Linking growth to environmental histories in central Baltic young-of-the-year sprat, Sprattus sprattus: an approach based on otolith microstructure analysis and hydrodynamic modelling

of 12

Please download to get full document.

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
12 pages
0 downs
Otolith microstructure analysis and hydrodynamic modelling were combined to study growth patterns in young-of-the-year (YoY) sprat, Sprattus sprattus, which were sampled in October 2002 in the central Baltic Sea. The observed ‘window of survival’,
  Linking growth to environmental histories in central Balticyoung-of-the-year sprat,  Sprattus sprattus : an approach basedon otolith microstructure analysis and hydrodynamicmodelling HANNES BAUMANN 1, *, HANS-HARALDHINRICHSEN, 2 RUDI VOSS, 2 DANIELSTEPPUTTIS, 2 WLODZIMIERZ GRYGIEL, 3 LOTTE W. CLAUSEN 4 AND AXEL TEMMING 1 1 Institute for Hydrobiology and Fisheries Science, Olbersweg 24,22767, Hamburg, Germany 2 Leibniz Institute of Marine Science, Du¨ sternbrooker Weg 20,24105, Kiel, Germany 3 Sea Fisheries Institute, ul. Kołła˛taja 1, 81-332, Gdynia, Poland 4 Danish Institute for Fisheries Research, Charlottenlund Castle,DK-2920, Charlottenlund, Denmark ABSTRACT Otolith microstructure analysis and hydrodynamicmodelling were combined to study growth patterns inyoung-of-the-year (YoY) sprat,  Sprattus sprattus , whichwere sampled in October 2002 in the central BalticSea. The observed ‘window of survival’, approximatedby the distribution of back-calculated days of firstfeeding (DFF), was narrow compared to the extendedspawning season of sprat in the Baltic Sea (mean± SD  ¼  22 June ± 14.1 days) and indicated that onlyindividuals born in summer survived until October2002. Within the group of survivors, individuals bornlater in the season exhibited faster larval, but morerapidly decreasing juvenile growth rates than earlierborn conspecifics. Back-calculated  larval  growth ratesof survivors (0.48–0.69 mm day ) 1 ) were notablyhigher than those previously reported for average lar-val sprat populations, suggesting that the YoY popu-lation was predominantly comprised of individualswhich grew faster during the larval stage. Daily meantemperatures, experienced across the entire YoY pop-ulation, were derived from Lagrangian particle simu-lations and correlated with (1) detrended otolithgrowth and (2) back-calculated, daily somatic growthrates of survivors. The results showed that abruptchanges in ambient temperature can be detected inthe seasonal pattern of otolith growth, and that highertemperatures led to significantly faster growththroughout the entire age range of YoY sprat. Key words : Baltic sprat, hydrodynamic modelling,otolith microstructure analysis, temperature-dependent growth, young-of-the-year INTRODUCTION The Baltic sprat,  Sprattus sprattus , has become themost abundant, commercially exploited fish species inthe Baltic Sea (ICES, 2004). Over the past decade, notonly absolute recruitment strengths but also inter-an-nual recruitment variability have reached historicallyhigh levels (ICES, 2004). Although the positive trendin sprat stock abundance is generally explained by aregime shift from a cod-dominated to a clupeid-dom-inated system (Ko¨ster  et al. , 2003), the recent, strongfluctuations in recruitment success remain insuffi-ciently understood.Previous field studies in the Baltic Sea haveinvestigated the sprat egg and larval stages (e.g. Voss,2002; Nissling  et al. , 2003), because they are consid-ered critical for recruitment variability (Ko¨ster  et al. ,2003; MacKenzie and Ko¨ster, 2004). However, aninherent limitation of larval studies is that averagecharacteristics (e.g. growth rate, hatch date) are un-likely to reflect the small fraction of individuals thatwill eventually survive until recruitment (Sharp,1987). Therefore, a better understanding of processesinfluential to sprat recruitment may require comple-mentary studies that describe characteristic traits insuccessful sprat recruits.Otolith microstructure analysis is a promising toolto accomplish this, as it provides a wealth of infor-mation about present and past characteristics in larvaland juvenile fish (Stevenson and Campana, 1992). Itstwo prerequisite assumptions, daily periodicity of increment deposition and coupled otolith and somatic * Correspondence . e-mail: hannes.baumann@uni-hamburg.deReceived 21 February 2005Revised version accepted 20 July 2005  FISHERIES OCEANOGRAPHY  Fish. Oceanogr. 15:6, 465–476, 2006   2006 The Authors. doi:10.1111/j.1365-2419.2005.00395.x 465  growth, have been confirmed for sprat, at least afterthe onset of exogenous feeding (Alshut, 1988; Shields,1989). Yolk-sac sprat larvae, on the other hand, do notseem to deposit regular micro-increments. Thus, forsprat survivors, otolith microstructure analysis can beused to back-calculate the day of first feeding (DFF) asa proxy for hatch day (Valenzuela and Vargas, 2002)and to reconstruct age- and day-specific histories of otolith growth as a proxy for somatic growth. Thesedata may later serve as a basis for comparing charac-teristics of survivors with those of non-survivors sam-pled earlier in the season (e.g. Allain  et al. , 2003).The present paper focuses on sprat survivors thatwere caught in October 2002 as young-of-the-year(YoY) or 0-group individuals in the Bornholm Basinand adjacent coastal areas (Fig. 1). The BornholmBasin has been shown to be an important spawningground for sprat in the central Baltic (Ko¨ster  et al. ,2001), where the spawning season usually extendsfrom March to August (Elwertowski, 1960). Sprat eggsand newly hatched larvae develop in intermediatewater depths of 45–65 m (Wieland and Zuzarte, 1991),whereas feeding sprat larvae typically occur in surfacewaters (Voss, 2002). Depending on the circulationpattern, larvae and juveniles may either be retained inthe basin or advected into shallower nursery grounds,where they likely stay until the end of the growingseason. Later, YoY sprat are thought to join adultschools during their over-wintering migration into thedeep basins (ICES, 2004).One of the main factors influencing growth andsurvival of larval and juvenile fish is ambient tem-perature (Heath, 1992). Given the extended spawningseason of sprat, it is likely that fish hatching at differenttimes of the year will experience considerably differenttemperature conditions. To understand how tempera-ture affected the past growth of survivors, it is necessarytoreconstructtheirtemperaturehistoriesandlinkthemto growth patterns inferred from otolith microstructureanalysis. The most obvious approach, i.e. taking directtemperature measurements from field surveys, is typic-ally hampered by the relatively low temporal resolutionof survey data and the uncertainty about the spatialdistribution of individuals at a given time.An alternative approach is to use a realistic hydro-dynamic circulation model, where the average drift of larval and juvenile cohorts is simulated by means of passive Lagrangian particles seeded into the model do-main. Apart from the spatial distribution, such modelsare also able to provide temperature data associatedwith particles on a daily (or even hourly) basis. La-grangian studies have been used successfully to identifyspawning and nursery grounds (e.g. Allain  et al. , 2003;Hinrichsen  et al. , 2003), to develop coupled biophysi-cal IBMs (e.g. Werner  et al. , 1996), or to reconstructdaily environmental histories at the level of individualfish (e.g. Baumann  et al. , 2003). In the present ap-proach, however, Lagrangian particles are not used toinvestigate individual drift patterns but to derive aspatially integrated, average index of daily temperatureconditions, which the majority of survivors potentiallyexperienced throughout their first months of life.By applying these approaches, the main goals of thisstudy were first to determine the potential ‘window of survival’ for YoY sprat in 2002 based on the back-calculated dates of first feeding and, secondly, tocombine otolith analysis with hydrodynamic model-ling in order to describe environmentally influencedchanges in the growth of survivors. MATERIAL AND METHODS Field sampling and otolith analysis Young-of-the-year sprat were sampled in October 2002from two research vessels participating in the annualBaltic International Acoustic Surveys (BIAS). From18 to 23 October 2002, 15 hauls were conducted bythe RV  Baltica  (Poland) in central and southern partsof the Bornholm Basin (Fig. 1), using a standard  Figure 1.  Study area with bathymetry shading (inset BalticSea) and positions of YoY-sampling sites between 7 and 23October 2002. Numbers refer to otoliths randomly selectedfor analysis. Grey dots show seeding positions of 618Lagrangian drifters used in the hydrodynamic circulationmodel. AB, Arkona Basin; B, Bornholm Island; BB,Bornholm Basin; GD, Gdansk Deep.466  H. Baumann  et al.   2006 The Authors,  Fish. Oceanogr. ,  15:6,  465–476.  pelagic trawl net with a 22-mm mesh opening in thecod end. On 7 and 8 October 2002, three hauls withcatches of YoY sprat were conducted by the RV  Argos (Sweden) in northern areas of the Bornholm Basin(Fig. 1) using a similar trawl type. From each haul, upto 20 YoY sprat were randomly selected and immedi-ately preserved in 95% ethanol. The preservative wasreplaced approximately 3 weeks after sampling.Prior to otolith extraction, sprat were individuallymeasured (nearest millimetre) for total (TL) andstandard length (SL) and assigned a unique identifi-cation number. From each specimen, both sagittalotoliths were removed and mounted individually onmicroscopic slides with a drop of Crystal Bond  thermoplastic glue (Structure Probe Inc., WestChester, PA, USA). All otoliths were subsequentlyground from one side and, after re-heating and repo-sitioning, from the other side with a 3- l m lapping film(266 ·  Imperial PSA, 3M Deutschland GmbH, Neuss,Germany) until all increments were sufficiently vis-ible. Irrespective of left or right, the otolith with themost distinct increments was chosen for analysis.Measurements took place under 400 ·  magnificationwith a digital camera (Leica  DC300, 3132  ·  2328pixels, Leica Camera AG, Solms, Germany) connec-ted to an image analysis system ( IMAGEPRO  Plus 4.5.1,Media Cybernetics Inc., Silver Spring, MD, USA)allowing for a theoretical resolution of 0.078  l m pix-el ) 1 . All increments were measured along the sameaxis from core to post-rostrum of the otolith (Fig. 2).Depending on the size of the otolith, this axis wascovered by up to four consecutive and overlappingimage sections. Each section was photographed four toseven times in different focal planes that were latermerged into a single multi-frame image. Multi-frameimages (.tiffs by default) are simple but effective meansto digitally ‘focus’ through an otolith’s microstructureduring measurements, which generally improved theconfidence of interpretation. During measurements,the reader also judged the quality of each interpretedotolith section according to a scale from 1 (best) to 5(worst), and the worst nine otoliths (  10%) were laterexcluded from the analysis.A subset of 48 randomly selected otoliths was readtwice by the same reader to estimate precision usingthe coefficient of variation (CV) method (Campana,2001). Mean CV across all re-read fish was 2.7%, and10 specimens were later excluded because age esti-mates differed by more than 5% from the mean. Whenavailable, the second reading was always preferred overthe first, assuming a learning curve. In total, 102 out of 121 read otoliths were used in the analysis. Precisionin increment counts was also assessed between twoexperienced but independent readers on another oto-lith subset, which consisted of sprat juveniles caught atvarious occasions and stages (  N  ¼  18). The significantlinear regression ( P  < 0.001) between the two read-ings explained 94% of the overall variability and had aslope of 0.92, which was not significantly differentfrom 1 (95% confidence  ¼  0.8–1.04). The mean CVacross all independently read fish was 4.2%.In the majority of YoY otoliths, a pronounced shiftfrom weak and indistinct to sharp and well-definedincrements was typically observed after the innermostfour increments (see results). This shift was assumed tocorrespond to the transition from a non-daily incre-ment formation during the yolk-sac stage to dailyincrements deposited after the onset of first feeding,and only the latter increments were included in theanalysis. Therefore, otolith-derived age estimates pre-sented in this study refer to individual ages in daysafter first feeding (DFF).Daily somatic growth rates (SGR) of YoY sprat werederived from individually back-calculated lengths-at-age estimated with the ‘biological intercept method’(Campana, 1990), where the otolith radius at firstfeeding corresponded to the distance between the coreandthefourthincrement.Forthestandardlengthatfirstfeeding a value of 5 mm was assumed, based on Voss et al. (2003)whofoundpreyinthegutsof4–6 mmspratlarvae from the Bornholm Basin. The biological inter-  Figure 2.  Polished sagittal otolith of a juvenile sprat aged84 days after first feeding (DFF). All increments werecounted and measured along the same axis from core to post-rostrum (arrow). Linking growth to environmental histories in YoY sprat  467   2006 The Authors,  Fish. Oceanogr. ,  15:6,  465–476.  cept method is independent of the slope in the otolith–fish size regression, but assumes linearity in individualotolith–fish size trajectories (Campana, 1990).Because mean and variance of increment widthsvary with age (Pepin  et al. , 2001), we used age-detrended otolith data to analyse seasonal patterns inotolith growth of the entire YoY population. Individ-ual increment widths were standardized to zero meanand unit deviation as in Baumann  et al.  (2003) andrepresent the daily growth anomaly of a specimen at agiven age relative to the population. Hydrodynamic model Likely experienced temperatures of YoY survivors werederived from the Baltic Sea circulation model of Leh-mann (1995) and Lehmann and Hinrichsen (2000),which is based on a free surface Bryan–Cox–Semtnermodel (Killworth  et al. , 1991). The model domainencompasses the entire Baltic Sea with a realistic bot-tom topography. The horizontal resolution is 5 km, avalue corresponding to approximately half the internalRossby radius inthe BalticSea (Fennel,1991), which isnecessary to fully resolve mesoscale motions (e.g. ed-dies). Sixty vertical levels are specified with a thicknesschosen to best represent the different sill depths in theBalticSea.Themodelwasinitializedin1979withmeantemperature and salinity fields and forced by actualmeteorological data, available at the Swedish Mete-orological and Hydrological Institute (SMHI, Norrko¨-ping) for a time series of 24 yr (1979–2002). Simulatedthree-dimensional velocity fields extracted from thecirculationmodelwerethenusedtoderivedriftroutesof Lagrangian particles seeded into the model domain.Along these trajectories, the model provided tempera-ture data for particles at 6-h time steps.Particles were released into the model to simulatethe average temperature history of the YoY populationfrom the DFF until catch in October 2002. Importantassumptions of the approach include that all survivorscaught in and around the Bornholm Basin srcinatedfrom this area, and that integrated particle drifts werereflective of the average transport pattern experiencedby the YoY cohorts, at least on a broad spatial scale.All particles were released inside the 40-m isobath of the Bornholm Basin (Fig. 1), based on average sprategg distributions (Ko¨ster, 1994) that were assumed aproxy for the spatial distribution of first-feeding spratlarvae. All drifters were seeded and forced to remainwithin the 5–10 m depth layer, because feeding spratlarvae predominantly occur in surface waters and ap-pear not to migrate vertically (Voss, 2002; STORE,2003). Five particle cohorts were released between 31May (day 151) and 10 July 2002 (day 191), corres-ponding to the DFF distribution of YoY survivors thatwas back-calculated from otoliths (see results). Each of these five larval ‘pulses’ consisted of 618 particles,which were seeded in regular spatial intervals of about5 km. Depending on the seeding date, drifters weretracked through the model domain for a period of 105–145 days, until all positions were finally recorded on23 October (day 296). For each release date, likelyexperienced temperature was estimated as the dailymean across all 618 particles, irrespective of individualhorizontal positions in the model domain. The fivetemperature curves obtained, however, were not sig-nificantly different between drifter cohorts 1–3 (re-leased 31 May, 10 and 20 June,  P  ¼  0.145) and 4–5(released 30 June, 10 July,  P  ¼  0.53) and were there-fore pooled. The average daily standard deviation of temperature across all 618 particles was 1.1  C.Daily temperatures were assigned to daily back-calculated SGRs of individuals according to their ageafter DFF. All individuals with DFFs prior to 30 Junewere assigned to temperatures from averaged particleruns 1–3, whereas all other specimens were related toparticle simulations 4–5. For example, SGR on day 30of an individual with DFF on 25 June was assigned totemperature on 25 July, taken from averaged particlecohorts 1–3. It should be noted that daily modelledtemperatures have the character of an index, which isassumed to reflect the temporal but not the spatialvariability in environmental conditions experiencedby YoY sprat from the Bornholm Basin. A back-tracking of individual fish from catch to the day of firstfeeding in order to resolve the spatial differences wasnot attempted, given the size and age of YoY sprat inOctober 2002 (>7.5 cm TL, see results). RESULTS  Age, length and growth of YoY survivors In October 2002, YoY sprat formed a distinct peak inthe length distribution of trawl catches with highestfrequencies occurring in the 9–9.5 cm TL class(Fig. 3). Age 1+ sprat were clearly recognizable bylengths exceeding 11 cm TL. Abundance of YoY spratwas high only in the southern parts of the BornholmBasin, in contrast to trawls in the northern areascontaining very few individuals <11 cm TL (Fig. 3).Because length frequencies of YoY sprat selected forotolith analysis compared well to catch length distri-butions (mean ± SD TL oto  ¼  9.23 ± 0.48 cm,mean ± SD TL trawl  ¼  9.17 ± 0.46 cm,  P  ¼  0.18), theanalysed sub-sample was considered representative forYoY sprat caught in October 2002 (Fig. 3). 468  H. Baumann  et al.   2006 The Authors,  Fish. Oceanogr. ,  15:6,  465–476.  After grinding and polishing, daily otoliths incre-ments could be identified relatively well (e.g. Fig. 2),although certainty of interpretation differed along themeasurement axis. First increments were found at adistance between 5 and 9  l m from the core(mean ± SD  ¼  7.9 ± 1.17  l m), usually after an iri-descent check mark that possibly corresponded to thehatch day. This check was typically followed by four(range 3–5) weak and indistinct structures, after whicha pronounced shift occurred to clear and unambiguousincrements that were interpreted as daily rings formedafter the onset of exogenous feeding. These incrementsincreased rapidly in width and were broadest between25 and 55 days after DFF with mean ± SD widths of 7.9 ± 1.25  l m, followed by steadily decreasing incre-ment widths until sampling in October. Meanwidth ± SD of the last three full increments prior tocatch was 1.6 ± 0.55  l m (Fig. 4).Mean ± SD age of YoY sprat was 119 ± 14.1 daysafter DFF corresponding to a mean date of first feedingon 22 June 2002 (Fig. 4). Mean DFFs were not signi-ficantly different between individuals sampled in thenorthern and southern areas of the Bornholm Basin(DFF ± SD north  ¼  24 June ± 16.5 days, DFF ±SD south  ¼  21 June ± 12.1 days,  P  ¼  0.38). Withinthe limited age and length range sampled, significantlinear relationships ( P  < 0.01,  N  ¼  102) were foundfor age versus length (TL  ¼  0.26 ± 61.5 age,  R 2 ¼ 0.59) and mean somatic versus mean otolith growth,estimated as the average increase in SL (SG) andotolith radius (OG) per day (SG SL  ¼  0.12 ± 0.10 OG, R 2 ¼  0.70). Residuals of the latter relationship did notshow significant deviations from linearity, indicatingan isometric otolith size–fish size relationship.When split into five cohorts of fortnightly DFFintervals (Fig. 4), YoY survivors showed markedlydifferent patterns of otolith growth. During the first25 days after DFF (assumed to correspond to the larvalstage), increments of individuals born later in theseason increased more rapidly in width than those of earlier born conspecifics. As juveniles, later born YoYsshowed more rapidly decreasing increment widthscompared to earlier born individuals (Fig. 4). Signifi-cant differences in increment width between DFFcohorts 2 and 5 (25 May–7 June and 6–19 July,respectively) were found for ages 4–27, 42–50 and>74 days after DFF (one-way ANOVA per increment, P  < 0.05). Later born survivors also tended to havegreater maximum increment widths than earlier bornindividuals (Fig. 4).Corresponding to the rapid initial increase inincrement width, average SGR (±SD) across all sur-vivors increased from 0.23 ± 0.07 mm day ) 1 on day 1to 0.92 ± 0.17 mm day ) 1 on day 25 after DFF,  Figure 3.  Comparison of sprat length distributions betweentrawl catches done by the RV  Baltica  in the southern half of the Bornholm Basin (15 hauls, 18–23 October 2002) andthose done by RV  Argos  in the north (three hauls, 7–8October 2002) and of YoY sprat selected for otolith micro-structure analysis.  Figure 4.  Distribution of back-calcula-ted dates of first feeding (DFF) for YoYsprat caught in October 2002 in thecentral Baltic Sea (bars) and meanincrement widths of these survivorsgrouped into five fortnightly periods(lines). Linking growth to environmental histories in YoY sprat  469   2006 The Authors,  Fish. Oceanogr. ,  15:6,  465–476.
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks