Paper of the Month: Recruitment dynamics in fish stocks – how important are interactions between cohorts?
Recruitment dynamics in fishes are shaped by many factors, including cannibalism and competitive interactions between cohorts.
In this study, the incidence of such negative intra-specific interactions in recruitment dynamics was investigated for commercially relevant fish stocks in the Atlantic Ocean using autocorrelation analysis of residuals from Ricker stock-recruitment models. This provides an easy and fast method of detecting negative intra-specific interactions for any stock with sufficient data available.
Cannibalism and competition between cohorts are assumed to be important for recruitment success and partially explain the common recruitment fluctuations in many fish stocks. For instance, in Northeast Arctic cod (Gadus morhua) it is established that strong year classes can suppress survival of the following year classes, particularly when other food sources are scarce in years of low capelin (Mallotus villosus) abundance. However, besides a few well studied cases, little is known about the relevance of these interactions on a broader scale. To explore this question, the recently published study used an efficient and easy to apply method to quantify the relevance of such interactions in time series data of 112 fish stocks from the Atlantic Ocean.
The study uses the residuals of Ricker stock-recruitment models fitted to time-series of spawning stock biomass and recruitment and analyses the autocorrelations among them for different year lags. This is an easy and generally accessible method for detecting negative intra-specific interactions, yet it is also a conservative method, especially when stocks have undergone periods of low abundance or are subject to other strong density-dependent or -independent influences. As a result, positive autocorrelations of recruitment residuals were dominating for cohorts of low year lags and the majority of stocks did not show evidence for negative intra-specific interactions. However, negative autocorrelations were found in a number of stocks for cohorts 3 to 5 years apart and confirmed the expectations in several stocks, including Northeast Arctic cod.
Figure legend: Autocorrelation of recruitment residuals from fitted Ricker model. For each time lag for one to five years, the value of the autocorrelation function for each available stock is shown as a thin horizontal bar and a Gaussian density kernel shows the overall distribution for both non-significant (light grey kernel, left) and significant values (p < 0.05; dark grey kernel, right). The number of non-significant and significant autocorrelation at each time lag appear at the bottom of the plot.
Significantly negative autocorrelations indicate that cannibalism or intra-specific competition are important in those specific stocks, whereas positive autocorrelations suggest that environmental drivers may be important by affecting several cohorts within several years positively or negatively. Positive autocorrelations can also be caused by ageing errors during the stock assessment process, which can lead to substantial spillovers of large cohorts to preceding and subsequent cohorts.
Although significantly negative autocorrelations were only found in a low proportion of the studied stocks, our findings nonetheless underline that cannibalism and competition at high stock levels can have important consequences for determining sustainable biomass and harvest levels. The results of the study also highlight that recruitment dynamics are typically the combined result of various factors that can vary in their relevance over time, making it difficult to isolate specific factors from existing data. Environmental drivers can play a dominating role in most stocks and overshadow other effects, as indicated by the many positive autocorrelations. Furthermore, cannibalism and inter-cohort competition can vary in their intensity as a response to overall food availability within an ecosystem, as the example of the Northeast Arctic cod shows, diminishing the statistically detectable effects of each of them.
Also, cannibalism and competition are mostly relevant when stock sizes are large, yet most of the stocks studied have undergone long periods of low stock sizes and depletion. Inter-cohort interactions will become subsequently more relevant when stocks are well managed and at high levels, as recently observed in many Norwegian stocks. It can therefore be expected that future data will show clearer signals of inter-cohort competition.