Elsevier

Biological Conservation

Volume 184, April 2015, Pages 290-299
Biological Conservation

Using null models to identify under-represented species in protected areas: A case study using European amphibians and reptiles

https://doi.org/10.1016/j.biocon.2015.02.006Get rights and content

Highlights

  • A new approach is proposed to assess the effectiveness of protected areas (PA).

  • Null models are used to identify over and under-represented species.

  • We test how well European herpetofauna is represented by two PA networks.

  • Most of these species were not better represented than expected by chance.

Abstract

One of the main issues in conservation biology is assessing how much biodiversity is currently represented in protected areas (PA). Traditional approaches such as ‘gap analysis’ require the choice of arbitrary targets and thresholds that can greatly influence the obtained results. We present here a complementary approach that avoids typical methodological uncertainties being particularly useful when the aim is to explore differences in the effectiveness of PA networks in representing species with distinct features and varying range sizes. Firstly, we calculated how far the distribution of a species overlaps with a network. Then, null models were used to test if this value is significantly different from random expectations (i.e. compared with random species of the same number of occurrences), which allowed over and under-represented species to be identified. Using this approach, we aimed to determine how well amphibian and terrestrial reptile species in Europe were represented by two protected area networks: nationally designated protected areas (NPAs) and the Natura 2000 network (N2000). We also tested to see if there were any differences in species representation depending upon their conservation status, range size and distribution type. Although N2000 is more effective than NPAs, both PA networks performed poorly in representing European amphibians and reptiles, as the level of representativeness for most species (excepting reptiles in N2000) within these networks was either not significantly different or significantly lower than expected by chance. A combination of this approach with traditional gap analyses could provide valuable information to improve the future effectiveness of PAs.

Introduction

Europe is one of the most intensively studied areas of the world with respect to its biota (Hawksworth and Bull, 2008). Biodiversity of Europe includes about 1500 vertebrate species, 20–25,000 species of vascular plants and well over 100,000 invertebrate species (IUCN European Red List, 2010). Apart from this high diversity of flora and fauna, a large proportion of European species can be considered as endemic to the region: from 27% of terrestrial mammal species to 88% of freshwater molluscs (IUCN European Red List, 2010). However, it must be also considered that Europe is a continent where landscapes have been subjected to an almost unparalleled amount of human disturbance and industrial development (Hawksworth and Bull, 2008).

Biodiversity loss is one of the main environmental challenges facing the planet (Sala et al., 2000). The establishment of protected area (PA) systems is the most common approach to prevent this loss of biodiversity, and is the mainstay of current conservation policies (Gaston et al., 2008, Le Saout et al., 2013). Consequently, one of the main issues in conservation biology is assessing how much biodiversity is currently represented by PAs using various methods (Cabeza and Moilanen, 2001, Pressey, 1994, Rodrigues et al., 2004b). One of the most common is ‘gap analysis’, a planning approach based on the assessment of the comprehensiveness of existing protected area networks (Jennings, 2000, Scott et al., 1993) by overlapping the distribution maps of individual species, habitats or other attributes of biodiversity with the PA network (Campos et al., 2013, D’Amen et al., 2011, e Silva et al., 2014, Rodrigues et al., 2004a, Vasconcelos et al., 2012). This allows exploration of the amount and nature of biodiversity falling within or outside these PAs and the identification of elements that need further protection through systematic conservation planning (Margules and Pressey, 2000).

However, this approach involves two major methodological problems when deciding if a species is “represented” in a PA. Firstly, the resolution of species distribution data is often coarser than that of PAs. Reliable species distribution data are usually available as regular grids on maps (e.g., distribution atlases), especially in broad scale studies, while reserves are represented as vectorial data, making it difficult to know whether species recorded in a particular grid cell (or spatial unit) actually occur within PAs. Thus, matching PAs with coarse planning units for species requires the definition of arbitrary rules to assign PAs to these planning units (Araújo, 2004), and subjectivity in the choice of a specific threshold to consider a cell as part of the PA would affect the results of species representation (see Abellán and Sánchez-Fernández, 2015, Araújo, 2004). The second uncertainty lies in the choice of a representation target to consider a species as represented by a PA network (Vimal et al., 2011). Frequently, a species is considered as represented or covered in gap analyses when it is present in at least in one, or at most a few, of the cells identified as a PA (e.g. Araújo et al., 2007, Rodrigues et al., 2004b). Although this is the most traditional measure of effectiveness (Rodrigues et al., 1999), it may be misleading, especially if the species are only represented at sites that are inadequate to ensure their persistence (Carvalho et al., 2011a, Gaston et al., 2002, Sánchez-Fernández et al., 2013). Hence, recent papers claim for the need to assess sensitivity of reserve effectiveness assessments to data and decision-rules (e.g. see Abellán and Sánchez-Fernández (2015) for an example using the same dataset than here), as the obtained results could be completely different depending on (i) the threshold used to solve the mismatch between the spatial resolution of data for species and PAs (Araújo, 2004) and (ii) the different conservation targets used to consider a species as represented by a PA network (Vimal et al., 2011). Furthermore, such targets, criteria and thresholds usually focus on just a part of the entire distribution of a species, are arbitrary and sometimes difficult to defend, especially when conservation priorities are appraised against competing land uses (Young et al., 2005).

A complementary form of evaluation that avoids fixed targets would be to explore the representation of the entire distribution of a given species in a PA network. Then, null models could be used to test the departure of the species’ representation in PAs from that expected by chance (based on the same number of planning units with occurrences of the species taken at random), which would allow identifying if the species is over and under-represented by the PA network. Although null models have been used to assess the effectiveness of protected areas, this was always done with the purpose of estimating the number of species that would be expected to be represented by chance given a determined number of cells, using the target and threshold as explained above (e.g. Araújo et al., 2007, Popescu et al., 2013).

We present here such an alternative and complementary approach to identify under or over-represented species by protected areas, avoiding traditional methodological uncertainties and providing a value of significance using null models. As an example, using this approach, we determined how well amphibian and terrestrial reptile species in Europe were represented by two protected area networks: nationally designated protected areas (NPAs) and the Natura 2000 network (N2000). We also tested to see if there were any differences in species representation depending on their conservation status, range size, type of distribution and if they were endemic to Europe or not. We expected that threatened endemic species with narrow distribution in Europe would be better represented in PAs than the remaining species and also than random species with the same number of occurrences. We focused on amphibians and reptiles because (i) they are two of the groups with the most comprehensive and updated information on their distribution in Europe (Sillero et al., 2014); and (ii) most of their species are threatened worldwide (e.g. Alford and Richards, 1999, Böhm et al., 2013, Gibbons et al., 2000, Houlahan et al., 2000), and especially in highly human-transformed landscapes such as European territory (IUCN European Red List, 2010).

Section snippets

Species data and study area

Species distribution data for both reptiles and amphibians were obtained from the recently available New Atlas of European Amphibians and Reptiles (NA2RE; Sillero et al., 2014; see also http://na2re.ismai.pt) which covers all European herpetofauna. This study provides taxonomical information and maps species distributions on an equal-area grid with cells of 50 × 50 km based on the Universal Transverse Mercator (UTM) projection and the Military Grid Reference System (MGRS). We used part of NA2RE

Results

The MPO between amphibians and both NPAs and N2000 ranged from 0.8% to 54.5% (mean 16.95 ± SD 9.73%) and from 13.3% to 47.4% (21.95 ± 7.42%), respectively (Table 1). Values for reptiles were quite similar, ranging from 2.9% to 60.6% (mean 16.99 ± 10.55%) and from 14.4% to 79.5% (27.84 ± 10.12%) for NPAs and N2000, respectively (Table 1). We found that only 19 (29.7%) and 22 (34.4%) out of the 64 amphibian species included in the analyses were better represented within the NPAs and N2000, respectively,

Discussion

The approach presented here offers a complementary view to traditional gap analysis in order to assess how well biodiversity is represented by PA systems. One of the main advantages of our approach over traditional gap analyses is that it allows obtaining a continuous value of representativeness for each one of the species from the degree of spatial overlap between its entire distribution in the study area and the protected area system, thus avoiding the need to select an arbitrary

Concluding remarks

Our approach avoids two major typical methodological uncertainties usually involved in gap analysis assessments; namely, the definition of rules (i) to assign PAs to PU, and (ii) to identify species as represented within a given PAs network. The method presented here provides complementary information to traditional assessments of protected areas’ effectiveness or gap analyses, and it is particularly useful especially when the aim is to explore differences in the effectiveness of PA networks in

Acknowledgments

We thank Salvador Carranza for his interesting comments on previous versions of the manuscript and Melissa Crim and Ana María Soria for checking the English. We also thank two anonymous referees for their constructive comments. D.S.-F. was supported by a postdoctoral grant (Juan de la Cierva program) from the Spanish Ministry of Economy and Competitiveness.

References (44)

  • M.B. Araújo et al.

    Climate warming and the decline of amphibians and reptiles in Europe

    J. Biogeogr.

    (2006)
  • M.B. Araújo et al.

    The effectiveness of Iberian protected areas in conserving terrestrial biodiversity

    Conserv. Biol.

    (2007)
  • M. Cabeza et al.

    Design of reserve networks and the persistence of biodiversity

    Trends Ecol. Evol.

    (2001)
  • F.S. Campos et al.

    Threatened amphibians and their conservation status within the protected area network in Northeastern Brazil

    J. Herpetol.

    (2013)
  • S.B. Carvalho et al.

    Incorporating evolutionary processes into conservation planning using species distribution data: a case study with the western Mediterranean herpetofauna

    Divers. Distrib.

    (2011)
  • M. D’Amen et al.

    Protected areas and insect conservation: questioning the effectiveness of Natura 2000 network saproxylic beetles in Italy

    Anim. Conserv.

    (2013)
  • P.G. Dimitrakopoulos et al.

    Questioning the effectiveness of the Natura 2000 special areas of conservation strategy: the case of Crete

    Global Ecol. Biogeogr.

    (2004)
  • V.D.N. e Silva et al.

    Formulating conservation targets for a gap analysis of endemic lizards in a biodiversity hotspot

    Biol. Conserv.

    (2014)
  • European Environment Agency, 2012. Protected areas in Europe: an overview. EEA Report 5/2012,...
  • K. Gaston et al.

    Persistence and vulnerability: retaining biodiversity in the landscape and in protected areas

    J. Biosci.

    (2002)
  • K.J. Gaston et al.

    Protected areas in Europe: principle and practice

    Ann. NY Acad. Sci.

    (2008)
  • J.W. Gibbons et al.

    The global decline of reptiles, Déjà vu Amphibians

    Bioscience

    (2000)
  • Cited by (30)

    • Combining spatial dependence occupancy models and conservation gap analyses to promote species conservation: A case study with a threatened semi-aquatic mammal

      2022, Biological Conservation
      Citation Excerpt :

      All indices showed that N2000 and NPh failed to protect the most suitable habitat for G. pyrenaicus, despite the latter being the strictest conservation tool in the French Pyrenees. Even if about 40% of high occupancy sections overlapped with N2000 category, we highlighted that this overlap is worse than a random selection of sections to protect (Sánchez-Fernández and Abellán, 2015). Last, about 25% of sections protected by at least one PA category are highly suitable for G. pyrenaicus.

    • A functional perspective for global amphibian conservation

      2020, Biological Conservation
      Citation Excerpt :

      Amphibians are undergoing an extinction crisis; out of the 7876 living extant species (Frost, 2018), nearly 40% are currently threatened with extinction (Hoffmann et al., 2010; Jenkins et al., 2013; Pimm et al., 2014; Stuart, 2004). On top of that, it is well documented that the global network of protected areas (PAs) could be much more efficient than they really are today in protecting threatened amphibians (Nori et al., 2015; Rodrigues et al., 2004; Sánchez-Fernández and Abellán, 2015). Besides, more than 40% of non-threatened amphibians are still poorly known by science, and presumably most of them face the same threats (Hof et al., 2011; Howard and Bickford, 2014; Morais et al., 2013; Nori and Loyola, 2015; Nori et al., 2018).

    • Vertebrate biodiversity on indigenous-managed lands in Australia, Brazil, and Canada equals that in protected areas

      2019, Environmental Science and Policy
      Citation Excerpt :

      Habitat loss and degradation due to human land conversion are key threats to global biodiversity that, to date, have been addressed mainly by expanding protected areas (PAs) globally (Maxwell et al., 2016; Sala et al., 2000). However, this approach has severe limitations because many existing PAs have limited overlap with the geographic ranges of most of the world’s threatened species, as PAs have often been placed in regions with relatively low economic value and biodiversity (Rodrigues et al., 2004; Sánchez-Fernández and Abellán, 2015; Venter et al., 2014). In several regions, PAs overlap the ranges of endemic, high-priority species less often than expected if they had been located randomly (Nori et al., 2015; Sánchez-Fernández and Abellán, 2015).

    View all citing articles on Scopus
    View full text