Elsevier

Science of The Total Environment

Volume 625, 1 June 2018, Pages 1596-1605
Science of The Total Environment

Modelling landscape constraints on farmland bird species range shifts under climate change

https://doi.org/10.1016/j.scitotenv.2018.01.007Get rights and content

Highlights

  • Climate-only and climate plus landscape models yielded different future species range shifts.

  • Different models also predicted contrasting geographical patterns of change in species richness.

  • Habitat farmland specialists tend to expand, whereas generalist to retract under the same scenarios.

  • Predicted expansions under climate change scenarios are severely constrained by predicted landscape changes.

  • Incorporating landscape factors is crucial to forecast range shifts of farmland habitat specialists.

Abstract

Several studies estimating the effects of global environmental change on biodiversity are focused on climate change. Yet, non-climatic factors such as changes in land cover can also be of paramount importance. This may be particularly important for habitat specialists associated with human-dominated landscapes, where land cover and climate changes may be largely decoupled. Here, we tested this idea by modelling the influence of climate, landscape composition and pattern, on the predicted future (2021–2050) distributions of 21 farmland bird species in the Iberian Peninsula, using boosted regression trees and 10-km resolution presence/absence data. We also evaluated whether habitat specialist species were more affected by landscape factors than generalist species. Overall, this study showed that the contribution of current landscape composition and pattern to the performance of species distribution models (SDMs) was relatively low. However, SDMs built using either climate or climate plus landscape variables yielded very different predictions of future species range shifts and, hence, of the geographical patterns of change in species richness. Our results indicate that open habitat specialist species tend to expand their range, whereas habitat generalist species tend to retract under climate change scenarios. The effect of incorporating landscape factors were particularly marked on open habitat specialists of conservation concern, for which the expected expansion under climate change seems to be severely constrained by land cover change. Overall, results suggest that particular attention should be given to landscape change in addition to climate when modelling the impacts of environmental changes for both farmland specialist and generalist bird distributions.

Introduction

Many studies have attempted to estimate the future effects of global environmental changes on biodiversity (e.g., Thuiller et al., 2005, Araújo et al., 2006, Garcia et al., 2011). Many of these studies examine the effect of climate change alone, leaving aside the effects of non-climatic drivers (e.g., Harfoot et al., 2014, Morelli and Tryjanowski, 2015, Titeux et al., 2016). However, the effects of climate change on biodiversity can be influenced by interactions with other components of global change (e.g., Clavero et al., 2011, Hof et al., 2011, Maxwell et al., 2016), particularly with land use changes and related effects on other pressures such as water regime (e.g., Jetz et al., 2007, Rosenzweig et al., 2008, Thuiller et al., 2014a; Newbold et al., 2016).

Both climate and land cover changes are considered major drivers of global biodiversity change (Sala et al., 2000, Jetz et al., 2007, de Chazal and Rounsevell, 2009). However, climate is often regarded as the most important driver at large spatial extents and coarse spatial resolutions (e.g., Thuiller et al., 2004a, Luoto et al., 2007, Triviño et al., 2011). The relative contribution of climate and land cover on future species range shift projections remain poorly explored (Pearce-Higgins and Green, 2014; but see studies from Table S1, Appendix A). Previous studies have found that land cover can be correlated with climate and that including land cover variables did not improve the accuracy of species distribution models, as expected (e.g., Seoane et al., 2003, Thuiller et al., 2004a, Triviño et al., 2011, Reino et al., 2013). Although climate and land cover are generally correlated, however, climate does not necessarily fully control land cover, which may be affected by a number of additional factors such as soil type, topography, socio-economic contexts and policies (Veldkamp and Lambin, 2001, Ribeiro et al., 2014, Ribeiro et al., 2016). In addition, because climate and land cover often play key roles at different spatial scales (Pearson et al., 2004), they are likely to show different geographical patterns of change and hence may affect different regions in a distinct way. Moreover, climate and landscape drivers may interact in their effect on species geographical range, because the two drivers may have different effects on different groups of species (Opdam and Wascher, 2004, Sohl, 2014, Jarzyna et al., 2015).

Most studies modelling the consequences of changes in the structural component of landscape have ignored potentially important processes related to landscape fragmentation (e.g., Vallecillo et al., 2009, Triviño et al., 2011), although they have been shown to be determinant for some species (Jarzyna et al., 2015). This is the case, for instance, of some farmland bird specialists that were shown to be very sensitive to habitat fragmentation at several spatial scales (Reino et al., 2009, Reino et al., 2013). There is a well-established idea that generalist species tend to cope better with environmental changes than specialist species (Gilman et al., 2010, Clavel et al., 2011, Davey et al., 2012, Lurgi et al., 2012, Case et al., 2015). However, at the same time, some studies point to an idiosyncratic nature of species responses to climate change, making it difficult to draw generalizations (e.g., Mair et al., 2012, Moritz and Agudo, 2013, Sohl, 2014). For example, in a recent study, Princé et al. (2015) found that the relative sensitivity of farmland bird specialists and generalists to climate and land cover changes varied among the different global change scenarios that were considered.

Here we model the relative importance of changes at the landscape level on range shifts predictions under future environmental change scenarios, aiming at bringing new insights on the interplay of three component of the environment: biosphere, atmosphere and anthroposphere. We focused on farmland birds in the Iberian Peninsula, considering both climate change and changes in land cover and landscape structure, mainly as the result of land abandonment and changes in agricultural practices, associated with three socioeconomic scenarios for the period of 2021 to 2050. We hypothesize that taking into account changes in landscape composition and structure will potentially strongly affect predictions of farmland bird geographical ranges under climate change scenarios. We also expect that the potential impacts of landscape changes on farmland bird geographical ranges is dependent on the degree of habitat specialization (Clavel et al., 2011), namely the association to farmland landscapes. The overarching goal of this study is thus to examine the proposition that landscape changes should be accounted for when forecasting the effects of environmental changes on the distribution of species highly sensitive to landscape structure.

Section snippets

Data

We used distributional records for 21 Iberian farmland bird species (Table 1), obtained from the most recent breeding bird atlas from Spain (Martí and Moral, 2003) and Portugal (Equipa Atlas, 2008), reporting the occurrence of bird species in 5923 10 × 10 km resolution UTM cells. These are the highest-resolution bird distribution data available for Iberia. Farmland birds selected for this study include species with different degrees of habitat specialization to open habitats, because these seem to

Model performance

The incorporation of landscape variables in the climate-based models consistently improved model's discrimination ability, as measured with mean cross-validation AUC values (Wilcoxon signed rank test, p < 0.001). However, the contribution of landscape variables did not increase substantially the discrimination ability of models, with percent (%) of improvements varying from 0.06% for Calandrella rufescens to 4.68% for Circus pygargus (Table 1). The species Landscape Specialization Index was

Discussion

Our results confirm the hypothesis that the inclusion of landscape variables in species distribution models strongly affect range shift predictions of Iberian farmland birds, despite a generally low contribution to models' performance. Overall, the resulting species distribution models predict that habitat specialists will tend to expand their range, whereas generalists will tend to retract under climate change scenarios. However, in many cases, the inclusion of landscape variables in the

Conclusions

This study underpins the need to consider landscape composition and structure when modelling species range shifts under future climate scenarios. This is particularly the case for habitat specialists, which are strongly constrained by habitat availability and configuration. In addition, our models show that specialist species (many with relevant European conservation concern, Table S2) produce less optimistic predictions when landscape changes are also accounted for. The interplay between

Acknowledgments

This study was funded by the Portuguese Ministry of Science, Technology and Higher Education and the European Social Fund, through the Portuguese Foundation of Science and Technology (FCT), under POPH - QREN - Typology 4.1, through the grants SFRH/BPD/93079/2013 (L.R.) and the contract (IF/01304/2015) (PS) under the IF Researcher Programme and through the project PTDC/BIA-BIC/2203/2012-FCOMP-01-0124-FEDER-028289 by FEDER Funds through the Operational Programme for Competitiveness Factors –

References (70)

  • J. Seoane et al.

    The effects of land use and climate on red kite distribution in the Iberian peninsula

    Biol. Conserv.

    (2003)
  • A. Veldkamp et al.

    Predicting land-use change

    Agric. Ecosyst. Environ.

    (2001)
  • M.B. Araújo et al.

    Climate warming and the decline of amphibians and reptiles in Europe

    J. Biogeogr.

    (2006)
  • M.B. Araújo et al.

    Biodiversidade e Alterações Climáticas/Biodiversidad y Alteraciones Climáticas

    (2012)
  • J. Clavel et al.

    Worldwide decline of specialist species: toward a global functional homogenization?

    Front. Ecol. Environ.

    (2011)
  • M. Clavero et al.

    Climate change or land-use dynamics: do we know what climate change indicators indicate?

    PLoS One

    (2011)
  • P. Clifford et al.

    Assessing the significance of the correlation between two spatial processes

    Biometrics

    (1989)
  • C.M. Davey et al.

    Rise of the generalists: evidence for climate driven homogenization in avian communities: avian diversity, homogenization and warming climate

    Glob. Ecol. Biogeogr.

    (2012)
  • N. Dendoncker et al.

    A statistical method to downscale aggregated land-use data and scenarios

    J. Land Use Sci.

    (2006)
  • V. Devictor et al.

    Distribution of specialist and generalist species along spatial gradients of habitat disturbance and fragmentation

    Oikos

    (2008)
  • S.M. Eglington et al.

    Disentangling the relative importance of changes in climate and land-use intensity in driving recent bird population trends

    PLoS One

    (2012)
  • J. Elith et al.

    A working guide to boosted regression trees

    J. Animal Ecol.

    (2008)
  • A.H. Fielding et al.

    A review of methods for the assessment of prediction errors in conservation presence/absence models

    Environ. Conserv.

    (1997)
  • Equipa Atlas

    Atlas das aves nidificantes em Portugal (1999–2005)

    (2008)
  • ESRI

    ArcGIS Desktop: Release 10

    (2011)
  • A. Estrada et al.

    Forecasting large-scale habitat suitability of European Bustards under climate change: the role of environmental and geographic variables

    PLoS One

    (2016)
  • European Environmental Agency

    CORINE Land Cover, 2000

    (2002)
  • J.H. Friedman

    Greedy function approximation: a gradient boosting machine

    Ann. Stat.

    (2001)
  • J.H. Friedman et al.

    Additive logistic regression: a statistical view of boosting

    Ann. Stat.

    (2000)
  • R.A. Garcia et al.

    African vertebrate species under warming climates: sources of uncertainty from ensemble forecasting

    Glob. Chang. Biol.

    (2011)
  • M.B.J. Harfoot et al.

    Emergent global patterns of ecosystem structure and function from a mechanistic general ecosystem model

    PLoS Biol.

    (2014)
  • T. Hastie et al.

    The Elements of Statistical Learning: Data Mining, Inference, and Prediction

    (2001)
  • B.A. Hawkins et al.

    Energy, water, and broad-scale geographic patterns of species richness

    Ecology

    (2003)
  • R.J. Hijmans et al.

    Package ‘dismo’

  • C. Hof et al.

    Climate change, pathogens and land use change additively threat amphibian diversity

    Nature

    (2011)
  • Cited by (0)

    View full text