Monitoring coastal fish through remote underwater camerasdetectability, abundance, vulnerability and future applications

Resumen

Managing recreational fishing is particularly challenging because of (1) its growing relevance, particularly along the Mediterranean coast, and (2) the scarcity of data available. The relevance of recreational fishing is nowadays indisputable, but catches, fishing mortality, fishing effort, abundance and almost all the descriptors of this activity are unknown or poorly known in most of the cases. Thus, population dynamics of the target species remains unknown too. This fact precludes the design, implementation, and evaluation of any scientifically based management plan. Therefore, there is an urgent need for developing observation methods that supply objective, precise, and accurate data on recreational fishing. This thesis is aimed to fulfil this gap by exploring the capabilities of underwater cameras. The technological advances experienced by underwater cameras has been impressive in the recent years. Nowadays, image quality, battery life, durability, memory capability or cost are no longer limiting factors. However, the underwater images must be distilled into the quantitative data needed for designing proper management plans in accordance with the underlying ecological processes. Connecting images and data is far to be trivial. Specifically, I explored how to deliver data for two particularly challenging variables: fish abundance and fish vulnerability to angling. In the case of fish abundance, the first step was to combine unbaited cameras with horizontal field view with one of the most popular methods currently used for estimating abundance of many costal fish: scuba divers counting fish along transects (i.e., underwater visual censuses, or UVCs). I demonstrated how to combine cameras and UVCs for exploring and accounting for any environmental or fish dependencies on fish detectability and, more important, I proved that once fish detectability has been estimated, fish abundance could be precisely and accurately estimated using only cameras. Provided that the number of cameras and the deployment time are not limiting factors, this fact opens the possibility of estimating fish abundance at spatial and temporal scales relevant for managing recreational fisheries. The second step was to solve two pending problems: (1) the area surveyed by unbaited cameras with horizontal view cannot be precisely estimated and (2) the statistical method developed for the cameras renders abundance values at the scale of the reference method, which is, in this case, UVC. In spite of being the common standard, it is also recognized that UVC may introduce some biases when estimating fish abundance. Accordingly, a new design of unbaited cameras with vertical view, a new sampling protocol, and a new statistical analysis were developed. I demonstrated that this new framework produces estimates of fish abundance more precise and more accurate. The third step was to demonstrate the applicability and feasibility of monitoring fish abundance at large spatial and temporal scale. As a proof-of-concept, the abundance of a small serranid targeted by recreational fishers (Serranus scriba) was successfully estimated along more than 100 km of the South coast of Mallorca. Moreover, this design allowed identifying the main ecological drivers that are correlated with fish abundance. For example, I demonstrated that fish abundance is negatively correlated with exposure to fishing. Concerning fish vulnerability, I used baited cameras for disentangling the correlational patterns of a surrogate variable (latency time until a fish attacks the bait) with several variables related with the complex underlying processes. The results evidenced that fish-fish (social) interactions play a relevant role in the odds a specific fish has of being captured. The results reported strongly suggest that a mechanistic understanding of the processes shaping fish vulnerability should be unravelled in order to improve the design of appropriate management plans. Otherwise, recreational fishing may lead to populations with a large percentage of non-vulnerable fish, which is not only ecologically undesirable but also affects fisher’s satisfaction. Finally, I have in-depth explored one of the ecological outcomes of vulnerability. The existing theoretical frameworks hypothesize that non-vulnerable fish may depict smaller reproductive potential than vulnerable fish. Accordingly, I have developed a method for scoring fish vulnerability, and I have applied it to design experimental groups in captivity, emulating populations with a different average vulnerability. The results demonstrate that neither the number of eggs laid, nor the seasonal spawning pattern is related with fish vulnerability. Contrasting, the quality of the eggs quality of non-vulnerable fish seems to be higher toward the end of the spawning season than that of vulnerable fish. Provided that egg quality is affecting survival and dispersal capability, the outcomes of this finding should be explored and accounted for when designing spatial management plans. Overall, this thesis has achieved its primary objective: to develop feasible and reliable sampling frameworks based on underwater cameras that produce accurate and precise data ready to be used for assessing recreational fishing and other poor-data fisheries.