Measuring biodiversity in marine ecosystems

Article metrics

The use of catch data to determine indicators of biodiversity such as 'mean trophic level' does not adequately measure ecosystem changes induced by fishing. Improved ways to assess those changes are required. See Letter p.431

Accurate indicators of biodiversity are essential for managing exploited marine ecosystems. The currently most widely adopted indicator is the 'mean trophic level' of catches, the position of a specific species in the food chain (trophic level) averaged over all the species in the catch. Declines in catch mean trophic levels have been interpreted as showing shifts in ecosystem diversity from high-trophic-level predators to lower-trophic-level species. But are indicators based on catch data accurately depicting what is happening to an ecosystem? This question has now been addressed by Branch and co-workers on page 431 of this issue1.

Catch databases from marine fisheries are a reflection of economic, biological, ecological and technological factors. As a result, some species are unduly emphasized in the catches, distorting their true occurrence in the ecosystem. Additionally, catch databases, or more correctly 'reported catches', might not reflect the full extent of exploitation. Discarded bycatch, recreational fisheries and rare species are difficult to monitor and are therefore often not fully represented in the data. Finally, the databases themselves are often organized around political jurisdictions and do not necessarily encompass the entire ecosystem. Nevertheless, catch databases are easily accessible and have relatively comprehensive species composition. So, despite the drawbacks, they remain attractive for formulating diversity indicators such as indices of mean trophic level.

Branch et al.1 examined how useful these databases really are. They did this by comparing the mean trophic level of catches with the mean trophic level of ecosystems (mean trophic level weighted by the estimated true abundance of species in the ecosystem), using two avenues of research.

First, they collated 25 existing and well-documented marine-ecosystem models, representing regions in the Northern and Southern Hemispheres, over a wide range of latitudes. For each model, components encompassing the existing fisheries of the region had already been incorporated. Time series of catch and abundance were projected under four fishing scenarios: 'fishing down', in which higher trophic levels were fished to depletion followed by the advent of fishing on lower trophic levels; 'fishing through', in which there was an expansion of fishing from some higher trophic species to other higher and lower trophic species; fishing 'based on availability', in which those species that were most abundant and accessible were exploited first, followed by expansion to less available and abundant species; and 'increase to overfishing', in which exploitation rates of all species gradually increased until they were overfished. The simulation projections were used to compute catch- and abundance-weighted trophic indices and compare their time series.

Branch and colleagues' second method was to compare catch- and abundance-weighted mean trophic levels for individual ecosystems. They used relative abundance from trawl surveys from 29 ecosystems, representing regions in the Northern and Southern Hemispheres, five continents and various latitudes, to calculate ecosystem (abundance-weighted) trophic indices. Additionally, estimates of absolute abundance from a database of 242 single-species stock assessments were also used to compute abundance-weighted trophic indices.

The results showed an inconsistent relationship between catch- and abundance-weighted trophic indices. In other words, catch-weighted trophic indices are not generally indicative of the changes in trophic level of the ecosystem. For example, simulated trophic indices from the ecosystem models, as depicted in the top two rows of the authors' Figure 1 (page 431), showed that in some cases the decline in ecosystem mean trophic level (blue lines) was more rapid than that of the catch mean trophic level (red lines), particularly when 'fishing down' occurred. In other cases, the change in mean trophic level of either the catch or the ecosystem was hardly noticeable, yet many species were depleted.

When the individual ecosystems were examined, almost half of the comparisons between catch mean trophic level and ecosystem mean trophic level from trawl or stock-assessment data were found to be negatively correlated. In particular, the relationship between catch and ecosystem trophic level tends to break down when fishing is not distributed across all portions of the ecosystem. On the face of it, then, the way forward is to use abundance-weighted rather than catch-weighted indices.

However, abundance databases have their limitations, too. Although stock-assessment estimates of abundance are considered to provide the best available data2,3, the suite of species for which such assessments are done are limited, being driven by economic and management considerations rather than ecological factors. Trawl survey data provide relative abundance estimates that are skewed by differential susceptibilities of the species and sizes to the sampling gear. Additionally, surveys are not normally designed to sample top predators. The results of Branch et al. highlight the need to expand research to estimate abundance through stock assessments of a broader range of species and more extensive trawl surveys.

But is there still some utility in using catch-weighted mean trophic levels? Perhaps so. Branch and colleagues' results1 suggest conditions in which they might be useful (for example, to indicate major shifts in exploitation patterns). Additionally, catch-weighted trophic level might be used as a 'policy-triggering' tool rather than as a monitoring index — that is, a major change in catch mean trophic level would trigger more detailed research and/or more precautionary management strategies. Indeed, it can be argued that this is exactly how catch-weighted mean trophic levels have been used previously, in that they have provoked consideration of broad ecosystem policy issues. However, further simulation research is needed to evaluate which management actions are most effective for specific ecosystems. Branch et al. have provided the basis for doing that.


  1. 1

    Branch, T. A. et al. Nature 468, 431–435 (2010).

  2. 2

    Polacheck, T. Mar. Policy 30, 470–482 (2006).

  3. 3

    Worm, B. et al. Science 325, 578–585 (2009).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Powers, J. Measuring biodiversity in marine ecosystems. Nature 468, 385–386 (2010) doi:10.1038/468385a

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.