from large overwintering Unidentified W,S 5.5-22.6 Calanus spp. (copepodite V-VI) to smaller, earlier de- Brachyuran larvae velopmental stages (copepodite I-V) of the next Calanus Unidentified S,A - generation. Cirripedian larvae Capelin from Placentia Bay fed on smaller copepods Unidentified cyprids W,S - during the winter than capelin from other areas (Fig. 6, Unidentified nauplii S 0.4-0.5 Larvaceans Kruskal Wallis test, χ2 = 48.2, d.f = 3, p < 0.001). Ex- Oikopleura sp. S - amination of identified samples showed Temora spp. Pteropods and Metridia spp. were much more abundant than Limacina sp. S,A - Calanus spp. in the winter diet of Placentia Bay capelin. Molluscan larvae The proportion of Calanus in the diet of capelin in Unidentified Prosobranchs W,S,A 0.2-0.6 Placentia Bay increased in the spring, and this was re- Unidentified Lamellibranch A 0.2 flected in the increased size of copepods observed in Chaetognaths stomachs in spring compared to winter (Fig. 6, Mann Sagitta sp. W,S,A 19.0-30.0 Fish larvae Whitney U test, U = 924, Z = –5.56, p < 0.001). De- Ammodytes sp. S 32.4-46.9 spite the increase of importance of Calanus in spring, Gadus morhua L. S 12.2-20.0 the diet of capelin in Placentia Bay still consisted of Fish eggs smaller copepods than in other areas (Fig. 6, Kruskal Unidentified S,A - Wallis test χ2 = 71.5, d.f = 4, p < 0.001). This was be- Diatoms cause few large Calanus (copepodite V) were observed Unidentified W,S,A 0.1-0.2 in capelin stomachs from Placentia Bay in spring and *Copepod lengths are prosome lengths only. Lengths of all there was a higher proportion of smaller species such other prey items are total lengths. as Metridia, Temora and Pseudocalanus than in other O’Driscoll – Capelin feeding 169 Fig. 2. Length related selectivity of prey groups by capelin. Length frequency of capelin from which stomachs were analysed is shown in the top left panel. Percentage occurrence of empty stomachs (second panel in left column) is the proportion of all stomachs within a 10-mm length class that contained no food. In all remaining panels percentage occurrence is the proportion of non- empty stomachs within a length class that contained a particular prey group. areas. Perhaps because of the paucity of large copepods, winter capelin from Labrador were larger than fish from larger capelin from Placentia Bay fed mainly on hyperiid Trinity Bay, Northeast Grand Banks and Placentia Bay amphipods in spring. Fifty six percent of non-empty (Fig. 7). By the spring, capelin from Labrador, Trinity stomachs from large (> 140 mm) capelin sampled in Bay, Northeast Grand Banks and the Avalon Peninsula Placentia Bay in spring were dominated by amphipods were similar in length (Fig. 7), indicating more rapid compared to 20 % of stomachs where copepods were growth in Trinity Bay and the Northeast Grand Banks dominant. In other areas copepods remained important than in Labrador or movement of larger fish into these for larger capelin. For example, in Trinity Bay in spring areas from the north. Capelin from Placentia Bay were copepods were the dominant item in 70 % of non-empty generally similar in size to fish from Trinity Bay and stomachs from capelin > 140 mm. the Northeast Grand Banks in winter, but were signifi- cantly smaller than capelin from all other areas in spring SEASONAL AND SPATIAL DIFFERENCES IN CAPELIN SIZE (Fig. 7). Autumn age data are not presented because There were spatial differences in length-at-age of capelin sample sizes were small and it was difficult to separate (Fig. 7). In all areas size of capelin increased between immature, maturing and recovering (spent) fish in au- winter and spring, consistent with growth (Fig. 7). In tumn samples. 170 Sarsia 86:165-176 – 2001 Fig. 3. Length related selectivity of copepods by capelin. Eight panels show percentage occurrence of eight 1-mm size classes of copepods based on measurements of prosome length. Occurrence of each copepod size class is expressed as a percentage of stomachs within a 10-mm capelin length class that contained copepods of any size (Fig. 2). DISCUSSION Diet composition of capelin from Newfoundland wa- 1997; this study) and may also vary interannually ters in 1999 was broadly similar to that reported in pre- (Gerasimova 1994), seasonally (Vesin & al. 1981; vious feeding studies in the Northwest Atlantic Astthorsson & Gislason 1997; this study) and spatially (Templeman 1948; Kovalyov & Kudrin 1973; Chan & (Naumenko 1984; this study). Carscadden 1976; Vesin & al. 1981; Gerasimova 1994; Euphausiids were less important in this study than Marchand & al. 1999) and in other northern regions previously reported, occurring in only 11 % of non- (Naumenko 1984; Ajiad & Pushchaeva 1992; Huse & empty capelin stomachs. This proportion was lower than Toresen 1996; Astthorsson & Gislason 1997). In this in earlier studies of capelin diet in Newfoundland wa- study, as in previous work, the diet was dominated by ters. Kovalyov & Kudrin (1973) recorded euphausiids copepods particularly Calanus spp. Other groups were in 44 % of stomachs with food in March-June 1972, also present including hyperiid amphipods, euphausiids, while Gerasimova (1994) found euphausiids in 11-28 % larvaceans and chaetognaths. The contribution of these of non-empty capelin stomachs from the Grand Banks other groups appears to increase with capelin size (Vesin in April-May 1987-1990. In other areas euphausiids & al. 1981; Ajiad & Pushchaeva 1992; Gerasimova seem to play an even more important role in capelin 1994; Huse & Toresen 1996; Astthorsson & Gislason diet, making up 80-90 % of the diet (by weight) in parts O’Driscoll – Capelin feeding 171 Fig. 4. Seasonal and spatial differences in importance of major prey groupings from six areas in Newfoundland waters. Area locations are shown in Fig. 1 and sample sizes are provided in Table 1. Missing bars indicate no samples were collected. Percentage dominance is the proportion of stomachs in which the prey group contributed the highest volume of stomach contents. of the Bering Sea (Naumenko 1984) and over 95 % of foundland waters by Kovalyov & Kudrin (1973), but the stomach content weight of large (12-14.9 cm) cape- higher than previously observed on the Grand Banks lin in the Barents Sea (Ajiad & Puschaeva 1992). Con- (5-15 % of non-empty stomachs, Gerasimova 1994) and versely, hyperiid amphipods were a relatively impor- also higher than in other northern regions (e.g. Iceland, tant contributor to the diet of capelin in this study, oc- Astthorsson & Gislason 1997). curring in 30 % of non-empty stomachs. This was simi- Previous studies in Newfoundland waters (Temple- lar to the proportion reported for capelin from New- man 1948; Kovalyov & Kudrin 1973; Gerasimova 1994) 172 Sarsia 86:165-176 – 2001 Fig. 5. Relationship between log-transformed copepod length and log-transformed capelin length. have reported cannibalism of capelin eggs and larvae. For example, Templeman found capelin eggs made up ~90 % of the diet of spawning and post-spawning capelin sampled near-shore. No identifiable capelin eggs Fig. 6. Seasonal and spatial variation of copepod sizes in or larvae were observed in capelin stomach contents capelin stomachs. Standardised copepod sizes are residuals examined in this study. Cannibalism is probably only about the fitted regression in Fig. 5 and so are corrected for capelin length selectivity. Boxes show 25th to 75th percen- important in spawning areas during or immediately fol- tiles divided by the median. Whiskers show the range of sizes. lowing spawning. W = winter, S = spring and A = autumn. The seasonality of capelin feeding observed in New- foundland waters in 1999 was consistent with previous reports (Chan & Carscadden 1976; Vesin & al. 1981; sonal pattern in capelin feeding was probably related to Astthorsson & Gislason 1997). Levels of feeding were seasonal patterns in abundance of zooplankton prey. lower in winter than in spring and autumn. The sea- Continuous plankton recorder (CPR) records from the Table 4. Spatial and seasonal differences in composition of capelin diet in Newfoundland waters. Symbols indicate whether occurrence of a prey group in capelin stomachs was higher (+), lower (–), or the same (0) as expected occurrence assuming equal availability of prey across time and space. Expected occurrences were calculated based on overall size selectivity of prey groups (Fig. 2) and observed length frequencies of capelin in each area and season. Significance levels were based on two-sided binomial tests of probability (*** = p < 0.001, ** = p < 0.01, * = p < 0.05, NS = p > 0.05). There was no correction applied for multiple tests. Area Season Empty Hyperiids Larvaceans Molluscan Chaeto- Fish Copepods Euphausiids Pteropods larvae gnaths larvae Labrador Winter + *** – *** – NS + *** –* – NS – NS – NS – NS Spring + *** + NS + NS 0 NS –* + *** + NS – NS 0 NS Autumn – *** +* + NS + NS –* – NS – NS – NS – NS Funk Island Autumn – NS + NS – ** – NS – ** – NS + ** + *** – NS Bank Trinity Bay Winter + *** – ** + NS – NS – NS 0 NS – NS – NS 0 NS Spring – NS + *** – *** + NS + NS + NS – NS –* + NS Avalon Spring – *** +* + NS – ** + *** + *** 0 NS – NS – NS Peninsula NE Grand Winter + NS + NS –* – NS – NS – NS – NS + ** 0 NS Banks Spring + *** + NS – *** –* + NS – NS 0 NS 0 NS – NS Autumn 0 NS + NS + *** + ** – NS 0 NS + NS – NS 0 NS Placentia Winter + *** 0 NS – NS – NS – NS 0 NS + NS – NS 0 NS Bay Spring – *** – *** + *** + NS – *** – *** –* +* + *** O’Driscoll – Capelin feeding 173 Fig. 7. Spatial variation in mean length-at-age of capelin in Newfoundland waters in winter (January) and spring (May-June). Error bars are + 2 SE. Plots are segregated by sex and maturity because there are differences in size between these groupings (Winters 1982). LA = Labrador, TB = Trinity Bay, NG = Northeast Grand Banks, AP = Avalon Peninsula, PB = Placentia Bay. Significance levels from non-parametric Kruskal Wallis tests are shown in the upper right of each plot (*** = p < 0.001, ** = p < 0.01, * = p < 0.05, NS = p > 0.05). 174 Sarsia 86:165-176 – 2001 Northwest Atlantic show most zooplankton species ex- larger-at-age in the spring than fish from Placentia Bay. hibit peak abundance in the spring or autumn, with Skjoldal & al. (1992) reported an interaction between greatly reduced abundance during the winter (Myers & zooplankton biomass (particularly Calanus finmarch- al. 1994). The timing of seasonal cycles in zooplankton icus Gunnerus), water temperature and capelin growth abundance varies between regions (Myers & al. 1994). in the Barents Sea. Low abundance of C. finmarchicus The relatively low levels of capelin feeding we observed in 1984 following an inflow of warm Atlantic water into off Labrador and on the Northeast Grand Banks in May- the Barents Sea contributed to low capelin growth rates June 1999 may reflect delayed development of zoo- and a decline in capelin abundance. Recovery of zoo- plankton biomass in these regions relative to areas fur- plankton abundance in the late 1980s was accompanied ther inshore and to the south (Trinity Bay, Avalon Pe- by an increase in capelin growth and biomass (Skjoldal ninsula and Placentia Bay). & al. 1992). Temporal and spatial differences in capelin diet prob- Similar large-scale fluctuations in growth and abun- ably reflect the underlying composition of the zoo- dance of capelin have been observed in Newfoundland plankton (Astthorsson & Gislason 1997). We did not waters in the 1990s. Since 1991 the mean length of ma- collect zooplankton samples, but most common zoo- ture age 3 and 4 capelin has decreased and the propor- plankton taxa reported from Newfoundland waters tion of fish mature at age 2 has increased resulting in a (Strong 1981) were found in capelin stomachs. An ex- decrease in the size of spawning capelin compared to ception was the cyclopoid copepods. Cyclopoids, par- the 1980s (Nakashima 1994; Carscadden & Nakashima ticularly Oithona similis Claus, were relatively abun- 1997). This reduction in size in the early 1990s occurred dant in zooplankton samples collected over the New- during a period of cooler water temperatures (Naka- foundland Shelf (Strong 1981; Pepin & Maillet 2000) shima 1994), but the size of spawning adult capelin has but were not observed in capelin stomachs during this remained low in the late 1990s despite warming waters study. Oithona similis were present in the diet of larval (DFO 2000). The spatial distribution of capelin off New- (5-14 mm) capelin from Newfoundland waters (Pepin foundland also changed dramatically in the early 1990s & Penney 1997), suggesting that their absence in this (review by Carscadden & Nakashima 1997). Acoustic study may be related to the limited length range of fish surveys, bottom-trawl surveys and cod-stomach-con- sampled. tent analysis all showed a shift in the distribution of The small size-at-age of capelin from Placentia Bay capelin towards the south and east. Capelin virtually and the relatively large size-at-age of capelin from Lab- disappeared from the northern part of their range off rador contradict the earlier observations of Winters Labrador (Carscadden & Nakashima 1997). At the same (1982) that capelin growth increased with increasing time capelin increased in areas such as the Flemish Cap water temperature from north to south. Differences in and Scotian Shelf where they were not common previ- length-at-age between Placentia Bay capelin and capelin ously (Frank & al. 1996). Capelin biomass in acoustic from other regions may be related to spatial variation surveys on the Northeast Newfoundland Shelf declined in diet, particularly in the consumption of Calanus spp. dramatically in 1990-1991 and has remained low Calanus species at high latitudes accumulate large re- (O’Driscoll & al. 2000). serves of lipids (mainly wax esters) in oil sacs (Sargent We hypothesise that observed changes in capelin & Falk-Petersen 1988). Early copepodite stages have growth in Newfoundland waters may be related in part relatively low levels of lipid, and most deposition of to spatial distribution and prey availability. Zooplankton wax esters occurs during stages IV and V (Sargent & biomass in autumn is highest over the deep channels Falk-Petersen 1988). As a result of this accumulation and shelf break of the northern Newfoundland shelf of lipid, large late-stage Calanus have a higher energy (Anderson & Dalley 1997; Dalley & al. 2000). Biomass content (J g–1) than earlier stages of Calanus (e.g. Comita is lower to the south, especially on the plateau of Grand & al. 1966), other copepod species (e.g. Laurence 1976), Banks (Anderson & Dalley 1997; Dalley & al. 2000). and other zooplankton groups (e.g. Williams & Robins The composition of the zooplankton also varies between 1979). In Placentia Bay, where capelin were relatively areas (Pepin & Maillet 2000). Summer zooplankton small, few large Calanus were observed in capelin stom- communities in northern areas tended to contain a higher achs. Late-stage Calanus spp. are also rare in zooplank- proportion of Calanus than communities further south, ton samples collected in Placentia Bay during spring that were dominated by smaller copepods (Pepin & and summer (Paul Snelgrove pers. commn). In other Maillet 2000). A southerly distribution of capelin, as areas (Trinity Bay, Avalon Peninsula, Northeast Grand observed in the 1990s, may have resulted in poorer feed- Banks, Funk Island Bank, Labrador) Calanus with vis- ing conditions (especially a reduction in the availabil- ible oil droplets were common, particularly in the diet ity of high lipid Calanus spp.) and reduced growth. Un- of larger capelin. Capelin from these other areas were fortunately no suitable time series of zooplankton abun- O’Driscoll – Capelin feeding 175 dance are currently available to allow us to test this hy- ACKNOWLEDGEMENTS pothesis. We thank all the sea-going scientific staff and crew of the Spatial distribution of capelin in Newfoundland wa- CCGS Teleost, especially John Anderson and Fran Mowbray. ters appears to be returning to a more northern distribu- Phil Eustace aged the capelin samples. Gary Maillet, John tion (DFO 2000). In 1998 and 1999 capelin were ob- Anderson, Pierre Pepin and three anonymous reviewers pro- served in significant quantities in spring and autumn vided useful comments on this manuscript. Funding for this acoustic surveys off southern Labrador for the first time work came from the NSERC Industrial Chair in Fisheries since 1990 (O’Driscoll & al. 2000). We are monitoring Conservation and from a New Zealand Foundation for Re- the influence of these changes on capelin biology and search, Science and Technology Post-Doctoral Fellowship held by the first author. growth. REFERENCES Ajiad AM, Pushchaeva TY. 1992. The daily feeding dynamics Gerasimova OV. 1994. Peculiarities of spring feeding by in various length groups of the Barents Sea capelin. In: capelin (Mallotus villosus) on the Grand Bank in 1987- Bogstad B, Tjelmeland S, editors. Interrelations be- 90. Journal of Northwest Atlantic Fisheries Science tween fish populations in the Barents Sea. Bergen, 17:59-67. Norway: Institute of Marine Research. p 181-196. Huse G, Toresen R. 1996. A comparative study of the feeding Anderson JT, Dalley EL. 1997. The nekton of the coastal and habits of herring (Clupea harengus, Clupeidae, L.) and shelf waters of Newfoundland. Canadian Stock Assess- capelin (Mallotus villosus, Osmeridae, Müller) in the ment Secretariat Research Document 97/124. 30 p. Barents Sea. Sarsia 81:143-153. Astthorsson OS, Gislason A. 1997. On the food of capelin in Kovalyov SM, Kudrin BD. 1973. Soviet investigations on the subarctic waters north of Iceland. Sarsia 82:81-86. capelin in the Northwest Atlantic in 1971 and 1972. Bundy A, Lilly GR, Shelton PA. 2000. A mass-balance model International Commission for the Northwest Atlantic of the Newfoundland-Labrador Shelf. Canadian Tech- Fisheries Redbook 1973(III):121-126. nical Report of Fisheries and Aquatic Sciences 2310. Laurence GC. 1976. Caloric values of some North Atlantic 157 p. calanoid copepods. Fishery Bulletin 74:218-220. Carscadden J, Nakashima BS. 1997. Abundance and changes Marchand C, Simard Y, Gratton, Y. 1999. Concentration of in distribution, biology, and behavior of capelin in re- capelin (Mallotus villosus) in tidal upwelling fronts at sponse to cooler waters of the 1990s. In: Lowell the head of the Laurentian Channel in the St. Lawrence Wakefield Fisheries Symposium Series. Forage fishes estuary. Canadian Journal of Fisheries and Aquatic Sci- in marine ecosystems. Anchorage, Alaska: University ences 56:1832-1848. of Alaska Sea Grant College Program. p 457-468. Myers RA, Barrowman NJ, Mertz G, Gamble J, Hunt HG. Chan M, Carscadden J. 1976. The food and feeding of capelin 1994. Analysis of continuous plankton recorder data in (Mallotus villosus) in the Labrador area during autumn the Northwest Atlantic 1959-1992. Canadian Techni- 1973. International Commission for the Northwest At- cal Report of Fisheries and Aquatic Sciences 1966. lantic Fisheries Research Document 76/VI/20. 5 p. 246 p. Comita GW, Marshall SM, Orr AP. 1966. On the biology of Nakashima BS. 1994. The relationship between oceanographic Calanus finmarchicus XIII seasonal change in weight, conditions in the 1990s and changes in spawning be- calorific value and organic matter. Journal of the Ma- haviour, growth and early life history of capelin rine Biological Association of the United Kingdom (Mallotus villosus). Northwest Atlantic Fisheries Or- 46:1-17. ganization Scientific Council Studies 24:55-68. Dalley EL, Anderson JT, Davis DJ. 2000. Short term fluctua- Naumenko EA. 1984. Diet of Pacific capelin, Mallotus villosus tions in the pelagic ecosystem of the Northwest Atlan- socialis (Osmeridae) in the Bering Sea. Journal of Ich- tic. Canadian Stock Assessment Secretariat Research thyology 24(3):130-134. Document 2000/101. 47 p. O’Driscoll RL, Anderson JT, Mowbray FK. 2000. Abundance DFO. 2000. Capelin in Subarea 2 + Div. 3KL. Canadian De- and distribution of capelin from an acoustic survey in partment of Fisheries and Oceans Science Stock Sta- conjunction with the 1999 pelagic juvenile survey. In: tus Report B2-02. 8 p. Capelin in SA2 + Div. 3KL. Canadian Stock Assess- Frank KT, Carscadden JE, Simon JE. 1996. Recent excursions ment Secretariat Research Document 2000/71. of capelin (Mallotus villosus) to the Scotian Shelf and Pepin P, Maillet GL. Biological and chemical oceanographic Flemish Cap during anomalous hydrographic condi- conditions on the Newfoundland Shelf during 1998 and tions. Canadian Journal of Fisheries and Aquatic Sci- 1999 with comparisons to the 1993-97 observations. ences 53:1473-1486. Canadian Stock Assessment Secretariat Research Docu- Freund JE. 1988. Modern elementary statistics. 7th edition. ment 2000/111. 40 p. Eaglewood Cliffs, New Jersey: Prentice-Hall. 574 p. 176 Sarsia 86:165-176 – 2001 Pepin P, Penney RW. 1997. Patterns in prey size and taxo- Vesin J-P, Leggett WC, Able KW. 1981. Feeding ecology of nomic composition in larval fish: are there general size- capelin (Mallotus villosus) in the estuary and western dependent models? Journal of Fish Biology 51(Sup- Gulf of St. Lawrence and its multispecies implications. plement A):84-100. Canadian Journal of Fisheries and Aquatic Sciences Sargent JR, Falk-Petersen S. 1988. The lipid biochemistry of 38:257-267. calanoid copepods. Hydrobiologia 167/168:101-114. Williams R, Robins D. 1979. Calorific, ash, carbon and nitro- Skjoldal HR, Gjøsæter H, Loeng H. 1992. The Barents Sea gen content in relation to length and dry weight of ecosystem in the 1980s: ocean climate, plankton, and Parathemisto gaudichaudi (Amphipoda: Hyperiidea) capelin growth. ICES Marine Science Symposia in the North East Atlantic Ocean. Marine Biology 195:278-290. 52:247-252. Strong KW. 1981. Section V - seasonal occurrence and distri- Winters GH. 1982. Life history and geographical patterns of bution of zooplankton in waters over the Grand Banks growth in capelin, Mallotus villosus, of the Labrador of Newfoundland. In: Grand Banks Oceanographic and Newfoundland areas. Journal of Northwest Atlan- Studies Volume 2. Report prepared for Mobil Oil Canada tic Fisheries Science 3:105-114. by MacLaren Plansearch. 145 p. Templeman W. 1948. The life history of the capelin (Mallotus Accepted 26 September 2000 – Printed 14 September 2001 villosus O. F. Müller) in Newfoundland waters. Bulle- Editorial responsibility: Tore Høisæter tin of the Newfoundland Government Laboratory 17:1- 151.