Community management indicators can conflate divergent phenomena: two challenges and a decomposition‐based solution

Abstract

Summary1. Community indicators are used to assess the state of ecological communities and to guide management. They are usually calculated from monitoring data, often collected annually. Since any given community indicator provides a univariate summary of complex multivariate phenomena, different changes in the community may lead to the same response in the indicator. Sampling variation can also mask ecologically important trends. 2. This study addresses these challenges for community indicators, with a focus on the large fish indicator (LFI), internationally used to report status of marine fish communities. The LFI expresses 'large' fish biomass as a proportion of total fish biomass and is calculated from species-size-abundance data collected on trawl surveys. We develop new methods to decompose the contributions of species, sampling locations and season to trends over time in the LFI, and highlight consequences for assessment and management. 3. Our results showed that both species and locations made divergent contributions to overall trends in the LFI indicator, with contributions differing by several orders of magnitude and in sign. Only small proportions of species and locations drove overall LFI trends, and their contributions changed with season (spring and autumn surveys). To assess significance of component trends, a resampling method was developed. Our method can be generalized and applied to many other community indicators based on survey data. 4. Synthesis and applications. Our new method for decomposing community indicators and generating confidence intervals makes it possible to extract much more information on what drives a 'headline' indicator, providing a solution to challenges arising from multiple possible interpretations of changes in the indicator and from sampling variation. Analysis of the effects of indicator components on headline indicator values is recommended, because the results allow assessors and managers to identify and interpret how divergent factors (e.g. species, sampling locations and seasons) contribute to the headline indicator value.

Sampling effects on the effectiveness of ecological indicators in detecting fishery-induced community changes

“…Besides, fisheries survey may be conducted in certain time/season each year, while seasonality is ignored in the structure of the "mizer" model (Scott et al 2014;Datta and Blanchard 2016). The locations of survey programs may substantially influence the observation of EIs (Adams et al 2017), which is not accounted for. In addition, our simulations assume species-specific catchability and a knife-edge type of selectivity to approximate the typical sigmoid selectivity of trawls as a result of data limitation, whereas the selectivity curve of trawl may be in different shapes, e.g., the dome selectivity (Sampson and Scott 2012).…”

“…In this context, our conclusions on fishing variability are D r a f t constrained to an idealized change of fishing effort, whereas the impact of fishing variability may depend on the nature of the underlying transition as well as the diverse characteristics of fisheries. Moreover, if the fluctuations of environment variables and recruitment variability are taken into account, the effectiveness of EIs may further degrade and the interpretation may be complicated as the different drivers of the ecosystem may lead to the same response (Adams et al 2017). Therefore, the quantitative results of this study should be used as caveats rather than references in practice.…”

“…We distinguish between sampling uncertainty caused by sampling errors and sampling bias in this study (Adams et al 2017). The former may arise from the variability of sampling operations, which is considered stochastic; and the latter may D r a f t arise from sampling methods, such as the selectivity of survey gears, which is considered as systematic bias.…”

Climate change alters fish community size‐structure, requiring adaptive policy targets

“…mean trophic level, Adams et al, 2017). The causes of temporal trends in community-level indicator time series can thus be ambiguous, resulting from complex, multivariate environmental phenomena that can affect fish communities alongside fishing pressure and climate change (Adams et al, 2017;Shin et al, 2005). These aspects constitute challenges to the practical application of the LFI in support of fisheries management and indeed to that of any community-level ecological indicator (Branch et al, 2010).…”

“…the European MSFD, EC, 2008). As a consequence, the setting of targets for policies affecting fisheries management is often carried out without an assessment of how climate change affects the indicators we use to monitor resources (Adams, Jennings, & Reuman, 2017). Policy assessment cycles (e.g.…”