Complex interplays among population dynamics, environmental forcing, and exploitation in fisheries

Abstract

The patterns of variations in fisheries time series are known to result from a complex combination of species and fisheries dynamics all coupled with environmental forcing (including climate, trophic interactions, etc.). Disentangling the relative effects of these factors has been a major goal of fisheries science for both conceptual and management reasons. By examining the variability of 169 tuna and billfish time series of catch and catch per unit effort (CPUE) throughout the Atlantic as well as their linkage to the North Atlantic Oscillation (NAO), we find that the importance of these factors differed according to the spatial scale. At the scale of the entire Atlantic the patterns of variations are primarily spatially structured, whereas at a more regional scale the patterns of variations were primarily related to the fishing gear. Furthermore, the NAO appeared to also structure the patterns of variations of tuna time series, especially over the North Atlantic. We conclude that the patterns of variations in fisheries time series of tuna and billfish only poorly reflect the underlying dynamics of these fish populations; they appear to be shaped by several successive embedded processes, each interacting with each other. Our results emphasize the necessity for scientific data when investigating the population dynamics of large pelagic fishes, because CPUE fluctuations are not directly attributable to change in species' abundance.

Fig. 1.

Bootstrap estimates of the mean dissimilarity between the wavelet spectra of each species versus their spatial repartition expressed through the southern/northern ratio (a) and between the wavelet spectra from southern and northern areas (b). The southern/northern ratio is computed for each species as the number of time series from southern areas divided by the number of time series from northern areas.

Fig. 2.

Cluster tree of the wavelet spectra for the CPUE time series in the Eastern Canary Coastal province on the west African coast. The wavelet spectra decompose the variance of time series over time (x axis) and frequencies (y axis), enabling one to follow the time evolution of the relative importance of frequencies in the signal. The colors gradient, from dark blue to dark red, codes for low- to high-power values. The wavelet spectra were then compared, and a dissimilarity matrix was produced (see Materials and Methods). The cluster tree was obtained by using the dissimilarity matrix on which flexible clustering was applied. The CPUE time series analyzed are plotted in black lines at the top of the corresponding wavelet spectra.

Fig. 3.

Clusters trees of wavelet spectra in the provinces that displayed several gears. The cluster trees were obtained by using the dissimilarity matrix constructed with the wavelet spectra of the CPUE time series, on which flexible clustering was applied. LL, longline; BB, baitboats; PS, purse seine; TA, trap; TO, troll.

Fig. 4.

Representation of the successive modulations of signal that shape the fisheries time series. The dynamics induced at the population level is first influenced by the ecosystem and climate. The signal is then modulated depending on the geographic location and on the local properties, also influenced by the climate. Finally, the different fishing gears and dynamics also constitute a source of modulation.