Determine biological management reference points

Introduction

When this project was originally specified, the purpose of this task was to determine biological management reference points. The specification of the project pre-dated the adoption and use of reference points by ICES in relation to the precautionary approach to fishery management. As a result of the considerable developments in this field, both scientifically and institutionally, it is not considered appropriate to propose revised reference points on the basis of the results of this project. The main justification for this is that reference points cannot be established solely on the basis of information on the recruitment process, but that a wide variety of other factors also need to be taken into account. The precautionary approach is framed around a rather simple model of stock dynamics, i.e. that there is a specific value of spawning biomass below which recruitment is impaired, and that fishing mortality is the only external factor which influences the size of the spawning stock. Such a simple approach is open to criticism, particularly in view of the information becoming available from process studies conducted in Task 1 to 5. It should be noted, however, that the simplicity of this conceptual model has a number of advantages from a management perspective. Firstly, the two indicators, SSB and mean fishing mortality, are readily available from age-based stock assessments, so their estimation does not require data or modelling beyond what is already routine. Secondly, the simplicity

Thirdly, the implication that SSB is influenced only by fishing mortality is often not far from the truth for heavily exploited stocks, with the added justification that fishing mortality is usually the only factor influencing SSB which fishery managers can seek to manage. Potential criticisms of the conceptual model underlying the ICES implementation of the precautionary approach are: firstly there is the implicit assumption that SSB is an adequate measure of the stock’s reproductive potential, which is obviously not the case – see Task 1 and 6. Secondly, there is the assumption that there is a fixed level of SSB below which recruitment is impaired. This in turn implies that recruitment is determined only by reproductive potential; another point which is not consistent with the results of process studies conducted under Task 2 to 5. Thirdly, reference points are different in single and multispecies contexts. This is explored in detail in the present section by simulation studies deploying single and multispecies prediction models.

Results and Discussion

For cod in the Central Baltic both target and limit reference points have been defined with the former (Bpa) corresponding to a spawning stock biomass (SSB) of 240,000 t and the latter (Blim) as 160,000 t. Bpa is adopted from the former MBAL estimate derived by a Ricker stock-recruitment relationship (1976-1994) as the SSB at which 50% of the maximum recruitment (age-group 2) originate. Blim is according to ICES (1998/ACFM:10) derived from Bpa * exp(-1.645*σ) with σ being the standard error of the biomass estimate from the international bottom trawl survey. Spawning stock biomass has been determined with a constant maturity ogive (used up to 1997 in the standard assessment) and constant weight at age for the periods 1976-1982 and 1991-1993. The precautionary fishing mortality (Fpa) has been determined as 0.65 leading to an SSB corresponding to the 10% lower fractile of SSB’s above Bpa, based on medium-term simulations conducted by ICES. A Beverton and Holt stock recruitment relationship assuming log normal errors fitted to year-classes 1981-1995 has been applied in the simulations, thus omitting extraordinary high recruitment originated in preceding years. This period selection was justified by changed environmental conditions leading to on average lower recruitment in latter period (see CORE 1998). The simulations utilized an average weight at age, maturity ogive and exploitation pattern determined for the period 1995-1997. This takes into account updated maturity ogives, but no trends in weight at age detected for Baltic cod. Likewise, the skewed sex distribution in the stock is not considered, with an increase in the female proportion at increasing age which results in higher egg production by SSB’s consisting of older individuals. A revision made by ACFM based on the 5% percentile of Fmed resulted in Fpa of 0.6 being officially adopted as precautionary reference point. The limit fishing mortality Flim corresponds to Fmed: 0.96 derived from a stochastic stock recruitment relationship covering year-classes 1966-1995 applying weight at age in the stock as described above, but period specific maturity ogives (averages over 5 years) up to 1994 and afterwards yearly data. The weight at age, combined maturity ogives and exploitations patterns utilized in the equilibrium calculation part are averages over most recent three year.

The determination of biological reference points for cod in the Central Baltic considers only to a limited extent the data available to characterize the reproductive potential of the stock. Firstly, both biomass reference points were calculated with constant maturity ogives, although test suggests application of yearly or periodic maturity at age will enhance the SSB as a measure of egg production. Even more promising is the utilization of female SSB or further the potential egg production based on the predicted relative fecundity in relation to food availability, i.e. mainly sprat and juvenile herring dynamics. Secondly, weight at age in the stock from

international bottom trawl surveys are available only for certain years and age-groups. Here an alternative, could be to use the multispecies weight at age in the catch data for age-groups 3+. These data have been tested against survey derived weight at age data and have the advantage that they are available for any given quarter, year and Subdivision and thus may also be applied for spatially explicit modelling approaches.

The problem of deviations in sex specific weight at age however has to be addressed by an analysis of the international trawl survey database. Updated period specific combined maturity ogives were considered in the stock recruitment relationships applied in medium-term projections to determine Fpa and for the determination of Fmed from stock recruitment plots, however the principal problem in the weight at age is the same as above. Thirdly, all medium-term projections applied by the assessment WG or ACFM utilized average combined maturity and weight at age over a certain time period, in the best case adding stochasticity, however neglecting density dependent feedback mechanisms and longer-term time trends in dependence of environmental conditions. As demonstrated structural uncertainties in projection models with respect to coupled consumption, growth and maturation processes impact substantially on biological reference points. Clearly an enhanced understanding of the linkage between prey availability, weight and maturity at age and realized fecundity is needed to improve medium-term projection methodology. Finally, effects of the stock structure on the quality of egg production affecting fertilization success, egg buoyancy and larval growth have been demonstrated, but have not been considered in any stock simulation studies yet.

Changes in environmental conditions largely affecting the developmental success of cod early life history stages have been considered in the Fpa determination by the Baltic Fisheries Assessment WG. The stock recruitment relationship has been divided into two different regimes. At present, only stock recruitment data from 1981 onwards are applied, representing unfavourable conditions for egg and larval survival. The earlier period represents a regime characterized by favourable oxygen conditions caused by frequent inflow events increasing the probability of high survival success. Simulations with a medium-term projection programme, explicitly incorporating environmental conditions, i.e. the reproductive volume and sprat stock size as a predator on cod eggs, have been conducted earlier by ICES. The reproductive volume used in the projections was a historic average with random variability superimposed. Similarly, the average sprat stock size was introduced. The latter procedure may be significantly enhanced by incorporation of an environmentally sensitive stock recruitment function into a multispecies prediction programme, e.g. MSFOR implemented in 4M. This also would allow to simulate fish prey availability to be coupled to growth, maturation and fecundity of cod. With respect to oxygen conditions, the situation is more difficult, as the reproductive volume is not predictable for more than one year in advance, as large-scale atmospheric processes responsible for inflow events cannot be forecasted. For medium- to long-term simulation of stock development under different hydrographic conditions, scenario modelling is at present the only solution.

For sprat the reference point Blim (200000 t) corresponds to MBAL, while Bpa (275000 t) is set to Blim*1.38.

Flim is not defined. Fpa set at 0.40, being the average Fmed in recent years, allowing for variable natural mortality. The latter procedure was tested for stability by computing various fishing mortality reference points for a range of natural mortality values. Several sets of average natural mortality rates were adopted referring to present, low, medium and high levels of M. Weight at age and selection patterns were assumed as averages over the years 1997–1999. The mean natural mortality was added to F reference points to derive Z

similar pattern is observed with the Z reference points. F0.1 and Z0.1 behaved in a reverse manner, both increasing with increasing natural mortality. The assumption in the above calculations of the same weights and selections for different levels of natural mortality may not be realistic as with higher mortality higher weights may be expected (density dependent effects). So, additional estimates of Zpa points were computed on basis of natural mortalities, weights and selection patterns from the corresponding earlier periods. The estimated Zmed values are considerably higher than the estimates of Zmed referring to the same periods but basing on weight and selection pattern from most recent years. Clearly changes in weight at age have to be considered in the determination of biological reference points. This may be achieved by applying simple weight at age – stock abundance relationships. Tests on the impact of variable maturity ogives revealed a more limited impact on biological reference points.

Reference points for fishing mortality based on single species yield and SSB calculations are difficult to use when natural mortality depends on the abundance of the predators and their alternative prey. For example, multispecies predictions conducted suggest that the cod stock in the Baltic should be reduced to a very low level of biomass in order to allow benefit from the higher productivity of herring and sprat, its major prey, when considering yield solely in quantity. Such a result stresses the need for incorporating socio-economic considerations in the definition of target reference points. Management advice based on biomass reference points will also differ. In the single species situation the combinations of cod and pelagic fishing effort for which the equilibrium spawning stock biomass of the three species is above the biomass reference points forms a rectangular area. When biological interaction is taken into account the limits of this area becomes curved. Reference limits for forage fish cannot be defined without considering changes in the biomass of their natural predators. Likewise, reference limits for their predators cannot be defined without considering changes in the biomass of their prey. Also considering variability in predation mortality by computing total instead of fishing mortality reference points for sprat did not result in stable alternative management objectives.

1. Viable egg production for Baltic cod and sprat

1.1. Estimate how the spatial distribution and stock composition of cod and sprat vary temporally and seasonally in relation to stock size and environmental conditions

Introduction

Cod and sprat aggregate in the deep Baltic basins to spawn, having largely overlapping spawning areas and spawning times (Bagge et al., 1994; Parmanne et al., 1994). However, they show different spawning behaviour vertical distribution and diurnal migration patterns (e.g. Tomkiewicz et al., 1998; Orlowski, 2000). Sprat is a major prey of the top-predator cod (e.g. Sparholt, 1994), however, sprat also preys on early life stages of cod (Köster and Schnack, 1994), both processes depending heavily on the fine- to meso-scale spatial and temporal predator/prey overlap (Neuenfeldt, 2002; Köster and Möllmann, 2000a). Furthermore, hydrographic conditions conducive for cod and sprat egg survival (Nissling, 1994, Wieland et al., 1994) and food supply for larval (Hinrichsen et al., 2002) as well as juvenile and adult life stages (Köster et al., 2001a) may vary considerably not only between, but also within spawning areas. Thus, the distribution of the spawning effort within and between spawning areas appears to be of importance for the reproductive success. This subtask utilises research survey data to test the hypothesis that the small-scale vertical and meso- to large-scale horizontal distribution of cod and sprat vary with stock sizes and hydrographic conditions. The studies cover a description of the stock structure of cod and sprat populations and their vertical and horizontal distribution during spawning periods in relation to environmental conditions, specifically addressing:

a) within survey and within Sub-division distribution, to study the small- to mesocale hydrographic processes affecting the vertical and horizontal distribution,

b) intraannual and within Sub-division distribution, to resolve the aggregation pattern in spawning season and areas,

c) intraannual and between Sub-division distribution, to describe larger scale shifts in distribution between seasons in dependence of ambient environmental conditions considering fishing activities,

d) interannual and within Sub-division distribution, to determine the impact of changing environmental conditions on aggregation patterns within spawning areas, considering also stock structure and related maturation processes,

e) interannual and between Sub-division distribution and abundance characterizing varying population development and large scale shifts in the distribution of the stocks.

Fine- to meso-scale vertical/horizontal distribution of cod and sprat during spawning periods in relation to environmental conditions

Based on a series of project related research surveys the hypothesis was tested that the vertical distributions of cod, herring and sprat, and consequently their distributional overlap, can be described by hydrographic stratification. Field activities comprised combined demersal/pelagic trawl and hydrographic surveys (July 1999) as well as hydroacoustic/hydrographic surveys (May/June 1999 and 2001, August 2001).

cod, sprat and herring as well as their distributional overlap at spawning time. Tomkiewicz et al. (1998) on basis of CORE (1998) results suggested that Baltic cod prefer salinities >11 during spawning in the Bornholm Basin.

Furthermore, Baltic cod avoid oxygen saturation < 28% in experiments (Plante et al., 1998), but voluntarily enter waters with oxygen saturation as low as 16 % for short feeding excursions (Claireau et al., 1995a). While cod stay in highly saline water within and below the halocline throughout the day, sprat and herring undertake diel vertical migrations during spring and summer, swimming downwards at dawn and upwards at dusk, and form schools during daylight hours (Orlowski, 2000; 2001).

Fishery independent estimates of pelagic stocks in the Baltic have been carried out by hydroacoustics since the end of the 1970s. The results from annual autumn surveys have been used for calibrating (“tuning”) VPAs (Virtual Population Analyses) of area aggregated standard assessments (ICES, 2002) as well as area dis-aggregated approaches (Köster et al., 2001b). In some years hydroacoustic surveys were additionally conducted in the Southern and Central Baltic in May. The main target species was sprat since the largest fraction of the herring population during this time of the year concentrates in coastal areas for spawning, i.e. outside the area covered by the hydroacoustic surveys (Aro, 1989). Corresponding spring hydroacoustic surveys were conducted in May/June 1999 and 2001 in Sub-divisions 24-28. Additionally a hydroacoustic survey was performed in August 2001 covering the most important spawning area of cod in the Central Baltic, i.e. the Bornholm Basin, during peak spawning time in order to investigate the spatial overlap of clupeid predators and cod early life stages as their prey (Köster and Möllmann, 2000a).

Interannual variability in meso-scale horizontal distribution patterns of cod and sprat during in relation to environmental conditions

Environmentally defined thresholds and preferences for occurrence can determine the species distributional overlap volume in the Baltic Sea, because it is physically heterogeneous. The Baltic has the property of a permanent vertical stratification of temperature, oxygen concentration, and salinity (Matthäus, 1995). It is semi-enclosed, and consists of several basins with one very tight connection to the North Sea (Dietrich et al., 1975).

High saline and oxygen rich water is only added to the system by rare North Sea water inflow events. The permanent halocline separates low saline surface water (7.5 to 8 PSU) from high saline deep water (11-18 PSU, varying between basins). In periods without salt-water inflow the oxygen conditions below the halocline deteriorate and the halocline depth increases (Matthäus and Franck, 1992).

Both experimental (e.g. Schurmann and Steffensen, 1994a, 1994b) and field studies (e.g. D’Amours, 1993;

Tomkiewicz et al., 1998) suggest that cod avoid oxygen saturation below 25 %. Tomkiewicz et al. (1998) found, that Baltic cod at the same time prefer salinities above 11 PSU during spawning in the Bornholm Basin. Gröhsler et al. (2000) (see activity 1 above) reported that in the Bornholm Basin sprat is mainly concentrated below the halocline.

Demersal trawl hauls with accompanying temperature, salinity and oxygen saturation measurements were conducted in February/March 1994-95 and 1997-99. The catch per unit of effort data from these hauls were used to test the hypotheses that species density at a given location is related to oxygen saturation, salinity, temperature and slope of the seabed.

The distribution of sprat in the Baltic depends largely on winter conditions, i.e. the availability of water volumes above 4°C (Rechlin, 1967), found normally within and below the permanent halocline in the deep Baltic basins.

The lower range of the vertical distribution is restricted by the oxygen concentration in the bottom water (see

above). Shifts in wintering distribution from the Gdansk to the Gotland Basin have been reported in relation to encountered hydrographic conditions (Vasilieva, 1996). After wintering, the sprat shoals start spawning in deep water, but orient themselves into upper water layers as vertical mixing is increasing the temperature in the intermediate water (Parmanne et al., 1994). In summer, after finalisation of spawning activities, feeding migrations of sprat to shallower coastal waters take place in intensity and direction depending on the hydrographic conditions as well as the age structure of the stock (Shvetsov and Gradalev, 1989).

A comprehensive analysis of distribution patterns in relation to the environmental conditions within and between Sub-divisions 26 and 28 was conducted on basis of Russian/Latvian hydroacoustic surveys conducted in September/October and occasionally also May/June 1987-2001. The objective of the investigation was to estimate how the spatial distribution and stock composition of sprat in the Central Baltic vary temporally and seasonally in relation to stock size and environmental conditions.

Interannual variability in large-scale horizontal distribution, abundance and population structure in relation to hydrographic features

Direct abundance estimates of Baltic cod (recruits and adults) are based on demersal trawl surveys (ICES, 2000a). There have been some ad hoc attempts to use hydroacoustic to measure Baltic cod spawning aggregations and spatial distribution on spawning grounds, but it has not been a standard practice.

Hydroacoustic have also been used to estimate Baltic cod 0-group abundance in the pelagic phase within the CORE project (Nielsen and Lehmann, 1996), but without any major success. The main body of trawl survey data originates from regular national and in recent years, internationally co-ordinated trawl surveys in the western Baltic since 1978 and in the eastern Baltic since 1980 (ICES, 2000a). The basic data have now been compiled into a Baltic International Trawl Survey (BITS) database containing about 9500 records on single haul basis (ICES, 2000a) in a special format. The major part of the data is originating from the first quarter of the year and a smaller part of the data is from the fourth quarter. Over the years, this database has been used to analyse the overall interannual variation of Baltic cod abundance for stock assessment and to a lesser extent to estimate the spatial distribution pattern. In recent years, the Baltic Fisheries Assessment Working Group has used the

Hydroacoustic have also been used to estimate Baltic cod 0-group abundance in the pelagic phase within the CORE project (Nielsen and Lehmann, 1996), but without any major success. The main body of trawl survey data originates from regular national and in recent years, internationally co-ordinated trawl surveys in the western Baltic since 1978 and in the eastern Baltic since 1980 (ICES, 2000a). The basic data have now been compiled into a Baltic International Trawl Survey (BITS) database containing about 9500 records on single haul basis (ICES, 2000a) in a special format. The major part of the data is originating from the first quarter of the year and a smaller part of the data is from the fourth quarter. Over the years, this database has been used to analyse the overall interannual variation of Baltic cod abundance for stock assessment and to a lesser extent to estimate the spatial distribution pattern. In recent years, the Baltic Fisheries Assessment Working Group has used the