Improving understanding of spatial ecology through network analysis of acoustic monitoring data

Understanding movement is important for defining animal spatial ecology and ensuring effective management and conservation. Accordingly, data on animal movement patterns, connectivity and habitat use have become crucial elements in management and conservation decisions. Large-scale movements of marine individuals are monitored using new tracking technologies such as acoustic monitoring. These new technologies often produce large amounts of high quality data, so data generation is no longer a challenge, however, data analysis and modelling are emerging issues. Arrangement of acoustic receivers into arrays or grids (i.e., as a "network") fits well with the use of an innovative approach: Network Analysis. Network Analysis is a powerful tool for examining the structure of complex interacting systems that are represented as a network characterized by connections between nodes. The use of Network Analysis to look at animal spatial ecology in the marine environment is in its early stages with only a few studies completed. Consequently, the potential of Network Analysis in studying animal spatial ecology using acoustic monitoring data is largely unexplored. However, this approach has been intensely used in other areas, including landscape ecology, and the results have proven incredibly useful for management and conservation. By combining acoustic monitoring and Network Analysis, researchers may be able to study the spatial ecology of species in the marine environment. Therefore, this project aimed to determine the contribution of Network Analysis in understanding marine animal spatial ecology using acoustic monitoring data. Literature analysis suggested that Network Analysis can help characterise marine animal spatial ecology in new ways, providing many tools to understand the complex interaction between animals and their environment. The multi-disciplinary nature of Network Analysis provides the researcher with convenient tools to understand the complexity of movement at different scales, compare movements between individuals or between species, and investigate the effect of environmental factors on the movement. The reviewed techniques were tested on acoustic monitoring data from six predator species. Field work was conducted along the north-east coast of Queensland, Australia. Two arrays of 67 and 48 acoustic transmitters deployed in the central section of the Great Barrier Reef passively tracked six predator species from 2008 to 2014. Two nearshore sharks (pigeye shark (Carcharhinus amboinensis) and spottail shark (Carcharhinus sorrah)), two reef sharks (silvertip shark (Carcharhinus albimarginatus) and grey reef shark (Carcharhinus amblyrhynchos)), and two carangid teleosts (giant trevally (Caranx ignobilis) and golden trevally (Gnathanodon speciosus)) were selected to determine efficacy of the Network Analysis method to contribute to the understanding of marine animal spatial ecology. To investigate the utility of Network Analysis in identifying core use areas and compare the results with traditional analysis, a case study using C. amboinensis and C. sorrah was conducted. Comparison of traditional analysis (kernel utilization distribution, KUD) and Network Analysis demonstrated that both methods provided similar results for identifying core use areas (50% KUD equivalent), but that Network Analysis tended to overestimate general use areas (95% KUD equivalent) compared to kernel-based methods. Furthermore, frequent bidirectional movements within core use areas were identified by Network Analysis, indicating the importance of movement corridors within or between core areas. Movements between acoustic receivers outside core use areas were less frequent and unidirectional suggesting transiting movements. Therefore, Network Analysis may be a practical alternative or companion to traditional home range metrics by providing useful data interpretation that allows for a comprehensive picture of animal movement, including identifying core use areas and pathways used. To test if Network Analysis could provide valuable information on functional connectivity in offshore reef habitats, a case study using C. ignobilis, C. amblyrhynchos and C. albimarginatus was conducted. Network modelling was used to examine and compare the structure of intra-reef movements to four simulated theoretical networks. All three species exhibited networks with properties of small-world and scale-free structures with rapid and direct intra-reef movements and high numbers of interconnected patches (i.e., area covered by acoustic receivers). These two characteristics have been identified in a variety of complex networks and explain how species may respond to habitat loss or disturbance. All three species also displayed consistent behaviour within reefs with a power-law node degree distribution suggesting Lévy-walk-like searching patterns. Furthermore, analyses of the networks revealed >75% of patches within reefs were important for either resources or connectivity for all three species. Receivers important for resources and for connectivity varied between species and reefs, and their locations were often found on opposite sides of the coral reefs. Consequently, network modelling provided insight into intra-reef predator movements that may assist in the development of effective management at an individual reef scale. To compare to Network Analysis results, the effects of biological and environmental variables on C. ignobilis monthly space use, daily presence and hourly depth use were investigated using traditional techniques. Using a linear modelling approach, temporal changes in movement patterns of C. ignobilis were explored to determine if individuals exhibited predictable movement patterns. Caranx ignobilis typically remained at their capture reef with 98.8% of detections recorded at these locations. Individuals were recorded in the study site for periods from 9 to 335 days (mean = 125.9) with a mean residency index of 0.53, indicating movements away from the reef or out of detection range occurred on the scale of days. Inter-reef movements from only three individuals were recorded which coincided with the summer full moon, and may have been related to spawning behaviour. Environmental drivers were correlated with daily presence and hourly depth use of C. ignobilis but had little influence on monthly space use. There was little or no effect of fish size on space use, presence and depth use. The results of this study reveal that individuals may be site attached and that environmental parameters play a role in observed movement patterns related to depth and presence. Finally, Network Analysis was used to examine the movement patterns of C. ignobilis and G. speciosus in inshore habitats. Tagged individuals were present in the study region between 30 to 394 days (mean ±SD = 166 ±116) with a mean residency index of 0.7 (±0.1 SE). Notable inter-annual variation occurred with individuals detected on more days, visiting more receivers, moving more frequently, and being more resident in some years than others. In addition, movement patterns differed between species, with C. ignobilis being detected on fewer days, using less receivers and moving less than G. speciosus. Network analysis revealed a combination of factors including ontogeny, foraging niche, and habitat influences may explain differences in space use between species. These results highlight unique behaviours between co-occurring and closely related species, and enhance our understanding of animal interactions in inshore habitats. This project demonstrated that by using Network Analysis, researchers studying the spatial ecology of marine animals can unlock a wide array of a species' behaviour. Using a single method, movement pattern, connectivity and space use of six predator species were investigated within an inshore and offshore habitat, revealing a range of movement strategies. Spatial and temporal partitioning and shifting of habitats both between and within species were found for all six species. Decreases in intra- and interspecific competition for resources, difference in foraging needs, decrease in risk of predation, response to environmental changes, or a combination of the above are possible explanations for the observed range of movements. This highlights that mechanisms behind movement patterns are complex and variable not only between but also within species and has important implications for management and conservation purposes. Finally, Network Analysis provides a toolbox of methods that can be used to assess consequences of habitat fragmentation and anthropogenic and natural disturbances and help design and evaluate the effectiveness of management and conservation plans. Network Analysis provided rapid assessment of species movement within studied areas that allows prioritisation of key patches and movement corridors for potentially creating marine reserve and maintain movement corridors of marine species. Therefore, Network Analysis is advantageous for guiding and assessing management measures as it allows for assessment of species movement and for prediction the consequences of anthropogenic and natural disturbances by testing a variety of species at different scales and under multiple scenarios