THORROLD: LIFE HISTORY PARAMETERS IN HERKLOTSICHTHYS CASTELNAUI 



growth increments on the sagitta provided the 

 closest estimate of age in larval H. castelnaui. 



Age at initial increment deposition was not deter- 

 mined in this study. Initial growth increments have 

 been shown to be deposited prior to egg hatching, 

 at hatching, just after hatching, and at onset of 

 exogenous feeding (Brothers et al. 1976; McGurk 

 1984; Kingsford and Milicich 1987). All temperate 

 clupeids studied have initiated ring formation at 

 yolk-sac absorption (Geffen 1982; Lough et al. 1982; 

 McGurk 1984; Re 1984) from 3 to 5 days after hatch- 

 ing. Although no work has been published on oto- 

 lith formation in tropical clupeids, a comparatively 

 high water temperature, and hence a rapid devel- 

 opmental rate, suggests that endogenous reserves 

 would be quickly exhausted (Houde 1974). It was 

 thus assumed here that the first otolith increment 

 is laid down at hatching, and otolith counts were 

 assumed to be a direct measure of age. Violation of 

 this assumption will have led to biased estimates of 

 Lq, the size at hatching, and ^o. the specific growth 

 rate at hatching, in the Laird-Gompertz model. The 

 magnitude of absolute and specific growth rates re- 

 main valid. The age at which the growth rates were 

 calculated will, however, have a systematic error 

 corresponding to the time from hatching to initial 

 increment formation. 



Standard length increased as a logarithmic func- 

 tion of otolith radius. Linear (e.g., Riceetal. 1985), 

 logarithmic (Nishimura and Yamada 1984; Tsuji and 

 Aoyama 1982), and some combination of the two 

 functions (Jenkins 1987) have been reported in the 

 literature. A close correlation between standard 

 length and otolith growth at a daily level implies that 

 the width of any growth increment is a measure of 

 instantaneous growth (Campana and Neilson 1985). 

 The smoothly monotonic relationship between stan- 

 dard length and otolith radius presented here sug- 

 gests that it may be valid to reconstruct individual 

 growth histories by examination of growth incre- 

 ment spacings in sagittae of H. castelnaui. 



Both Laird-Gompertz and von Bertalanffy growth 

 curves adequately fitted the length-at-age data. The 

 growth trajectory of larval H. castelnaui indicates 

 that growth is rapid for the first two to three weeks, 

 but slows after this period. Growth may become 

 asymptotic after this point, as predicted by the 

 single cycle Laird-Gompertz model, or alternative- 

 ly, enter a new growth stanza during juvenile life, 

 as has been reported for Herklotsichthys quadri- 

 maculatus (Williams and Clarke 1983). 



Data presented here for larval H. castelnaui af- 

 fords good comparison with some temperate clupeid 

 species, where growth rates have also been elu- 



cidated using the otolith increment technique. Ini- 

 tial growth rates of 0.5-0.6 mm/d in H. castelnaui 

 are as high as any recorded for clupeid larvae in the 

 field. Similar growth estimates have been reported 

 off South West Africa, where Sardinops ocellatus 

 larvae grow linearly at rates of approximately 0.7 

 mm/d (Thomas 1986). Growth estimates of 0.2-0.4 

 mm/d after this initial burst are closer to those 

 presented for Clupea harengus from the northern 

 Atlantic (Townsend and Graham 1981; Lough et al. 

 1982; Henderson et al. 1984). Growth rates of lar- 

 val H. castelnaui may reflect higher ambient water 

 temperatures, as both S. ocellatus and C. harengus 

 have a higher L„ and hence higher predicted 

 growth rates (Ricker 1975). 



Spawning periodicity in H. castelnaui was ap- 

 parently correlated with the quarter moon phases. 

 Lunar-synchronized spawning has been reported in 

 salmoniform, atheriniform, tetraodontiform, and 

 perciform fishes (Taylor 1984). Most fish species 

 with lunar-spawning rhythms spawn on or around 

 the new or full moon (e.g., Lobel 1978; Middaugh 

 et al. 1984), although spawning in French grunts, 

 Haemulonjlavolineatum, also appears to be coupled 

 with quarter moons (MacFarland et al. 1985). It 

 should be noted that results presented here may be 

 subject to some systematic error in ageing. If, for 

 example, initial increment formation occurs some 

 time after hatching, then birth dates will have been 

 consistently underestimated. MacFarland et al. 

 (1985) hypothesized that currents favorable for set- 

 tlement may account for fertilization and recruit- 

 ment events peaking on the quarter moon. My 

 results suggest that spawning occurs with some 

 semilunar periodicity, but the time of initial incre- 

 ment formation needs to be determined before 

 relating spawning events to moon phases and possi- 

 ble tidal influences on egg and larval distributions. 



The most significant advantage of using otolith 

 ageing techniques is the ability to produce individual 

 rather than population statistics. Although it has 

 been possible to fit a growth equation to the length- 

 age data presented here, there is also an amount 

 of variability surrounding the curve. This variabil- 

 ity may, at least in part, be a sampling artifact 

 caused by methodological problems. Inaccurate age 

 determinations may be caused by nondaily deposi- 

 tion of rings under some conditions (e.g., Geffen 

 1982), or failure to detect all rings within an otolith 

 due to the resolution problems of light microscopy 

 (Campana et al. 1987). Conversely, if the data are 

 accurate, variable growth rates on small spatial (tens 

 of meters) and temporal (days) scales are detectable 

 by otolith analysis. It is often tacitly assumed in lar- 



81 



