FISHERY BULLETIN: VOL. 83, NO. 4 



the early 1970's. In 1971 and 1972, field technicians 

 collected samples predominantly from females with 

 the "adult" or fused color pattern (Perrin 1970). 

 Beginning in 1973, field technicians were instructed 

 to collect samples nonselectively with respect to size 

 and sex. Operationally, this meant working-up speci- 

 mens in the order in which they appeared on the 

 deck of the tuna vessel. Sampling methods and 

 laboratory procedures are described in detail by Per- 

 rin et al. (1976). Sample locations are shown in 

 Figure 1. 



Reproductive tracts of mature and nearly mature 

 females were preserved in the field for laboratory 

 examination. In 1971 and 1972, the definition of 

 "mature and nearly mature" was not explicit. In 

 1973, "mature and nearly mature" was defined as 

 individuals with "mottled" or "fused" developmen- 

 tal color phases (Perrin 1970). Because females in 

 the younger "speckled" color phase occasionally were 

 found to be pregnant, "mature and nearly mature" 

 was redefined operationally (beginning in 1974) as 

 specimens >150 cm total length (TL, measured from 

 tip of rostrum to fluke notch). Laboratory examina- 

 tion of preserved ovaries was used to determine the 

 presence of corpora from past ovulations. Pregnancy 

 was determined by visual examination of the uterus 

 (in later years, fetuses >30 cm TL were removed and 

 measured in the field). Mammary glands were slit 

 and checked in the field for the presence of milk. 



In addition to the above life history information, 

 field technicians collected data pertaining to condi- 

 tions under which the samples were taken. Informa- 

 tion used in this report includes the observer's 

 estimate of the size of the school from which the 

 sample was taken, the duration of the chase before 

 the net was set, the number of dolphins known to 

 be killed during fishing operations, and the 

 geographic location at which the sample was taken. 



METHODS 



Three indices of female reproduction are con- 

 sidered in this paper: the percent pregnant, the per- 

 cent lactating, and the percent mature. Tbmporal 

 trends in these three indices were examined by 

 regressing annual means against year (weighting by 

 the inverse of binomial variances). 



In calculating the percentages of mature females 

 that were pregnant and that were lactating, 

 specimens were used only if both ovaries were col- 

 lected and if at least one corpus of ovulation (cor- 

 pus albicans or corpus luteum) was present. 

 Previously, 1971 and 1972 samples were excluded 

 from calculation of percent pregnant because of 



undersampling of younger females with a mottled 

 color pattern (Perrin et al. 1977). This was not deem- 

 ed necessary in this study, because in 1973-83 

 samples the percent pregnant for mature mottled 

 females (31.1%) was essentially the same as that for 

 mature females with a fused color pattern (31.4%). 



In calculating the percentage of females that were 

 sexually mature, two different criteria were used for 

 determining maturity. In the majority of cases both 

 ovaries were examined, and the presence of one (or 

 more) corpus of ovulation was taken as evidence of 

 sexual maturity. If ovaries were not collected or ex- 

 amined (which was true for about 30% of females 

 over 150 cm TL), a length criterion was used for 

 maturity. Samples from 1971 and 1972 were ex- 

 cluded from these analyses because sampling was 

 not random in those years. i 



"Length at attainment of sexual maturity" was 

 determined by the method used by Perrin et al. 

 (1977). Based on the sample for which ovaries were 

 examined, this length was estimated as the length 

 at which the number of longer immature individuals 

 equals the number of shorter mature individuals. For 

 the northern stock, the length at the onset of sex- 

 ual maturity was determined independently for each 

 year 1974-83 (176.5, 177.5, 177.0, 177.0, 178.0, 177.5, 

 179.0, 178.5, 180.0, and 182.0 cm, respectively). In 

 1973 the decision to collect ovaries was not based 

 on specimen length. The apparent trend in these 

 data yields a significant regression (P = 0.0008); 

 hence, regression estimates were substituted for an- 

 nual estimates for 1974-83, with an extrapolation to 

 1973. These values were 175.6, 176.1, 176.6, 177.1, 

 177.6, 178.0, 178.5, 179.0, 179.5, 180.0, and 180.5 cm, 

 respectively, for 1973-83. For the southern stock, in- 

 sufficient data exist to calculate a length at attain- 

 ment of sexual maturity for individual years, hence 

 the collective value was used for all years (175.0 cm). 



Six factors were examined to determine whether 

 annual changes in the above percentages of pregnant 

 females were caused or affected by changing biases 

 in the sampling methods. These factors include 1) 

 geographical provenance, based on two strata (Fig. 

 1) which roughly correspond to the historical tuna 

 fishing grounds (inside the Commission Yellowfin 

 Regulatory Area, CYRA) and a more recently ex- 

 ploited area (outside the CYRA); 2) the quarter of 

 the year; 3) the length of chase, or the time between 

 sighting the dolphin school and capture (net set); 4) 

 the observer's estimate of the dolphin school size 

 (only available since 1973); 5) the number of dolphins 

 known to have been killed in the set; and 6) the total 

 number of tons of tuna caught in the set. 



The selection of these six factors was guided to 



658 



