Scottish Marine and Freshwater Science Volume 5 Number 10: Updating Fisheries Sensitivity Maps in British Waters

READ BEFORE USING THESE SENSITIVITY MAPS

Caveats Regarding use of 0 Group Distribution Maps

a) Lack of Environmental Data for Coastal Areas

The GIS layers based on oceanographic models used for modelling 0 group fish distribution do not cover the inner coastal areas of some parts of Britain's coastline. This is most noticeable in the west coast.

For the model to run correctly and the statistical processes to make sense all the environmental layers have to be clipped (cover the same spatial extent) to the same spatial layer. In this case, the hydrographic layers are the ones that create the spatial gaps due to lack of data close to the coast.

These are shown as the white areas in the probability of presence of aggregations maps. It is important to highlight that these "white" areas are not included in the model prediction hence no outputs are available for these areas.

MSS is aware that this is an important issue and is already seeking methods to resolve it.

Hence these maps should be used in combination with alternative information that describes the species' nursery areas around the areas that are missing from these outputs. A source for this information for cod around the Scottish coast is:

Gibb, F., Gibb, I. & Wright, P. (2007). Isolation of Atlantic cod (Gadus morhua) nursery areas . Marine Biology, 151(3): 1185-1194.

b) Using the Prediction Probability Maps

The sensitivity maps show the probability of finding 0 group aggregations in a huge area around the UK, larger than the extent of UK territorial waters. This has allowed the development of models to be more robust but also can be inconvenient for some species when high abundances of 0 group aggregations occur outside of UK waters. This can be compounded by the lack of local data within the area of interest.

In some species, for example whiting, the probability of presence of aggregations of 0 groups in UK waters can appear relatively less probable due to high probability of aggregations in waters outside of UK. It is advisable to use a local interpretation of the model outputs.

c) The Maps Represent Aggregations of Fish in their First Year of Life

They do not represent "nursery areas" as described in the precursor to this report, Coull et al. (1998). "Nursery areas" can comprise a larger spread of ages and sizes. These representations are of 0 group fish, fish that are in the first year of their life.

d) Herring Larvae Aggregations Model, Performance Evaluation

The timing between the act of spawning and the surveying of these herring larvae aggregations mean that the larvae have been drifting in the currents for an unknown period of time. This makes it difficult to find a connection between the explanatory variables used in the models and the probability of presence of aggregations. This problem has a direct effect on the herring larvae aggregation model's performance compared to the predictive strength of other species' models.

Also, the parameters that predominantly drove this model were counter-intuitive to what could be expected given the life history of this species. For example: distance to gravel, the preferred substrate for herring spawning, did not give a strong signal in these predictions.

For these reasons, the herring larvae maps are only presented here as a first approach to updating the spawning areas, and should not be published and used as spawning maps for herring.

These maps of herring larvae aggregations may benefit from further applications of this model once additional environmental layers that, for instance, better describe herring larvae dispersal or give a stronger predictive signal can be included in the modelling process.

e) Species-specific Issues

Insufficient data on 0 group fish were available to perform the species distribution modelling approach on saithe and ling, these two species have not been included in this report.

Hake and anglerfish outputs represent presences and absences of 0 group fish as insufficient data were available to apply the species distribution modelling approach to aggregations of 0 group fish of these two species.