An important issue for marine ecologists and managers is the projection of future spatial distributions of marine populations. Projecting spatial distributions can be a useful but only if they are given with a known and sufficiently high level of confidence. These uncertainties can arise for the observation process, conceptual and numerical model formulations, parameter estimates, model evaluation, appropriate spatial and temporal scales, and the adaptation of living systems. To analyze different sources of uncertainty and the ways they are considered in current studies, 75 publications for 2005–2009 were selected and the frequency of considered type of uncertainty was calculated. What was found was that there is little attention to many sources of uncertainty except for parameter estimates. Unless certainty can be better accounted for, such projections may be of limited use for managerial purposes.–Alyshia Silva
Planque, B., Bellier E., Loots, C. 2011. Uncertainties in Projecting Spatial Distributions of Marine Populations. ICES Journal of Marine Science 68, 1045–1050.
Spatial distributions define the geographical extent of marine population, as well as the abundance of the individuals or density within these geographical boundaries. Projecting spatial distributions for marine populations is becoming a more difficult task and a chief concern for managers, conservationists, and human communities that depend on marine resources. In particular, measuring uncertainty is important because spatial distributions should only be useful if they are given with a high-level of confidence. Major sources of uncertainty are related to the observation process, conceptual and numerical model formulations, parameter estimates, model evaluation, appropriate spatial and temporal scales, and the adaptation of living systems.
The observation process is the way we perceive the marine world, which is already a very limited methodology due to filtered observation instruments and a lack of a uniform method to observe adequately. Our representation of this world is therefore inadequate and incomplete. Conceptual model formulations are mental representations of the processes that control the spatial distribution of marine populations. These models are becoming increasingly difficult to use because environmental conditions can no longer be compared to observable past climatological phenomenon. Numerical implementations within a conceptual model can represent functional relationships, deal with interactions, non-linearity, and complexity in general, and can accommodate various statistical distributions. However, they do not outperform other methods under every circumstance. A model evaluation provides an objective way of measuring model performance and validation on independent datasets is the most robust approach. Spatial and temporal scales are important to understand the distribution and abundance of organisms and inference will be weaker when based upon vague notions of scale than if precise notion of scale is used. The adaptability of living systems is also highly questionable considering that we have never seen these effects before in the history of human-kind. Predicting future changes based upon past observations is highly uncertain, however, ecosystems are highly adaptive and have a strong dependence on historical contingencies.
The author conducted a literature survey that includes the following words and their variations: spatial, fish, distribution, benth, geography, habitat, sea, ocean, marin, and model. These articles are restricted to a period from 2005 to March 2010 within the fields of marine and freshwater biology, oceanography, and fisheries. Seventy-five articles were then selected that presented models that were used or could be used for the projection of spatial distribution of marine populations. Within each article, uncertainty within the previously mentioned criteria was assessed. Overall, little attention is given to the various sources of uncertainties in models and consequently to uncertainties in the resulting projections. Only 5 of the 75 studies explicitly accounted for the observation process in the model design; conceptual model uncertainty only accounted for one of the studies surveyed; uncertainty in the appropriateness of the numerical formula is addressed in one-fourth of the articles; parameter uncertainty was accounted for in 69%; 45% of the model evaluation was based upon visual comparison of predicted and observed distributions were infrequent; spatial and temporal scales were defined before modeling in 45% of the literature; only 4% discussed possible implications of ecological adaptability for projected changes.
Uncertainty in spatial projections has been poorly considered in marine ecological research, indicating that the current projections in marine biota distributions are likely poorly reliable. There is an explicit trend in handling various sources of uncertainty in model projections but a more extensive study would be required to confirm this. Highly uncertain or inaccurate projections could negatively harm managerial and conservation efforts, erasing what “success” we have created if future studies show that these uncertainties are indeed too great to ignore.