A new test for density-dependent survival: the case of coastal cod populations

Ecology, June, 1999 by Ottar N. Bjornstad, Jean-Marc Fromentin, Nils Chr. Stenseth, Jakob Gjosaeter

INTRODUCTION

Fish stocks are known to fluctuate extensively over large spatial and temporal scales (e.g., Laevastu 1993, Cushing 1995). The early stages in the life cycle are believed to be critical in determining year class strength (May 1974, Cushing 1995). Survival has been related to abiotic factors, such as changes in temperature, salinity, wind field, and currents (e.g., Cushing 1982, 1995); human exploitation (Hutchings 1996, Myers et al. 1996, Cook et al. 1997); and biological processes, such as predation, competition, and cannibalism (Caley et al. 1996). Sundby et al. (1989) reported evidence suggesting that juvenile mortality in the northeast Atlantic cod is density dependent. Myers and Cadigan (1993) concluded that most of the marine demersal fish populations are regulated through density dependence that takes place during the juveniles stages.

There is, however, no consensus on the functional form of density dependence in fish (see, for example, Ricker 1954, Shepherd and Cushing 1990, Myers and Cadigan 1993). Indeed, uncertainty about functional forms for biotic interactions (such as density dependence and functional responses) is common in many areas of population ecology (see, for example, Abrams 1982, Hanski 1991, Boutin 1995, Pascual et al. 1997). In the following we develop a method that allows the estimation of density-dependent survival without making a priori assumptions about functional forms. From this, we develop a formal test of density dependence. When the test is applied to synthetic data, it is found to have good power and to be relatively resistant to measurement errors. We apply the method to census data of cod (Gadus morhua) populations along the Norwegian Skagerrak coast.

THE LIFE CYCLE OF THE NORWEGIAN SKAGERRAK GOD

Fromentin et al. (1997) demonstrated significant periodic fluctuations in time series of the Norwegian Skagerrak cod population. The period was [approximately]2-1/2 yr, and the authors hypothesized that the periodicity might be related to interactions between the two main juvenile cohorts, leading to density-dependent mortality. We investigate the long-term data sampled in two different areas along the Norwegian Skagerrak coast in order to test for density-dependent survival. Firstly, we will provide a summary of the population ecology of the cod.

The life cycle of cod (Gadus morhua L.) can be divided into four main stages [ILLUSTRATION FOR FIGURE 1 OMITTED]: eggs ([approximately]2 wk), larvae ([approximately]3 mo), juveniles (2 yr), and adults (2 yr). Eggs are buoyant and hatch near the surface 1 or 2 wk after the spawning, which usually occurs in early March along the Norwegian Skagerrak coast. The larvae stay in the water column and feed on zooplankton, mainly copepod nauplii and copepodites. The fish larvae metamorphose into juvenile fishes around May-June in this area. The younger stage of these juveniles constitutes the O-group, which lives on their nursery ground but settles and feeds on the bottom when they are [approximately]3-5 cm long. The egg and larval stages are variable among years because of starvation and expatriation (Skreslet 1989; see also Cushing 1995) and changes in temperature, salinity, wind field, and currents (Koslow and Tompson 1987, Ellersten et al. 1989, Dickson and Brander 1993, Ottersen and Sundby 1995). Juveniles grow at the bottom for [approximately]2 yr after which maturation occurs. The maturation time of 2-3 yr in the Norwegian Skagerrak cod is short compared to the North Sea and the Arcto-Norwegian cod (Gjosaeter et al. 1996). Tagging experiments have indicated that the Norwegian Skagerrak cod is relatively isolated, with limited migration (individuals appear fjord specific) and limited interchange with individuals of other nearby areas (such as the open sea population from the Skagerrak [Danielssen 1969]). The young juveniles (O-group), the older juveniles (1 -group; 1 - 1/2 year old), and the adults are, thus, considered to be sympatric or parapatric. The adults are, however, generally found in deeper water than the two juvenile cohorts (Gjosaeter 1990, Gjosaeter et al. 1996; see also Dalley and Anderson [1997]).

THE DATA

The time series on cod have been assembled as part of the Flodevigen survey in which fish communities at numerous fixed locations along the Norwegian Skagerrak coast have been censused every September/October since 1919. The sampling has been carried out with beach seines of standardized design since the onset of the study (for more details, see Tveite [1971], Johannessen and Sollie [1994]). The abundances of the two main juvenile cohorts of cod (the O-group and the 1-group) are well represented in the catch. The O-group individuals in a given year correspond to the 1-group cohort in the succeeding year [ILLUSTRATION FOR FIGURE 1 OMITTED]. Because the sampling started a bit later than 1919 in some stations in the southwestern (SW) area [ILLUSTRATION FOR FIGURE 2 OMITTED], we study the census data of juvenile cod sampled from 1921 to 1994 (barring the five war years 1940-1944) at 38 fixed stations. A second set of 37 stations in the northeastern (NE) area were sampled from 1936 to 1994 (excluding the war years; [ILLUSTRATION FOR FIGURE 2 OMITTED]).