Zooplankton Distribution and Dynamics in a North Pacific Eddy of Coastal Origin: I

Journal of Oceanography, Vol. 58, pp. 725 to 738, 2002

Zooplankton Distribution and Dynamics in a North Pacific Eddy of Coastal Origin: I. Transport and Loss of Continental Margin Species

1 2 DAVID L. MACKAS * and MOIRA D. GALBRAITH

1Institute of Ocean Sciences, Fisheries and Oceans Canada, Sidney, B.C., V8L 4B2, Canada 2101-2527 Quadra St., Victoria, B.C., V8T 4E1, Canada

(Received 14 November 2001; in revised form 11 April 2002; accepted 11 April 2002)

Zooplankton from coastal/continental margin environments can be transported long Keywords: distances seaward into the subarctic North Pacific by the large (100Ð200 km diam- ⋅ Zooplankton, eter) anticyclonic eddies that form annually in late winter along the eastern margin ⋅ Alaska Gyre, ⋅ of the Alaska Gyre. One recurrent region for eddy formation is off the southern tip of eddy. the Queen Charlotte Islands (near 52°N 132°W). Eddies from this source region (termed ‘Haida eddies’) propagate westward into open ocean waters during the subsequent 1Ð3 years, often to about 140°W, occasionally to mid gyre. Each eddy contains a core of anomalously low density water, and produces an upward doming of the sea surface detectable by satellite altimetry, thereby aiding repeated ship-based sampling. The zooplankton community in the eddies is a mixture between shelf/slope species (transported from the nearshore formation region) and subarctic oceanic species (which colonize the eddy from the sides and below). This paper reports sequential observations (late winter, early summer and fall seasons of 2000, and early summer and fall of 2001) of the abundance and distribution of continental-margin zooplankton in the Haida eddies that formed in late winters of 2000 and 2001. Shelf-origin species declined in abundance over time. Species that appeared to have a continental slope origin sometimes declined but sometimes persisted and flourished. Transport and retention within the eddy appeared to be especially effective for species that undergo diel vertical migration.

1. Introduction physical characteristics and behavior are increasingly Near the end of most winters, several large anticy- well-described (e.g. Crawford, 2002). Each eddy contains clonic eddies form along the eastern margin of the a core of anomalously low density water of nearshore ori- subarctic Pacific. Some of these eddies, especially those gin. For the Haida eddies, Whitney and Robert (2002) between 50°Ð60°N, subsequently detach from the coast show that subsurface TS and nutrient characteristics at and move seaward into the Alaska Gyre. Two recurrent the core are very similar to winter water properties in formation regions (Fig. 1) are off Baranof Island, Alaska Hecate Strait (inshore of the Queen Charlotte Islands). (roughly 57°N 136°W) and off the southern end of the The presence of a lens of low density water produces a Queen Charlotte Islands, Canada (roughly 52°N 132°W). strong anticyclonic geostrophic circulation (surface cur- To label and track individual features, we have adopted a rents 20Ð50 cm sÐ1) around the eddy. Within the eddy, source-year naming convention: eddies originating off there is a 50Ð100 m or larger downward displacement of Baranof Island are called ‘Sitka’ eddies, those from off subsurface isopycnals which extends to depths >1000 m, the Charlottes are called ‘Haida’ eddies. For example, and an upward doming of the sea-surface (10Ð30 cm). ‘Haida 2000’ or ‘H2000’ formed off the Charlottes in Eddy diameter, and the vertical amplitude of the sea sur- JanuaryÐFebruary 2000. face elevation anomalies associated with the eddies, are Although the physical mechanism(s) of initial eddy both large enough that the eddies can be detected and formation are not yet fully understood, their subsequent tracked by satellite. This relatively recent capability has greatly enhanced our awareness and understanding of the * Corresponding author. E-mail: [email protected] eddies. It also made repeated ship-based time series sam- Copyright © The Oceanographic Society of Japan. pling of a given eddy logistically feasible.

725 regions, either dispersed in the general pelagic environment, or associated with isolated shallow offshore habi- tats such as seamounts and islands. How do nearshore- origin zooplankton get there? How likely are they to sur- vive and reproduce? The focus of this paper is on how much and what kinds of ‘coastal’ zooplankton were carried seaward by the Haida 2000 and Haida 2001 eddies, their distribution within the eddies, abundance comparisons vs. near-shore and ‘non-eddy’ Alaska Gyre locations sampled at approximately the same dates, and the even- tual fate of the near-shore taxa as the eddies age. A later paper will examine the interaction of Haida eddies with ‘offshore’ zooplankton species.

2. Methods

2.1 Sampling locations and dates The zooplankton community in and near the Haida 2000 eddy was sampled during five multidisciplinary surveys, and over a total time span of 19 months (Table 1). The final two surveys (June and September 2001) also sampled the Haida 2001 eddy. Figure 2 shows net tow Fig. 1. Map of the eastern subarctic Pacific and its continental and eddy locations during each survey. Sampling loca- margin. tions relative to the eddy were selected for each survey using near-real-time maps of sea-surface elevation anomalies derived from TOPEX-Poseidon and ERS-2 satellite altimeters by the Colorado Center for Astrodynamics The horizontal extent (100Ð200 km diameter) of the Research (http://www-ccar.colorado.edu/~realtime/glo- eddies is large enough that there is appreciable meridi- bal-real-time_ssh/). In each eddy survey, our basic sam- onal cross-eddy difference in Coriolis force, which causes pling pattern (black triangles in Fig. 2) consisted of a sec- the eddy to propagate westward. For both Haida and Sitka tion or sections across the midline of the eddy, with a Eddies, the westward movement carries them away from horizontal station spacing 18.5 or 27.75 km. Each station the North American continental margin and seaward into included a vertical net tow for zooplankton, a CTD pro- the oceanic Alaska Gyre at speeds of 0.5Ð2 km dÐ1. Over file to 1000 or 2000 m using a SeaBird 911, the subsequent 1Ð3 years, the seaward penetration into transmissometer, and fluorometer mounted on a 24 bottle the Alaska Gyre by the Haida Eddies is often to about rosette frame (for water properties, pigments, and macro- 140°W, occasionally to mid gyre. Eddies originating north nutrients), a mixed-layer GO-FLO bottle sample (for trace of about 60°N, where the orientation of the coastline is metal micronutrients), and, if at night, a surface neuston more zonal, travel farther (to 180° and beyond) and some- tow for fish and macroinvertebrate larvae. Additional sam- what faster (about 2.5 km dÐ1) but usually remain along pling was done at selected sites (indicated by black cir- the northern boundary of the Alaska Gyre (Crawford et cles in Fig. 2: eddy center, eddy margin = within the band al., 2000). of strongest anticyclonic currents and steepest dynamic Improved knowledge of the size, annual rate of height gradient, and an ‘outside-the-eddy’ control loca- formation, and physical characteristics of Haida and Sitka tion). This included day and night vertically-stratified eddies has stimulated additional questions about their zooplankton net tows, euphotic zone primary productiv- ecological and biogeochemical importance. Coastal and ity and PAR profiles, and dissolved iron profiles to 1000 offshore parts of the subarctic Pacific are very different m. Current velocities associated with the eddy were mea- environments, not only in terms of differing water depth sured or estimated using an underway acoustic doppler and temperature-salinity characteristics, but also differ- current profiler (RDI 150 kHz), dynamic height data from ing limiting factors for primary productivity, and differ- the CTD surveys, sea-surface elevation patterns from the ences in size spectrum, life history strategies, and spe- satellite altimetry maps, and (in 2001) satellite-tracked cies mix within phytoplankton, zooplankton, and pelagic drifter buoys drogued at 15 m depth. and demersal fish and macroinvertebrate communities. Because we wanted to compare the zooplankton com- Yet ‘nearshore’ species are frequently present in offshore munities within the eddy to the communities in the coastal

726 D. L. Mackas and M. D. Galbraith Fig. 2. Sampling locations during the five 2000 and 2001 surveys of Haida eddies, superimposed on contour maps of sea-surface height anomaly from TOPEX-Poseidon and ERS-2 satellite altimeters (maps courtesy the Colorado Center for Astrodynamics Research website). Areas of yellow and red are anticyclonic eddies. Black triangles indicate locations of bongo tows, larger black circles indicate major stations at which day and night BIONESS profiles were also done.

‘source’ and Alaska Gyre ‘destination’ end-member en- Gyre ‘offshore’ data were from the seaward portion of vironments, we also use zooplankton data from other more Line P (48°49′ N 128°40′ W to 50°N 145°W), from June distant locations that were sampled near the dates of the and September 2001 ‘reference’ stations located between eddy surveys. The comparison ‘nearshore’ data were from H2000 and H2001, and from a June 2001 sampling line established time series lines along the northern Vancou- extending east along 54°30′ N from H2000 toward the ver Island continental shelf and slope (49°Ð52°N). Alaska coast.

Zooplankton Dynamics in a Coastal-Origin Eddy 727 Table 1. Zooplankton sampling in Haida eddies, and at coastal and offshore comparison stations. Two net types were used: bongo nets for vertically-integrated samples from all sites, and BIONESS for day and night vertically-stratified oblique tows (7 strata/tow) at key eddy center, eddy margin, and nearby non-eddy control sites. For spatial comparisons between regions, and as transects across individual eddies, data from BIONESS tows were also vertically integrated 0Ð150 m. See Fig. 1 for map of locations, text for detailed net specifications.

2.2 Zooplankton sampling and processing methods sure sensor and General Oceanics flowmeter interfaced Zooplankton sampling during each of the five sur- to a CTD mounted on the BIONESS. The entire sample vey periods is summarized in Table 1. Vertical hauls were from each BIONESS tow stratum was preserved in 10% done at all stations using a bongo net. Tow depths (150m- formalin. to-surface at eddy and other offshore locations; near-bot- Preserved samples were returned to the laboratory tom-to-surface or 250m-to-surface at the continental mar- for identification and enumeration. The entire sample was gin locations) followed established protocols for NE Pa- scanned for abundance of large and/or rare taxa. The sam- cific offshore and coastal time series (Mackas, 1992; ple was then quantitatively sub-sampled using a Folsom Goldblatt et al., 1999). The bongo net consists of two splitter for counts of small abundant taxa. Total counting cylindrical-conical nets mounted on a central towing effort was ≥400Ð500 individuals per sample, sufficient frame and weight. Each net has a 0.25 m2 mouth area, a to give an expected subsampling error of ≤20% for the filtering area:mouth area ratio of 11.5, and 0.23 mm ap- dominant species. erture black mesh. Volume filtered is estimated from a TSK flowmeter in one side of the net frame. For each 2.3 Statistical tests tow, the catch from one net was preserved in 10% forma- We used different statistical methods to examine lin for taxonomic identification and enumeration; the heterogeneity of horizontal distributions at two different catch from the other net was quick frozen or live-sorted spatial scales. for various gravimetric and chemical analyses. Bongo ¥ Large-scale comparisons between “near-shore” tows were collected both day and night. However, for the (continental shelf and slope), “in-or-near-eddy”, and “non- majority of taxa considered in this paper, diel migration eddy” offshore locations. In this case, we wanted to know was either weak or confined to within the upper 150 m. if concentrations of individual taxa within the eddy were Exceptions are noted in the subsequent text. higher, lower, or similar to concentrations in the near- To resolve vertical distributions and day-night dif- shore origin and offshore dissipation regions. Regional ferences in these distributions, we also collected paired classes for the samples correspond to the columns and day-and-night stratified oblique tows with a BIONESS designations shown in Table 1. We tested for pairwise instrumented multiple net sampler at selected stations between-region heterogeneity using the non-parametric (circles in Fig. 2). BIONESS net mouth area, mesh size, Kruskal-Wallis rank sum test. and filtration ratio are similar to the bongo net, however ¥ “Within-eddy” spatial structure. In this case, we the BIONESS is towed obliquely at a horizontal speed of wanted to know if and how concentrations changed as 2.5Ð3 kts. Standard depth strata for the BIONESS tows we moved from the eddy center toward and into the sur- were 250Ð150, 150Ð100, 100Ð75, 75Ð50, 50Ð25, 25Ð10, rounding open-ocean waters. We considered the size and and 10Ð0 m. Data from the upper six BIONESS strata shape of the eddy to be defined by the pattern of anticy- (0Ð150 m) were also vertically integrated to compare clonic geostrophic circulation around the eddy. Current against the bongo tow results. Depth and cumulative vol- streamlines follow contours of constant geopotential ∆Φ ume filtered were continuously monitored with a pres- anomaly = “dynamic height” ( 5m/1000m, derived from

728 D. L. Mackas and M. D. Galbraith σ Fig. 3. Density ( t) vs. depth from north to south across the Haida 2000 eddy during the June 2000 (a) and September 2000 (b) surveys. Subsurface isopycnals are depressed at the center of the eddy. The major change between June and September was the progressive thermal stratification of the upper 50 m.

CTD data, but also closely following the sea-surface el- both warmer and slightly fresher than at similar depths in evation contours mapped in Fig. 2). Local geopotential the surrounding ocean. Subsurface isopycnals at the center anomaly (high at the center of the eddy) was used to in- of the Haida eddies normally rebound gradually toward dex proximity to the eddy center. Taxa that are most abun- the surface as the eddy ages and weakens (Whitney and dant near the center of an eddy will therefore tend to have Robert, 2002). H2000 maintained approximately the same positive covariance with geopotential anomaly, while taxa diameter, surface elevation anomaly, and internal most abundant at the extreme periphery of the eddy will isopycnal depths from spring through autumn 2000 (see have negative covariance. To test significance of abun- Figs. 2 and 3), but both sea-surface elevations (Fig. 2) dance trends, we calculated within-survey, within-eddy and internal isopycnal depths at the center of the eddy parametric and rank correlations between log (abundance) flattened by about 35% between autumn 2000 and early and local geopotential anomaly. summer 2001. Even in its first year, H2001 was a weaker feature than H2000, with smaller sea-surface elevation 3. Results and Discussion (right hand panels of Fig. 2) and shallower isopycnal depths. 3.1 Eddy hydrography In both H2000 and H2001, density and temperature We provide here only brief descriptions of water in the upper 100 m were strongly affected by seasonal properties and currents in the upper 500 m of the H2000 warming (the September time period shown in Fig. 3(b) and H2001 eddies, for comparison to observed vertical was sampled near the seasonal maximum of thermal strati- and horizontal distributions of zooplankton. The general fication in H2000). Consequences of the summer-autumn physical characteristics of Haida eddies, and the tempo- stratification include reduced frictional coupling between ral evolution of these characteristics as the eddies age, the surface and subsurface layers, and weak horizontal are described in greater detail by Crawford and Whitney gradients of water properties above the seasonal (1999), Crawford (2002), and Whitney and Robert (2002). thermocline. In particular, surface water of the eddy, and Figure 3 shows the vertical distribution of density any surface resident organisms, are easily advected from σ ( t) across the H2000 eddy during the June and Septem- the top of the eddy by Ekman transport and inertial cur- ber 2000 surveys. Below 100 m, the physical structure of rents generated by strong wind events. Supporting evi- H2000 was relatively similar in the two survey periods. dence for this is provided by trajectories of drifters Subsurface isopycnals near the center of the eddy showed drogued at 15 m depth. Six drifters deployed inside ed- a strong downward displacement, due to water that was dies in June and September 2001 all had relatively short

Zooplankton Dynamics in a Coastal-Origin Eddy 729 , and non-eddy Alaska Gyre , and non-eddy

gin, the Haida 2000 eddy

in the upper 150 m). “>” and “<” indicate between-region

Ð3

Taxa shown in the table were selected because they are indica- Taxa

allis rank sum tests (in Feb.-00, we obtained too few samples to

With the exception of the “coastal inlets and straits” group (always With

< 0.05 by Kruskal-W

fshore source locations.

, June and September 2000 surveys.

samples, during February tors of potential coastal or of absent or rare), these taxa were numeric and/or biomass dominants in one or more of the surveyed regions and time periods. Numeric table entries are within-region mean abundances (# m

differences which were significant at differences allow an effective rank comparison among regions). allow an effective

Table 2(a). Zooplankton abundance comparisons between continental mar Table

730 D. L. Mackas and M. D. Galbraith fec-

Alaska Gyre re-

< 0.05 by Kruskal-

gin, H2000 and H2001 eddies, and

ferences which were significant at

Taxa as in Table 2(a). Numeric table entries are within-region mean abundances Table as in Taxa

in the upper 150 m). “>” and “<” indicate between-region dif

Ð3

gions, during June and September 2001. (# m

Wallis rank sum tests (in September 01, because of bad weather we obtained too few samples from H2001 to allow an ef Wallis tive rank comparison vs. the other regions).

Table 2(b). Zooplankton abundance comparisons between continental mar Table

Zooplankton Dynamics in a Coastal-Origin Eddy 731 eddy residence times (days to a few weeks, W. Crawford, ranged from absent to hundreds mÐ3 (Table 2). Average pers. comm). abundances in eddy samples were usually intermediate: lower than at the continental margin sites, but often much 3.2 Zoogeographic classification as ‘near shore’ vs. ‘oce- higher than in the non-eddy Alaska Gyre samples. Occa- anic’ zooplankton taxa, and differences in average sional samples along the outer half of Line P (P16, P20) abundance between coastal, eddy, and oceanic en- and to the northeast of H2000 in September 2001 were vironments exceptions. They had abundances of the shelf species, We selected species for detailed analysis/discussion overall zooplankton species ranking, and water proper- based partly on their pattern of occurrence in the present ties that were more similar to the nearshore and eddy sam- data set, but primarily on previously published descrip- ples than to the other Alaska Gyre stations. Pending avail- tions of mesozooplankton community composition in off- ability of additional geochemical tracer information, we shore areas of the Alaska Gyre (e.g. Marlowe and Miller, do not yet fully understand the reason for this. Although 1975; Miller et al., 1984; Goldblatt et al., 1999), and along the altimeter maps show additional features with posi- the continental margins of southern Alaska (Cooney, 1986; tive sea-surface-elevation anomalies (Fig. 2), these were Incze et al., 1997); British Columbia (Mackas, 1992); weak in comparison to H2000. Locations of these posi- Washington (Landry and Lorenzen, 1989); and Oregon tive elevation anomalies were not coincident with the (Peterson and Miller, 1977). anomalous Line P stations. Another possibility is that high Any zooplankton carried seaward from the B.C. con- concentrations of continental margin taxa are remnants tinental margin are likely to be from one of the following of the very intense H1998 eddy, which took a southwest- four zoogeographic groups: erly track that crossed Line P near 137°W. ¥ ‘Neritic’ (very nearshore) species. Taxa in this Within the ‘boreal shelf’ species group, group normally complete their life cycle, and reach their Pseudocalanus mimus was the most common species. It maximum abundance, either within inner-coast estuaries often ranked as the second or third most abundant copepod and straits or along the innermost part of the outer-coast in both H2000 and H2001 samples (less than the tiny continental shelf. Examples include the copepods Acartia cyclopoid Oithona similis in all seasons, and less than hudsonica (=A. clausi in earlier publications) and Paracalanus parvus in September 2000). In the June sur- Pseudocalanus moultoni, the ctenophore Pleurobrachia veys, abundances of P. mimus and the dominant subarctic bachei, and the scyphomedusa Aurelia sp. (probably copepod Neocalanus plumchrus were similar, although Aurelia labiata, M. Arai, pers. comm.). Taxa from this N. plumchrus accounted for much greater biomass because group were rare in our continental shelf comparison sam- of its larger body size. ples; absent or present at trace levels in our Haida eddy ¥ ‘Southern’ species abundant over the B.C. con- samples (Aurelia juveniles in June 2000, P. moultoni in tinental slope, but endemic to lower latitude regions. Sur- June 2001), and absent in all samples from the oceanic face currents and prevailing winds along the B.C. shelf Alaska Gyre (Table 2). In all regions, abundance estimates break and slope are strongly seasonal: equatorward dur- are unreliable because of their rarity. ing the summer upwelling season, and poleward during ¥ ‘Boreal Shelf’ species endemic to NE Pacific winter (Freeland et al., 1984). Timing and intensity of continental shelf environments (roughly 42°Ð60°N in the the transitions vary from year to year (Thomson and Ware, Alaska Gyre, and also on the Bering Sea shelf). These 1996), but the late-winter season when eddies form fol- are often the dominant zooplankton taxa in terms of both lows immediately after several months of strong poleward abundance and biomass at locations between the B.C. transport. Zooplankton endemic to the California conti- outer coast and the continental shelf break, but are also nental margin and more offshore waters (N. Pacific Tran- common in samples collected up to 100 km seaward of sition Zone) are carried northward along the continental the shelf break. Many have diel and/or ontogenetic mi- margin by the winter currents, with the somewhat gration strategies that aid retention and population main- counterintuitive consequence that ‘warm water’ tenance in a strongly advective shelf environment zooplankton are often most abundant off the Oregon and (Peterson et al., 1979). Interannual variations in abun- British Columbia coasts in the winter season (Peterson dance are associated with interannual differences in the and Miller, 1977; Mackas, 1992). As with the endemic balance between winter poleward and summer boreal shelf species, interannual variations in alonghore equatorward transport (Mackas et al., 2001; Peterson et transport produce strong interannual variations in abso- al., 2002). Examples include the copepods Acartia lute and relative abundance (Mackas et al., 2001). Exam- longiremis, Pseudocalanus mimus, and Calanus ples of taxa in this group include the copepods marshallae, and the euphausiid Thysanoessa spinifera. Paracalanus parvus and Mesocalanus tenuicornis, the Their abundance in individual samples from H2000, planktonic tunicate Salpa fusiformis/aspera, and the the- H2001, and non-eddy Alaska Gyre comparison sites cate pteropod Clio pyramidata. Of this list, only

732 D. L. Mackas and M. D. Galbraith Paracalanus is neritic within its home range. The remain- migrate or be entrained into the eddy from a relatively der have continental margin abundance maxima off Brit- nearshore source region. Their horizontal abundance pat- ish Columbia not because they prefer nearshore environ- terns are affected both by when and where the eddy ments, but because their poleward winter transport is crossed the corresponding nearshore ‘source’ region, and stronger over the slope than further offshore. Our by subsequent horizontal transport interacting with ver- BIONESS samples showed diel vertical migration by both tical positioning behavior of that species in the water col- salps and pteropods, but (contrary to previous reports, e.g. umn. In the remainder of this section, we show that the Wormuth, 1981) most pteropods remained in the upper species with the most nearshore source are initially lo- 75 m both day and night. All taxa in this group were abun- cated closest to the core of the eddy. Continental slope dant in the eddies during at least one survey (Table 2). and offshore species subsequently either accrete around Paracalanus was the numeric dominant in September the rim of the eddy, or in the case of the ontogenetic 2000; salps were biomass dominants at some of the June migrators such as the Neocalanus spp., enter the eddy 2001 H2001 stations, and large Clio were overwhelm- from below as the eddy propagates seaward. We further ingly the biomass dominants in September 2001 samples hypothesize that taxa which avoid the surface layer are from both H2000 and H2001. more likely to remain abundant near the center of the eddy. ¥ ‘Oceanic’ species abundant over the B.C. con- 3.3.1 ‘Boreal shelf’ species tinental slope, but primarily endemic to the North Pa- Pseudocalanus mimus was in most samples the most cific Subarctic Gyres. In most years, subarctic oceanic abundant of the ‘boreal shelf’ species. In February 2000, taxa are the numerical and biomass dominants, not only its absolute abundance was low and consisted mostly of over the B.C. continental slope, but also in the oceanic adult females, but it accounted for nearly 50% of all Alaska Gyre including the Haida eddies. On occasion, copepods. Abundance increased by nearly an order of substantial numbers can also be transported onto the continental shelf by on-shore flow patterns (Cooney, 1986). Examples include the copepods Neocalanus plumchrus, N. cristatus, N. flemingeri, Metridia pacifica, the chaetognath Eukrohnia hamata, the pteropod Limacina helicina, and the euphausiids Euphausia pacifica and Thysanoessa inspinata. The Neocalanus spp. undergo strong seasonal ontogenetic migration. Spawning is at depths between 400Ð2000 m, and the annual cohort of early juveniles enters the surface layer in early spring. N. plumchrus and N. flemingeri complete their growth cycle in the upper 50 m, and leave the upper layer by early summer; N. cristatus normally reside below the surface mixed layer but persist in the upper 150 m to early autumn. Metridia and the euphausiids undergo deep (>150 m range) diel vertical migration. Limacina also migrate vertically but (like Clio) in our samples mostly remained within the upper 150 m. Source location (slope vs. open ocean) for eddy populations of these species is unfortu- nately almost completely cryptic based solely on species identity. However, it is possible that organic bio-markers or isotopic composition (e.g. Perry et al., 1999) may provide clues about where they fed and grew. Abundances were in general similar in the eddies and in other offshore locations. Exceptions to this generalization, and apparent eddy effects on local vertical and horizontal distribution, retention, and growth will be described in a later Fig. 4. Horizontal distribution of Pseudocalanus mimus rela- paper. tive to the H2000 eddy during June 2000. a) Abundance (# mÐ3 in the upper 150 m) by developmental stage vs. lo- 3.3 Horizontal and vertical distributions of ‘coastal-ori- cation (averages at sites with multiple net tows), from north gin’ zooplankton within the eddy (ED10) to south (ED10) across the eddy. b) Scatter plot of log P. mimus abundance (# mÐ3 in the upper 150 m summed As noted above, zooplankton taxa in the ‘neritic’, over all developmental stages) vs. geopotential anomaly ‘boreal shelf’, and ‘southern’ species groups are likely to ∆Φ ( 5m/1000m) for each net tow.

Zooplankton Dynamics in a Coastal-Origin Eddy 733 magnitude between February and June, and the June popu- innoculation of the eddy from dense P. mimus populations lation included many pre-adults, indicating successful along the Southeast Alaska continental margin. reproduction. In the first spring-summer season for each Acartia longiremis was also abundant in first year eddy (H2000 in June 2000, H2001 in June 2001) P. mimus eddy samples, although less so than P. mimus (Table 2). horizontal abundance maxima were near the centers of the eddies (Figs. 4(a) and 5). Consistent with this spatial pattern, P. mimus abundance was strongly correlated with local geopotential anomaly (Fig. 4(b), parametric and rank-correlations both +0.72, p < 0.01). Total abundances were higher in the stronger H2000 eddy. Vertical distribution (Fig. 6) showed a strong but short range diel migration within the upper 250 m. During the night (only about 5 hours duration in the June surveys), all developmental stages were in the upper 25 m. During the remainder of the day, all P. mimus stages left the upper 25 m and were concentrated between 25 and 100 m depth. During the subsequent surveys of H2000 (September 2000, June 2001, and September 2001), total abundances were low (similar to non-eddy locations elsewhere in the Alaska Gyre), and the population was spread much more evenly across the eddy (little or no correlation with geopotential anomaly). In September 2001, abundances in H2000 were higher than in the June 2001 survey and were highest at the northeast margin of the eddy, suggesting possible re-

Fig. 6. Night (left side) vs. daytime (right side) depth distribu- Fig. 5. Horizontal variability of Pseudocalanus mimus in H2001 tions of Pseudocalanus mimus, Acartia longiremis, and during the June survey. Abundance and sample location map Paracalanus parvus, from vertically-stratified BIONESS (red circles, diameter proportional to abundance, which tows. Pseudocalanus migrated down in daylight, ranged 1Ð88 mÐ3) is superimposed on a surface height Paracalanus did not migrate, Acartia had a weak reverse anomaly map of H2001 similar to those shown in Fig. 2. diel migration, but remained above the strong pycnocline.

734 D. L. Mackas and M. D. Galbraith Nearly all Acartia remained in the upper 50 m both day 3.3.2 ‘Southern’ species and night, although there may be a short-range reverse The next group of three species are ‘southern-source’ diel migration (down at night, up in daytime) into and tracers: at times abundant over the B.C. continental slope, out of the upper 10 m (Fig. 6). Location of the horizontal but more usually endemic to lower latitude oceanic and abundance maximum differed among surveys. In June continental margin regions. Of these, the small copepod 2000, highest Acartia horizontal abundance was at the Paracalanus parvus was the most abundant. In all regions, south margin of the eddy (Fig. 7(a)), and this trend was Paracalanus was most abundant in the September 2000 uncorrelated with the domed along-transect variation of survey (Table 2). Like Acartia longiremis, it remained local geopotential anomaly. In September 2000, the high- near the surface both day and night (Fig. 6). Horizontal est abundance was near the middle of the eddy (Fig. 7(b)) abundance varied erratically, but was on average higher and abundance was positively correlated with geopotential in the southern half of the eddy (Fig. 8). Correlations with anomaly (parametric and rank correlations 0.62 and 0.66 geopotential anomaly were non-significant. Mesocalanus respectively, p < 0.05). Like Pseudocalanus, H2000 abun- tenuicornis (not shown) was also most abundant in the dances in the eddy’s second year were greatly reduced September 2000 survey, but remained moderately abun- (Table 2), and became relatively uniform in cross-eddy dant into 2001. It had diel vertical migration and hori- distribution. zontal distribution patterns similar to P. mimus: upper 25 Calanus marshallae is the final copepod in our list m at night, 10Ð50 m in daytime, horizontal abundance of ‘boreal shelf’ tracers. It was rare in all eddy surveys maximal near the center of the eddy, and positively cor- (Table 2). Horizontal distributions and changes over time related with local geopotential anomaly (parametric and (not shown) were generally similar to Pseudocalanus rank correlations +0.6 to +0.66). mimus. We suspect that low abundance in the eddies is Salps were much less abundant than the copepods, due to the deeper depth distribution of overwintering but because of their large body size strongly dominated C. marshallae; relatively few would be in the upper wa- the biomass in several of our June 2001 samples. Highest ter column while eddies were near the coast. salp abundance and biomass was over the continental slope, but salps also dominated the biomass in several samples located around the rim of the H2001 eddy. Salps were very rare or absent in H2000, and along Line P, ex- cept at station P20 (138°40′ W). The June 2001 P20 net tow was one of the anomalous Line P samples mentioned earlier which contained unexpectedly high abundance of several taxa with southerly and/or continental margin origin (most notably Pseudocalanus mimus and Clio pyramidata). In September 2001, the shelled pteropod Clio pyramidata was a dominant member of the eddy zooplankton community (Table 2). Like the salps in June 2001, its horizontal abundance maxima tended to be within the eddy but around the eddy margin rather than at

Fig. 7. Abundance of Acartia longiremis (# mÐ3 in the upper 150 m) across H2000 during June (a) and September (b) Fig. 8. Abundance of Paracalanus parvus (# mÐ3 in the upper surveys. 150 m) across H2000 during the September surveys.

Zooplankton Dynamics in a Coastal-Origin Eddy 735 Fig. 10. Change in average abundance of Pseudocalanus mimus in H2000 (circles) vs. eddy age. Error bars indicate within- time period standard error (±20Ð35%), dashed line shows Fig. 9. Night (left side) vs. daytime (right side) depth distribu- overall average abundance at non-eddy ‘oceanic’ compari- tions of Clio pyramidata in H2000, September 2001 (frac- son sites. tion of water column total in depth interval, three body size classes).

For example, salps were much more abundant in the June 2001 samples from H2001, than in the June 2000 sam- the eddy center (highest at about 30Ð40 km radial dis- ples from H2000; Clio was much more abundant in 2001; tance from the eddy center, and at intermediate local Pseudocalanus mimus was less abundant in first summer geopotential anomalies of about 12.5 m2sÐ2). Clio and the samples from H2001 than in first summer samples from subarctic pteropod Limacina helicina were short-range H2000. diel migrants in our samples (most individuals remaining But there were also trends over time within a single in the upper 75 m both day and night), although both gen- eddy. Declines with eddy age are particularly strong for era migrate to >200 m depth in the Atlantic and elsewhere the shelf origin species. Although average abundance of in the Pacific (Wormuth, 1981, 1985), BIONESS profiles P. mimus in H2000 was initially close to P. mimus abun- (e.g. Fig. 9) showed pronounced daytime abundance dance along the continental margin (90Ð200 mÐ3), eddy maxima between 35Ð50 m, with somewhat variable night abundance rapidly declined toward average spring-sum- migration upward into the upper 10 m. During the Sep- mer ‘open ocean’ concentrations (Fig. 10). The decrease tember 2001 survey, Clio was by far the dominant con- between June 2000 and June 2001 was very close to nega- tributor to zooplankton biomass (and 100Ð150 kHz acous- tive exponential, and spanned two orders of magnitude, tic backscatter) within the 25Ð75 m depth stratum. On although it was followed by a partial recovery in Sep- stormy nights, echosounder data suggested that night up- tember 2001 (also seen at nearby oceanic comparison ward migration of this scattering layer was suppressed, sites). Combined loss rates from the eddy (dispersal and perhaps because strong wind mixing in the upper 10Ð20 predation) apparently greatly exceeded long term repro- m was disruptive of their mucus-web feeding strategy. ductive success within the eddy. Year-to-year declines in the abundance of the other shelf species (Acartia 3.4 Temporal trends in abundance, and implications for longiremis, Calanus marshallae, and Thysanoessa seeding of oceanic environments with coastal spinifera) were also large (Table 2). Paracalanus parvus zooplankton, fish and macroinvertebrates was substantially less abundant in September 2001 than The zooplankton communities in the eddies showed in September 2000. In general, the near-shore origin taxa large survey-to-survey differences in abundance of indi- did relatively well in their first growing season in the eddy, vidual species, and in community dominance hierarchy. but failed to maintain or increase their population into Some of this temporal variability is expected based on the following year. seasonal life histories. For example, Neocalanus spp. were Two strong exceptions to this generalization are the abundant only in June, Paracalanus was most abundant shelled pteropods Clio pyramidata and Limacina helicina. at the end of the summer. Much of the temporal variabil- Of all the zooplankton taxa, these appeared to have been ity is probably due to interannual differences in what spe- the most successful at colonizing the H2000 and H2001 cies are initially and subsequently seeded into the eddies. eddies (in the Atlantic, Wormuth, 1985, reported a simi-

736 D. L. Mackas and M. D. Galbraith larly successful colonization of Gulf Stream cold core late autumn and winter of 2000Ð2001. We found larvae rings by Limacina inflata). Clio and Limacina have trans- of several shelf- and slope-resident fish and invertebrates Pacific oceanic distributions (Limacina more subarctic, in relatively high abundance in both our neuston and Clio more subtropical), but often attain high abundances BIONESS tows within the year-one eddies (Perry et al., in the California Current system (McGowan, 1967). in prep.). The same taxa were much less abundant at our Whether their source region is continental margin or oce- non-eddy reference stations. For these taxa, the required anic is therefore uncertain. However, both Limacina and residence and retention time is a much smaller fraction Clio were often more dominant within the eddies than in of their life cycle: they can and do reach suitable settle- either of the source/sink regions. For example, in June ment environments within a few months. Successful eddy- and September 2000, and in June 2001, within-eddy av- mediated colonization of offshore island environments erage abundances of Limacina were 2Ð10 fold greater than may therefore be less challenging and more common than in the comparison ‘coastal’ or ‘oceanic’ samples (Table colonization of the Alaska Gyre by coastal zooplankton. 2(a)). Somewhat surprisingly, the ‘subtropical’ pteropod Clio pyramidata appeared in substantial numbers in the 4. Summary more southerly Line P samples starting in June 2000, and Large anticyclonic eddies such as the Haida and Sitka in all regions by 2001. By September 2001, it had be- eddies affect zooplankton populations in the oceanic come the dominant contributor to total zooplankton Alaska Gyre in several ways. One of these is an annual, biomass in both the H2000 and H2001 eddies, as well as although spatially-scattered supply of shelf- and slope- being very abundant along the British Columbia conti- origin species to offshore regions. During the first sum- nental slope (Table 2(b)). Within the eddies, many of the mer after eddies leave the coast and move offshore, Clio were individually very large (~1.5 cm shell length). nearshore tracer species are often the dominant or sub- In consequence, although average abundance of Clio dominant zooplankton within the eddies. In most cases, within the eddies was somewhat lower than the continen- their abundances decline rapidly between first- and sec- tal margin average (Table 2(b)), Clio biomass in the ed- ond-year eddies. Species that avoid the surface layer for dies was 3Ð10 fold greater within the eddies than else- most or all of each day appear best able to be retained where. within the eddy core. One of our primary interpretations from the above observations is that vertical distribution behavior is an Acknowledgements important key to population retention within the eddy. In This work was funded by the Canadian Department most cases, species that are either diel migrants or avoid of Fisheries and Oceans ‘Haida Eddies’ Strategic Science the immediate surface layer both day and night (Clio, project. Many individuals participated in this Pseudocalanus, Mesocalanus, Metridia pacifica) multidisciplinary project. However some deserve special achieved and sometimes maintained high average abun- mention. The officers and crew of the research ship J. P. dance in the eddies. Surface dwelling taxa (Acartia, Tully got us and our instruments to the sampling loca- Paracalanus) were more uniformly or randomly distrib- tions, despite sometimes difficult working conditions. uted across the eddy diameter. Because they spend most D. Yelland, J. Dower and H. Maclean did many of the of their time in the surface mixed layer (roughly the up- BIONESS and bongo net tows. Carol Lalli confirmed per 15Ð25 m), their horizontal transport includes not only pteropod species identifications. Our zooplankton data- the ‘closed’ geostrophic circulation around the eddy, but base was designed and is maintained by S. Romaine. Near also intermittently strong (20Ð50 cm sÐ1) wind-driven real-time satellite altimetry from TOPEX/ERS-2 was pro- currents that can shift the surface layer laterally, relative vided by the Colorado Center for Astrodynamics Re- to the underlying geostrophic eddy. In contrast, taxa which search, with further processing by J. Cherniawsky and spend most of each day below the mixed layer are largely W. Crawford. Data on eddy hydrography were mostly isolated from the surface wind-driven transport. processed by M. Robert and F. Whitney. W. Peterson, What do our observations of zooplankton distribu- S. McKinnell, M. Tsurumi and an anonymous referee pro- tion and population dynamics imply for meroplanktonic vided helpful comments on the manuscript. larvae of fish and macroinvertebrates? One of our inter- ests in the eddies is their role in transporting seed References populations of fish and benthic invertebrates to isolated Cooney, R. T. (1986): The seasonal occurrence of Neocalanus environments such as offshore seamounts. H2000 moved cristatus, Neocalanus plumchrus, and Eucalanus bungii very quickly (within 4 months) from the continental mar- over the shelf of the northern Gulf of Alaska. Cont. Shelf gin to the near vicinity of Bowie Seamount (Figs. 1 and Res., 5, 541Ð553. Crawford, W. R. (2002): Physical characteristics of Haida Ed- 2). H2000 then remained close to Bowie for about 5 dies. J. Oceanogr., 58, this issue, 703Ð713. months, before again moving rapidly to the northwest in

Zooplankton Dynamics in a Coastal-Origin Eddy 737 Crawford, W. R. and F. Whitney (1999): Mesoscale eddies luscs in the California Current region. CalCOFI Atlas 6, vii aswirl with data in Gulf of Alaska Ocean. EOS, Trans. Amer. + 218 pp. Geophys. Union, 80, 365Ð370. Miller, C. B., B. E. Frost, H. P. Batchelder, M. J. Clemons and Crawford, W. R., J. Y. Cherniawsky and M. G. G. Foreman R. E. Conway (1984): Life histories of large grazing (2000): Multi-year meanders and eddies in the Alaskan copepods in the subarctic Pacific Ocean. Prog. Oceanogr., Stream as observed by TOPEX/Poseidon altimeter. Geophys. 13, 201Ð243. Res. Lett., 27, 1025Ð1028. Perry, R. I., P. A. Thompson, D. L. Mackas, P. J. Harrison and Freeland, H. J., W. R. Crawford and R. E. Thomson (1984): D. R. Yelland (1999): Stable carbon isotopes as pelagic food Currents along the Pacific coast of Canada. Atmos. Ocean, web tracers in adjacent shelf and slope regions off British 22, 151Ð172. Columbia, Canada. Can. J. Fish. Aquatic Sci., 56, 2477Ð Goldblatt, R. H., D. L. Mackas and A. G. Lewis (1999): 2486. Mesozooplankton community characteristics in the NE Peterson, W. T. and C. B. Miller (1977): Seasonal cycle of subarctic Pacific. Deep-Sea Res. II, 46, 2619Ð2644. zooplankton abundance and species composition along the Incze, L. S., D. W. Sifert and J. M. Napp (1997): central Oregon coast. Fish. Bull., 75, 717Ð724. Mesozooplankton of Shelikov Strait, Alaska: Abundance Peterson, W. T., C. B. Miller and A. Hutchinson (1979): Zona- and community composition. Cont. Shelf Res., 17, 287Ð305. tion and maintenance of copepod populations in the Oregon Landry, M. R. and C. J. Lorenzen (1989): Abundance, distribu- upwelling zone. Deep-Sea Res., 26, 467Ð494. tion, and grazing impact of zooplankton on the Washington Peterson, W. T., J. Keister and L. Feinberg (2002): Effects of shelf. p. 175Ð210. In Coastal Oceanography of Washing- the 1997–98 El Niño event on the hydrography and ton and Oregon, ed. by M. R. Landry and B. M. Hickey, zooplankton off the central Oregon coast. Prog. Oceanogr. Elsevier, New York. (in press). Mackas, D. L. (1992): Seasonal cycle of zooplankton off south- Thomson, R. E. and D. M. Ware (1996): A current velocity in- western British Columbia: 1979Ð1989. Can. J. Fish. Aquatic dex of ocean variability. J. Geophys. Res., 101, 14,297Ð Sci., 49, 903Ð921. 14,310. Mackas, D. L., R. E. Thomson and M. Galbraith (2001): Whitney, F. and M. Robert (2002): Structure of Haida eddies Changes in the zooplankton community of the British Co- and their transport of nutrient from coastal margins into the lumbia continental margin, 1985Ð1999, and their covariation NE Pacific Ocean. J. Oceanogr., 58, this issue, 715Ð723. with oceanographic conditions. Can. J. Fish. Aquatic Sci., Wormuth, J. H. (1981): Vertical distributions and diel migra- 58, 685Ð702. tions of Euthecosomata in the northwest Sargasso Sea. Marlowe, C. J. and C. B. Miller (1975): Patterns of vertical Deep-Sea Res., 28, 1493Ð1515. distribution and migration at Ocean Station ‘P’. Limnol. Wormuth, J. H. (1985): The role of cold-core Gulf Stream rings Oceanogr., 20, 824Ð844. in the temporal and spatial patterns of euthecosomatous McGowan, J. A. (1967): Distributional atlas of pelagic mol- pteropods. Deep-Sea Res., 32, 773Ð788.

738 D. L. Mackas and M. D. Galbraith