MARINE ECOLOGY PROGRESS SERIES Vol. 153: 229-238, 1997 Published July 10 Mar Ecol Prog Ser

Demographic consequences of within-year variation in recruitment

Catherine A. Pfister*

Department of Zoology, NJ-15. University of Washington, Seattle, Washington98195. USA

ABSTRACT: I explored the consequencesof variance in the timing of recruitmentto early growth and survivorship in a guild of that occupy intertidalareas on the outer coast of Washington state (USA).When I quantified the appearanceof new recruits every month for 3 yr. I found seasonal peaks in recruitment for the 3 most common speciesin this system, although fishes recruited throughout the sprlng and summer months. Mark-recapturestudies revealed espec~allystrong seasonal growth pat- terns in 2 of the 3 , with increased growth rates durlng the summermonths. Thiseffect was most pronounced for young-of-the-year fishes.As a consequence,lndiv~duals of all specles reach reproduc- tive size more rapidly and benrfit from higher surv~vorshipto reproductive size when they recruit ear- lier in the spring and summer.These demographic consequencesof with~n-yearrecruitment vandtion appear driven by large-scale oceanographicevents, since recruitmentin all 3 specles can be correlated with upwelling andlor sea surface temperature. Although these results suggest that within-season recruitment variation,via its effect on juvenile performdnce, may affect the population trajectoriesof these nearshore fishes, previous researchin this system also ~ndicatesa strong role for adult survivor- ship patterns. Thus, understanding thenet effect of pelagic events in drivlng population dynamicsof organisms with complex life cycles requires integration of both pelagic and benth~chfe history events.

KEY WORDS: . Intertidal fishes Up\uelling Recruitment vanatlon . Timing of recruitment . Complex life cycles . Benthi.c-pelagiccoupling Demographicvariability

INTRODUCTION 1977, McFarland et al. 1985, Victor 1986). On the other hand, there is usually a strong seasonal component The larval stage of marine organisms continues to to recruitment in temperate and occasionally tropical represent a challenge to ecologists who seek to charac- areas (e.g. Victor 1983, Caffey 1985, Raimondi 1990, terize the components of population growth of marine Morgan & Christy 1995, Pfister 1996). In temperate organisms and the processes that influence multi-spe- areas, marine fishes and typically recruit cies assemblages. Variability in the number of young in pulses, usually in the spring and summer months. that establish in a population, referred to here as Although there may be well-defined peaks when recruitment, has the potential to determine abundance the majority of individuals return to the adult popula- patterns for that organism for a long time in the future. tion, there is often a period of months over which For organisms whose adult stage inhabits nearshore recruits establish (Loosanoff 1964, Yoshioka 1982, areas, the completion of the larval planktonic stage Caffey 1985, Gaines & Roughgarden 1985, Raimondi and successful reestablishment on the shore is essen- 1990, Doherty 1991, Bingham 1992, Kneib 1993, Secor tial. In some organisms, especially in tropical reef en- & Houde 1995). This spread could be due to variation vironments, reproduction and thus recruitment may be in the timing of spawning, variation in the duration a continuous process with only slight seasonality (Sale in the , differential survival during the plank- tonic stage, or any combination of these, and remains

'Present address:Dept of Ecology and Evolution, 1101E. 57th a major barrier to an integrated understa.nding of St., Unlverslty of Chicago, Chicago,Illinois 60637, USA. life histories which include planktonic and benthic E-mail: [email protected] phases.

O Inter-Research 1997 Resale of full article not permitted 230 Mar Ecol Prog Ser 153: 229-238, 1997

Variability in recruitment may reflect variability in season. These studies demonstrate how differences in oceanographic events such as uptvelling and spring the timing of early life history events can affect the algal blooms, since the establishment of some marine future performance of individuals. organisms is often cued to predictable oceanographic I explored whether variability in the timing of re- or lunar events. For example, both the timing of cruitment, correlated with large-scale oceanographic spawning and the recruitment of marine organisms events, affects the performance of individuals during can occur in lunar cycles (e.g.Kornnga 1947, Johannes their first months as part of the benthic population. I 1978, Connell 1985, Babcock et al. 1986, Kneib 1986, used 3 yr of demographic information on 2155 marked Robertson et al. 1990, Morgan & Christy 1995), and individuals in a guild of tidepool fishes in tandem with tidally forced internal waves have been associated automated buoy data to ask: (1) do large-scale oceano- with the return of reIatively high densities of fish and graphic events correlate with the timing of fish recruit- propagules to shore (Shanks 1983). ment? and (2) does variation in the timing of recruit- Sissentvine (1984) summarized correlative studles be- ment affect individual growth, survivorship, and the tween the recruitment of exploited temperate finfish time interval to first reproduction? My analyses indi- stocks and a number of physical factors such as wind cate that oceanographic effects on planktonic events direction, air temperature, and the intensity of coastal have predictable consequences for early benthic life upwelling; the recruitment success of commercially histories. exploited species in the northeast Pacific was often positively corie!atcd with indices of upwe!!ing. Al- though upwelling events can supply necessary sea- STUDY SYSTEM sonal resources to larvae, they can also be associated with offshore Ekman transport, which may be detri- The natural hlstory of intertidal sculpins (Order Scor- mental to organisms which need to return to shore. For paeniformes) has been described in detail elsewhere example, the recruitment success of on the (Pfister 1993, 1995, 1996). Briefly, with the exception of central California, USA, shore was negatively corre- a 1 to 2 mo larval period in the plankton (Washington lated with an index of upwelling (Roughgarden et al. et al. 1984), sculpins spend the majority of their life in 1988). intertidal habitats, including tidepools and surge chan- Despite an increasing number of studies docu- nels. Intertidal sculpins recruit during the spring and menting variability in larval supply and recruitment summer months. As adults (>35 mm standard length, of marine organisms, we still know very little about SL), they seem to inhabit a restricted area of the shore; whether temporal variation has a lasting a.nd de- individuals are routinely caught in a single or small tectable effect on population~ (but see Jones 1990, group of tidepools (Richkus 1973, Pfister 1995). Their Doherty & Fowler 1994, Pfister 1996). Whether intra- homing capabilities facilitate this site fidelity (Green annual variability in recruitment is important to indi- 1971, 1973, Yoshiyama et al. 1992, author's pers. obs.). viduals depends upon how differences in timing affect As many as 6 to 8 species of primarily intertidal growth and survivorship. Perhaps it is easiest to under- sculpins can be found at a variety of locales in the stand how benefits would accrue to an individual that northeast Pacific and all appear to reproduce in late recruits earlier than others. Establishing earlier than winter and early spring; recruitment occurs from competitors has been shown to favor individual fishes spring to summer (Hart 1973, Pfister 1996). Little is in intra- and interspecific encounters for shelter (Shul- known about the planktonic larval stage. Although man. 1985). In thermally stressful intertid.al habitats in there is one report that newly hatched larvae school the Gulf of California, barnacles that recruited earlier close to the substrate (Marliave 1986) potentially and obtained better microsites had a higher survivor- remaining close to their parental source, many individ- ship than those that recruited later (Raimondi 1990). uals live in extremely exposed, high energy habitats Priority effects may also benefit individuals who need where tidal and wind-driven currents may result in to grow through a perlod of vulnerability to predators long distance dispersal. The genetic data for intertidal (Shulman et al. 1983). Additionally, in an area where sculpins (gel electrophoresis data; Waples 1987, Yoshi- prey resources may have been limited, larval mummi- yama & Sassaman 1987) provide little evidence for chog fish that estctblished earlier had a higher sur- population substructure on scales of hundreds of kilo- vivorship than those that were manipulated expen- meters in the northeast Pacific. m.entally to establish later (Kneib 1993). In contrast, In the northeast Pacific, a predictable oceanographic later season recruitment coincided with higher rates of event is the spring transition which initiates upwelling survivorship and growth for a Hawaiian coral reef in March or April (Strub et al. 1987, Strub & James damselfish (Booth 1995), although these individuals 1988). Alongshore surface currents shift to become pre- were more vulnerable to predati.on in the subsequent dominantly southward and upwelling persists through- Pfister: Variation in recruitinent 23 1

out the summer. Sea surface temperature (SST, in "C) to facilitate marking large numbers of fishes in the also varies seasonally in the northeast Pacific and field. temperature has often been thought to be a correlate of Consequences of timing of recruitment to growth recruitment (e.g. Grossman 1982).Thus, I focussed on and survivorship. I examined the consequences of upwelling and SST when looking for correlations variable timing in recruitment for early growth and among oceanographic events and variability In recruit- survivorship. From data on marked individual fishes ment. and marked cohorts of fishes, I estimated the length of The recruitment data reported here are from Tatoosh time it took a new recruit of each species to attain Island, Washington, USA. Tatoosh Island (48.40°N, reproductive size and the probability that a new recruit 124.73" W) is separated by a 0.6 km water gap from the survived to reproductive size when it entered the pop- northwest tip of the Olympic Peninsula. It is an ulation at varying points in the recruitment season. exposed rocky intertidal site, encountering directly the Mark-recapture studies often resulted in fishes that predon~inantly westerly swells and winds. There are were uniquely distinguishable by a combination of 3 species of intertidal sculpins routinely found on characters that included species, gender, length and Tatoosh Island. Their rank abundance by biomass in the color and position of the paint mark. From these mid intertidal pools is: globiceps, Oligocot- marked individuals, I was able to determine individual tus maculosus, and C. embryum. Information on biotic growth rates in mm mo-' for 384 individual fishes of 3 interactions between intertidal sculpins and spatial species. I tested whether size, species identity, and patterns in recruitment is presented elsewhere (Pfister seasonal effects were important to growth using an 1995, 1996). ANCOVA (analysis of covarlance) with size as the covariate. I divided the year into a winter (Oct-Mar) and a summer (Apr-Sep) season for these analyses, METHODS where the summer season brackets the period of upwelling at these northerly latitudes (Parrish et al. Temporal recruitment patterns. I censused tidepool 1981).I then used a winter and a summer linear regres- sculpins by draining a tidepool partially, mixing in a sion equation for each species to simulate the size tra- few drops of the anesthet~cQuinaldine, and then dip- jectory for 2 to 3 yr of an individual that recruited in netting any fish that came swimming to the surface, different months during the recruitment season. From while continually pouring water into crevices to flush these s~mulationsI estimated the number of months it out flshes. The fishes were transferred to a bucket with would take for an individual to attain approximate untainted seawater where they rapidly recovered from reproductive size (35 mm SL) when it recruited from any ill effects of Quinaldine. I then drained the remain- May until September. ing water from the pool, thoroughly searched the pool In addition to examining the growth rate of fishes again, and refilled the pool with seawater. that recruited at various times, I also estimated the From September 1990 to September 1993 I quan- probability of surviving to 35 mm SL as a function of tified the numbel- of sculpins that had recruited in recruiting in different months dung the recruitment the previous month by marking and recapturing scul- season. Survivorship rates for juveniles (135 mm SL) pins at approximately l mo intervals in 9 tidepools. I were estimated from the loss of marked fishes during marked fishes by injecting small dots of colored, monthly intervals from the 9 Tatoosh Island tidepools acrylic paint (Ceramcoat, Delta Technical Coatings, over 3 yr I estimated survivorship 2 ways and found Inc., Whittier, CA 90601, USA) subcutaneously on the the results to be concordant (Pfister 1996). The sur- ventral side of the fish just adjacent to the anterior por- vivorship estimates used here are from a log-decay tion of the anal fin. Injections were usually made with function, where individuals were not tallied as dead a 26l~~G needle, although 301/2 C needles were some- unless they failed to ever reappear again in subse- times used for fishes that were <25 mm SL. Sculpins quent censuses. The remarkable site fidelity of these were given 3 color dots which represented the tidepool species resulted in repeated recaptures, sometimes as they were from, the date of their capture and their size much as 3 yr later Using these mean monthly survivor- class. Fishes were a priori divided into 8 size classes ship rates, I estimated the probability that an individ- represented by 1 of 8 colors: <25,26-30,31-35, 36-40, ual recruiting at 20 mm SL would survive to reproduc- 41-50, 51-60, 61-70, and >?0 mm SL. The smallest tive size as a function of recruiting from May through size class (525 mm SL) represents those fishes that September. recruited within the previous month and the smallest Physical correlates of recruitment timing and indi- 3 size classes represent pre-reproduct~ve juveniles. vidual growth and survivorship. I tested whether an Since these fishes tend to recruit and grow in cohorts, I index of upwelling or mean SST ("C) correlated with chose this method over an individual marking method the supply of new recruits to the shore at monthly inter- marEcol Prog Ser 153: 229-238, 1997

RESULTS +- C. globiceps Recruitment patterns and correlates

Fig la shows the pattern of appearance of new individuals 525 mm SL through time for 1991 through 1993 on Tatoosh Island. Since these individuals are probably 51. mo in age (based on individual growth data, see below), Fig la provides a good indication of the number of new recruits on a monthly basis. The abun- m m m m m dance of recruits peaks in the spring/summer 888$+++++?qq,,,,,,q2 - months, with Clinocotfus globiceps generally 7c%2:s17$p?I>~Q>~-$=Q>=S$~~$~~~~~ 18- peaking in May, C. embryum in June and Oligo- precededyears,cottus maculosusthe that peak of 0. recruitmentin maculosus. July and forAugust. C. globiceps In all 3

212 Mean SST and upwelling also cycled season- $iiiyb0) p- ally over this 3 yr period [Fig. l b, c, respectively). - I examined whether temperature or upwelling 8- values correlated with the number of recruits us-

6 r1,,,,,11,1.1,11,,,1,,,,,,,,,,,,,,11,~ ing cross correlation analyses. I specifically

C asked whether there were correlations among 50 - recruitment and these oceanographic events P 0- with lags or leads of up to 3 mo. In other words, -.- a, g -50- were prior temperature and upwelling important I to the magnitude of recruitment? 2-100- macu1osus recruitment was positively correlated UW. C-150- with temperature at no lag and at intervals of 1 or 2 mo earlier (Fig. 2a). maculosus recruit- -200 ...... 0. ment was also positively correlated with temper-

Fig. 1. Oligocottus maculosus, Clinocottus globiceps,and C. ature the month. This pattern is 'Or- embryurn.(a) Pattern of appearance of new individuals S25 mm SL roborated in Fig. lar b where peaks in SST through time for 1991 through 1993 on Tatoosh Island, (b) mean coincide with peaks in 0. maculosus recruit- monthly sea surface temperature, and (cl the monthly upwelling ment, In contrast, Clinocottus globiceps recruit- index (m3 s-' per 100 m of coastline) ment correlated negatively with temperature at periods from 2 and 3 mo earlier (Fig. 2b). C. em- vals for each species. Monthly mean values for SST ("C) bryum recruitment showed no correlations with tem- were obtained from Buoy no. 46041 at 47.4"N and perature (Fig. 2c). 124.S0W, 1.0" to the south and 0.23" to the east of Oligocottus maculosus recruitment was positively Tatoosh Island (or approximately 50 km to the south- correlated with the upwelling index at no lag and at east).This buoy is operated by the National Data Buoy lags of 1, 2 and 3 mo (Fig. 3a).In contrast, upwelling in- Center (Mississippi, USA); data are logged once per dices simultaneous with and 1, 2 and 3 mo later than second during an 8.0 min averaging period per hour Clinocottus globiceps recruitment correlated positively and mean monthly SST values are reported in the (Fig. 3b), although there was also a significant nega- 'Mariner's Weather Log'. Monthly intervals are the ap- tive correlation with upwelling 3 mo earlier There propriate time scale to estimate correlations since this were no correlations between upwelling and C. em- was also the frequency of sampling for fish recruitment. bryum recruitment. These correlations are consistent Upwelling indices for 48"N and 125" W were provided with the patterns seen in Fig. 1: peaks in upwelling are electronically by the National Oceanographic and At- concurrent with both 0. maculosus and C, globiceps mospheric Administration (Pa.cific Fisheries Environ- recruitment, but tend to precede 0 rnaculosus recruit.- ment Group, Pacific Grove, CA). Values are reported as ment and follow peak C. globiceps recruitment. an index in m3 S-' per 100 m of coastline and integrate One concern with doing a large number of correla- over a monthly period using surface atmospheric pres- tive analyses is the expectation that some correlations sure fields to estimate wind-induced coastal upwelling will be significant due only to chance. I used the tem- Pfister: Variation in recruitment 233

a. 0. rnaculosus 0.6 a. 0. maculosus

0.8 l I

-0.8-0.6

-3 -2 -1 0 1 2 3 b. C. globiceps :L0.8 , 9

W 0 c C ernbryurn

-0.6 I Number of lags between variables (mo.) -3 -2 -1 0 1 2 3 Number of lags between variables(mo.)

Fig. 2. (a) Oljgocoltus rnaculosus, (b) Cljnocottus globiceps, Fig. 3. (a) Oligocottus maculosus. (b) Clinocottus globiceps, and (c) C. ernbryum. Cross correlation coefficients between and (c) C. embryum. Cross correlation coefficients between recruitment and mean sea surface temperature (SST). The recruitment and an index of upwelling. The correlation coef- correlation coefficient was estimated contemporaneously (no ficient was estimated contemporaneously (no lag) and using lag) and uslng temperature values obta~ned1 to 3 mo before pressure estimates obtained 1 to 3 mo before or after the or after the recruitment estimate. Asterisks indicate statistical recruitment estimate. Asterisks indicate statistical signifi- significance: 'p < 0.05; "p < 0.01 cance: 'p 0.05; "p < 0.01

perature and upwelling data for 21 correlation tests embryum growth rates were significantly affected by and should thus expect 1 significant result (at p < season, with summer growth rates greater than those 0.05) for each test purely by chance. For correlations of winter (Table 2, Fig. 4). C. globiceps growth rates among upwelling and number of recruits, I found 9 might have been affected by season, although the rela- significant results, while correlations using tempera- tionship was only marginally significant (Table 2, ture showed 8 significant results. Thus, in both cases, Fig. 4) Size and season explained 57 % and 40% of the I found evidence for more significant relationships total variation in growth rate for C, embryum and 0. among these variables than would be expected by maculosus respectively (Table 2). In contrast, only 11 % chance.

Table 1 ANCOVA testing whether individual growth rates Growth rates differ among Oligocott~~smaculosus, Clinocottus globiceps and C. en~bryuniusing slze (mm SL) as a covariate Growth rates differed significantly among species when size was used as a covariate (Table 1).Analyzing Source SS df M S F P species separately indicated that size explained a sig- Species 208.1 2 104.1 25 9 <0.001 nificant amount of the variation for Oligocottus macu- Size 206.9 1 206.9 51.4 <0.001 losus and Clinocottus embryum growth rates, but not Error 1528.4 380 4.0 for C. globiceps (Table 2). 0. maculosus and C. 234 Mar Ecol Prog Ser 153: 229-238, 1997

Table 2. Two-way ANOVAs testing the effect of size (mm SL) a. 0. rnaculosus and season (summer vs winter) on species growth rates

Source SS df MS F P

Oligocottus maculosus, r2 = 0.399 Size 81.9 1 81.9 33.3 <0.001 Season 64.8 1 64.8 26.4 <0.001 Size X Season 25.6 1 23 6 9.6 0.002 Error 373.4 152 2.5 Clinocottus globiceps, ? = 0.105 Size 5.2 1 Season 20.1 1 Size X Season 7.2 1 b. C. globiceps Error 785.9 138

C. embryum, r2 = 0.567 10 Size 43.9 1 Season 7.4 1 Size X Season 5.0 1 Error 87.6 60

of the variability in C. globiceps growth is explained by size and season. The interactmn between slze and sea- son for 0. rnaculosus and to a lesser extent for C. embryum (p = 0.068) indicated that smaller individuals c. C. ernbryum of those species respond more positively to summer 6- - 0 growth conditions. In bther words, adult growth rates 0 5 - 00 benefit less during summer conditions than juvenile . 4- O'QCOO a, growth rates. 3 - 2 - Consequences of timing of recruitment to growth 1 - and survivorship 0 - -1 - Recruitment earlier in the season favored rapid -2 ( l' 'l'I ' I ' I . I' 'l growth and higher survivorship to reproductive size 0 10 20 30 40 50 60 70 in all 3 species (Fig. 5). Using the equations for sum- Initial Standard Length (mm) mer and winter growth in Table 3 (based on the analyses in Tables 1 & 2), the number of months it Fig. 4. (a) Oligocottus maculosus, (b) Clinocottus globiceps, and (c) C. cmbryum. Individual growth rates as a function took for individuals of each species to reach reproduc- of initial standard length in both summer (0) and winter tive size increased as the summer progressed. For (0) months. (- - -) Best linear fit to summer growth rates; example, an Oligocottus maculosus that recruited in (-) best linear fit to wlnter growth rates

Table 3. Linear regressions of growth rate (G, in mm mo-l) as related to size (in mm SL) in both the summer season (Apr-Sep) and the winter season (Oct-Mar) for each of the 3 species

1Species Summer Winter

1~~lgocottus maculosus

Clinocottus globiceps

C. embryum Pfister: Variation in recruitment 235

a. 0.maculosus also declines. For Oligocottus maculosus, recruits in September experience a 87.5 % decline in survivorship to reproductive size compared with individuals that arrive in June. Again, seasonality has a less drastic affect on Clinocottus globiceps: recruits in September experience a 61.9% decline in survivorship to repro- ductive size compared with individuals that arrive in June. The probability of surviving to reproductive size peaks in May for C. enlbryum and declines continu- ously thereafter. May June July Aug Sept + From Fig. 5, one can bracket roughly the best months for future performance (including survivorship and growth) of recruits of each species. Oligocottus maculosus recruits do best from May to July, Clinocot- tus globiceps in May and June and C. embryum in May only.

DISCUSSION

Consequences of recruitment timing n May June July Aug Sept

The timing of peak recruitment can differ between years (Fig la), both affecting the probability that an individual survives to reproduce and determining whether an individual will realize sufficient growth to reproduce in its first winter (Fig. 5). For all species, recruitment earlier in the summer appears to be most beneficial when performance criteria of early growth and survivorship are used. Individuals that recruit later in the summer pay survivorship and growth costs, including a delay in first reproduction of 1 yr. However, one species (Oligocottus maculosus) rarely recruits Fig. 5. (a) Oligocottus maculosus, (b) Clinocottus globiceps. early in the summer, generally not appearing until July and (c) C. embryum. Probability that an individual recruiting or August. Although July recruits would suffer no at 20 mm SL survives to 35 mm SL (approximate size of first reproduction), and length of time it takes an individual to growth or survivorship costs, recruits in August or later attain reproductive size as a function of what month it recruits would. This 'mismatch' between what appears to be in the spring and summer. Survivorship probabilities are best for individuals of 0. maculosus and what I observe based on Pfister (1996). The times required to grow to repro- in the field is difficult to interpret, but may be linked to ductive size are based on the linear regression equations in Table 3 trade-offs in timing with other phases of the life cycle such as adult breeding and larval release. Similarly, Booth (1995) showed that although recruitment and June would be 35 mm SL by September, whereas one enhanced survivorship and growth coincided in the that recruited in September will not reach that size late summer for an Hawaiian coral reef damselfish, until April. Since 0. maculosus can reproduce at late-settling fish were smaller in the winter due to min- 35 mm SL and reproduction occurs in the winter imal winter growth rates and were more vulnerable to months (author's pers. obs.), then 0. maculosus that . recruit in August and September are unlikely to The variability in arrival .of recruits to the reproduce in their first year (Fig. 5a). In contrast, shore may be driven, in part, by variation in oceano- Clinocottus globiceps recruits are likely to be 35 mm graphic factors such as upwelling and SST. Maximum SL by their first winter, unless they recruit in Septem- SST coincided with Oligocottus maculosus recruitment ber, an infrequent event (Fig, la). (Fig. 2a), while peaks in upwelling coincided with re- Because individuals of all species take longer to cruitment in 0. maculosus and Clinocottus globiceps. reach reproductive size when recruitment occurs later Both temperature and upwelling can be characterized in the season, their survivorship to reproductive size by high interannual variation, wlth the timing of the 236 Mar Ecol Proy Ser 153: 229-238, 1997

peak value varying among years. For example, sea- Temporal variation in ecological models sonal maxlmum values for mean monthly SST occurred in July in 1992. during August in 1991 and 1993, and Variability In the temporal dynamics of organism during September in 1990 (Fig. tb). Similarly, the abundance has been incorporated into models that strength and duration of upwelling can vary among predict specles CO-occurrence and persistence (Ches- y~drs,although Fig. lc shows relatively constant peak son 1983, Schmida & Ellner 1984, Warner & Chesson upwelling from 1990 to 1993. Prevlous studies In tem- 1985, Chesson & Huntley 1988). Examination of many perate environments have also shown a strong corre- published recruitment studies indicate that interan- latlon between large-scale oceanographic events and nual recrultment vanation, assumed to be important in recruitment (Parrish et al. 1981, Shanks 1983, Sls- the above models, is widespread in marine taxa (e.g. senwine 1984, Roughgarden et al. 1988). Here, I have Hjort 1914, Loosanoff 1946, Thorson 1950, Coe 1956, shown that variation in these oceanographic events Denley & Underwood 1979, Sale et al. 1984, Caffey can also affect strongly the early llfe history events of 1985, Connell 1985, Shulman 1985, Raimondi 1990, newly recruited individuals. Doherty 1991, Pfister 1996). Temporal varlation wlthin Recru~tment in each species correlated differently a season has also been demonstrated (Loosanoff 1946, with SST and upwelling, with maximum upwelling Yoshioka 1982, Caffey 1985, Gaines & Roughgarden often preceding Oligocottus maculosus recruitment 1985, Raimondi 1990, Doherty 1991, Bingham 1992, and following Clinocottus globiceps recruitment. The Kneib 1993). iriielspeciiic diiierences in col-relaiioils with oceano- Whether this within-seasoi; variation affec:s popula- graphic events and recrultment suggest either: (1) tlon trajectories or species interactions depends also physiologically, these flsh species respond differently upon the nature of post-recruitment processes. I have to upwelling events or temperature; or (2) these spe- demonstrated here that fishes that recruit earller max- cies are distnbuted differently (spatially or temporally) imize the time of growth durlng the more agreeable in the water column as larvae and encounter oceano- summer conditions, reaching reproductive size In their graphic phenomena distinctly. Currently, these hypo- first breeding season, and the extent of this response theses are indistlnguishable, slnce so little is known varies among species. However, these effects may be about the dispersal and distribution patterns of these mlnor compared with events that affect benthic adult larval fishes. performance in tldepools (Pfister 1996). Using slze- Although interannual variability can characterize structured models of population growth m these inter- both oceanographic events and correlated recruitment tidal sculplns, I found that variability in recruitment success, the causal factors driving interannual differ- events was not transmitted reliably to the adult popu- ences in recruitment of these fishes remains unknown. lation. Instead, natural variation in adult survivorship A multitude of causal factors, alone or in concert, may can affect population size to a greater extent than be important, including how oceanographic events recruitment variability, and population growth can be mlght affect the variability in the time of spawning most sensitive to survivorship in the adult stages. (e.g. Secor & Houde 1995),differences in larval growth Other studles of marine flsh demography also suggest and survivorship (e.g. Kneib 1993), and variability in that events during adult stages can strongly affect pop- larval transport (e.g. Shanks 1983). ulation dynamics. For example, a number of studies Clearly, events in the early llfe history and luvenlle have demonstrated interspecific competitive interac- stages of an organism have the potential to influence tions among marine fishes (e.g.Hixon 1980, Sweatman life history strategies (e.g.Stearns 1976) and the long- 1985, Schmitt & Holbrook 1986, 1990, Robertson 1996), term population dynamics of a species (e.g. Lewontin includlng the fish species discussed in this study (Pfis- 1965). Here, I demonstrate that oceanographic factors, ter 1995). includlng seasonality, via their effect on the timing of Although the results reported here indicate oceano- recruitment, influence the early demography of inter- graphic correlates of early demographic events in tidal sculpins. Previously, I showed that competitive marine fishes, these results do not Imply that intertidal interactions among fishes also affect growth and eml- sculpin population dynamics are driven solely by thew gration rates of juveniles (Pflster 1995). We are increas- early events. The extent to which pelagic and recruit- ingly understanding more about the juvenile stages of ment events versus subsequent processes dunng later marine fishes, including how growth and survivorship stages influence population dynamics will depend are influenced by conspeciflcs (e.g. Kneib 1987, Jones upon whether density dependent processes occur, the 1988, Forrester 1990, Booth 1995),predators (Wnght et lifespan of the specles, and the extent to whlch recruit- al. 1993), and physical variables such as temperature ment success and survivorshlp covary (e.g. Benton & (Rutherford & Houde 1995) and tidal cycles (Kneib Grant 1996). In order to integrate pelagic and benthic 1993). events in the life hlstory of marme organisms, we need Pfister Varlati.on in recruitment 237

to understand not only what processes are important in Grossman GD (1982) Dynamics and organizat~onof a rocky different life history stages, but also the net demo- intertidal fish assemblage: the persistence and resilience graphic effects these processes have. of taxocene structure. Am Nat 119:611-637 Hart JL (1973) Pacific fishes of Canada Bull Fish Res Bd Can 180 Acknowledgements. I thank J. T Wootton, P. N~tschke,R T. Hixon MA (1980) Competitive interactions between Califor- Pa~ne,W. P. Sousa and 2 anonymous rcviewcrs for comments nia rccf fishes of the genus En~blotoca. Ecology 61: on the nlanuscript. I am grateful to the Pacific F~sher~csEnvi- 918-931 ronment Group of NOAA for the upwelllng data. ,411 other Hjort J (1914) Fluctuations in the great fisheries of northern data collection was funded by the Lerner Gray Fund for Europe. Rapp P\! Reun Con Int Explor Mer 20:l-228 hlarine Research, Sigma Xi Grants-in-Aid, and NSF grant no. Johanncs RE (1978) Reproductive strategies of coastal marine OCE 91-15760 to R. T Paine. Thc hliller Foundation for Rasic f~shesin the tropics. Environ Biol Fish 3:65-84 Research In Science and the Andrew W. Mellon Founci.~tion Jones GP (1988) Experimental evaluation of the effects of provided support during the preparation of the manuscript. habitat structure and compet~tiveInteractions on juveniles The Makah Tribal Nation permitted access to the study area of two coral reef fishes. J Exp Mar Biol Ecol 123.115-126 and Big Salmon and Eagle Air assured safe transport. Jones GP (1990)The Importance of recrultment to the dynam- ics of a coral reef fish population. Ecology 71:1691-1698 Kneib RT (1986) Size-specific patterns in the reproductive LITERATURE ClTPD cycle of the killlflsh Fundulus heteroclltus (Pisces: Funuli- dae) from Sapelo Island, Georgia USA. Copeia 1986. Babcock RC, Bull GD, Harrison PL, Heyward AJ, Oliver JK, 342-351 M1allace CC, Willis BL (1986) Synchronous spawnlngs of Kneib RT (1987) Predation risk and use of intertidal habitats 105 scleractinian coral species on the Great Barrier Reef. by young fishes and shrimp. Ecology 68:379-386 Mar Biol90:379-394 Kne~bRT (1993) Growth and mortahty in successive cohorts Bcnton TG, Grant A (1996) How to keep fit in the real world: of f~shlarvae within an estuarine nursery. Mar Ecol Prog elasticity analyses and selection pressures on life histories Ser 94-115-127 in a variable environment. Am Nat 147:115-139 Korringa P (1947) Relationship betwecn the moon and penod- Bingham BL (1992) Life histories in an epifaunal community: ;city in the reeding of marine . Ecol Monogr 17- coupling of adult and larval processes. Ecology 73: 347-381 2244-2259 Le'wontin RC (1965)Selection for colon17,ing ability. In: Baker Booth DJ (1995) Juvenile groups in a coral-reef damselfish: HG, Stebh~nsGL (cds) Thr genetlcs of colonizing ability. densits-dependent effects on indlvldual fitness and popu- Arildemlc Press. Ncw York, p 77-92 lation demography. Ecology 76:91-106 Loosanoff VL (1964) Variations in time and intensity of set- Caffey IIhl 11985) Spatial and temporal variation in settle- tling In the starfish. Asterias forbesi, in Long Island Sound ment and recrultment of intertidal barnacles. Ecol hlonogr durlng a twenty-flve year prriod. Biol Bull 126:423-439 55:313-332 Marli~iveJB (1986) Lack of planktonic dispersal of rocky Chesson PL (1983) Coexistence of competitors In a stochastic intertidal fish larvae. Trans Am Fish Soc 115:149-154 environment: the storage effect. Lect Notes B~omath52: McFarland WN, Brothers EB, Ogden JC, Shulman MJ, 188-198 Bermingham EL, Kotchian-Prentiss NM (1985) Recruit- Chesson PL, Huntly N (1988) Community consequences of ment patterns in young french grunts, Haemulon flavolin- life-history tra~tsin a va~iableenvironment. Annls Zoo1 eatum (family Haen~ulidae)at St. Croix, U.S.V.I. Fish Bull Fenn 255-16 C'S 83:413-426 Coe WR (1956) Fluctuations in populations of littoral marine Morgan SG, Christy JH (1995) Adaptive significance of the ~nvertebrates.J Mar Res 15:212-232 timing of larval release by crabs. Am Nat 145:457-479 Connell J (1985) The consequences of variation in initial set- Parrish RH. Nelson CS, Bakun A (19811 Transport mecha- tlement versus postsettlement mortality in rocky intertidal nisms and rcproductive success of fishes in the California communities. J Exp Mar Biol Ecol93:ll-45 Curl-ent. Biol Oceanogr 1:175-203 Llenley EJ, Underwood AJ (1979) Experiments on factors Pfister CA (1993) The dynamics of flshes In intertidal pools influencing settlement, survival, and growth of two spe- Disstrtation, llnivers~tyof Washington, Seattle cies of barnacles in New South Wales. J Exp Mar Biol Ecol Pfister CA (1995) Estimating compet~tloncoeffic~ents from 36:269-293 census data a test with field manipulations of tidepool Doherty PJ (1991) Spatial and temporal patterns in recruit- fishes Am Nat 146:271-291 mcnt. In: Sale PF (ed)The ecology of fishes on coral reefs. Pfister CA (1996) The role and importance of recruitment Academic Press, San Diego, p 261-293 variability to a guild of fishes. Ecology 77: Doherty PT. Fowler T (1994) An empirical test of recruitment 1928-1941 limitation in a coral reef fish. Sciencc 263:935-939 Raimondi PT (1990) Patterns, mechanisms, consequences of Forrester GE (1990) Factors influencing the juvenile demog- vanability in settlement and recruitment of an intertidal raphy of a coral reef fish. Ecology 71:1666-1681 . Er01 Monogr 60:283-309 Gaines S, Roughgarden J (1985) Larval settlement rate: a R~chkusWA (1973) A quantitdtive study of intertidcpool leading determinant of structure In an ecological commu- movement of the wooly sculpin Cl~nocottusanaljs. h4ar nity of the marine intertidal zone. Proc Ndtl Acad S~IUSA Biol 49~277-284 82:3707-3711 Robertson DR (1996)Interspecific competition controls abun- Green JM 11971) H~ght~de movements and hornlng behavior dance and habitat use of territorial Caribbean damsel- of the tidepool sculpin Oligocottus maculosus. J Fish Res fishes. Ecology 77:885-889 Bd Can 28:383-389 Robertson DR, Peterson CW, Brawn JD (1990)Lunar reproduc- Green JM (1973)Ev~dence for hom~nqIn the mosshead sculpin tive cycles of benthic-brooding reef fishes: reflections of (Clinocottus globiceps).J Flsh Res Bd Can 30:129-130 larval biology or adult biology? Ecol Monogr 60:311-329 Mar Ecol Prog Ser 153: 229-238. 1997

Roughgarden J, Games S, Posslngham H (1988) Recruitment Strub PT, Allen JS, Huyer A, Smith RL (1987) Large-scale dynamlcs in complex ltte cycles Science 241.1460-1466 structure of the spring transition in the coastal ocean of Rutherford ES, Houde ED (1995) The influence of tempera- western North America. J Geophys Res 92(C2):1527-1544 ture on cohort-spenfic growth, survival, and recrultment Strub PT, James C (1988) Atmospheric conditions during the of stnped bass, Marone saxahlis, la~aeIn Chesapeake spring and fall transitions in the coastal ocean off western Bay US Natl Mar Ftsh Serv F~shBull 93:315-332 United States. J Geophys Res 93(C12):15561-15584 Sale PF (1977) Maintenance of high divers~tym coral reef fish Sweatman HP (1985) The influence of adults of some coral cornmunlties Am Nat 111 337-359 reef fishes on larval recruitment. Ecol Monogr 55:469-485 Sale PF, Doherty PJ. Eckert GJ. Douglas WA, Ferrell DJ Thorson G (1950) Reproductive and larval ecology of marine (1984) Large scale spatial and temporal variation In re- bottom invertebrates. Biol Rev 251-45 cruitment to fish populatlons on coral reefs. Oecologia 64 Victor BJ (1983) Recruitment and population dynamics of a 191-198 coral reef fish. Science 219:419-420 Schm~daA, Ellner S (1984) Coexistence of plant species wlth Victor BJ (1986) Larval settlement and juvenile mortality in a similar niches. Vegetatto 58 29-55 recruitment-limited coral reef fish population. Ecol Schm~tt RJ, Holbrook SJ (1986) Seasonally fluctuating Monogr 56145-160 resources and temporal variablhty of lntenpeafic compe- Waples RS (1987) A multispecies approach to the analysis of t~hon.Oecologla 69.1-1 l gene flow in marine shore fishes. Evolution 41:385-400 Schmitt RJ, Holbrook SJ (1990) Population responses of surf- Warner RR, Chesson PL (1985) Coexistence mediated by perch released from compet~t~on.Ecology 71.1653-1665 recrultment fluctuations: a field guide to the storage Secor DH, Houde ED (1995) Temperature effects on the hm- effect. Am Nat 125:769-787 tng of Stnped Bass egg production, larval vlabll~ty,and Washington BB, Moser HG, Laroche WA, Richards WJ (1984) recru~tmentpotential in the Patuxent Rlver (Chesapeake : development. Ontogeny and systemat- Bay) Estuaries 18.527-544 ~csoi fishes. Special Publication no. l, h~leticanSociety of Shanks AL (1983) Surface shcks assoc~atedwtth tidally forces Ichthyologists and Herpetologists lnternal waves may transport pelag~clarvae of benttuc Wright RA, Crowder LB, Martin TH (1993) The effects of pre- Invertebrates and f~shesshoreward Mar Ecol Prog Ser 13 dation on the sunrival and size-distribution of estuarine 311-315 fishes: an experimental approach. Environ BioI Fish 36: Shulman MJ (1985) Coral reel flsh assemblages intra- and 29 1-300 interspec~fic compet~tton for shelter s~tes Environ Biol Yoshioka PM (1982) Role of planktonic and benthic factors in Fish 13 81-92 the population dynan~icsof the bryozoan Membranipora Shulman MJ, Ogden JC. Ebersole JP, McFarland WN, MllIer membranacea. Ecology 63:457-468 SL. Wolf NG (1983) Pnority effects In the recrultment of Yoshiyama RM, Gaylord KB. Puppart MT. Moore TR, Jor- juvenlle coral reef fishes. Ecology 64 1508-1513 dan JR, Coon CC, Schalk LL, Valpey CJ, Tosques I (1992) S~ssenwmeMP {1984) Why do fish populat~onsvary? In. May Homing behavior and site fidelity in intertidal sculpins RM fed) Explo~tation of marme communlues. Springer- (Pisces: Cottidae). J Exp Mar Biol Ecol 161): 115-130 Verlag. Berlm, p 59-94 Yoshiyama RM, Sassaman C (1987) Geographical patterns of Stearns SC (1976) Llfe-history tactics a review of the Ideas Q allozyrnic variation in three species of intertidal sculpins. Rev Biol 51:3-47 Environ Biol Fish 20:203-218

Th~sart~cle was presented by K Sherman (Sen~orEdjtonal Manuscript first received; June 19, 1996 Ad v~sor),Narragansett, Rhode Island, USA Revised version accepted-Apr118. 1997