Climate-induced variability in Calanus marshallae populations

Calanus marshallae (Frost, 1974) is an important component of the zooplankton community over the south-eastern Bering Sea middle shelf, and is the only large copepod that reproduces there (Vidal and Smith, 1986). The Bering Sea ecosystem supports important commercial fisheries and a high biomass of marine mammals and seabirds (National Research Council, 1996). Nauplii and copepodites are important prey of larval and juvenile fish; for example, C. marshallae copepodites constituted 63% of the prey items consumed by age-1 walleye pollock (Theragra chalcogramma) during the summer in the eastern Bering Sea (Grover, 1991). Variability in copepod production therefore potentially affects the Bering Sea ecosystem and its commercial fisheries (e.g. Walsh and McRoy, 1986; Napp et al., 2000). The life history of C. marshallae populations in the south-eastern Bering Sea has been described previously (Smith and Vidal, 1986). Stage 5 copepodites (C5) are believed to overwinter on the shelf, moult to the adult stage in late winter–early spring and produce one or two cohorts each year. Egg production has been reported to begin before (Naumenko, 1979; Smith and Vidal, 1986), during (Vidal and Smith, 1986), and after (Flint et al., 1994) the spring phytoplankton bloom. Spring concentrations of C. marshallae were higher during a cool year than during a warm year (Smith and Vidal, 1986). These years (1980 and 1981), however, do not represent the full range of environmental conditions that occur in the region. The south-eastern Bering Sea is directly influenced by atmospheric forcing, which shows large interannual fluctuations correlated with the North Pacific circulation, El Niño–Southern Oscillation, and Pacific Decadal Oscillation (Niebauer et al., 1999; Overland et al., 2001; Stabeno et al., 2001). This forcing is manifested regionally through the Aleutian Low Pressure System, which affects local winds and surface heat flux, and therefore the formation and movement of sea ice (Ohtani and Azumaya, 1995; Niebauer, 1998). The south-eastern shelf is a marginal ice zone, subject to great variability in the timing and extent of seasonal sea-ice cover (Wyllie-Echeverria and Wooster, 1998). Ice is not formed in situ on the south-east shelf but is advected over it by north-east winds, sometime between November and June; the presence, extent and persistence of ice cover fluctuate with interannual variations in these winds and with local water temperature. Sea-ice dynamics dominate spring physical processes over the shelf, directly affect the development of the phytoplankton bloom and are thought to influence the population dynamics of higher trophic levels (Napp et al., 2000; Hunt et al., 2002). An early phytoplankton bloom occurs when the ice is advected over the shelf during March or April, while in the absence of ice the bloom is delayed until May or June (Stabeno et al., 2001). The biomass of scyphomedusae over the south-eastern Bering Sea shelf increased dramatically during the 1990s, concurrent with increased extent and persistence of ice cover; the authors hypothesized that this was because early life stages of jellyfish depend on ice-associated phytoplankton blooms and the consequent high secondary production (Brodeur et al., 2000). Poor recruitment of walleye pollock, Theragra chalcogramma, in cold years has been attributed to cannibalism resulting from overlapping distributions of larvae/juveniles and adults, which avoid the coldest water (Wyllie-Echeverria and Wooster, 1998; Wespestad et al., 2000). We examined springtime populations of C. marshallae on the middle shelf (between 50 and 100 m isobaths) during 5 consecutive years that spanned the full range of temperature and ice extent observed during the past 34 years (Stabeno et al., 1998; Wyllie-Echeverria and Wooster, 1998). Data on C. marshallae reproduction and abundance were synthesized with environmental data to investigate how C. marshallae populations respond to short-term climate variability. Understanding the linkages between environmental forcing and C. marshallaepopulation dynamics is a first step towards understanding how this ecosystem may respond to longer-term changes in climate. Collections were made between February and September 1995–1999 over the middle shelf of the south-eastern Bering Sea (Figure 1; Table I). Shipboard observations were made from the NOAA ship ‘Miller Freeman’, and time-series of physical and fluorescence data were obtained from moored instruments. We assumed that within each year the population sampled in May was the same as that sampled in April. This assumption, based on sluggish circulation over the shelf, is probably valid for the middle shelf, where mean current speed is <1 cm s−1 (Coachman, 1986). Patterns of copepodite abundance and stage composition in 1998 and 1999, when our temporal coverage was most complete, do not contradict this assumption. Mesozooplankton was sampled using bongo nets(1995, 1997–1999), and a Tucker trawl equipped with Clarke–Bumpus net frames (1996). Copepod nauplii were sampled only in April 1996, using a CalVET net. Bongo samplers comprised 60 and 20 cm diameter frames fitted with 333 and 153 μm mesh nets, respectively. The 20 cm bongo was attached to the towing wire 1 m above the 60 cm frame. A SeaCat CTD was located 1 m above the 20 cm frame to telemeter net depth. Double oblique tows were made to ~5–10 m from the bottom. Clarke–Bumpus nets (153 μm mesh, 25 cm diameter) were nested inside each 1 m2 Tucker trawl net (333 μm mesh). The trawl was towed from 5 m off the bottom to the base of thethermocline (net 1) and from the base of the thermocline to the surface (net 2), using wire angle and length of wire to calculate gear depth. Data from the depth-discrete Tucker trawl samples were integrated over the water column for comparison with those from the bongo tows. Vertical tows with CalVET nets (53 μm mesh) for copepod nauplii were made to 60 m depth. Calibrated flowmeters were used to estimate the volumes filtered by each net. All samples were preserved in 5% formalin:sea water. Copepodites were enumerated at the Polish Plankton Sorting and Identification Center, Szczecin, Poland. Copepodite stages 3–6 (C3–C6) were assumed to be quantitatively retained by the 333 μm mesh, while C1 and C2 were counted from the 153 μm mesh samples (Incze et al., 1997). Sorted subsamples were returned to the Alaska Fisheries Science Center in Seattle. April 1996 CalVET samples for Calanus nauplii were processed in Seattle. In Seattle, female C. marshallae were measured and gonadal maturity was assessed. Approximately 50 females from each station were randomly selected to determine length and gonadal maturity. The specimens were cleared using a 1:1 mixture of ethanol (70% solution) and glycerine. The state of oogenesis was determined using the following criteria, which are slightly modified from Tourangeau and Runge (Tourangeau and Runge, 1991): stage 1—no oocytes formed, small ovary visible; stage 2—anterior oviducts with single rows of small, clear oocytes (~50 μm); stage 3—anterior oviducts filled with clear, amorphous oocytes (~70 μm); stage 4—ventral loop of anterior oviducts filled with translucent oocytes (~100 μm); stage 5—ventral loop of anterior oviducts filled with opaque, round, oocytes (~150 μm); stage 6—posterior oviducts with rows of opaque, round oocytes (~150 μm). Females in stages 4–6 were considered to be reproductively mature. Birth dates and dates of previous moults were estimated for copepodite stages collected in April and May using Campbell’s temperature-dependent growth function for Calanus finmarchicus under satiating food concentrations at temperatures of 4–12°C (Campbell et al., 2001b). Bering Sea in situ temperatures were averaged over depth and then time (February–April) from mooring data. The water column was well mixed and its temperature changed little during that period. Campbell’s function assumes that food does not limit growth, so development times in nature may be longer than our estimates. Calculated development times for C. marshallae using Campbell’s equation were comparable with laboratory and field observations (Table II). The calculated duration of C1–3 combined was 20 days at 3°C. Vidal and Smith’s field-derived estimate was 21 days at similar temperatures (Vidal and Smith, 1986). Development time to naupliar stage 3 at 3°C in our laboratory was 9.6 ± 1.3 days (unpublished data), compared with 9.4 days predicted by Campbell’s equation. A subsurface biophysical mooring was deployed in February of each year at Site 2 [Figure 1; (Stabeno et al., 2001)]. It was replaced in May by a surface mooring. Water temperature was measured every 5–10 m from near surface to bottom with either Seacat SBE 16-03 sensors or Miniature Temperature Recorders (MTR). Wet Labs fluorometers at depths of 5–11 m generated a time-series of fluorescence used to determine the timing of the spring phytoplankton bloom. Ice cover along longitude 169°W was digitized from weekly NOAA ice charts (S. Salo, Pacific Marine Environmental Laboratory, personal communication). South-eastern Bering Sea shelf summer bottom temperature data were collected from temperature recording devices mounted on the head rope of the bottom trawl gear, deployed during annual bottom trawl surveys sampling ~350 stations over the south-eastern Bering Sea shelf. The bottom trawl station grid extended from 54.6°N to 61°N latitude, and covered the outer and middle shelf domains; only middle shelf temperatures were used (G. Walters, NOAA/National Marine Fisheries Service, personal communication). Copepod abundances were compared among years and among months within years using analysis of variance (ANOVA) on natural log-transformed data, and multiple comparisons were made using a Bonferroni adjustment. Spearman’s rank correlation test was used to analyse relationships among the mean abundance of C. marshallae, date of first appearance of C1, and environmental variables. Results from the present study were compared with data from spring 1980–1981 (Smith and Vidal, 1986). Diverse climate, sea ice and phytoplankton bloom conditions were observed during 1995–1999 in the south-eastern Bering Sea (Table I). Mean summer bottom temperatures ranged from 3.4 to 0.8°C during 1996, 1998, 1997, 1995 and 1999 (ordered from warmest to coldest). Ice was advected over the middle shelf mooring as early as February (1996, 1998, 1999) and as late as mid-March (1995 mid-March (1997) (Figure 2). Ice cover remained at the mooring for as little as 1 week in 1996, and as long as 1 month in 1995 (Figure 2). In years with ice cover after February (1995, 1997 and 1999), fluorescence increased relatively early, just after the ice arrived (Figures 2 and 3A). In 1996 and 1998, when ice retreated early, fluorescence remained low until May, when the water column to warm and (Figures 2 and 3A). Smith and Vidal made observations during a cool year with ice present times at the mooring and in which was warm and year [Figure (Smith and Vidal, of adult female C. marshallae among years (Table In April and May, concentrations of females were high in 1995 compared with years In May, concentrations in 1999 were than in of females were similar between April and May in years 1999, when the of females in May the two years for which February samples were the moult from to adult before long before the spring bloom. In February of 1998 and 1999, and of copepodites sampled were The were were over a long well before the spring phytoplankton bloom in 1999, when fluorescence to in late and at females were present in February in 1998 and in 1999), April in years 1996 when were and females were in May of years (Table calculated using Campbell’s equation for the copepodite stages sampled in April and May ranged from as early as to the of March (Figure for nauplii were collected only during April 1996, when Calanus nauplii were present in concentrations of ± 3 well before the of the phytoplankton bloom in production was copepodites were during a relatively time (Figure and two copepodite stages were present during a sampling period. The 1997, relatively among in but were The of copepodite cohorts through stages over spring and summer was in 1998, when temporal sampling coverage was most The timing of copepodite recruitment with the spring phytoplankton bloom (Figure and Table An was 1998, when were in April, before water column fluorescence increased in The first appearance of C1 as early as March and as late as the of May (1996). Spring copepodite stage development was relatively following cold conditions early in the year (1995, compared with warm or The stages in May 1995 and 1998 were and In May of 1996, 1997 and 1999, C1 was the Copepodite and the timing of between years (Figure In 1995 and 1998, when ice extended of abundances were high in copepodites were present during in 1996, but high concentrations of copepodites were sampled in 1996, recruitment after our May samples were In 1997 and 1999, abundances remained low the abundances of copepodites not high abundance (Figure Mean April ± ± and May ± concentrations of females in 1995 were higher than in the years May concentrations of copepodites in 1995 ± were not higher than in 1998 ± when there were high of copepodites were with high abundances of females in but relatively females a large in combined with data from 1980–1981 (Smith and Vidal, 1986), as mean shelf bottom temperature was to the Pacific Decadal the of the spring phytoplankton bloom in years of cold bottom the timing of copepodite recruitment was in years of cold bottom ice extent and early bloom the of copepodite abundance was in years of most sea ice extent (Table sampling spanned the full range of sea ice and temperature conditions that in the south-eastern Bering Sea over the past 34 years (Wyllie-Echeverria and Wooster, 1998). or Bering Sea ice extent by and has remained while spring temperatures over the Bering Sea increased by between and et al., 1997). The from to 1999 is considered a of the Pacific Decadal this of warm however, 1995 and 1999 were the years in comparable with conditions before the while 1996 and 1998 were in 1998 were influenced by climate forcing of a temporal the El Niño–Southern Oscillation et al., 2001). The timing and of C. marshallae production on the south-eastern Bering Sea shelf was among the years of our study environmental The first copepodites were as early as March and as late as the of May and there was a range in spring copepodite abundances among data was relatively in its and temporal coverage and of It does not for of the between climate and C. marshallae population and relationships may that were not from only 5 years of data, the variability of the these clear production over a long well before the spring spring copepodite recruitment over a with the spring the between of females and copepodites was the timing of the spring bloom and copepodite recruitment was in cold and the spring abundances of C. marshallae in years of sea The moult to adult female before and well before the spring in the 2 years for which February data. In the C. marshallae and moult to maturity the same time as in the Bering Sea 1998). the the moult occurs months before during a when surface water is the the shelf 1998). In the Bering C. marshallae is thought to overwinter on the shelf, for the of and the as length and are in these two and do not the that in populations copepodites at the same the timing of the moult is C. marshallae has a in the Bering Sea shelf and In production is with high phytoplankton concentrations with and there are each 1998). In the Bering the spring was before the bloom in most In only one is by the Bering Sea two were in a year (Smith and Vidal, 1986). is to from the data for the and Bering Sea In the samples were collected relatively in a while in the was in a where is considered to be low and is been observed for C. which reproduces during the spring phytoplankton bloom in but before the bloom in and et al., et al., 1998; et al., on our that Bering Sea C. marshallae may life history with populations of the same in the than it does for example, C. a shelf population that of the bloom and over a long duration and et al., 1998). This study of the and marshallae to the of the C. finmarchicus is our production well before the spring bloom and over a long at females oocytes in February 1998, and in February In production in February of females in the were reproductively and an of (unpublished Calanus marshallae nauplii were present in April 1996, a month before the bloom The copepodite stages that were most in April and May were well before the spring phytoplankton bloom in most to our these data that reproduction early and copepodites before the spring bloom the at in 1998, and probably in years of high copepodite that the of annual recruitment sampling of the same in late May and June when bloom and nauplii to the copepodite of the and of early, production is to example, et that production by Sea C. finmarchicus was to production during and after the bloom et al., the low of females was by the large of with observations of female C. marshallae as C. and Calanus nauplii in the south-eastern Bering Sea in (Naumenko, We calculated that production as early as and that a large of females were and in that female C. marshallae in the south-eastern Bering Sea be and before the bloom. of copepodites until the spring that production be important for Bering marshallae production is important to Bering Sea C. marshallae, then of In production before the bloom has been to be by as and ice Tourangeau and Runge, and and Runge, Runge and et al., 1998; et al., determined at in the marshallae production at phytoplankton concentrations above but were at concentrations 60 We little the food in the Bering under the A estimate of prey before the bloom is et al., nauplii (Napp et al., these are then at female growth may be by phytoplankton that are for or by (e.g. and 1998). are a of to early and over a long period. that C. marshallae for low production by over a long duration before the spring bloom may be the of a been to in the laboratory et al., et al., et al., in situ relationships between and copepod et al., 2002). A that only affect copepod and that the of be by of food 2002). It is not prey items C. marshallae during the spring bloom over the Bering Sea shelf. and two that copepod in laboratory dominate the stages of the bloom over the Bering Sea shelf, but be during a bloom in the of Alaska and et al., 1996). may early that C. finmarchicus in the sea early to to by In the south-eastern Bering reproduction may be to that with conditions in an subject to great interannual variability 2000). We that at one of the conditions is that recruitment to C1 with the of the spring phytoplankton bloom. A recruitment may occur at the naupliar or early through one of a of example, C. finmarchicus in the Sea well before the but net population growth does not occur until because of high attributed to cannibalism by and copepodites before phytoplankton prey were and 2001). Calanus in the Sea in or but nauplii do not the first until food concentrations during the spring bloom and 1998). of food on C. finmarchicus on were most during late naupliar and early copepodite stages (Campbell et al., et that C. finmarchicus in the Sea from March to but copepodites do not the stages before the spring and then a clear et al., 2001). that this may be the of or food of copepodite stages compared with nauplii and late stage Bering Sea C. marshallae, the recruitment to be the first copepodite were over a long February until at copepodites were during a relatively there were two copepodite stages at spring sampling Copepodite recruitment was with the spring phytoplankton Smith and (Smith and Vidal, that the life of C. marshallae is so that most growth occurs during the bloom. An was 1998, a cold spring with a February ice when early copepodites were in April, before fluorescence increased in is a for phytoplankton biomass and an of zooplankton and 1998 be an of a in which fluorescence was an a understanding of the recruitment to the and of the of and during this period. In of the 5 years that however, the in fluorescence to with conditions to C1 A recruitment at the C1 with the between and copepodite that than may be most important in the timing and of C. marshallae copepodite the of copepodite recruitment with the spring there be in the food or nauplii and to a between these the has to be for Bering Sea C. marshallae populations and 2001)]. The of a between climate and copepod population dynamics has previously been for example, higher abundances of C. finmarchicus are correlated with low sea surface temperatures in the of et al., and a in zooplankton biomass is with increased temperature et al., 1998). fluctuations in the north-east well with the and 1998). the for these relationships are and and 1998). linkages between climate and are to be in the Bering Sea and the of but the to be (Brodeur and and et al., 2000; Napp et al., 2002). Sea ice extent was the of spring copepodite abundance in our data does not an for this but ice cover affect copepod recruitment through its on food or Ice cover affect the bloom which were not to Ice may be an important of food for Runge et that in C. increased at the of a bloom of ice and that ice and from the and in the water were a of for spring copepod production in et al., 1991). The of sea ice may in on example, as walleye pollock (Theragra to avoid the where C. marshallae when sea ice is present (Wyllie-Echeverria and Wooster, 1998). a annual variability in the with copepod population C. marshallae over a long well before the spring phytoplankton bloom. Spring copepodites are over a relatively with the bloom. timing and of copepodite recruitment are among The abundance of is not to but is in years with the most ice This study how longer-term through on sea ice and may the dynamics of the Bering Sea zooplankton of stations sampled by year and month of stations sampled by year and month development times for C. marshallae on the middle shelf development times for C. marshallae on the middle shelf Mean and gonadal maturity of C. marshallae females over the Bering Sea middle shelf Mean and gonadal maturity of C. marshallae females over the Bering Sea middle shelf of C. marshallae population timing and abundance of May copepodites to environmental correlation of C. marshallae population timing and abundance of May copepodites to environmental correlation South-eastern Bering Sea middle shelf, of mooring and zooplankton sampling and months sampled are in Table 1995 samples were collected from from the and 1999 from and of ice cover from at mooring Development of C. marshallae copepodite populations in to the spring phytoplankton bloom and water column during shows temperature data from mooring. Ice was present during of coldest temperature data. fluorescence at m to for each year from et al., composition and estimated development time of copepodites sampled in April and to C1 and stages was using Campbell’s temperature-dependent growth to stages sampled in April and May estimated time at stage and C1 for that example, C2 sampled in April 1995 were in late and to C1 in early April. of the spring phytoplankton bloom is by Mean abundance and stage composition of copepodites between February and data We the and of the NOAA ship ‘Miller for field and and for sampling We are to Stabeno and and mooring and Salo, and climate data. and two made on the The Polish Sorting counted and This was in by through Bering Sea and is of Fisheries