Large-scale tagging experiments are required to provide the level of information (fishery exploitation rates and population size) necessary for tuna stock assessments of tropical tunas in the Western and Central Pacific Ocean.
Tagging data has the potential to provide much information of relevance to stock assessment, either by way of stand-alone analyses or, preferably through integration with other data directly in the stock assessment model. Stock assessments of the main targeted tuna species, skipjack (Katsuwonus pelamis), yellowfin (Thunnus albacores), and bigeye (Thunnus obesus) are based on the integrated analysis of catch, effort, size composition and tagging data, using a length-based, age-structured, spatially-explicit fish population dynamics and stock assessment model known as MULTIFAN-CL.
The advantages of integrated assessments are numerous, and include the ability to combine multiple data sources, the ability to estimate important population dynamics processes such as growth, natural mortality, fishing mortality and movement simultaneously, and more realistic quantification of the uncertainty in the integrated estimates of key management quantities. Furthermore, the spatially explicit nature of the model enables sub-sets of the overall model domain to be examined to provide information on sub-regional trends in stock size and fishery impacts. The general objective of a tag-release-recapture experiment is to establish an “experimental” population of individuals tagged with numbered tags that can then be monitored and modelled over time through recaptures by the fishery.
Assessments of the main species targeted by tuna fisheries of the region have in the past been based mainly on the results of large-scale tagging experiments carried out in 1977-81 (Skipjack Survey Assessment Project) and 1989-92 (Regional Tuna Tagging Project). The projects have generated data for "snap-shot" assessments of the skipjack and yellowfin stocks, but more limited information on bigeye and South Pacific albacore. With a range of data becoming available from port sampling and observer programmes, the continued improvement in logbook data coverage and the cooperation of several fishing nations in providing historical fisheries data, it has been possible to consider the application of more data-intensive stock assessment methodologies.
With funding support from several donors, an integrated, length-based, age-structured model referred to as MULTIFAN-CL has been developed for application to tuna stock assessment. The initial version of this model was applied to South Pacific albacore, but is now in various stages of application to yellowfin, skipjack and bigeye tuna as well.
The main advantages of this modeling approach are:
Movement & stock structure
The illustration, and less commonly, the estimation of movement rates or probabilities is one of the classical uses of tagging experiments. In their simplest form, observed tag movements can be used to gain an overall impression of movement and the relationship, or lack thereof, between particular geographical locations.
For example, the movements of tagged tuna, observed from a long history of tagging experiments show the clear linkages and continuity of tuna stocks In the Pacific Ocean.
(Figure below (Left Panel: yellowfin; Right Panel: bigeye).
While such observations are useful, they can sometimes create a biased view of the extent and frequency of long-distance movement by tuna because they emphasize the long-distance movements. The release of large quantities of tagged fish over wide spatial areas allows for more detailed analyses that correct this bias. For example, the more detailed analyses for skipjack and yellowfin tuna tagging data found that the median lifetime displacement of skipjack tuna is 420-470 nmi and for yellowfin tuna, 337-380 nmi, demonstrating that while spectacular long-distance movements of thousands of nautical miles frequently occur in tropical tuna, the vast majority of observed tag movements are much less.
The rate of natural mortality is a critical parameter for stock assessment and one that is often poorly known. It typically cannot be estimated from fisheries data (such as catch, effort and size- or age-composition) alone. The estimation of fishing mortality is of course a key element of the stock assessment. Again, it is a parameter that is often poorly determined from fisheries data alone. The integration of tagging data into the assessment, along with other critical information on the tag-reporting rate(s), can greatly assist the estimation of both natural and fishing mortality. The advantage of tagging data over regular fisheries data is that we know the initial population size of the tagged population (the release numbers) whereas this is not normally known for the untagged population. The observed rate of tag attrition (Figure below) then provides a direct measure of total mortality, which can be disaggregated into its components given knowledge of initial release numbers and rates of tag loss from other sources.
Tagging data are also routinely used to provide information on tuna growth rates. Currently, information on growth from tagging data, in the form of length-increment and time at liberty observations, is not routinely integrated into stock assessment models, although there is potential to do so. Rather, tagging data are currently (i) analysed externally to the assessment model and the estimated parameters incorporated as fixed parameters or Bayesian priors in the stock assessment model; or (ii) used as an independent check on the growth parameter estimates derived from the data used in the assessment. The latter use is currently made of tuna tagging data in the SPC assessments (Figure below).
While the focus on the use of tagging data has been in integrated stock assessment models such as MULTIFAN-CL, a new class of model known as Spatial Ecosystem and Population Dynamics Models (SEAPODYM) have potentially far-reaching application in the understanding the dynamics of pelagic fish stocks and their relationship with the biological and physical environment. SEAPODYM attempts to model stock distribution and abundance in relation to the biological characteristics of the stock and their interaction with environmental conditions (e.g., water temperature, currents, oxygen concentration, primary productivity, tuna forage, etc). A feature of SEAPODYM models is their ability to predict relatively fine-scale stock distributions (Figure below) potentially providing information on sub-components of the stock. Currently, SEAPODYM models are parameterized by fitting to fishery data (catch, size composition), but work is now underway to extend this to both conventional and archival tagging data. These models, incorporating tagging data, could provide considerable new information of tuna stock variability in relation to environmental factors and fishing, from EEZ to basin scales.