Combining a locomotion indicator and data mining to analyze the interactive patterns between copepods and ciliates

a r t i c l e i n f o Interactions between zooplankton not only affect the world's carbon fixation but also have a direct impact on the yields of the fishing industry. Both copepods and ciliates have a crucial linkage role in constituting the marine food web. Analyzing the predator–prey interactions between these two species helps us to understand the productivity of oceans better. In this study, we explored the interactive patterns between copepods and ciliates and used the locomotion indicator net-to-gross displacement ratio (NGDR) to conduct quantitative analyses on the swimming patterns of copepods. We discovered that the movement trails of copepods are more distorted in undisturbed environments where the NGDR was significantly lower. In an environment where ciliates were present, the NGDR of copepods was significantly higher. This result indicated that the movement trails in the latter scenario were more linear and that the NGDR can clearly distinguish the swimming patterns of copepods. In addition, this study developed a qualitative motion description to be embedded in the interactive trail data of copepods and ciliates, which facilitates the data mining technologies needed to perform advanced analyses. The results of the association rules and decision tree analyses clearly demonstrated the interactive characteristics that could not be explored solely using locomotion indicators. The most obvious characteristic is that the swimming patterns employed by copepods to approach ciliates were either a downward vertical sinking or a horizontal movement. Upward movements for the copepods were not typically observed. More detailed swimming patterns of interactions between copepods and ciliates were revealed using the rule-based forms. Both copepods and ciliates play a critical role in the marine food web Analyzing the interactions between these two organisms remains an important issue to be explored because the results of such studies would help in understanding the fundamental mechanisms of the marine food web (Hwang and Strickler, 2001). Many studies of zooplankton behavioral patterns in small-scale environments use video recording to trace movement trails (Buskey et al. Analyses after acquiring trail data mostly rely on locomotion indicators to objectively illustrate the characteristics of the zooplankton movement (Chen et al. statistical analyses of the indicators summarize the specific movement patterns for certain species of zooplankton. Traditional data-oriented analyses, such as statistical inference, are hypothetico-deductive processes (Galitsky et al., 2007). These processes always presume relevant null and alternative hypotheses based on the literature or experts' knowledge and then …