vol 25 alheit etal

8/13/2019 Vol 25 Alheit Etal

http://slidepdf.com/reader/full/vol-25-alheit-etal 1/10

ALHEIT ET AL.: SPAWNING FREQUENCY AND SEX RATIO IN PERUVIAN ANCHOVY

CalCOFI Rep., Vol. XXV, 1984

SPAWNING FREQUENCY AND SEX RATIO INTHE PERUVIA N ANCHOVY, ENGRAULIS RINGENS’

JUERGEN ALHEIT

Program a Cooperativo Peruano-Aleman de Investigaci6n Pesquera (PROCOPA)lnstituto del M ar del Peru

Apartado 22Calao, Peru

V. HUGO ALARCON

lnstituto del Mar del Penj

BEVERLY J. MACEWICZ

National Oceanic and Atmospheric AdministrationNational Marine Fisheries Service

Southwest Fisheries CenterLa Jolla, California 92038

Apartado 22

Callao, Peru

ABSTRACTSpawning frequency was determined for the first

time for the Peruvian anchovy, Engraulis ringens, us-ing the incidence of postovulatory follicles. Theagreement between two independent frequency esti-mates, on e for females taken one day after spawningand another for females taken two days after spawn-ing, demonstrated the reliability of this method. InAugust-September, the main period of annual spawn-ing, 16.04 of the female population of the centraland northern anchovy stock off Peru spaw ned per day:i.e., every 6 .23 days the average female spawned anew batch of eggs. Hydration of ovaries began asearly as 0700 hours in the morning. Spawning oc-curred at night between 1800 and 0200 hours, reach-ing a maximum at about 2200 hours. Sex ratio was57 .9 females by weight . The vulnerabi l i ty of

females to the purse seine changed with their repro-ductive state. Females with hydrated ovaries whowere ready to spawn seemed to attract males and toform male-dominated ‘‘spawning schools” by segre-gating from “norm al” schools at night.

RESUMENLa frecuencia de puesta en la anchoveta peruana

Engraulis ringens ha sido determinada por primeravez, tomando como base la incidencia de foliculospost-ovulatorios. La concordancia entre dos estima-ciones independientes de esta frecuencia, obtenidas dehembras un y dos dias despuCs de la puesta, demues-tran la fiabilidad de esta mCtodo. Anualmente, elperiod0 principal d e la puesta abarca d e Agosto a Sep-t iembre, cuando la puesta diar ia comprendi6 el

16.04 de las hem bras de las poblaciones del norte ycentro de la regi6n peruana; es decir, cada 6 .23 diasuna hembra promedio pone una nueva remesa dehuevos. La hidrataci6n de 10s ovarios se inici6 tem-

~

‘Publication No. 15 of the “Cooperative Peruvian-Germ an Fisheries Research Proj-

ect” (PROCOPA)

[Manuscript received Apnl 3, 1984 1

prano, a eso d e las 0700 de la m afiana. La puesta seprodujo entre las 1800 y 0200 horas, alcanzando unmiiximo a las 2200 horas aproximadamente.

La proporci6n de sexos calculada en peso fue de57.9 de hembras. L a vulnerabilidad de las hembras ala pesca co n redes d e jareta varia con la fase reproduc-tora enque se encuentran. Hembras con ovarios hidra-tados, que estaban dispuestas para la puesta parecianatraer a 10s machos y formar cardtimenes de puestadominados por machos separAndose de 10s card&menes normales durante la noche.

INTRODUCTIONIn the past, estimating the spawning biomass of

multiple-spawning pelagic fish species like anchoviesand sardines posed serious problems because noadequate methods were available to determine the

spawning frequency, i.e., the fraction of the femalepopulation spawning per unit time. Recently, Hunterand Goldberg (1980) solved this problem for the north-ern anchovy, Engraulis morda x, by an ingenious tech-nique. They followed up the suggestion of Yamamotoand Yoshioka (1964) that spawning frequency could bedetermined by incidence of postovulatory follicles-the remnants of ovulated follicles.

Follicles are the layers of cells surrounding de-veloping oocytes in the ovaries. After ovulation, theyimmediately begin to deteriorate and are then calledpostovulatory follicles (Hunter and Goldberg 1980).Hunter and Gold berg identified postovulatory folliclesin histological sections of northern anchovy ovariesand classified the females into three groups: thosespawning on the night of capture (new, or day-0 post-ovulatory follicles), those spawning on the night pre-vious to capture (day-1 postovulatory follicles), andthose showing no evidence of recent spawning. Hun-ter and Goldberg then used the frequency of day-1postovulatory follicles as a measure of spawning fre-quency. They used anc hovy spaw ned in the laboratoryto develop this technique (Leong 1971). Because thetime of induced spawning was known, they could de-

43



ALHEIT ET AL .: SPAWNING FREQUENCY AN D SEX RATIO IN PERUVIAN ANCHOVYCalCOFI Rep ., Vol. XXV, 1984

scribe histological criteria for aging postovulatoryfollicles. These criteria were then applied to fieldpopulations.

Hunter and Goldberg collected their field sampleswith a small trawler, which could catch adult anchovy

only at night. They were therefore not able to obtaincomplete 24-hour time series of females with post-ovulatory follicles. T his could ha ve resulted in a biasin the estimation of spaw ning frequency, since Hunterand Goldberg (1980) had already shown that femalescaptured on the night of spawning were oversampled.

It is well known that prior to ovulation the ovariesof teleost fish take u p fluid , a process called hydration(Fulton 1898). Hunter a nd M acewicz (1980) suggestedthat if anchovies could be sampled during the day, itmight be possible to use the incidence of females withhydrated ovaries (‘ ‘hydrated females”) to determinespawning frequency. The advantage would be thattime and money could be saved because histologicalexamination would no longer be required.

In this paper w e estimate the spa wning frequency ofthe Peruvian ancho vy, Engrad is ringens, using Hun-ter and Goldberg’s (1980) method. We can answersome of the questions posed by Hunter and Goldbergand Hunter and Macewicz (1980) because we used adifferent technique to sample anchovy. In Peru,anchovy were co llected with a purse seiner and couldtherefore be sampled in the day as well as at night.Thus , f ema les wi th pos tovu la to ry fo l l i c l e s andfemales in the hydrated, prespawning stage could be

obtained for an entire 24-hour cycle. Any time-relatedbias in the incidence of females with postovulatoryfollicles would becom e obvious when a series of sam-ples taken at regular intervals over 24 hours was ex-amined. Thus, the use of a purse seine for the Peru-vian anchovy gave new insights into bias in estimatingspawning frequency. Because hydrated females weresampled during the day, the alternative approach ofusing incidence of hydrated females to estimatespawning frequency could also be evaluated. T hus theobjectives of this paper are to evaluate Hunter andGoldberg’s (1980) method for estimating spawningfrequency by using a 2 4-hour sam pling scheme and to

determine the effect of different sampling gear (trawlversus purse seine) on the estimates of spawning fre-quency and sex ratio of Peruvian anchovy schools.

The findings of Hunter and Goldberg (1980) en-abled the Southwest Fisheries Center, La Jolla, to de-velop the “egg production method” (Parker 1980;Stauffer and Picquelle 1980) for estimating spawningbiomass of the northern anchovy off California. Twoof the five parameters required for this method arespawning frequency and sex ratio. Th e data presentedhere were used to estimate the spawning biomass of

the Peruvian anchovy using the egg product ionmethod (Santander et al., in press).

METHODSAdult anchov ies were collected with a purse seiner.

The cruise was run from August 25 to September 17,1981. It began in the south and proceeded northw ards.The purse seine stations were usually located within20miles of the shore because of adverse weather co n-ditions and the scarcity of anchovy schools fartheroffshore. Most of the collections came from the south-ern part of the investigation area becau se weather con-ditions in the north prohibited use of the purse seine(Figure 1). The total number of collections was 49.

Immediately after captur e, the anchovy’s body cav-ity w as opened from the anus to the ventral fins. O nlylive specimens were processed because of the rapiddegeneration of the postovulatory follicles. The fishwere preserved in a 4 buffered formaldehyde solu-tion. S eawater was not used to dilute the formaldehy-de solution because it causes a white precipitate thatmakes reading the histological sections difficult.

Twenty mature females were collected at random

60

8

loo

12-

1L‘

1 I

80’ 78O 76’

Figure 1. Map of surveyed area and location of samples of anchovies.

44



ALHEIT ET AL. : SPAWNING FREQUENCY AND SEX RATIO IN PERUVIAN ANCHOVYCalCOFI Rep., Vol. XXV, 1984

for histological analysis. Standard m ethods were usedto process the ovaries. A small cube of about 0.125cm3 was cut from the center of the ovary. It w as dehy-drated in a series of alcohol solutions and em bedded inparaplast. The histological sections were cut at 6 k m

and s ta ined with hematoxyl in /eosin . A deta i ledaccount of the procedure is given in Alarc6n et al. (inpress). Th e criteria develope d by Hu nter and Goldberg(1980) for aging postovulatory follicles are based onthe stages of degeneration through which they pass.Because postovulatory follicles degenerate rapidly,their age can only be determined up to 50 hours afterspawning. The age classes for postovulatory folliclesused here are som ewha t different from those describedby Hunter and Goldberg (1980).

Th e fraction of fema les (sex ratio) was estimated foreach collection from a subsample consisting of thefirst 800 g of fish. The total body weight was used,because the gonad-free weight of males was not mea-sured. The total body weight of hydrated females wasadjusted (Santander et al., in press) for the excessweight of the hydrated ovary (Stauffer and Picquelle1980). Imm ature fish were included because it was notpossible to distinguish between mature and immaturemales. In order to attain exactly 800 g of fish it wasnecessary to use only that fraction of the w eight of thelast fish in the collection that completed the 800 g .Sample mean and variance were estimated accordingto Stauffer and Picquelle (1980):

n (n-1)

where

i = fraction of females by weight in percent incollection i

= average fraction of females by weight inpercent from all collections.

For estimating the spawning frequency (fraction of

mature females spaw ning per day) we used only post-ovulatory females taken at least 9 hours after peakspawning (2200 hours), to prevent bias arising fromsampling females during the time of day w hen they areactually spawning. Two independent 24-hour sets ofpostovulatory follicles could be separated: one set ofpostovulatory follicles with an age between 9 and 32hours after spawning an d another set of postovulatoryfollicles with an age between 33 and 56 hours. H ence-forth females with ovaries containing postovulatoryfollicles of 9-32 hours will be called day-1 females;

those having ovaries containin g postovulatory folliclesof 33-56 hours will be called day-2 females.

Assuming that sampling of females with hydratedovaries, day-1 fema les, or day-2 females is unbiased,then the spawning fraction for collection i is esti-

mated by:

Fi =- or r

where

mi = mh + m l i + mzi mar

mhi mli m i

mi mi mi

and where

mhi = num ber of hydrated fem ales in collectionmli = num ber of day-1 fema les in collection i

mzi = num ber of day-2 fem ales in collection i

mai = number of females that have not spawned

within the past 9 to 56 hours (includesfemales with postovulatory follicles of anage of less than 9 ho urs)

= num ber of mature fem ales in collection i

= spaw ning fraction in collection imi

Fi

The results for the Peruvian ancho vy were similar to re-

sults for the northern anchovy (Stauffer and Picquelle1980); hey indicated that hydrated females were over-sampled. To correct for this apparent oversampling,under the assum ption that the true fraction of hydratedfemales is the same as the fraction of day-1 or day-2females, mhi is replaced by mli mzi such that

2

mliF . =

mli + m i

2m l i + m2i ma;

m2ior

+ ml i m2i maili + m2i

where

Fi = corrected fraction of day- 1 or day-2 females incollection i

The estimates for mean and variance are given by:

j 7 = C m li m2i I:m li m2i

2 C my i

+m l i+mzi+ mai

and

45



ALHEIT ET AL.: SPAWNING FREQUENCY AND SEX RATIO IN PERUVIAN ANCHOVYCalCOFI Rep., Vol. XXV, 1984

where-F = average fraction of females spawning per day

from all collections

+ mli m2i + mai =correctedli + m2i

2m . =

YC

number of mature females in collection i

corrected per collection= average number of mature females

n

= number of collections.

RESULTS

Peak Spawning TimeTo age postovulatory follicles, one must determine

the duration of the daily spawn ing period and its mid-point. This goal can be reached in three ways: 1) byrecording time of incidence and frequencies of newpostovulatory follicles from samples of adult anchovyfemales; (2) by reco rding the decline of occurrence ofhydrated females; and (3) by recording time of occur-rence and frequencies of newly spawned eggs fromichthyoplankton samples . Hunter and Macewicz(1980) demonstrated clearly that the percentage ofnorthern anchovy females with hydrated oocytes de-clined steadily from 10 to 14 at 1800 hours to 0

at 2400 hours, and concluded that 2200 to 2300 hourswas the period of maximum spawning.

The data from the P eruvian anchovy did not provideas clear a picture, fo r two reasons. H unter and Mace-wicz (1980) collected all their samples at night be-tween 1800 and 0500 hours, whereas only 40 of thePeruvian samples were collected between 1800 and2300 hours, and only one sample was obtained be-tween 2300 and 0700 hours. In the Peruvian samples,females with new postovulatory follicles were re-corded fo r the first time at 1800 hours, but their num-bers were very low (Table 1).Th e numbers of femaleswith hydrated oocytes ranged between 7.5 and

51.7 from 0700 to 2030 hours and declined sharplyto 3.3 and 5 at 2130 to 2230 hours, respectively(Figure 2) . The best data for estimating the midpointof the nightly spawning period was that provided bythe ichthyoplankton survey. The occurrence of newlyspawned eggs in the water column demonstrated thatpeak spawning time of the Peruvian anchovy is be-tween 2200 and 2300 hours (Santander et al., inpress).

Combining all these data leads to the conclusionthat Peruvian anchovy spawning starts at sunset,

TABLE 1

Collections Containing Female Peruvian Anchovywith New Postovulatory Follicles

~~ ~

Time of No. of new

Collection no. day postovulatory follicles

46 18004 1930

12 193041 201525 210021 2230

1 0315

around 1800 hours, and that the period of maximumspawning is between 2200 and 2300 hours. It was notpossible to estimate the time when nightly spawningceases, but by analogy to the northern anchovy,spawning probably ceases around 0200 hours. Forconvenience, 2200 hours is taken as the midpoint ofthe daily spawning period in the following analysis.

Spawning FrequencyThe spawning frequency is the fraction of mature

females that spa wns per day. In theory, one should getthree independent estimates of this parameter: 1) thepercentage of females with hydrated oocytes, (2) thepercentage of females with day-1 postovulatory folli-cles, and (3) the percentage of females with day-2postovulatory follicles. Because females with hy-drated oocytes tend to be oversam pled, they cannot beused for this purpose. The age of postovulatory folli-cles can only be determ ined up to about 50hours after

spawning, because older postovulatory follicles maybe confused with other structures, such a s atretic folli-cles (Hunter and Goldberg 1980; Hunter and Mace-wicz 1980).

Because the northern anc hovy were sampled only atnight (Hunter and Goldberg 1980; Hunter and Mace-

1I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ' '

8 10 12 14 16 18 2 22 24

TIME OF DAY

Figure 2. Change of percentage of hydrated females in collections with timeof day.

46




wicz 1980; Stauffer and Picquelle 1980), investigatorscould use only day-1 postovulatory follicles to esti-mate spawning frequency. Our use of a purse seinerfor sampling the Peruvian anchovy permitted collec-tion of samples at any time of day. Consequently, we

obtained two independent sets' of data for estimatingspawning frequency (day-1 and day-2 females) andcompared them.

The way in which these two sets of data wereobtained is demonstrated in Table 2. It was assumedthat all anchovy females spawn at 2200 hours. Be-cause the earliest samp le was collected at 0700 hours,we considered that time the beginning of the day-1 andthe day-2 periods. Postovulatory follicles found at thistime are, according to their structure, assigned to oneof the following three groups: day-1, day-2, or olderpostovulatory follicles. They are given ages of 9, 33,

or more than 56 hours , respectively. T able 3 containsthe summary of all data: (1) date and time of eachcollection, (2) numbers of hydrated, day-1, and day-2females, (3) hours past spawning of each collection,(4) values corrected for oversampling hydratedfemales, and ( 5 ) percentage of hydrated, day-1, andday-2 females per collection.

Stauffer and Picquelle (1980) suggested that fe-males with new postovulatory follicles may also beoversampled, but lacked data with which to test this

2 0 1

7 1000 13W 16 19 2 2 M 0100 4 7M T lM E OF DAY.

9 12 15 18 21 24 27 30 33 HOURS AFTERSPAWNING.

Figure 3 Spawning frequencies calculated from consecu tive 24-hour periodsThe time of day indicates at what time each 24-hour period begins Thecorresponding hours after mean spawning time (2200 hours) are also indi-cated Standard deviations are added

hypothesis. To test it, we calculated the mean spawn-ing frequencies for con secutive 24-hour periods. F irst,we calculated the mean spawning frequency for thefirst 24-hour period starting at 0700 hours-9 hoursafter mean spawning time. Only postovulatory folli-cles within the age range from 9 to 32 hours wereincluded (Figure 3). T hen we calculated the spawningfrequency for the second 24-hour period, beginning at0800 hours and including postovulatory follicles with

TABLE 2

Time of Collection of Females and the Age of Postovulatory Follicles

Age of postovulatory follicles(hours past spawning)

Day- 1 Day-2

Time of day

1st day following. spawning

22.00Night of spawning (Peak spawning)

07 .00-07.59 9 3308 00-08.59 10 34

09.00-09.59 11 35

10.00-10.59 12 36

11.00-11.59 13 37

12.00-12.59 14 38

13 00- 13.59 15 39

14.00-14.59 16 40

15.00-15.59 17 41

16.00-16.59 18 42

17.00-17.59 19 43 18.00-18.59 20

19.00-19.59 21 45

20.00-20.59 22 46

2 1 00-21.59 23 47

22.00-22.59 24 48

23.00-23.59 25 49

00.00-00.59 26 5001.00-01.59 27 51

02 00-02.59 28 52

2nd day following 03 OO-03.59 29 53

spawning 04.00-04.59 30 5405 OO-05.59 31 5506.00-06.59 32 56

47




an age from 10 to 33 hou rs. We repeated these calcula-tions for a total interval of 48 ho urs, alw ays advancingby one hour, so that the last 24 -hour period starts againat 070 0 hours-33 hours after mean spawning timeand includes postovulatory follic les with an age from

33 to 56 hours. Theore tically, there should have been25 different 24-hour periods. However, because nocollections were taken at 1100 hours and only onebetween midnight and 0700 hours (Table 3) , only 17consecutive 24-hour periods are included in Figure 3.

The spawning frequencies calculated from these 17consecutive 24-hour periods showed no strong trendover the observed 48-hour in terval, and ranged from14.70 to 17.26 . We conclud e that females withpostovu latory follicles of 9 hours and older are equal-

ly available to the purse seine.If spawning frequency data for day-1 females and

for day-2 females are independent of each other andidentically distributed, they could be combined. T heircombined total would double the sample size and

TABLE 3

Data Summary

Hydrated females Day-1 females Day-2 females P , P 2

Time Collection Sex ratio Hours past No. m 1 1 m 2 , ma Day-I Day-2

of day No. Date ( females) No. (mhJ spawning (m,,) spawning (m2J fml,lm y,) fmz,/mY,l

07.15

07.2508.0008.3008.45

09.0009.0009.0009.1509.3010.00

10.1012.0012.0012.1012.3013.00

13.1013.1514.1514.1514.4515.3016.0016.1016.1516.3016.4517.1017.3518.0018.0018.0018.15

19.0019.1519.3019.3019.3019.3019.4020.1020. I521 oo

21 oo

21.1522.3022.5503.15

28

383326485

10

4229172252

257342718

5349115835

615

4020

3234454

8364631

241947

121655

45412547322137

1

5 .9 .

8 .9 .6.9.2.9.

12.9.26.8.28.8.9.9.

5.9.30.8.

1.9.16.9.25.8.17.9.7.9.2.9.

30.8.

16.9.12.9.28.8.17.9.7.9.

28.8.29.8.

8 .9 .31.8.25.8.

1.9.9.9.

16.9.27.8.

7.9.11.9.5 .9 .

1.9.30.8.25.8.26.8.28.8.29.8.16.9.9 .9 .8.92.9

11.9

5 .931.8

7 .925.8

52.76

55.3083.0168.7445.3244.9631.5133.3890.5054.4467.3345.8068.9563.6270.9649.8955.32

62.0459.2947.43

28.1870.5181.2732.5760.3826.3674.4278.7362.1271.9445.2136.1129.9061.15

60.3846.1355.3358.2812.6243.9365.5273.9315.8678.6879.2064.1879.6864.3651.42

11

30

0

5

11722

10

0

322

1

63

423

113251

30

0

26

13162

34132

17821

1920

01

1

2

55

150

0

2555351

10

50

0

15101

503015

201015551510255

15

0

0

1030

565801

152065108540105

951

0

0

5

510

9

91

1010111111

111112121414141415

1515161616171818181818191920202020

212121212121212222232323242429

0

32262

60

222656312644

10

913446

7121

5

14320

31

5

0

442395

33

3334343435

35353535363638383838393939404040414242424242434344444444

454545454545454646474747484853

0

31

4311

0

30

1

3350

22324161

35674341

40

1

2

0

40

3560

74241

1

0

3 .01.53.04.51.53.50

2.51 o

1.54.54.05 .5

1.51.52.04.53 .04.01 o3.05.02.04.05.0

5.55.02.05.51 o

3.00.53 .0

1 .55.5

1.53.00

3.03.05.5

0

5.5

4.02.03.55.0

3.0

9 9.0

11 20.017 21.514 23.06 19.56 10.5

6 16.518 18.013 20.5

8 11.017 21.58 21.5

10 22.07 23.57 11.5

11 15.513 19.07 20.5

12 21.09 21.07 10.0

11 20.08 23.0

11 17.011 23.0

7 22.09 25.5

I O 25.014 20.03 19.5

17 20.01 10.0

3 4.512 21.0

14 18.55 21.54 8.5

12 21.03 3.06 15.0

12 21.08 24.51 1 .0

23.512 24.016 22.012 22.59 24.0

0

,1500,0930

,0870.3077,1905,36360

,0976,1818,0930

,2791,2273,2553,2609.0645. lo53

.2927,1905,1905,10000

,3913,0588,1304

.I818,1569,2400,0500,35900500

.2000,2222,2381

.0541I860

.3529,09520

,2000,0476,2041

0

,1702,1667.0909.I333,3750

0

,1500,0465,1739,1538,0952

,06060

,14630

,0465,1395,1364,2128

0

,1290,1053,1463,0952,1905.1000,3000.0435,1765,2174,2727,2745.1600

,1500,2051

0500

,40000

,0476

,108 1,32560

.1905

0

.2000,2381,24490

,2979,1667.09W,1778,0417

12 21.0 ,2381 ,0476

48



ALHEIT ET AL .: SPAWNING FREQUENCY A ND S EX RATIO IN PERUVIAN ANCHOVYCalCOFI Rep., V ol. XXV, 1984

TABLE 4

Arithmetic Means and Probability Values p and q of theBinomial Distributions of Hydrated, Day-1, and Day-2 Females

Reproductive sta ge of

Hydrated 4.6122 .2306 ,7694Day-1 3.1633 ,1726 ,8274

Day-2 2.7143 ,1481 ,8519

females P 4

thereby reduce the variance of the estimate of spawn-ing frequency. If their occurrence were random, theobserved frequencies of day-1 and day-2 femalesshould follow a positive binomial distribution. To testthis assump tion, we applied the Kolmogorov /Smirnovtest (Siege1 1956). The frequencies of all three groupsof females in different reproduc tive stages are listed inTable 3. Table 4 gives arithmetic m eans and the prob-

ability values p andq

of the binomial distribution.The term k of the positive binomial distribution, themaximum number of individuals in a collection, was20; in other words, 20 females per collection wereincluded in the estimate of the frequency of hydrated,day-1, and day-2 females. The 5 significance levelD)or the Kolmogorov/Smirnov test is 0.1943, and

the estimated maximum differences (a)were 0.3628(hydrated), 0.1656 (day-1), and 0.1859 (day-2). Theseresults demonstrate that the frequencies of day-1 andday-2 females were not significantly different from thepositive binomial distribution at the 5 level, whereasthe hydrated fem ales do not correspond to the positive

binomial distribution. This becomes more obviouswhen the cumu lative percentages of the observed andexpected class frequencies of all three groups offemales are plotted (Figure 4).

If the spawning frequencies of day-1 and day-2females are to be combined, they must not be statis-tically different. To establish this we used a test of thed i f fe rence be tween pa i red independen t samples

;i;10I R W I N C C L A S S

Figure 4. Cumu lative requen cies of observed and expected (positive binomialdistribution) class freque ncies of females in d ifferent reproductive stages.The 49 collectionsof mature emale anchovies are distributed nto requencyclasses according to the num ber of females they have in a particular rep ro-ductive stage (hydrated, day-I, or day-2); for example, in the left panel,frequency class 0 contains all collections having 0 hydrated females.

T A B L E 5

Mean Percentage of Hydrated, Day-1 Day-2, and Day-1IDay-2Females with Variance V), Standard Deviation SD), and

Coefficient of Variation CV)

Reproductive stage

of females Mean V SD cvHydrated ,2306 1.219 8X ,0349 ,1515Day- ,1726 2.3248X ,0152 ,0883

Day-2 .I481 1.7224X ,0131 ,0886Day-1 & day-2

(combined) ,1604 1.0175X ,0101 ,0629

(Snedecor and Cochran 1967). The paired sampleswere the spawning frequency values of day-1 (PI) ndday-2 p 2 ) emales from each collection (Table 3).

The null hypothesis was that the mean difference (d)between F 1 and p2 equals zero. d was 0.032. Thet-test gave a value of 1.526 with n-1 = 48 degrees of

freedom. This shows that the null hypothesis cannotbe rejected at the 5 level of significance. Therefore,the two estimates can be comb ined, thus doubling thesample size.

The spawning frequency of the day-1 females was0.1726 with a variance of 2.3248 x a standarddeviation of 0.0152, and a coefficient of variation of0.0883 (Table 5 ) . The spawn ing frequency of the day-2 females was 0.1481 with a variance of 1.7224 x

a standard deviation of 0.0131, and a coefficientof variation of 0.0886. When these two data sets werecombined, the estimate of spawning frequency was

0.1604 with a variance of 1.0175x

lop 4, a standarddeviation of 0.0101, and a coefficient of variation of0.0629 (Table 5 ) . By combining the data we havereduced the coefficient of variation by nearly a third.This would be important when the estimate of spawn-ing frequency is used to calculate spawning biomassusing the egg production method (Parker 1980; Stauf-fer and Picquelle 1980), because by reducing thecoefficient of variation for spawning frequency onereduces the coefficient of variation for the biomassestimate (Santander et al., in press).

A spawning frequency of 16.04 (Table 5 ) meansthat in August/September 1981 the average mature

Peruvian anchovy female spawned a new batch ofeggs every 6.23 days. The high daily incidence ofhydrated females (23.06 ; Table 5 ) clearly indicatesthat hydrated females were oversampled. Comparingthe coefficient of variation for the incidence of hy-drated females with that of the day-1 and day-2females indicates that the number of hydrated femalesper sample was much more variable than were thenumbers of day-1 or day-2 females per sample.

Hydrated females also seem to be more vulnerableto the purse seine than other females. In order to test

49



ALHEIT ET A L. : SPAWNING FREQUENCY AND SEX RATIO IN PERUVIAN ANCHOVYCalCOFI Rep., Vol. XXV, 1984

YDRATED

0800 iom izw LOO 16 ram 2 ~ 0

I f I

TIME OF DAY

Figure 5. Chang e of perc enta ge of frequency of occu rrence of hydrated, day-1,and day-2 females in co llections with time of day.

whether the three groups of fem ales (hydrated, day-I,and day-2) show a diel chan ge in their vulnerability topurse seining, the percentage of females in each groupwas calculated for 2-hour intervals and plotted at themidpoints of the 2-hour interval (Figure 5 ) . The hy-drated females had the highest percentage except forcollections taken at about 1600 and at 2200 hours. T hevalue at 2200 hours (4 ) is low because most of thefish had completed spawning at that time. The veryhigh values at 1800 hours (33 ) and at 2000 hours(38 ) indicate that hydrated females are particularlyvulnerable at that time of day. At these hours, hydra-tion of the ovaries has reached its maximum, and theovaries are extremely heavy-in some cases up to25 to 30 of the female’s gonad-free body weight.The percentages of day-1 and day-2 females, on theother hand, remain relatively constant throughout theperiod, indicating that their vulnerability to capturedoes not change significantly with time of day.

Sex RatioThe sex ratios (percentage of females on a weightbasis) of all 49 collections showed large variationsranging from 12.62 to 90.50 females (Table 3).The average sex ratio was 56. 43 , with a variance of0.0007, a standard deviation of 0.0259, and a coeffi-

cient of variation of 0.045 9. Similar variability in sex

ratio of the northern anchovy has been reported byKlingbeil (1978), Hunter and Goldberg (1980), andStauffer and Picquelle (1980). Klingbeil demonstratedthat the greatest variability in the sex ratio occurs inthe months of peak spawning. Hunter and Goldberg

showed that sex ratio varied with spawning activity.We carried out the same analysis for the Peruviananchovy.

We grouped all 49 collections into three classesbased on their sex ratio. F or each of these three classeswe calculated the percentage of females in the follow-ing four spawning groups: (1) spawning on the night ofcapture (includes hydrated females and females withnew postovulatory follicles), (2) day-1 females, (3)day-2 females, 4) emales w ith no evidence of recentor imminent spawning (Table 6). In the male-dom-inated collections, which contained only 10 to 39females, 54 of the females had spaw ned on the nighof cap tu re . On the o the r hand , in the f emale -dominated collections, where 70 to 99 of the fishwere females, only 13 of the females had spawnedon the night of capture. Correlation coefficients forsex ratio and (1) number of hydrated fema les, (2) num-ber of day-I females, and (3) number of day-2 femaleswere - 0.6 6, 0.35, and 0.34 , respectively. Only thecorrelation coefficient of .6 6 for the hydratedfemales was significant at the 5 level. This showsthat as the proportion of males in a collection in-creases, so does the proportion of hydrated females.

In addition, sex ratio also seemed to change with

time 6f day. Most samples taken between 0700 and1800 hours had 50 to 70 females, whereas aftesunset (1800-2100 hours) the average ratio of femalesin the collections dropped below 50 . A similar pattern is described for the northern anchovy by Stauffeand Picquelle (1980). Average sex ratios (including alcollections) computed for morning, afternoon, andnight periods were 56.8 0 , 59.11 , and 54.25 , respectively (Table 7). These data show that the averagesex ratio (56.43 fem ales, N = 49) is probablybiased by including those collections that have morethan 30 hydrated females and were sampled at nightA m ore realistic sex ratio is computed by excluding al

night collections sampled between 1800 and 2300

TABLE 6

Sex Ratio and Percentageof Females in Different Reproduct ive Stages

Females

Sex ratio class Number of Spawning on day No evidence of N o . of females( females) collections of capture Day-1 Day- 2 spawning classified

10-39 9 54 8 8 29 180

40-69 28 19 16 13 52 56070-99 12 13 20 20 48 240

50



ALHEIT ET AL.: SPAWNING FREQUENCY AND SEX RATIO IN PERUVIAN ANCHOVY

CalCOFI Rep., Vol. XXV, 1984

TABLE 7

Sex Ratios at Different Times of Day~~~~~~ ~~ ~~~~~ ~ ~~

Excluding collections withTime of day All collection s >50% hydrated females >3 0 hydrated fema les

No. of of No. of of No. of of

collections females collections females collections females07.00-12.00 12 56.80 9 56.44 8 59.5612.00-18.00 18 59.11 16 60.30 14 60.2218 OO-23OO 18 54.25 13 63.59 12 65.23

Excluding collections with

hours, the time of peak variability in sex ratio. Fromthe remaining 30 co llections we calculated a sex ratioof 57.90 with a variance of 0.0010, a standard de-viation of 0.0311, and a coefficient of variation of0.0536.

DISCUSSION

The spawning frequency of the Peruvian anchovywas found to be 16.04 in August-September 1981.This means that during the peak spawning period,Peruvian female anchovy spawn a new batch of eggsabout once every six days. The spawning frequenciesreported for the northern anchovy off California aresomew hat lower: 14.5 in March-April 1980 (Stauf-fer and Picquelle 1980), 10.6 in February 1981 and12.5 in April 1981 (Stauffer and Picquelle2), and12.0 in January-March 1982 (Picquelle and Hew itt1982). Hunter and L o n g (1981) estimated that thenorthern anchovy spawns about 20 times per year;however, this has yet to be proven by an annual sam-pling program. These data on northern and Peruviananchovy clearly contradict earlier assumptions thatthese two and other multiple-spawning pelagic speciesspawn only two or three times a year. Consequently,spawning biomass of such fishes has been consider-ably overestimated because of a severe underestimateof total fecundity (batch fecundity times spawningfrequency).

Sex ratio is an important parameter for the appli-cation of the egg production method (Parker 1980),because of possible biases. The midwater trawl isreported to be a biased sampler with respect to sex

ratio and hydrated females (Hunter and Goldberg1980; Stauffer and Picquelle 1980). A higher-than-expected number of trawl sam ples had a high or a lownumber of females. In addition, hydrated femaleswere twice as numerous as day-1 females. Staufferand Picqu el le3 suggest tha t m ales and hydratedfemales segregate from other females at the hours of

*Stauffer, G.D ., and S.J. Picquelle. The 1981 egg production estimates of anchovy

spawning biomass. Unpublished manuscript, 29 p. NMFS, Southwest Fisheries Cen-ter, P.O. B ox 271, La Jolla, California 92038.

'Ibid.

peak spawning at a depth w here they are more vulner-able to the trawl.

The purse se ine samples also appear to be biased inrespect to sex ratio and num bers of hydrated fem ales.In our study, oversampling of hydrated females oc-curred in the early morning hours, when the onset ofhydration could be determined only by recording the

migration of the oocyte nucleus to the pole. At thistime, the oocytes have not increased perceptibly insize, and the ovary is only a small fraction of the totalfemale weight. Three different explanations are possi-ble for the oversampling of hydrated females in purseseines:

1. Hydration decreases the ability of female ancho-vies to avoid nets.

2. Females segregate vertically (by depth), andthose with hydrated oocy tes are more accessibleto the purse seine than other females.

3. Females segregate horizontally (by area), andthose with hydrated oocytes occur in differentareas than those without hydrated oocytes.

In any case, using the incidence of hydrated femalesto estimate spawning frequency for anchovy (sug-gested by Hunter and Macewicz 1980) seems an in-accurate procedure, even if day samples of anchoviescan be obtained.

The sex ratio in the 49 collections ranges from12.62 to 90 .50 fem ales, with an average of56.43 (Table 3). Night purse seine collections with ahigh percentage of hydrated females (with respect tothe female fraction) also contain a high percentage of

males. Hydrated females are oversampled both dayand night, but the co-occurrence of high percentagesof hydrated females and of males is recorded only atnight. It might be hypothesized that the high maleratio in these night collections is because hydratedfemales, which are about to spawn, are attractive toand surrounded by a high number of males (Hunterand Goldberg 1980). If the hypothesis is correct thathydrated females are caught more often (than ex-pected) because their vulnerability to the net in-creases, then only the hydrated females should be

51




oversampled and not the m ales as well. Thus it seemsmore likely that the hydrated fem ales segregate, eitherby depth or by area , from the “norm al” school, tak-ing a high percentage of m ales with them and forming“spawning schools” dominated by males.

The average sex ratio is biased because the nightcollections contain a high percentage of hydratedfemales. When these collections are om itted, the aver-age sex ratio rises to 57.9 0 . If the true sex ratio ofthe population is 50 female, the purse seine clearlyoversamples females, whereas the trawl seems toundersample them slightly (Stauffer and Picquelle1980; Picquelle and Hew itt 1982). Klingbeil (1978)reported similar findings when he compared the sexratios of anchovy obtained from commercial purseseiners (fema1es:males = 1.60:l) and from researchtrawlers (femalexm ales = 1.09:l). He suggested that

the male-dominated schools may not form the largedense aggregations required for effective purse sein-ing. This might also explain the difference in sexratios between the n orthern and the Peruv ian anchovycollections. The large Peruv ian purse seiner (270 tons)may have shot the seine when the echo sounder indi-cated a relatively large, dense, female-dominatedschool, whereas each California trawl sample col-lected fish from a much larger area. Thus the trawlsamples included more male-dominated spawningschools, which presumably are less dense and extendover a much larger area. Therefore, if spawningschools are segregated horizontally from normal

schools, the trawl may be a more suitable tool fordetermining the sex ratio of anchovies than a purseseine.

Interpreting the different effects of trawling andseining on sampling of hydrated females is difficult.They were oversampled by the Peruvian purse seinerin 1981 and by the California trawler in 1980 and 1981(Stauffer and Picquelle 1980; Stauffer and P icquelle4).However, hydrated females were not oversampled bythe California trawler in 1982, possibly because thetrawl may have fished a shallower depth (Picquelleand Hewitt 1982). The oversampling of hydratedfemales might be solved by a comparative study be-

tween a purse seiner and a larger midwater trawlerwhich allows fishing in depths not reached by a pursseine.

ACKNOWLEDGMENTS

We are indebted to J.R. Hunter (Southwest Fisheries Center), A.D. MacCall (Southwest Fisheries Center), and G .D . S tauffer (Northwest Fisheries Centerfor their critical reading of the manuscript and themany helpful comments.

This work was supported by the Peruvian-GermanFisheries Investigation Project (PROCOPA), which ifinanced by the German Agency for Technical Cooperation (GTZ).

LITERATURE CITEDAlarch, V.H., J. Alheit, and B.J. Macewicz. In press. Estimado d

frecuencia de desove y proporci6n sexual de la anchoveta peruana, E

gruulis ringens. Bol. Inst. Mar. Peru, Callao.Fulton, T.W. 1898. On the growth and maturation of the ovarian eggs o

teleostian fishes. Ann. Rep. Fish. Board Scotl. 16:88-124.

Hunter, J.R., and S.R. Goldberg. 1980. Spawning incidence and hatcfecundity in northern anchovy, Engradi s mordux. Fish. Bull., U.S

Hunter, J.R., and R. Leong. 1981. The spawning energetics of femanorthern anchovy, Engruulis mordux. Fish. Bull., U.S. 79:21.5-230.

Hunter, J.R., and B.J. Macewicz. 1980. Sexual maturity, hatch fecunditspawning frequency, and temporal pattern of spawning for the northeanchovy, Engruulis mord ux, during the 1979 spawning season. CalCoop. Oceanic Fish. Invest. Rep. 21:139-149.

Klingbeil, R.A. 1978. Sex ratios of the northern anchovy, Engruulis mo

dux off southern California. Calif. Fish Game 64:200-209.

Leong, 5. 1971. Induced spawning of the northern anchovy, Engruu

mordux Girard. Fish. Bull., U.S. 69:357-360.

Parker, K. 1980. A direct method for estimating northern anchovy, Egruulis mo r d a x spawning biomass. Fish. Bull., U.S. 78541-544.

Picquelle, S. J., and R.P. Hewitt. 1982. The northern anchovy spawnibiomass for the 198 2-83 California fishing season . Southwest FisheriCenter Admin. Rep . LJ-82-16, 37 p.

Santander, H., J. A lheit, and P.E. Smith. In press. Estimado de la bioma

desovante de la anchoveta peruana, Engruulis ringens, con el “Metodde Produccih de Huevos” en 1981. Bol. Inst. Mar Peru, Callao.

Siegel, S. 19.56. Nonparametric statistics. McGraw-Hill Book CompanLondon, 312 p.

Snedecor, G.W., and W.G. Cochran. 1967. Statistical methods. Iowa StaUniversity Press. Ames, 593 p.

Stauffer, G.D ., and S.J . Picquelle. 1980. Estimates of the 1980 spawnibiomass of the central subpopulation of northern anchovy. SouthweFisheries Center Admin. Rep. LJ-80-09, 41 p.

Yamamoto, K., and H. Yoshioka. 1964. Rhythm of development in toocyte of the medaka, Oryzius lutipes. Bull. Fac. Fish., Hokkaido Uni

77641-652.

1.5:.5-19.ee footnote 2 on page 5 1

52

vol 25 alheit etal

Documents