Curtis: Validation of a method for estimating annual fecundity for Hippocampus guttulatus 



329 



excluded from the analyses because eggs and small 

 embryos were in poor condition due to freezing and were 

 therefore difficult to count. For the same reason, it was 

 not possible to estimate clutch sizes by examining the 

 ovaries of captured adult females. Data were recorded 

 from a total of 1264 adult males and 1211 adult females 

 captured with the fishing gears (Erzini et al.-). 



Underwater visual censuses of tagged H. guttulatus 

 were also carried out on a 10 m x 10 m focal study 

 grid established in June 2001. The grid was located 5 

 m from the intertidal zone next to a permanent pier 

 located within the Ria Formosa Natural Park at a 

 depth of 1.3 to 6.1 m (additional details provided in 

 Curtis, 2006; Curtis and Vincent, 2006). Tagged sea- 

 horses exhibited high site fidelity to small home ranges 

 averaging 20 m- during multiple years (Curtis and 

 Vincent, 2006). The juveniles released by eight tagged 

 males on the grid were also counted and used to es- 

 timate brood size as in Vincent and Giles (2003). The 

 eight brooding males were placed in cages (0.3 m high 

 X 0.45 m wide, 0.5 mm mesh) in situ for a maximum of 

 48 h each between 18 July and 16 August 2002 (Curtis 

 and Vincent, 2006). The mesh allowed water and small 

 zooplankton to flow through the cages, while retaining 

 newborn juveniles. After juveniles were born, the cages 

 were slowly brought to the surface and emptied into 

 large tanks filled with seawater where the total num- 

 ber of juveniles was counted. Males and their juveniles 

 were released at their original capture site within 12 

 hours of birth. 



Three models for estimating annual fecundity 



The assumptions of rapid remating and continuous 

 reproduction during the reproductive season were exam- 

 ined by using three progressively realistic models to 

 estimate realized annual fecundity: the continuous 

 reproduction (CR) model, the intermittent reproduction 

 (IR) model, and the intermittent and seasonal reproduc- 

 tion (ISR) model. 



Continuous reproduction (CR) model 



In the simplest model, individuals were assumed to 

 reproduce continuously from the beginning to the end of 

 the reproductive season, from March to October (Bois- 

 seau, 1967; Reina-Hervas, 1989). Annual fecundity, /j,,., 

 was estimated by using mean estimates of the brooding 

 period, f^, and brood size, f^; 



on the grid using underwater visual census. In 2001 

 and 2002, 553 H. guttulatus (264 males, 254 females, 

 35 juveniles) were individually tagged with visible 

 implanted fluorescent elastomer (VIE tags, Northwest 

 Marine Technologies, Inc., Shaw Island, WA) (Curtis, 

 2006; Curtis and Vincent, 2006). A total of 163 under- 

 water visual censuses (2-5 person-hours/dive) were 

 carried out with SCUBA on the grid during two census 

 periods: from 17 July to 26 October 2001 and from 23 

 May to 12 September 2002. On average, one underwater 

 visual census was carried out every 1 to 2 days during 

 each of these census periods. During underwater visual 

 censuses, observers swam ~1 m above the substrate and 

 searched for tagged seahorses. The date, tag, life history 

 stage, sex, size (trunk length, converted to standard 

 length by using equations developed for H. guttulatus, 

 Curtis and Vincent 2005, 2006), and reproductive state 

 of all tagged seahorses were noted. The reproductive 

 state of adult males (full or empty brood pouch) was 

 determined by visual (and in ambiguous cases, manual) 

 inspection. 



Changes in reproductive state were plotted as a func- 

 tion of census date for each tagged male. Because of 

 high seahorse density on the grid ( l.S/m^; Curtis and 

 Vincent, 2006), it was not feasible to verify the status 

 of all individuals during each dive; therefore, f^ was 

 estimated by recording the minimum and maximum 

 possible durations of each brooding period, as graphi- 

 cally depicted in Figure 1. Unmated male H. guttulatus 

 occasionally mimic males that are close to releasing 

 juveniles by temporarily and completely filling their 

 pouches with water during reproductive displays to 

 females (J. Curtis, personal observ. ). Reproductive in- 

 teractions between male and female H. guttulatus gen- 

 erally occurred within two hours of sunrise and lasted 

 on average 4.8 min ±3.3 SD (standard deviation) (J. 

 Curtis, unpubl. data). Therefore putative brooding pe- 

 riods were included in the analysis only when a male 

 was observed a3 times with a full pouch during a pe- 

 riod bracketed by observations of an empty pouch. The 

 minimum and maximum possible brooding periods were 

 averaged within and among all males. 



Intermittent reproduction (IR) model 



This model expanded the continuous reproduction model 

 by incorporating a mean estimate of the interbrood inter- 

 val, <,j, which was assumed to be constant throughout 

 the reproductive season: 



U,,=f,>^tjt, 



(1) 



h.yr = fb-h I (h + t„), 



(2) 



where f/,,,^ 



the total number of young produced per 



male per year; and 



the duration of the reproductive season. 



where f,^ = the time elapsed between release of young 

 and deposition of a new clutch of eggs into 

 the male's brood pouch. 



In the CR model, f^ and tf^ were assumed to be constant 

 throughout t^, and the interbrood interval, f,,,, was 0. 

 The brooding period, f^, was estimated by monitoring 

 changes in the reproductive state of tagged adult males 



The interbrood interval was estimated by using a simi- 

 lar approach to that employed to estimate the brooding 

 period (Fig. 1): the maximum and minimum possible 

 durations of each interbrood interval were recorded and 



