Application of association analysis for identifying indicator taxa of vulnerable marine ecosystems in the Emperor Seamounts area, North Pacific Ocean

Reflecting the growing interest in ecosystem-based fishery management, deep sea bottom fisheries are being called upon to minimize adverse impacts on vulnerable marine ecosystems (VMEs), communities of marine organisms susceptible to anthropogenic disturbance. Many fishery management organizations have introduced indicator-based management measures for VME conservation, such as encounter protocols, in which VME indicator species and bycatch weight thresholds are assigned. If the bycatch amount of the indicator species in a fishing operation exceeds the predetermined threshold, the fishing vessel halts the fishing operation and moves a certain distance away from the encounter point. However, the representativeness of VME indicator taxa has not been evaluated quantitatively. In this study, we analyzed the co-occurrence of benthic animals collected by scientific bottom tow-net surveys in the Emperor Seamounts area, North Pacific Ocean, to characterize benthic communities in the area and to examine the ability of six candidate indicator taxa (gorgonians, Alcyonacea excluding gorgonians, Antipatharia, Scleractinia, Stylasterina, and Porifera) to represent the local benthic communities. Cluster analysis revealed four clusters of benthic communities, each of which includes both sessile and mobile benthos: (1) gorgonians- Scleractinia community with many mobile benthic taxa and Pisces; (2) Porifera-Stylasterina community with Polychaeta and Bivalvia; (3) Antipatharia-Alcyonacea (excluding gorgonians) community with Cephalopoda; and (4) Zoanthidea-Pennatulacea community with Crinoidae and Holothuroidea. The first cluster included the largest number of taxa and showed strong tendencies of co-occurrence, possibly reflecting the habitat-providing function of gorgonians and Scleractinia as well as the common environmental preferences of filter feeders, which constitute major components of the cluster. We used association analysis to identify VME indicator species in the study area. Association analysis reveals relationships between items in the form of association rules, where the occurrence of an "antecedent" {A} implies the co-occurrence of a "consequent" {B}; {A} and {B} contain items, in this case, taxa. Association analysis applied to the co-occurrence data extracted many effective association rules that include gorgonians or Scleractinia as the consequent and many benthic taxa as antecedents. These results demonstrate that gorgonians and Scleractinia are effective VME indicators in the study area because they co-occur with many other benthic animals and represent VME characteristics such as functional significance as habitat and structural complexity as well as fragility and slow recovery from physical damage.