332 



Fishery Bulletin 105(3) 



eggs for mating. Although there was no significant dif- 

 ference between the brooding period, fj, of males and 

 female interclutch interval, t^^ (21.06 6.6 days, «=102 

 intervals from 26 females; median = 20 days) (^-test, 

 df=180, ^=-0.36, P=0.72), females produced signifi- 

 cantly more clutches than males brooded during both 

 underwater visual census periods (2001: df=37, <=3.32, 

 P=0.002; 2002: df=62, t=l.\&, P<0.0001, Table 2). As- 

 suming that the duration of the reproductive season is 

 the same for both sexes and that the IR model is cor- 

 rect, the estimate of the interclutch interval indicates 

 that the average female prepared 11.63 (±3.65) clutches 

 per year. Temperature was not a significant predictor 

 of f,^, (H-'= 0.02, Fmo,|=1.66, P=0.2). 



Intermittent and seasonal reproduction (ISR) model 



The reproductive activity of males varied significantly 

 within years (/■2 = 0.78, P[2 i9, = 39.28, P<0.0001, Fig. 

 3A). Hippocampus guttulatus males reproduced during 

 most of the year and peaked in reproductive activity 

 from June through August. No reproductively active 

 males were captured from December 2000 to March 

 2001, or in February 2002. Male reproductive activity 

 was also significantly and positively correlated with 

 water temperature (r-' = 0.55, P|i,i2] = 13.32, P=0.004), 

 which varied seasonally from 10- 28''C (mean annual 

 temperature=18.2°C). There was no effect of lunar phase 

 on male reproductive activity (r2 = 0.01, F^^ 5oi = 0-26, 

 P=0.78), as inferred by Boisseau (1967). The fraction of 

 mature females that were preparing eggs also varied 

 seasonally (r-' = 0.21, Fjj i7j=3.59, P=0.05, Fig. 3B). Tem- 

 perature was weakly correlated with female reproductive 



activity (r- = 0.30, F^^ p2|=4-80, P=0.05), but lunar phase 

 was not (/^ = 0.04, ^,2 g"o|=1.15, P=0.33). 



Incorporation of seasonal variation in reproductive 

 activity into the ISR model resulted in estimates of 

 Sj, ,,^ and /'^ ,,^ that were 63.1% smaller than those pre- 

 dicted by the CR model. The ISR model predicted that 

 the total time an average male H. guttulatus spent 

 brooding embryos per year in the Ria Formosa was 

 90.4 days, giving an estimate of male annual spawn- 

 ing frequency, s^ ^,^, of 4.22 broods per year (Table 1). 

 Similarly, the total time females spent preparing eggs 

 per year was 13.51 days, giving an estimate of female 

 annual spawning frequency, s^. ,,,., equal to 5.23 clutches 

 per year. These estimates of annual spawning frequency 

 were at least four times greater than estimates from 

 the Arcachon Basin population. During the reproductive 

 season, which lasted 3.5 months (Fig. 4), the average 

 male had a full pouch for 20.7 days, indicating that, on 

 average, males in the Arcachon Basin population bred 

 only once a year (Table 1). 



Model validation 



The number of broods predicted by the CR model for the 

 2001 and 2002 census periods, s^^.^,,, was significantly 

 greater than observed directly with underwater visual 

 census (Table 2), indicating that this model produces 

 upwardly biased estimates of spawning frequency for 

 H. guttulatus. By contrast, the more realistic ISR model 

 produced expected values of s^^.^,„ and the total number 

 of days female spent with hydrated eggs, ^,. ,.p„, which 

 were within 95% confidence intervals for the observed 

 mean values on the grid in both 2001 and 2002. This 



