WROBLEWSKI ET AL.: SURVIVAL OF NORTHERN ANCHOVY LARVAE 



match between the interstorm duration and the 

 critical development period of the larvae, i.e., 

 the development to a stage vi^here starvation is 

 no longer a major factor. 



Survival of northern anchovy larvae in the sea 

 depends on the right type (50-100 |jl in size and 

 high nutritional quality) of prey being present in 

 abundance at first feeding (O'Connell and 

 Raymond 1970). Lasker (1975) maintained that 

 the dinoflagellate Gymnodinium splendens is 

 small enough to be ingested, has sufficient nutri- 

 tional qualities, and is abundant enough in sub- 

 surface layers to sustain first-feeding anchovy 

 larvae. Lasker regarded microzooplankton as 

 not abundant enough to contribute significantly 

 to larval survival during the first week of life. 



Let us assume for the moment that G. splen- 

 dens is the preferred prey of first-feeding an- 

 chovy larvae. Let us also assume that this dino- 

 flagellate species is present in the phytoplankton 

 community after a wind event (e.g., as observed 

 by Mullin et al. 1985). The concentration of this 

 motile organism after a wind event depends on 

 its ability to aggregate in the face of turbulent 

 mixing in the water column. 



Wroblewski (1984) showed that the vertical 

 diffusivity which just balances the dinoflagel- 

 lates' ability to aggregate is given by 



K, = H W, 



where H is the vertical scale of the dinoflagellate 

 layer and Wg is the swimming speed of the dino- 

 flagellate. Cullen and Horrigan (1981) found that 

 if nitrate is available in the upper water column 

 (as after a strong storm), G. splendens will mi- 

 grate to the surface into a layer several meters 

 thick. If the wind event is weak and nitrate is not 

 available in the surface layer, the dinoflagellates 

 will aggregate near the nitracline in a subsurface 

 layer. Assuming H is, b m and W^ is 1 m h~' 

 (Kiefer and Lasker 1975; Cullen 1985), G. splen- 

 dens should be able to aggregate against a diffu- 

 sivity of 14 X lO^^m^s-^ 



In our model, K,, quickly decays to a back- 

 ground diffusivity of 1 x 10""* m^ s ' 

 after cessation of wind forcing (Fig. 2b). Solar 

 heating adds buoyancy at the surface and the 

 turbulent kinetic energy in the mixed layer dissi- 

 pates through friction. Thus, first-feeding north- 

 em anchovy larvae should be able to subsist on 

 layers of G. splendens within a day after a 

 storm. The short period of starvation when G. 

 splendens are dispersed during the storm should 

 not result in a significant increase in larvae mor- 



tality, unless the storm continues for several 

 days. A few days after successful first feeding, 

 the developing anchovy larvae must include 

 microzooplankton in their diet. Our assumption 

 that anchovy larvae feed on microzooplankton is 

 appropriate when simulating larvae mortality 

 over a 15 d period from first feeding. 



The concentration of prey required for sur- 

 vival of anchovy larvae is controversial. The 

 laboratory feeding experiment of O'Connell and 

 Raymond (1970) indicated a prey concentration 

 > 800 copepod nauplii (~^ were required for 

 maximal survival. More recent laboratory exper- 

 iments (Houde and Schekter 1981; Munk and 

 Kiorboe 1985) suggest the minimum concentra- 

 tion of prey may be as low as 200 copepod nauplii 

 €"\ but of course the larvae would be ex- 

 pected to grow at a much slower rate (O'Connell 

 and Raymond 1970). 



Even concentrations of 200 nauplii €~^ are 

 high compared to estimates of average densities 

 of microzooplankton in the sea (Beers and 

 Steward 1967). However, as pointed out by 

 Owen (1989), maximal concentration of prey 

 rather than average concentration (estimated 

 from integrating net hauls or pump samples) is 

 the relevant quantity. 



The highest zooplankton concentration in the 

 model (at 3 m depth in Figure la) is about 4 ixg 

 atom N €"^ of which 25% or 1 n,g atom N 

 (~^ is larval anchovy food (e.g., copepod 

 nauplii). This is equivalent to about 14 \i.g N 

 £~^ of copepod nauplii. The nitrogen content 

 of Paracalanus stage I nauplii is about 5 ng 

 (Checkley 1980) and of Calanus is about 20 ng 

 (Comer et al. 1965; Mullin and Brooks 1967). 

 Thus, 14 fig N (~^ of copepod nauplii is 

 equivalent to 2,800 Paracalanus stage I nauplii 

 ("^ or 700 Calanus stage I nauplii f \ 

 These concentrations correspond roughly to the 

 prey densities used experimentally by O'Connell 

 and Raymond 1970), but are two orders of mag- 

 nitude higher than the median numbers m"^ 

 observed in pump samples from the sea taken by 

 Mullin et al. (1985) before and after a storm. 



The overestimation of zooplankton biomass in 

 the model is due to the omission of vertical loss 

 processes such as sinking of phytoplankton and 

 zooplankton fecal pellets which reduce the total 

 nitrogen available to the plankton ecosystem in 

 the upper water column (Walsh 1983; Checkley 

 1985). However, we would get the same re- 

 sponse in the mortality of larval anchovy pre- 

 sented above, if we altered the model to simulate 

 more realistic zooplankton concentrations, while 



395 



