Growth and development of Neocalanus flemingeri/plumchrus in the northern Gulf of Alaska: validation of the artificial-cohort method in cold waters

Abstract

In situ growth and development of Neocalanus flemingeri/plumchrus stage C1–C4 copepodites were estimated by both the artificial-cohort and the single-stage incubation methods in March, April and May of 2001–2005 at 5–6°C. Results from these two methods were comparable and consistent. In the field, C1–C4 stage durations ranged from 7 to >100 days, dependent on temperature and chlorophyll a (Chl a) concentration. Average stage durations were 12.4–14.1 days, yielding an average of 56 days to reach C5, but under optimal conditions stage durations were closer to 10 days, shortening the time to reach C5 (from C1) to 46 days. Generally, growth rates decreased with increasing stage, ranging from 0.28 day⁻¹ to close to zero but were typically between 0.20 and 0.05 day⁻¹, averaging 0.110 ± 0.006 day⁻¹ (mean ± SE) for single-stage and 0.107 ± 0.005 day⁻¹ (mean ± SE) for artificial-cohort methods. Growth was well described by equations of Michaelis–Menten form, with maximum growth rates (G_max) of 0.17–0.18 day⁻¹ and half saturation Chl a concentrations (K_chl) of 0.45–0.46 mg m⁻³ for combined C1–3, while G_max dropped to 0.08–0.09 day⁻¹ but K_chl remained at 0.38–0.93 mg m⁻³ for C4. In this study, in situ growth of N. flemingeri/plumchrus was frequently food limited to some degree, particularly during March. A comparison with global models of copepod growth rates suggests that these models still require considerable refinement. We suggest that the artificial-cohort method is the most practical approach to generating the multispecies data required to address these deficiencies.

Received June 16, 2005; accepted in principle August 25, 2005; accepted for publication November 14, 2005; published online November 25, 2005
 Communicating editor: R.P. Harris

INTRODUCTION

As grazers and nutrient recyclers, copepods play an important role in marine ecosystems linking primary production to upper trophic levels and accounting for up to 80% of the metazoan biomass in the marine environment (Kiørboe, 1998). Over the last few decades, copepod productivity has become a major focus of research necessitating precise measurement of copepod rate processes to fully understand the trophodynamics of marine ecosystems (Longhurst, 1984). Traditionally, egg production has been considered an easy and quick proxy for growth rate of all developmental stages, based on the assumption that adult copepods cease somatic growth and that equivalent growth continues as egg production (Sekiguchi et al., 1980; Berggreen et al., 1988; Runge and Roff, 2000). Increasingly, this assumption has been challenged (Peterson et al., 1991; Hutchings et al., 1995; Hopcroft and Roff, 1998; Richardson and Verheye, 1998; Calbet et al., 2000; Hirst and McKinnon, 2001). At present, our knowledge of somatic growth is incomplete, despite recent attempts at synthesis (Hopcroft et al., 1998; Hirst and Lampitt, 1998; Hirst and Bunker, 2003). This lack of knowledge is greatest for cold waters and particularly for species in the subarctic North Pacific.

Within the subarctic North Pacific, Neocalanus spp. are a major component of the seasonal zooplankton cycle (Miller, 1993; Mackas and Tsuda, 1999; Coyle and Pinchuk, 2003, 2005). Their abundance and large size make Neocalanus spp. an important prey species for many higher trophic levels (e.g. Kawamura, 1982; Willette et al., 1999; Moku et al., 2000; Hunter et al., 2002) especially for the productive salmon fisheries in the Gulf of Alaska (Weingartner et al., 2002; Armstrong et al., 2005; Cross et al., 2005). Each of the three species, Neocalanus plumchrus, Neocalanus flemingeri and Neocalanus cristatus, exhibits extensive ontogenetic vertical migrations and diapause in deep water over the winter. During early spring, they grow and develop in the surface layer while feeding on phytoplankton and microzooplankton (Dagg, 1993; Gifford, 1993). Their reproductive maturation and egg production take place in the deeper layers (>250 m) without feeding (Miller et al., 1984; Miller and Clemons, 1988). Neocalanus flemingeri and N. plumchrus are more abundant and common within the upper mixed layer (Miller, 1988; Coyle and Pinchuk, 2005) than N. cristatus and clearly divide the surface mixing layer seasonally (Mackas et al., 1993). Temporally, the newly recruited copepodites of N. flemingeri are observed in the surface mixed layer earlier in spring than those of N. plumchrus (Tsuda et al., 1999). Spatially, N. flemingeri is more common within inner-shelf waters on the northern Gulf of Alaska shelf than the other two species (Coyle and Pinchuk, 2005).

While we have an overall picture of the life cycle of N. flemingeri and N. plumchrus from numerous studies in the subarctic Pacific, few direct estimates of their vital rates exist. We are limited to the single-stage method-determined rates of C4 (duration) and C5 (growth) copepodites of N. plumchrus (Miller and Nielsen, 1988); natural cohort growth rates of C1–C5 copepodites of N. cristatus and N. plumchrus/flemingeri (Vidal and Smith, 1986); laboratory rates for C2–C4 copepodites of N. flemingeri (Slater and Hopcroft, in review) and three studies of egg production and naupliar development of N. cristatus, N. flemingeri and/or N. plumchrus (Fulton, 1973; Saito and Tsuda, 2000; Slater and Hopcroft, in review). Despite their importance, in situ somatic growth rates for younger copepodites of N. flemingeri are unavailable.

In the coastal Gulf of Alaska, it is generally difficult to track natural cohorts due to the region’s highly advective nature and the high sampling frequency required for robust estimates (Miller and Clemons, 1988; Miller and Nielsen, 1988; Miller, 1993). Alternatively, ‘single-stage’ and ‘artificial-cohorts’ incubations can be utilized; the former is labor intensive in the field, while the latter is most time-consuming post-cruise. In this study, we evaluate the utility of these two methods for a large species in a subpolar environment. This study, together with a concurrent laboratory study of egg production, development and growth rates of N. flemingeri (Slater and Hopcroft, in review), provides a fuller understanding of the role of N. flemingeri/plumchrus in the ecosystem of the northern Gulf of Alaska.

METHOD

The study area in the northern Gulf of Alaska has been sampled as part of the U.S. Northeast Pacific GLOBEC program (Weingartner et al., 2002). The region is characterized by a shelf of 100- to 300-m depth, with complex bathymetry and many deep-water coastal fjords and embayments (Fig. 1). In each of 2001, 2002 and 2003, six cruises were conducted in March, April, May, June/July, August and October, while in 2004 the April and August cruises were not undertaken. In 2005, only a single cruise occurred in May. Experimental work was carried out at four stations along the Seward line from inshore to just past the shelf break (i.e. GAK1, 4, 9, and 13), plus one station in the western inner passage of Prince William Sound (PWS—either KIP2 or nearby PWS2) where the depth is 500–800 m (Fig. 1). Water samples for assessment of ambient phytoplankton concentration at these stations were collected at multiple depths with 5-L Niskin bottles on a CTD rosette, serially size fractioned using 20-µm Poretics, 5-µm Nuclepore and GF/F filters, with frozen samples later analysed fluorometrically using techniques for chlorophyll a (Chl a) (Stockwell and Whitledge, unpublished data).

Fig. 1.

Map of sampling locations in the northern Gulf of Alaska study region.

Seawater for incubations at each station was collected by replicate CTD casts with a 12-place rosette of 10-L Niskin bottles equipped with 9-mm valves to facilitate draining. Collections were typically made within the upper mixed layer, usually from 5- to 20-m depth, but at inshore (GAK1) and PWS stations, the depths for seawater collection were occasionally greater to avoid salinities of <30 caused by melting snow and glaciers. Incubation seawater was prescreened through 100-µm Nitex placed over the ends of Tygon tubing while siphoning the bottles into 20-L soft-walled carboys. Once filled, carboys were stored in a large insulated fish tub (∼1 m³ capacity) rigged as flow-through incubators. The insulated lids of the incubators were fitted with numerous 8-cm plexiglass windows, and the lighting was reduced to ∼20% of ambient surface illumination. Food concentrations of incubation seawater at the beginning and the end of experiments were measured as size-fractionated Chl a using the same protocols and fluorometric techniques employed for monitoring activities.

At each of the experimental stations, copepods were collected using a 64-µm plankton net with 4-L cod-end hauled slowly from the surface to 50 m and back to the surface (∼20 m³ of water filtered) between 08:00 and 12:00 hours. Immediately upon retrieval, copepod collections were diluted using the prescreened seawater and placed into an incubator at ambient surface-water temperatures. Soon after, copepods were sorted into ‘artificial cohorts’ (Kimmerer and McKinnon, 1987; Peterson et al., 1991; Hopcroft and Roff, 1998; Hopcroft et al., 1998) by sequential passage through submerged screens of the following mesh sizes: 1800, 1300, 1000, 800, 600, 500, 400, 300, 200, 150 and 100 µm. The sample was constantly diluted with prescreened water at ambient seawater temperature, and as each cohort size class was created, it was placed into an incubator at ambient seawater temperature. Under ideal conditions, creating the cohort took 1 h and as much as 3 h when chains of large filamentous algae were abundant.

Prior to incubation, each size fraction was gently homogenized and evenly divided. One-half was concentrated and preserved in 5% buffered seawater formalin as the time zero sample (T-0) and the other half equally divided among several of the 20-L carboys previously filled with prescreened seawater. The number of carboys employed varied depending on the biomass of copepods being added. The labeled carboys were put back into the on-deck incubators and maintained at surface-water temperatures by running seawater. The temperature inside the incubators was recorded by Onset Tidbit loggers. Ship movement provided constant jostling and ‘mixing’ of the carboys. After 5 days, the carboys were screened through 45-µm mesh, copepods pooled by the original size fractions and preserved immediately in 5% buffered seawater formalin as the final sample (T-5). All preserved material was stained with Rose Bengal.

Concurrent experiments were carried out for N. flemingeri/plumchrus at the same stations by picking active and undamaged stage C1–C4 copepodites from additional 64-µm plankton net collections. At least 60, and up to 300, of each stage present were picked and incubated under the identical conditions as the artificial-cohort experiments. Ideally, the two species would have been separated for experimentation, but this proved impractical at sea for these early stages. Nonetheless, our impression is that samples were predominantly N. flemingeri, particularly for the earlier months and more inshore stations. For C5 copepodites, we were unable to observe molting or growth due to the long duration of lipid accumulation prior to diapause. To explore the effect of food enhancement on growth rate, additional single-stage experiments were carried out onboard in March and April of 2003 to which Isochrysis sp. (cell length 5–6 µm) and Pavlova lutheri (cell length 6–10 µm) were added.

In the laboratory, preserved copepods were identified and staged (Miller, 1988; Kobari and Ikeda, 2001), prosome lengths were digitally measured (Roff and Hopcroft, 1986), and the progression of the cohorts was determined by changes in the stage and body size. Separation of early copepodites of N. flemingeri and N. plumchrus also proved problematic in the laboratory, so the two species were grouped. Development time was calculated as 1/MR, where MR is the observed molt rate. Weights were predicted based on a prosome length (PL) to dry weight (DW) relationship: log₁₀[DW] = 3.56 × log₁₀[PL] – 2.32, where PL is in mm and DW is in mg (Slater and Hopcroft, in review). The instantaneous weight-specific growth rates (day⁻¹) within a given cohort, over the incubation time t (days), were computed from the equation g = (lnW_t – lnW₀)t⁻¹ (Runge and Roff, 2000). Recent concerns over growth rate errors using the molt rate method (Hirst et al., 2005) do not apply in this study, because we employed incubation periods not development time to estimate our growth rates.

The relative effects of initial copepodite stages, incubation temperatures and total Chl a concentrations on the growth rates were estimated by backward stepwise regression analysis (SAS system V8). We explored the explanatory power of Chl a measured both within our experiments and in the upper 30 m at the time of collection. We also explored the effect of log₁₀ transformation on growth rate and Chl a concentration, as this has been shown to linearize such data (Hirst and Bunker, 2003). For other analyses, we used the regression features within Sigmaplot (V8). When necessary, rates were standardized to 5°C using a Q10 of 2.70 for food-saturated broadcast-spawning copepods (Hirst and Bunker, 2003), which agrees with a previous estimate of 2.78 (Kleppel et al., 1996). To convert dry weight to carbon content (µg), we used a conversion factor of 0.44, the arithmetic mean of carbon content for N. flemingeri in the Gulf of Alaska (Miller, 1993).

RESULTS

Temperature and food resource

Neocalanus flemingeri/plumchrus stage C1–C4 copepodites were only present during March–May cruises. The onboard incubation temperatures were generally 5–6°C, similar to the in situ temperatures within the upper mixed layer. The notably higher temperatures in May of 2003 were due to a combination of a warmer year and the relatively late timing of that cruise (i.e. late May versus early May) (Fig. 2).

Fig. 2.

Chlorophyll a (Chl a) concentration at study sites averaged over the upper 30 m (bars) and incubation temperatures of experiments at sea in the northern Gulf of Alaska 2001–2005 (circles). Error bars are SD.

Average Chl a concentrations within the upper 30 m in the study area reflected the progression of the spring bloom. Generally, the lowest Chl a concentrations occurred in March, increased in April and peaked in May. Spatially, the highest Chl a occurred in PWS during April, in advance of the stations along the Seward Line (Fig. 2). The seasonal Chl a concentrations (mean ± SD) over four consecutive years across all sampling stations were 0.46 ± 0.19 in March, 2.45 ± 3.71 in April and 2.46 ± 2.02 in May. The spatial distribution of Chl a concentrations in this study area over 4 years typically declined offshore, with corresponding values (mean ± SD) of 3.12 ± 4.04 at PWS, 1.51 ± 1.63 at GAK1, 1.77 ± 2.56 at GAK4, 1.20 ± 1.34 at GAK9 and 1.05 ± 1.36 at GAK13. The monthly partitioning of size-fractionated Chl a concentration within the prescreened seawater during the sampling seasons was variable (data not shown). Generally, the larger particle Chl a (>20 µm) averaged 44.5% of the total, the 5- to 20-µm fraction 27.8% and the remainder (∼0.5–5µm) 27.7%.

Developmental time and growth rate

The development and growth of N. flemingeri/plumchrus C1–C4 copepodites estimated during five consecutive years by the artificial-cohort and single-stage methods showed similar patterns, although the artificial-cohort method appeared to produce more variable results (Fig. 3).

Fig. 3.

Stage durations (upper panels) and growth rates (lower panels) of Neocalanus flemingeri/plumchrus in the northern Gulf of Alaska 2001–2005. For the single-stage method (bars), initial stages are integers, while for the artificial-cohort method (circles) stage is the average of the population at time zero.

The single-stage method showed the most consistent patterns; only the first two copepodite stages were common in March, the first four copepodite stages were present, in April and only copepodite stages C3 and C4 (plus the non-incubated stage C5) were common in May. In March, the stage durations were >30 days, and growth rates were slow (∼0.05 day⁻¹), with the exception of PWS where growth and development were much faster, and late March 2004 which is more consistent with April observations. In April, the first three copepodite stages shared comparable stage durations of ∼10 days, with growth rates ranging from 0.10 to 0.15 day⁻¹, but spent ∼30 days at C4, with a relatively low growth rate of 0.04 day⁻¹. Overall, growth rates in April appeared to slow down with increasing stage. Similarly, the growth rates declined from C3 to C4 in May, with the C3 growth rate ∼0.16 day⁻¹ and C4 rate of ∼0.10 day⁻¹, both higher than those observed in April. These patterns become somewhat clearer when averaged by month across years (Fig. 4): within each month, stage duration increases with stage and growth rate declines with stage; rates in March are notably slower than other months, while rates in April and May are more similar. Overall growth rates for the first four copepodite stages ranged from 0.01 to 0.22, with a mean ± SE of 0.11 ± 0.01 day⁻¹.

Fig. 4.

Monthly mean stage duration and growth rate versus Neocalanus flemingeri/plumchrus copepodite stage estimated by artificial-cohort and single-stage methods in the northern Gulf of Alaska 2001–2005. Values plotted against initial stage and offset to improve interpretation. Error bars are SE.

The artificial-cohort method produced more variable results, but the pattern was still comparable and consistent with the single-stage method (Fig. 3). In March, only early copepodites of N. flemingeri/plumchrus occurred (in mesh sizes ranging from 400 to 800 µm). In April, C1–C4 stages occurred (in mesh sizes from 400 to 1300 µm). In May, early copepodites were virtually absent (and the mesh sizes ranged from 600 to 1800 µm). In March, the mean stage durations were ∼20–35 days and mean growth rates ∼0.05–0.07 day⁻¹. During April, copepodites experienced high growth (∼0.13 day⁻¹) and short stage duration (<15 days) with decreasing growth as the animals became larger and older. In May, growth rates remained high, but not significantly different from those in April, with a trend of decreasing growth with increased stage (Fig. 4). Overall, growth rates for the first four copepodite stages by artificial-cohort method were from 0.01 to 0.28, with a mean ± SE of 0.11 ± 0.01 day⁻¹.

Food-enhancement experiment

Food addition resulted in a modest enhancement of the growth rate for N. flemingeri/plumchrus (Fig. 5); the increases were significant (two-tail paired t-test, P < 0.001; Wilcoxon signed-rank test, α = 0.001). The average growth rate for C1 was increased by about 16% by a 25-fold increase in Chl a (to 4.89 mg m⁻³), for C2 an increase of ∼15% by a 20-fold increase of Chl a (to 4.27 mg m⁻³) and for C3 an increase of ∼8% by 10-fold increase of food concentration (to 3.25 mg m⁻³). There was only one paired experiment for C4, with a 28% increased growth rate from a 52-fold increase of Chl a (to 4.17 mg m⁻³). In general, copepods in food-enhanced experiments looked healthier than those without food addition.

Fig. 5.

Effect of food enhancement on growth rate of Neocalanus flemingeri/plumchrus in the northern Gulf of Alaska. Error bars are SE.

Body size and growth rate

Consistent with the relationships between growth rate and stage, growth rate on average declined with increased body size (Fig. 6; Table I). Simultaneously, the variability in growth rate declined with body size using both methods. Interestingly, when examined on a per stage basis, for both C1 and C2, there was a positive relationship (Table I, r² = 0.19–0.76) between growth rate and body size, regardless of the method. The relationships were also positive for C3 and C4 by the single-stage method (Fig. 6; Table I), while patterns were somewhat contradictory for the artificial-cohort methods. The poorer relationship by the artificial-cohort method arises because multiple stages exist in most initials; yet, we force this fractional average initial stage to the nearest whole-numbered stage in this analysis, thus blurring the underline patterns. Not surprisingly, combining the two methods generally resulted in a reduction of the variability explained, because weights at initial stage were not perfectly comparable between the methods. The existence of these positive relationships within a stage indicates that when growth rate is fastest, individuals within the stage tend to be larger.

Fig. 6.

Relationship between growth rate of Neocalanus flemingeri/plumchrus and the body weight (µg C individual⁻¹) within early copepodite stages estimated by artificial-cohort and single-stage methods in the northern Gulf of Alaska.

Table I:

Analysis of growth rate (Gr day⁻¹) versus body weight (BW μg C individual⁻¹) grouped by Neocalanus flemingeri/plumchrus copepodite stages for the northern Gulf of Alaska over 4 years

Group	Copepodite stage	Number of data	Corrected temperature	Gr = a + b × logBW			r2	P
				Intercept (a)	Slope (b)
Artificial-cohort	C1	21	5	0.101	0.328	0.503	<0.0005
	C2	34	5	0.068	0.108	0.194	0.009
	C3	37	5	0.125	−0.001	0.002	0.806
	C4	28	5	0.189	−0.069	0.252	0.006
	All	111	5	0.104	−0.011	0.022	0.124
Single-stage	C1	19	5	0.159	0.581	0.76	<0.0001
	C2	20	5	−0.099	0.409	0.66	<0.0001
	C3	15	5	−0.099	0.165	0.37	0.017
	C4	21	5	−0.133	0.102	0.51	0.0002
	All	76	5	0.107	−0.017	0.057	0.038
Combined	C1	40	5	0.121	0.381	0.504	<0.0001
	C2	54	5	0.031	0.174	0.288	<0.0001
	C3	52	5	0.097	0.010	0.002	0.765
	C4	46	5	0.064	−0.002	0.0002	0.923
	All	187	5	0.105	−0.014	0.035	0.010

Statistical analysis of growth rate

Developmental stage, temperature and Chl a were significant explanatory variables for the single-stage method (r² = 0.26, P < 0.0001), while for the artificial-cohort method (r² = 0.13, P = 0.0011) only stage and Chl a were significant (Table II). Combining both types of experiments resulted in slightly poorer fit (r² = 0.17, P < 0.0001). Untransformed data showed the same patterns and explained a similar degree of variation.

Table II:

Backward multiple regression analysis of the weight-specific growth rate regressed on the initial stage (Stg), incubation temperature (T °C) and chlorophyll a (Chl a) concentration (Chl μg L⁻¹) of Neocalanus flemingeri/plumchrus

Group	Dependent	N	\(\mathrm{Log}_{\mathrm{10}}g\mathrm{\ =\ }a_{1}Stg\mathrm{\ }+\mathrm{\ }a_{2}T\mathrm{\ }+\mathrm{\ }a_{3}\mathrm{log}_{10}Chl\mathrm{\ }+\mathrm{\ }a_{4}\)							r2 (P)
			a1 (P)	a2 (P)	a3 (P)	a4 (P)
Artificial-cohort	Stg, T, Chl	111			0.2903 (<0.0001)	−1.0818 (<0.0001)	0.1358 (<0.0001)
Single-stage	Stg, T, Chl	76	−0.1469 (0.0005)	0.0919 (0.0315)	0.3425 (0.0005)	−1.2894 (<0.0001)	0.2432 (0.0002)
Combined	Stg, T, Chl	187	−0.0820 (0.0049)	0.0440 (0.0910)	0.3388 (<0.0001)	−1.1468 (<0.0001)	0.1587 (<0.0001)

After removing the influence of temperature through Q10 standardization, N. flemingeri/plumchrus copepodite growth rates estimated by both the artificial-cohort and the single-stage methods showed significant Michaelis–Menten relationships to Chl a concentration (Fig. 7). Although we had begun this fitting exercise by stage, because we had shown it to be a significant variable in the previous analysis, curves for C1–C3 were very similar with high r², so we combined those stages. For the C1–3 group by both methods, 32–34% of variance in growth rate was significantly explainable by Chl a (P < 0.0001), with the maximum growth rates (G_max) of 0.17–0.18 day⁻¹ and half saturation Chl a concentrations (K_chl) of 0.45–0.46 mg m⁻³. At C4, Chl a accounted for 19% of variance in growth rate using the artificial-cohort method and 30% of variance in growth rate using the single-stage method, with G_max dropping to 0.08 day⁻¹ and K_chl remaining at 0.38–0.66 mg m⁻³. It is notable that both methodologies effectively predict the same underlying relationships to ambient Chl a concentration.

Fig. 7.

Functional relationship between temperature-corrected growth rates estimated by the artificial-cohort and single-stage methods and chlorophyll a (Chl a) for Neocalanus flemingeri/plumchrus in the northern Gulf of Alaska. Michaelis–Menten curves fitted for C1–C3 (solid line) and for C4 (dashed line).

DISCUSSION

This study provides the first comprehensive estimates of in situ developmental time and somatic growth rates for N. flemingeri/plumchrus. Overall, these rates are comparable with previous estimates. It is notable that the two methods employed in this study are comparable and reveal the same underlying patterns in growth and development rates.

Developmental time

Development times are in reasonable agreement with the few estimates for this and sibling species determined by incubation or through following natural cohorts in the field (Table III). On the basis of this study, the developmental time of N. flemingeri/plumchrus from the start of C1 to C5 averages ∼56 days. This is identical to a 56-day estimate for N. plumchrus in the Alaska gyre (Miller, 1993). Under ideal conditions, average development times for C1–C3 are ∼10 days each and for C4 ∼10–14 days for a total of 40–45 days, which is more comparable with the 46-day estimate in the southern Bering Sea (Vidal and Smith, 1986) and <45 days in the Oyashio region (Tsuda et al., 1999). It is notable that the only previous direct measurements of development rate (i.e. by incubation) for any Neocalanus spp. were the 21.3-day estimate for C4 N. plumchrus (Miller and Nielsen, 1988). All these estimates suggest that the recent indirect estimates of development time at colder temperatures deduced using Calanus spp. as an analogue (Saito and Tsuda, 2000) may be too fast.

Table III:

Growth rates corrected to 5°C and development times (in parentheses) for Neocalanus species in the subarctic Pacific

Species	Temperature (°C)	Growth rate and developmental time											Location	Source
		C1	C2	C3	C4	C5	Average
Neocalanus flemingeri/plumchrus	5.0–6.0	0.117 (14.1)	0.122 (13.1)	0.109 (12.4)	0.072 (13.5)		0.105 (13.3)	Gulf of Alaska	This study
N. flemingeri	5.0		0.14 (11)	0.12 (12)	0.07 (15)		0.11 (13)	Gulf of Alaska	Slater and Hopcroft (in review)
N. flemingeri						0.10		Gulf of Alaska	Miller and Nielsen (1988)
N. plumchrus				(24)	0.05 (21.3 or 25)	0.15		Gulf of Alaska	Miller and Nielsen (1988)
N. plumchrus				(13.4)				Gulf of Alaska	Miller (1993)
N. plumchrus	0.5–6.0	0.143 (8)	0.141 (10)	0.133 (12)	0.111 (16)	0.039	0.113 (11.5)	Southeastern Bering Sea	Vidal and Smith (1986)
Neocalanus cristatus	0.5–6.0		0.075 (17)	0.072 (20)	0.066 (23)	0.045	0.065 (20)	Southeastern Bering Sea	Vidal and Smith (1986)

In this region, the entire developmental time of N. flemingeri/plumchrus from egg hatching to C5 is 119–123 days, estimated by adding the stage durations of copepodite at C1–C4 from this study with the laboratory-estimated 63–67 days for naupliar stages at 5°C (Slater and Hopcroft, in review) assuming that laboratory measurements reflect the field values. Again, this value is in reasonable agreement with other estimates, e.g. 4 months in the western subarctic Pacific (Tsuda et al., 1999) and 3.5–4 months in the central Gulf of Alaska (Miller and Clemons, 1988). Despite some minor differences in water temperature, life-cycle timing is very similar between these three regions, with stage C1 appearing in late February to early March and stage C5 predominating by May.

Growth rate

Growth rates of N. flemingeri/plumchrus in this study are comparable with other Neocalanus spp. in the region (Table III). A comparison of available in situ and incubation estimated growth rates of Neocalanus spp. in the subarctic Pacific reveals an overall trend of decreasing growth with increasing developmental stages (Table III). This pattern for copepodite stages at C1–C4 compares well with our data ranging from 0.072 to 0.12 day⁻¹ in this study and the growth rates of 0.11–0.14 day⁻¹ for N. plumchrus in the southeastern Bering Sea (Vidal and Smith, 1986), although our average stage-specific rates for N. flemingeri/plumchrus are slower. The in situ growth rates at Station P in the Alaska gyre for N. plumchrus at C4 were 0.05 (Miller and Nielsen, 1988), consistent with our field-estimated growth rate of 0.072 day⁻¹ for that stage of N. flemingeri/plumchrus. Growth rate of C5 may increase to 0.10–0.15 day⁻¹ (Miller and Nielsen, 1988) or continue to decline (Vidal and Smith, 1986); hence it remains unclear what the growth rate of this stage may be. It is notable that the only previous direct measurements of growth rate (i.e. by incubation) for any Neocalanus spp. were those for C5 obtained by measuring weight increase within that stage (Miller and Nielsen, 1988). We were unable to estimate growth of this lipid-accumulating stage by our techniques due to its long stage duration.

Environmental variables and growth rate

Food and temperature are the important environmental variables for copepod growth (Mauchline, 1998; Hirst and Bunker, 2003). In aquatic ecosystems, Chl a has long been considered as a general index of food concentration, although for some species it may be a poor predictor of growth and fecundity (Hirst and Bunker, 2003; Bunker and Hirst, 2004). In this study, total Chl a appears to be a reasonable food index for N. flemingeri/plumchrus, as indicated by its relationship to growth rate. Neocalanus spp. have been shown to act primarily as suspension feeders on microsized (>20 µm) particles but capable of utilizing food down to 5 µm (Gifford, 1993; Kobari et al., 2003; Liu et al., 2005; Dagg et al., submitted for publication). At times when microzooplankton predominates, this component may form a larger proportion of the diet than phytoplankton (Gifford, 1993; Kobari et al., 2003), and this no doubt contributes to some of the scatter in our growth relationships with chlorophyll.

Growth of adult and juvenile copepods can often be food limited (Kimmerer and McKinnon, 1987; Peterson et al., 1991; Hopcroft and Roff, 1998). Food limitation appears to become more severe with increasing temperature (Hirst and Lampitt, 1998; Richardson and Verheye, 1998); thus food limitation is pervasive in warm waters and less common in cold waters. Globally, food limitation can affect adult growth more substantially than juvenile growth (Hirst and Bunker, 2003). In this study, the growth of N. flemingeri/plumchrus also appeared to be food limited. The predicted food saturation (G_max) growth rate for C1–C3 was 0.17–0.18 day⁻¹. While some rates exceeded these, many field values were below this saturated rate in March (prior to significant increase in phytoplankton biomass) and for later stages in May when the chlorophyll bloom remained variable in its timing. Food limitation is also suggested by our experimental addition of food, although the impact was relatively limited over the 5-day duration employed in this study. Interestingly, although food addition resulted in elevated growth rates, values remained well below G_max, although the Chl a concentrations were much higher than the K_m. This implies that either there is considerable ‘inertia’ in somatic growth rates within N. flemingeri/plumchrus (that requires time for a response to be realized) or the food offered (Isochrysis sp. and P. lutheri) were too small to be captured efficiently (Richardson and Verheye, 1999). Finally, food-limited growth in the field is also supported by comparing field values with laboratory-determined values (Table III), and it is likely to underlie the variability observed between stations and years in this study.

Comparison to global models

Growth is a key component in understanding the role of copepods in material flow and transformation in the sea; however, estimating in situ somatic growth rates is time consuming and effort intensive. In the past decade, attempts have been made to make predictions from a few easily measurable parameters such as temperature (Huntley and Lopez, 1992), temperature and body weight (Hirst and Sheader, 1997; Hirst and Lampitt, 1998) or temperature, body weight and food concentration (Hirst and Bunker, 2003). Predictive global models are necessary to make the estimation of growth, and hence secondary production over large spatial and temporal scales is feasible. There is, therefore, a constant need to test and refine such models by comparing them with new rates, such as those directly measured in this study, and exploring the utility of such models in specific study areas.

The simplest of these global models (Huntley and Lopez, 1992) considers only temperature, which has long been known to influence biological rates. It is clear that this model fails to capture the underlying form of growth rate observed in this study (Fig. 8). A more complex model, incorporating both temperature and body size (Hirst and Lampitt, 1998), more adequately reflects the patterns we observed (Fig. 8). The most complex models, which incorporate temperature, body size and chlorophyll concentration (Hirst and Bunker, 2003), come closer to our observations (Fig. 9), at least for their models of adult broadcasters and all their data combined. Nonetheless, the potential errors associated with these models can be large (Table IV). All the predictions by the models of Huntley and Lopez, Hirst and Lampitt and the two models of Hirst and Bunker (adult broadcasters and all data) noticeably underestimate by 10% up to 60%. In contrast, the model of Hirst and Bunker for juvenile broadcasters seriously overestimates by 3- to 8-fold the observations in this study (Table IV).

Fig. 8.

Comparisons of temperature-corrected Neocalanus flemingeri/plumchrus growth rates estimated by the artificial-cohort and single-stage methods in this study to those predicted at 5°C by models. Dashed line, Huntley and Lopez (Huntley and Lopez, 1992); solid line, Hirst and Lampitt (Hirst and Lampitt, 1998) equation for all data (adults and juveniles of both broadcast and sac-spawners).

Fig. 9.

Comparisons of temperature-corrected Neocalanus flemingeri/plumchrus growth rates in this study with those predicted by the Hirst and Bunker (Hirst and Bunker, 2003) model at 5°C (colored surface). Left column, artificial-cohort method; right column, single-stage method. (A) For juveniles broadcasters; (B) for adult broadcasters; (C) for all data combined.

Table IV:

Comparison of growth rates predicted by the global models at 5°C with Neocalanus flemingeri/plumchrus in situ growth rates corrected to 5°C

Data source	Percent of predicted to rates of this study
	C1	C2	C3	C4	C1–4
Huntley and Lopez (1992)	84a	66	71	117	79
Hirst and Lampitt (1998)	106a	54	36	37	56
Hirst and Bunker (2003)b	719	367	245	249	380
Hirst and Bunker (2003)c	54	40	47	64	49
Hirst and Bunker (2003)d	75	47	42	51	52

a

Non-significant at α = 0.05, two sample t test assuming unequal variance.

b

For juvenile broadcasters.

c

For adult broadcasters.

d

For all data.

The lack of predictive power of these equations is not surprising, considering that all these empirical models had few data on species living in cold waters around 5°C, especially for juvenile somatic growth, and almost no data from the subarctic Pacific. It is interesting to find that the predictions by the simple temperature-dependent model of Huntley and Lopez (Huntley and Lopez, 1992) are closer to the direct estimates (Table IV), in part because their relationship tends to represent maximum temperature-dependent rates. At the same time, this simple model fails to capture the body-size pattern present in our data. The Huntley–Lopez model may be even less satisfactory in oligotrophic waters (Calbet and Agusti, 1999), where it results in overestimation, because growth is seldom maximal due to food limitation.

The more recent models of Hirst and Lampitt (Hirst and Lampitt, 1998) are clearly improvements over the Huntley and Lopez model and have already been employed in a number of studies (Roman et al., 2000, 2002; Coyle and Pinchuk, 2003), but to our knowledge they have only been tested in a few cases (Richardson et al., 2001; Peterson et al., 2002; Rey-Rassat et al., 2004) where success has been variable. These equations have already been superseded by a more sophisticated effort (Hirst and Bunker, 2003). Both of these models generally indicate a negative relationship with body size, which agrees with the overall pattern for all data in our study. What is interesting is the positive relationship between growth and body size within a stage shown by our data. Although such a pattern can be demonstrated in the laboratory (Vidal, 1980a,b), it is not normally detectable in the field. Such a pattern arises because when populations are growing rapidly, size at stage is large because the animals have been well fed, while populations that have not been well fed grow slowly and hence are smaller at stage. The loss of clarity in these relationships with increasing stage probably arises because later stages are more likely to have experienced a mixture of favorable and unfavorable growth conditions during their lives, and hence they are more variable in size. These within-stage relationships are simply lost within the high variability present in the data sets used to construct global models; yet, in the case of Neocalanus, they are sufficiently strong that they can be used to predict growth from simple knowledge of size at stage.

Clearly, there is still a need to refine these models, particularly in polar and subpolar environments. This requires the continued generation of data on growth rates from a wide variety of species. Initially, the artificial-cohort method was developed to address the difficulty of sorting by stage, continuously reproducing copepod populations of smaller species (Kimmerer and McKinnon, 1987). The application of the artificial-cohort technique, and its modifications, has been successful in warmer waters (Kimmerer and McKinnon, 1987; Peterson et al., 1991; Hopcroft et al., 1998; Campbell et al., 2001; McKinnon and Duggan, 2003), but this study represents the first validation of the technique in colder waters. Although the technique yields more variability than the single-stage method, it ultimately reveals the same underlying relationships with food resources. The single-stage method, while arguably superior, is only practical for larger species where stages may be more readily separated. Furthermore, a significant attraction of the artificial-cohort method is that it can be routinely performed at sea, whereas the single-stage method requires working conditions suitable for live sorting. Last, the artificial-cohort method provides data simultaneously on all the dominant species present in a collection, while it is difficult to prepare single stages of more than one species concurrently. Thus, the artificial-cohort method appears to be the most practical method for local estimation of copepod community production and the refinement of more global models.

ACKNOWLEDGEMENTS

We thank the captain and crew of the R/V Alpha Helix, as well as Amanda Byrd, Mike Foy and Alexei Pinchuk for assistance in experimental setup and execution. Alexei Pinchuk also provided invaluable assistance by terminating experiments still running post-cruise. Cheryl Clarke provided significant laboratory support. Terry Whitledge kindly provided ambient chlorophyll a concentrations from the GLOBEC LTOP program. This is contribution number 260 of the US GLOBEC program, jointly funded by the National Science Foundation and the National Oceanic and Atmospheric Administration under NSF Grant OCE-0105236.