5.1 Estimation of predation on early life stages of cod and sprat by planktivorous fish
Introduction
Subtask 5.1 addresses the hypotheses that: 1) cod recruitment in the Baltic is presently controlled by sprat and herring through predation on early life stages and 2) cannibalism on eggs and larvae is a controlling factor for sprat stock development. To test these hypotheses we conducted following activities:
a) evaluate available sprat and herring diet composition databases with respect to ichthyoplankton predation,
b) perform a new stomach sampling and analysis programme covering different month of the spawning season as well as different areas,
c) estimate daily rations of sprat and herring based on evacuation experiments and diel sampling programmes,
d) resolve predator abundances in cod and sprat spawning areas during spawning time,
e) compare daily consumption rates by predator populations to standings stocks and production estimates of prey.
Based on stomach content analyses of sprat and herring, individual daily predation rates on cod and sprat eggs and larvae were updated for years 1997 to 1999. A discrepancy between predator stock sizes derived by spatially disaggregated MSVPA runs and hydroacoustic surveys was observed, hampering the estimation of sprat population consumption. Thus, an exercise was conducted to estimate the sprat population size in the Bornholm Basin by using the daily egg production method. This exercise lead to the estimation of alternative predator population sizes used in the quantification of the predation pressure on early life stages by sprat and herring.
Results and Discussion
To estimate the sprat predator stock size by means of the egg production method, egg abundance, stock structure, sexual and gonadal maturation, spawning frequency and the batch fecundity of sprat throughout the sprat spawning season were simultaneously investigated to obtain an estimate of the population size at spawning time. The results confirmed the population estimate from the hydroacoustic survey in May/June 1999, but contrasted spatially down-scaled results from an area dis-aggregated MSVPA runs.
Clear interannual differences were encountered in the amount of sprat eggs in sprat stomachs, with a reduction by one order of magnitude from 1990-1992 to 1993-1999, although considerable quantities were available in the plankton. For herring, a similar tendency of decreasing amounts of fish eggs in the diet is obvious for spring and early summer dates, however, far less pronounced. An explanation for the different feeding behaviour in the beginning of the 1990’s is a changed vertical overlap between predator and prey,
due to significant alterations in the hydrographic regime of the Bornholm Basin after the major inflow in early 1993 influencing both predator and prey.
The observed decline in predation on cod eggs by sprat in spring and early summer during most recent years is more difficult to explain. A similar change in the vertical overlap of predator and prey due to hydrographic conditions, as concluded for sprat eggs, alone does not explain the decline in cod egg consumption by sprat. Variations in the abundance of other fish eggs, first of all sprat being most abundant, are likely to contribute to the reduction in cod egg predation by sprat as these may act as a trigger for switching from zooplankton to fish eggs as prey.
Comparing consumption by the sprat population to sprat egg abundance suggests cannibalism to be an important source of sprat egg mortality in the Bornholm Basin, while consumption by herring is of minor importance. For cod eggs, predation pressure by sprat reduced during the 1990s caused by the shift of cod spawning time into summer month, as sprat leave the central basin after finalizing their spawning activity in July. Additionally, remaining sprat are distributed more shallow in the water column selecting cladocerans as prey in the intermediate water instead of copepods and cod eggs within and below the halocline. However, herring returning from their spawning to their open sea feeding grounds in June/July, not showing the same vertical distribution, replaced sprat partly as predator of cod eggs, still exaggerating a substantial predation pressure. This result is independent of the revision of the predator population size, as major deviations were encountered in sprat and not in herring population estimates.
Cod and sprat larvae were obviously not substantially affected by herring and sprat predation, independent of the area and date of investigation. This can be partly explained by a limited vertical overlap of prey and predator and partly by a size selective predation by herring preying nearly exclusively on larger larvae. Only newly hatched larvae concentrate in relatively high quantities within or below the halocline, where they are available as prey to herring and sprat aggregating in these water layers while feeding (but note the change of sprat vertical distribution in summer enhancing the vertical overlap with larger larvae).
5.2 Extend the time series of predation mortalities and recruitment of cod and sprat as well as estimates of spawning stock sizes, structures and fishing mortalities
Introduction
Before STORE, the time series of stock sizes of cod and sprat derived by the MSVPA covered the time period 1977-1996. This limited the examination of variations in stock and recruitment relationships due to environmental and species interaction processes as well as fisheries actions, especially as years with outstanding high recruitment were excluded, e.g. for cod 1976. Thus, multispecies assessment data were compiled to cover the time span 1974-1998, and in a second work step revised and updated to 2000. Based on these extended data sets, and updated consumption rates based on revised ambient temperatures an MSVPA run covering 1974-2000 was performed to supply modelling activities in Task 6 with stock abundance, stock structure as well as predation and fishing mortality rates of cod and sprat in the Central Baltic.
An examination of cod stomach content variability in different quarters, years and Sub-divisions in relation to hydrographic conditions, nutritional condition, gonadal maturation and prey availability was conducted using long-term biological data sets (1967-1990).
Results and Discussion
Spawning stock biomass of Eastern Baltic cod derived by the MSVPA run shows a pronounced increase from 1977 to 1980, remaining on a high level during the first half of the 1980s, afterwards declining to lowest level on record in 1992, showing a restricted intermediate increase in the mid 1990s being presently close to the historic minimum. This is well in agreement with the respective estimates from single species VPA.
Higher deviations between standard and multispecies SSB estimates are obvious for the beginning of the 1980s. These differences are caused by lower mean weight at age in the stock applied in the MSVPA runs, as derived stock numbers are rather similar for age-groups 2+. To get a further indication of the reliability of the MSVPA, abundance indices of age-group 2+ from the international bottom trawl surveys were correlated to corresponding MSVPA estimates showing a high correlation. Repeating the exercise for recruitment estimates at age 2 showed a good agreement between MSVPA and singlespecies output. Fishing mortality rates determined by MSVPA and the standard assessment show similar time pattern, with the single species assessment estimating in general slightly higher values. Predation mortalities of 0-, 1- and 2-group cod are in the same order of magnitude than derived by earlier MSVPA runs.
The estimated spawning stock biomass of sprat showed a pronounced decline from the mid 1970s to the early 1980s, a trend that is slightly less pronounced in the standard assessment. Deviations in sprat recruitment estimates are apparent especially for the early years of the time series. Correlating the MSVPA derived sprat stock size (age-group 1+) with the abundance estimate from the Latvian/Russian hydroacoustic survey in Subdivisions 26 and 28, revealed a very close agreement. The MSVPA derived fishing mortality rates follow rather well the general trend in F estimates from the standard XSA. Predation mortalities of sprat showed a continuous decline from mid 1970s to early 1990s being rather constant afterwards.
Spawning stock biomass estimates of Central Baltic herring derived by the MSVPA run show a continuous decline, which is to a large extend caused by reduction in weight at age. A high variability is indicated by the hydroacoustic estimate of age group 1+. In contrast to cod and sprat the comparison of stock size estimates from MSVPA and hydroacoustic surveys shows more variability with the hydroacoustic survey indicating a more pronounced peak of herring abundance in early 1990s due to technical and area coverage problems.
Recruitment at age 1 derived by the MSVPA does not show a pronounced peak in the early 1990s, but rather a declining trend since the early 1980s. In contrast to the good agreement in abundance estimates, the SSB shows larger deviations between the multispecies and the standard assessment. The reason is a difference in mean weight at age, i.e., the MSVPA uses the mean weight in the 1st quarter, whereas the annual mean weight was applied in the standard assessment. The estimated fishing mortality rates obtained from MSVPA and standard assessment are rather similar both indicating a doubling of fishing mortality from early to late 1990s. Predation mortality follows closely the time trend described for sprat. However, a substantial difference between the species is, that predation mortalities of adult herring is very low, reaching seldom 0.1 per year.
Conducted investigations confirmed a relationship between energetic condition of cod and food supply. The liver condition index of cod and abundance of clupeids in the diet of cod during autumn/winter and spring
was positively correlated. Differences in the diet according to sex- and maturity, liver weight and utilization of energy for gonadal growth and spawning were observed. Ripening females on the spawning grounds were hardly feeding on clupeids. The liver condition index of females increased with maturation stage until peak spawning. Feeding intensity and liver condition index of maturing males decreased earlier than of females at the final gonad ripening stage. Cod of both sexes nearly ceased feeding when being in spawning condition with the liver condition index declining.
In 1967-1977, at favourable oxygen/temperature conditions, the gonadal development in pre-spawning time (January) was closely related to gonadal maturation during the main spawning season (March-April). During the 1980s the spawning population of cod was influenced by poor hydrographic conditions - low oxygen saturation and temperature, with gonadal development in pre-spawning time largely decoupled from the situation at spawning time. Changes in distribution and age composition of the spawning population of cod are associated with a decline in proportion of males on the spawning grounds. The diverging developments in the proportion of maturing cod in pre-spawning and spawning seasons indicated that hydrographic conditions likely were the cause of the delay in gonadal maturation during the stagnation period in the 2nd half of the 1980s and the 1990s.
5.3 Develop area dis-aggregated estimates of recruitment, related juvenile predation mortalities as well as spawning stock sizes and structures
Introduction
The primary objective of this Subtask was to define regional reproductive success in relation to size and structure of the originating spawning population, explicitly considering species interactions, i.e. cod cannibalism and predation on clupeids as well as area specific exploitation by the fishery. The approach to conduct spatially dis-aggregated MSVPAs performed in CORE has been validated. Trends in population sizes from research surveys showed a good agreement with the MSVPA estimates for cod and sprat, but not for herring. The relative horizontal distribution between areas determined by MSVPA runs and research surveys, revealed indications of drift of early life stages and migrations of adult cod and sprat. Migration might influence the estimates derived for the different stock components, but does not affect the overall stock sizes and key parameters estimated. This is obvious from integrating the results of the MSVPA runs over the different sub-areas, which resulted in close agreement with the standard assessments for the Central Baltic.
Based on these findings the approach appears to be suited for estimating regional spawning stock sizes and produced recruitment, especially as this procedure is at present the only feasible way of assessing the Baltic fish stocks spatially dis-aggregated. Consequently, area dis-aggregated MSVPA runs were conducted for Sub-divisions 25, 26 and 28 containing the major spawning areas of the Central Baltic cod and sprat, utilizing the revised and extended multispecies assessment data covering 1974-1999 (Subtask 5.2).
An investigation of the feasibility and necessity of an implementation of a statistically based spatial multispecies framework in the Baltic has been handled within the modelling section 6.1 in connection to the implemented modifications of the MSVPA and the developed multispecies stock production model.
Results and Discussion
The performed area dis-aggregated MSVPA runs confirmed distinct trends in population abundance, spawning stock biomass, recruitment and predation mortalities of cod, herring and sprat in different areas of the Central Baltic. A comparison of cod, sprat and herring SSB and recruitment derived from area aggregated, dis-aggregated MSVPA and standard assessment revealed rather similar results in absolute estimates as well as time trends. This confirms the finding that assessing the cod and sprat stock components separately does not change the perception derived by the standard stock assessments for the entire Central Baltic, independent whether migration between different Sub-divisions occurs.
A comparison of direct stock abundance information from surveys to multispecies results revealed again a good coherence in time trends for cod and sprat, but not for herring. Correlations between cod recruitment at age 1 derived by trawl survey and MSVPA revealed highly significant relationships for Sub-division 26 and 28, but not for 25. This is interpreted as evidence, that recruits originating from Sub-division 25 inhabit nursery areas located in Sub-division 26 (see Subtask 1.1) and explains also why recruitment indices derived by the survey were higher in the latter area, although the spawning stock was in general lower and environmental conditions for cod egg survival were unfavourable (see Subtask 2.3).
For sprat the contribution of Sub-division 25 to the combined stock in the Central Baltic was found to be considerably higher in the MSVPA output than in the October hydroacoustic surveys, while in Sub-division 28 the situation is opposite. Hydroacoustic surveys in May/June (see Subtask 1.1) showed highest sprat concentrations in the Bornholm Basin as well, thus fitting to the distribution pattern obtained by MSVPA, but being less pronounced. This may indicate a different distribution of sprat during spawning and feeding periods, potentially caused by migrations between Sub-divisions (see Subtask 1.1). The conducted comparison between spring and autumn hydroacoustic surveys is, however, based on material sampled in the 1980’s, a period of low sprat stock size. Indications exist that this seasonal distribution pattern may have changed with increasing sprat stock sizes (see Subtask 1.1). Comparing recruitment of age-group 0 determined by MSVPA and hydroacoustic surveys revealed significant relationships in all areas, serving as an independent test of determined recruitment pattern. Deviations in absolute recruitment strength are, however, obvious when comparing age-group 1 estimates derived by MSVPA and hydroacoustic survey in autumn. Here a similar trend with higher recruitment in Sub-division 25 estimated by the MSVPA and a higher recruitment determined by the hydroacoustic survey in Sub-division 28 was encountered, suggesting similar drift processes as in cod.
Application of the derived area specific results in stock recruitment models under Task 6 depended critically on the behaviour of recruits when attaining sexual maturity, i.e. whether they show homing behaviour or conduct spawning migrations in dependence of hydrographic conditions. Analyses conducted under Subtask 1.1 revealed indications that this assumption is fulfilled at least for cod.
In contrast to cod and sprat hydroacoustic survey estimates were not correlated to MSVPA results in herring, independent of the area considered. Difficulties encountered to validate the herring stock developments are caused by the heterogeneity of the stock structure, extensive migratory behaviour as well as problems to assess the different components by research surveys. Also the present approach of separation into divisions does not solve this problem, as different stocks inhabit same areas, and the borders of Sub-divisions are not equivalent with borders of stock distributions.