est 1 X 10"'' g. The gonad index equals the (gonad 

 weighty body weight) x 100. 



Results 



Although tides in the Rhode River subestuary ex- 

 hibited an approximately semidiurnal rhythm, fluc- 

 tuations in measured high tide level were not corre- 

 lated with predicted high tide level (product moment 

 correlation of log transformed data for daily predict- 

 ed and measured high tide, r = 0.111; Student's 

 ^test for r i^ 0, P > 0.2; Fig. 1). Moreover, time 

 series spectral analysis (Thrall and Engelman 1981) 

 showed that measured tides exhibited only a very 

 weak peak in spectral density at a frequency of about 

 12.5 d rather than the strong peak of 7.5 d exhibited 

 by predicted tidal cycles. Cross correlation between 

 predicted and measured tides in bivariate spectral 

 analysis showed low and variable coherence {P > 

 0.05). Similarly, linear association of the two vari- 

 ables was weak, with variable phase shifts and vari- 

 able coefficients necessary to fit one variable to the 

 other. In addition, the occurrence of tides sufficiently 

 high to flood the marsh fringing the creek (i.e., tides 

 > 46 cm) did not occur more frequently at night than 

 during day (x^ test, P > 0.2), nor more frequently 

 during any particular lunar phase (x^ test, P > 0.4). 

 High tides did not occur consistently during any 2-h 

 period of the 24-h day at the quarters of the new or 

 full moons (x^ test, P > 0.4). Thus, water levels in the 

 estuary were neither predictable in amplitude within 

 the tidal range nor related in any obvious cyclical 

 pattern to the lunar cycle. 



However, both female and male F. heteroclitus 

 showed distinct semimonthly cycles in readiness to 

 spawn from May through August (Fig. 1). Females 

 and males also exhibited a distinct semimonthly cycle 

 of gonad size, although the cycle of male gonad index 

 was not as pronounced as that of females, due to the 

 small changes in size of the testes during spawning 

 (Fig. 1). The cycles of spawning readiness in males 

 and females were usually in phase with the lunar 

 cycle. Comparisons of the observed frequencies of 

 spawning readiness with frequencies predicted by 

 cycles with perfect semilunar periods and the same 

 amplitudes showed no significant differences in 27 of 

 37 d of observation for females and 33 of 37 d of 

 observation for males (x^ test, P < 0.05). Observed 

 frequencies of spawning readiness over the entire 

 study period were not significantly different from 

 frequencies predicted by the perfect semilunar cycles 

 for either females or males (x^ test, P > 0.2). 

 Although six out of six peaks of spawning readiness 

 for males occurred during the 7 d surrounding new 



or full moons, for females only four of the six peaks 

 occurred during the first or last quarters of the 

 moon, indicating that for the small numbers of peaks 

 occurring during a reproductive season, readiness to 

 spawn does not coincide with new and full moons 

 (Fisher's Exact Test for number of spawning peaks 

 occurring within the specified period, P = 0.227; 

 Fig. 1). However, if the timing of the peaks in readi- 

 ness to spawn is considered to lag 3.5 d after the new 

 and full moons, then all peaks for males and females 

 occurred within the 7 d surrounding the lagged 

 period, indicating a significant synchronous semi- 

 lunar cycle (Fisher's Exact Test on number of peaks 

 occurring within the specified period, P < 0.01). 

 Female spawning was correlated with male spawn- 

 ing (product moment correlation of arcsine trans- 

 formed spawning frequencies, r = 0.695; Student's 

 ^test for r ¥= 0, P < 0.001). However, neither female 

 nor male readiness to spawn was correlated with the 

 measured tides (product moment correlation of arc- 

 sine transformed frequencies of spawning readiness 

 with log transformed high tide measurements, r = 

 0.184 for females and r = 0.272 for males; Student's 

 ^tests for r # 0, P > 0.2). Thus the semilunar cycles 

 of spawning readiness of both sexes appeared to be 

 synchronized, but not to be related to the tidal 

 regime of the estuary. 



The reproductive season of F. heteroclitus in the 

 Rhode River-Muddy Creek estuarine system oc- 

 curred from late April to September, when water 

 temperatures were above 17° C (pers. obs.). Mean 

 gonad indices of female and male samples declined 

 during the season, both at the peak and at the spent 

 phases of the semimonthly cycles (Female Gonad In- 

 dex = -0.124 Day + 14.1 and Male Gonad Index = 

 - 0.035 Day + 3.83; Student's ^test for slopes ¥=0,P 

 < 0.05; (Fig. 1)). The decline of mean gonad index 

 during the season reflected two statistics. First, an 

 increasing percentage of the population failed to in- 

 itiate gonad recrudescence during successive cycles. 



Figure 1.- Lunar phase, tidal levels, and reproductive cycles of 

 Fundulus heteroclitus from May to August 1982 in a tidal creek on 

 central Chesapeake Bay. Full moons (open circles) and new moons 

 (solid circles) are indicated. Daily high water levels above mean low 

 water are shown for measurements at the study site and for tidal 

 levels predicted by National Oceanographic and Atmospheric Ad- 

 ministration (1981). Horizontal lines indicate water level which 

 floods the marsh adjacent to the Creek. The percentages of female 

 and male fish which were ready to spawn are shown for each sample 

 throughout the spawning season (solid lines), along with 

 hypothetical spawning cycles with the same amplitudes and perfect 

 semilunar periods in phase with the new and full moons (dashed 

 lines). Gonad indices (mean ± SE) for females and males are also 

 plotted for each sample. 



468 



