Elsevier

Biological Conservation

Volume 213, Part B, September 2017, Pages 341-350
Biological Conservation

Assessing the suitability of diversity metrics to detect biodiversity change

https://doi.org/10.1016/j.biocon.2016.08.024Get rights and content

Highlights

  • Synthetic communities are used to explore the behaviour of 12 biodiversity metrics.

  • Metrics behaviour is assessed under different scenarios of biodiversity change.

  • Abundance-weighted heterogeneity metrics are unsuitable for biodiversity monitoring.

  • Sørensen similarity index and geometric mean abundance show desirable properties.

  • Separate metrics of species composition and abundance should be reported.

Abstract

A large number of diversity metrics are available to study and monitor biodiversity, and their responses to biodiversity changes are not necessarily coherent with each other. The choice of biodiversity metrics may thus strongly affect our interpretation of biodiversity change and, hence, prioritization of resources for conservation. Therefore it is crucial to understand which metrics respond to certain changes, are the most sensitive to change, show consistent responses across different communities, detect early signals of species decline, and are insensitive to demographic stochasticity. Here we generated synthetic communities and simulated changes in their composition according to 9 scenarios of biodiversity change to investigate the behaviour of 12 biodiversity metrics. Metrics showed diverse abilities to detect changes under different scenarios. Sørensen similarity index, arithmetic and geometric mean abundance, and species and functional richness were the most sensitive to community changes. Sørensen similarity index, species richness and geometric abundance showed consistent responses across all simulated communities and scenarios. Sørensen similarity index and geometric mean abundance were able to detect early signals of species decline. Geometric mean abundance, and functional evenness under certain scenarios, had the greatest ability to distinguish directional trends from stochastic changes, but Sørensen similarity index and geometric mean abundance were the only indices to show consistent signals under all replicates and scenarios. Classic abundance-weighted heterogeneity indices (e.g. Shannon index) were insensitive to certain changes or showed misleading responses, and are therefore unsuitable for comparison of biological communities. We therefore suggest that separate metrics of species composition, richness, and abundance should be reported instead of (or in addition to) composite metrics like the Shannon index.

Introduction

In a period of rapid global change, monitoring biodiversity changes is key to detect early warning signals of decline, infer the causes of such decline, and develop effective conservation strategies to mitigate it (Ash et al., 2009, Balmford et al., 2003, Balmford et al., 2005, Buckland et al., 2005, Butchart et al., 2010, Gregory et al., 2005, Nichols and Williams, 2006, Tittensor et al., 2014). The multifaceted nature of biodiversity (Gaston, 1996, Purvis and Hector, 2000) is studied through a large number of metrics. Different metrics measure different components of biodiversity such as species richness, abundance, evolutionary history (i.e. phylogenetic diversity; Faith, 1992), and functional traits (Mason et al., 2005). However, as no single metric captures all relevant aspects of biodiversity, none of them taken individually can provide a full picture of the patterns of change. Further, metrics can even be misleading if considered individually. For instance, the geometric mean abundance can increase if rare species increase in abundance, while total abundance is decreasing (Schipper et al., 2016). Similarly, invasive species can increase species richness or functional and phylogenetic diversity, while having negative impacts on the abundances of native species (Thomas, 2013, Winter et al., 2009). The rate and direction of change in a metric may also depend on idiosyncrasies in the state of the initial community, and/or natural ecological succession. Moreover, in addition to directional changes in biodiversity, species relative abundances may fluctuate over shorter time frames due to demographic stochasticity or competitive and predator-prey dynamics. This “noise” can confound the signal of interest (i.e. directional change in response to a specific driver).

The choice and response of biodiversity metrics may strongly affect our interpretation of biodiversity change and, hence, prioritization of resources for conservation (Gaston and Spicer, 2004, Purvis and Hector, 2000). Thus, it is crucial to understand how alternative metrics respond to specific changes, which metrics are the most sensitive in order to detect early signals of biodiversity decline, and which ones respond consistently to changes. Empirical datasets allow investigating how metrics change in space and time, but have several limitations. These include the limited number of possible scenarios and communities represented, and the lack of control on the underlying cause of change, the likely co-existence of several mechanisms of decline (e.g., decline of habitat specialists due to the loss of their habitat type and decline of large species due to overexploitation). This complicates the attempts to link the behaviour of a diversity metric to a definite mechanism of biodiversity change. Virtual datasets allow full control of both the community composition and the mechanism of decline, and thus allow the comparison of the relative responses of the diversity metrics (Zurell et al., 2010) by simulating ecological processes under alternative scenarios (Dornelas, 2010, Lamb et al., 2009, Münkemüller and Gallien, 2015, Olden and Poff, 2003, Supp and Ernest, 2014).

In this study, we explored the behaviour of a set of diversity metrics under different scenarios of biodiversity change. To this end, we generated synthetic communities and simulated changes in their composition to investigate the responses of the metrics. We recorded how metrics changed over time under each scenario, and identified those that were most sensitive to these community changes and showed a consistent response irrespective of the state of the original community. We also assessed non-linearity in metrics responses, and their effect on our ability to detect early warning signals of biodiversity change. Finally, we measured the signal-to-noise ratio (SNR) of the metrics under each scenario to compare the metrics' ability to detect directional changes in biological communities.

Section snippets

Virtual dataset

We assumed a landscape area of 10,000 km2 consisting of two habitats, one dominant and one secondary. For convenience we will refer to these habitats as forest and grassland, respectively. The size of the landscape was chosen such that it was large enough to allow each species to form a population from ~ 15 to > 50,000 individuals. Forest covered a random proportion between 0.7 and 0.9 of the entire landscape.

We generated 150 species, and randomly assigned to each a diet, body mass, population

Metric behaviour under alternative scenarios

The diversity metrics exhibited different temporal trends under the nine scenarios of biodiversity change (Fig. 2, Fig. 3, Fig. A1, Fig. A2, Fig. A3, Fig. A4, Fig. A5, Fig. A6, Fig. A7, Fig. A8). Under the “Uniform decline” scenario, where all species decreased by the same number of individuals and rare species went extinct first, all metrics showed a decrease, especially species richness, functional richness and functional dispersion (Fig. 2). The “Proportional decline” scenario, where all

Discussion

Simulating biodiversity change through time allowed us to explore the behaviour of a set of biodiversity metrics and assess their suitability for monitoring biodiversity change, including declines in species' abundances that can be of conservation concern. Richness-based metrics require presence data, which is less time-consuming and costly to collect than abundance data (Costello et al., submitted for publication). Knowing which species are present, particularly those that are ecologically

Acknowledgements

We thank R D Gregory and another anonymous reviewer for providing constructive comments on earlier versions of the manuscript. This article is based upon work from COST Action ES1101 “Harmonising Global Biodiversity Modelling” (Harmbio), supported by COST (European Cooperation in Science and Technology).

References (69)

  • A. Balmford et al.

    The Convention on Biological Diversity's 2010 target

    Science

    (2005)
  • T.M. Blackburn et al.

    A critical assessment of the form of the interspecific relationship between abundance and body size in animals

    J. Anim. Ecol.

    (1997)
  • U. Brose et al.

    Consumer-resource body-size relationships in natural food webs

    Ecology

    (2006)
  • J.H. Brown et al.

    Toward a metabolic theory of ecology

    Ecology

    (2004)
  • S.T. Buckland et al.

    Monitoring change in biodiversity through composite indices

    Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci.

    (2005)
  • S.T. Buckland et al.

    The geometric mean of relative abundance indices: a biodiversity measure with a difference

    Ecosphere

    (2011)
  • S.H.M. Butchart et al.

    Global biodiversity: indicators of recent declines

    Science

    (2010)
  • M. Cardillo et al.

    Multiple causes of high extinction risk in large mammal species

    Science

    (2005)
  • CBD

    Strategic Plan for Biodiversity 2011–2020. Montreal

    (2010)
  • M.J. Costello et al.

    Amphipod fauna of the sponges Halichondria panicea and Hymeniacidon perleve in Lough Hyne, Ireland

    Mar. Ecol. Prog. Ser.

    (1987)
  • M.J. Costello et al.

    Distinguishing essential and fundamental biodiversity variables

    Biol. Conserv.

    (2016)
  • F. Courchamp et al.

    Rarity value and species extinction: the anthropogenic allee effect

    PLoS Biol.

    (2006)
  • J. Damuth

    Interspecific allometry of population density in mammals and other animals: the independence of body mass and population energy-use

    Biol. J. Linn. Soc.

    (1987)
  • M. Dornelas

    Disturbance and change in biodiversity

    Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci.

    (2010)
  • M. Dornelas et al.

    Assemblage time series reveal biodiversity change but not systematic loss

    Science

    (2014)
  • J.A. Estes et al.

    Trophic downgrading of planet Earth

    Science

    (2011)
  • K.J. Gaston

    Biodiversity: A Biology of Numbers and Difference

    (1996)
  • K.J. Gaston et al.

    Biodiversity: An Introduction

    (2004)
  • M. Gilpin et al.

    Minimum viable populations: processes of species extinction

  • J.C. Gower

    A general coefficient of similarity and some of its properties

    Biometrics

    (1971)
  • R.D. Gregory et al.

    Wild bird indicators: using composite population trends of birds as measures of environmental health

    Ornithol. Sci.

    (2010)
  • R.D. Gregory et al.

    Developing indicators for European birds

    Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci.

    (2005)
  • M. Hoffmann et al.

    The impact of conservation on the status of the world's vertebrates

    Science

    (2010)
  • S.P. Hubbell

    The Unified Neutral Theory of Biodiversity and Biogeography

    (2001)
  • Cited by (84)

    View all citing articles on Scopus
    View full text