Spatial patterns and trends in abundance of larval sandeels in the North Sea: 1950-2005

Early recruitment indices based on larval fish data from the Continuous Plankton Recorder (CPR) have the potential to inform stock assessments of Ammodytes marinus in the North Sea. We evaluate whether the CPR data are reliable for sandeel larvae. Spatially, CPR larval data were comparable with catches by dedicated larval samplers (Gulf and bongo nets) during ICES coordinated surveys in 2004 and 2009. ICES data are also used to explore environmental influences on sandeel distributions. Temporally, CPR data correlate with larval data from plankton surveys off Stonehaven (1999–2005), with sandeel 0-group trawl data at the east Fair Isle ground (since 1984), and with recruitment data (since 1983) for the Dogger Banks stock assessment area. Therefore, CPR data may provide an early recruit index of relative abundance for the Dogger Banks assessment area, where the majority of the commercial catch of A. marinus is taken, and the Wee Bankie area that is particularly important for seabird foraging. While warm conditions may stimulate the production of sandeel larvae, their natural mortality is typically greater, in the Dogger Banks and Wadden Sea areas, when the larvae are hatched in warm years and/or with abundant 1-year-old sandeel that are likely to be cannibalistic.

Comparison of Zooplankton Sampling Performance of Longhurst-Hardy Plankton Recorder and Bongo Nets

Data on the abundance and biomass of zooplankton off the northwestern Portuguese coast, separately estimated with a Longhurst-Hardy Plankton Recorder (LHPR) and a Bongo net, were analysed to assess the comparative performance of the samplers. Zooplankton was collected along four transects perpendicular to the coast, deployments alternating between samplers. Total zooplankton biomass measured using the LHPR was significantly higher than that using the Bongo net. Apart from Appendicularia and Cladocera, abundances of other taxa (Copepoda, Mysidacea, Euphausiacea, Decapoda larvae, Amphipoda, Siphonophora, Hydromedusae, Chaetognatha and Fish eggs) were also consistently higher in the LHPR. Some of these differences were probably due to avoidance by the zooplankton of the Bongo net. This was supported by a comparative analysis of prosome length of the copepod Calanus helgolandicus sampled by the two nets that showed that Calanus in the LHPR samples were on average significantly larger, particularly in day samples. A ratio estimator was used to produce a factor to convert Bongo net biomass and abundance estimates to equate them with those taken with the LHPR. This method demonstrates how results from complementary zooplankton sampling strategies can be made more equivalent.

ARE ARTIFICIAL NEURAL NETS AS SMART AS A RAT

The origins of artificial neural networks are related to animal conditioning theory: both are forms of connectionist theory, which in turn derives from the empiricist philosophers' principle of association. The parallel between animal learning and neural nets suggests that interaction between them should benefit both sides.

The Computational Complexity of Dominance and Consistency in CP-Nets

We investigate the computational complexity of testing dominance and consistency in CP-nets. Previously, the complexity of dominance has been determined for restricted classes in which the dependency graph of the CP-net is acyclic. However, there are preferences of interest that define cyclic dependency graphs; these are modeled with general CP-nets. In our main results, we show here that both dominance and consistency for general CP-nets are PSPACE-complete. We then consider the concept of strong dominance, dominance equivalence and dominance incomparability, and several notions of optimality, and identify the complexity of the corresponding decision problems. The reductions used in the proofs are from STRIPS planning, and thus reinforce the earlier established connections between both areas.

Decision Making by Credal Nets

Credal nets are probabilistic graphical models which extend Bayesian nets to cope with sets of distributions. This feature makes the model particularly suited for the implementation of classifiers and knowledge-based systems. When working with sets of (instead of single) probability distributions, the identification of the optimal option can be based on different criteria, some of them eventually leading to multiple choices. Yet, most of the inference algorithms for credal nets are designed to compute only the bounds of the posterior probabilities. This prevents some of the existing criteria from being used. To overcome this limitation, we present two simple transformations for credal nets which make it possible to compute decisions based on the maximality and E-admissibility criteria without any modification in the inference algorithms. We also prove that these decision problems have the same complexity of standard inference, being NP^PP-hard for general credal nets and NP-hard for polytrees.