Elsevier

Ecological Indicators

Volume 86, March 2018, Pages 1-8
Ecological Indicators

Original Articles
Does a correlation exist between environmental suitability models and plant population parameters? An experimental approach to measure the influence of disturbances and environmental changes

https://doi.org/10.1016/j.ecolind.2017.12.009Get rights and content

Highlights

  • Maxent models fit better if disturbance parameters are included.

  • Distances from streets and villages are important contributors to Maxent models.

  • Maxent model suitability is correlated with plant population parameters.

  • Population parameters are influenced by climate change and presence of trails.

  • Maxent may represent a useful tool for disturbance estimations and predictions.

Abstract

Due to increasing human pressures, there is a need to understand how environmental and anthropogenic disturbances could affect the conservation status of endangered plant species. When information on distribution or population parameters is poor, Species Distribution Models (SDMs) may offer a valuable additional source to assess threats and to evaluate conservation options. In this sense, if the output of SDMs represents the relationships between habitat features and species occurrence, SDM results can also be related to other key parameters of population. For the endangered yellow gentian, we tested the relationship between six field population parameters and the suitabilities obtained by SDMs with natural and limiting parameters (i.e. proxies of disturbances). Specifically, the six population parameters were the surface area covered by each population, the number of vegetative and reproductive individuals per population, the density of reproductive and all individuals per plot and the proportion of reproductive individuals. Thus, threats were evaluated by testing if relationships between population parameters and suitabilities increased when proxies of disturbances were included in models.

Best-fitting models resulted when all natural and human disturbance variables were included. In addition, results show relationships between suitability and population parameters only when disturbance parameters were used for SDMs. When the index related to the sensitivity to climate change was included in SDMs, the density of all individuals and number of reproductive plants were lower than in low suitability sites, suggesting that climate change is likely already challenging the ability of yellow gentian to bloom and germinate. Otherwise, we observed a decrease of the extent of localities in areas with higher suitability obtained through the index related to trail intensity. This confirmed the positive effect of conservation activities, which were mainly implemented in the proximity of trails.

Using a thoroughly studied plant species as a straightforward example, we showed the potentiality of SDMs to inform on population parameters instead of only discriminate species presence or absence. This information can suggest the use of disturbance parameters when specific SDMs aim to support strategic decisions in management and conservation.

Introduction

It is frequently assumed, but rarely tested, that models of environmental suitability (‘Species Distribution Models’, hereafter SDMs) may provide useful indices of environmental quality or other species-specific information, either from an ecological or conservation perspective (Bean et al., 2014). In general, SDMs correlate species occurrence with environmental data (e.g., topography, soil, climate) in order to predict the probability of presence on a map and thus to inform about potential species’ spatial occurrence (e.g. Sousa-Silva et al., 2014, Koch et al., 2017), guide field surveys (e.g. Pearson et al., 2007, Fois et al., 2015), predict impacts of climate and habitat changes (e.g. Fois et al., 2016, López-Tirado and Hidalgo, 2016, Bosso et al., 2017), species invasions (Pěknicová and Berchová-Bímová, 2016, Bosso et al., 2016, Dullinger et al., 2017) and support strategic decisions in management and conservation (e.g. Unglaub et al., 2015, Fois et al., 2016, Smeraldo et al., 2017).

For practical applications, the numerical outputs of statistical SDMs have often been simplified to indices of environmental suitability, ranging from 0 (unsuitable) to 1 (optimal). Otherwise, if the output of SDMs represents a relationship between a species and its environment, it would be possible that SDM results were related not only to the probability of occurrence, but also to other key parameters of populations. Being based on environmental characteristics, such indices of suitability can easily and interestingly be compared with measures in the field. For instance, some authors demonstrated that environmental suitability, obtained through presence-only SDMs, can also be associated to demographic and population parameters such as abundance (e.g. VanDerWal et al., 2009, Bean et al., 2014), reproductive success (e.g. Brambilla and Ficetola, 2012, Swab et al., 2015) and apparent survival (e.g. Weber and Grelle, 2012, Bean et al., 2014). Otherwise, because several studies did not find the expected link between environmental suitability and species demography, such tests should be more in deep evaluated (Thuiller et al., 2010, Unglaub et al., 2015, Weber et al., 2017). Indeed, several ecological processes can lead to deviations from this relationship between demographic parameters and the environmental suitability (Pulliam, 2000, Thuiller et al., 2014). Competitive interactions could, for example, exclude a weak competitor from its optimal environmental conditions, while it might persist in more extreme environments that the dominant competitors cannot occupy (Pulliam, 2000, McGill, 2012).

In the Mediterranean Basin, where this study is focused, plant diversity particularly shares its heritage with several human activities that have had profound, often negative consequences for plant species distribution, abundance and dynamic (Lavergne et al., 2005, Fois et al., 2017). In particular, climatic anomalies (e.g. Malcolm et al., 2006, López-Tirado and Hidalgo, 2016) and human related factors, such as land use change, overgrazing and overharvesting, have been identified as main threats leading to extinction or population decreases in narrowly distributed plant species (Lavergne et al., 2005, Fenu et al., 2017).

According to previous authors (e.g., Tôrres et al., 2012, Weber and Grelle, 2012, Weber et al., 2017), a relationship between population parameters and outputs of SDMs is common; also, it was proved that such relationship could increase if key limiting parameters (i.e. proxies of disturbances) were added to natural environmental variables (Weber et al., 2017). Nonetheless, it is still unclear which method and predictors will provide better population parameters predictions (Thuiller et al., 2010, Weber et al., 2017). Accordingly, we contributed to the validation of simple and easily applicable SDMs as a predictive tool for the evaluation of threats by measuring differences between models with only natural environmental parameters and models with human disturbance parameters.

In this paper, we studied the relationship between common variables used in SDMs and population parameters of yellow gentian (Gentiana lutea L. subsp. lutea) in Sardinia. As no information on the genetic relationships among each group of individuals is currently available, populations were defined on the basis of geographic separation. We tested our hypotheses with this plant species, because it has been thoroughly studied in recent years and the distribution, biology, phenology and ecology at regional level, as well its conservation status, were partially already published by the authors (Cuena-Lombraña et al., 2016, Cuena-Lombraña et al., 2017, Fois et al., 2015, Fois et al., 2016). We used this information to extrapolate and spatialize human-induced disturbances related to climate change, human accessibility and livestock grazing in order to model the environmental suitability of yellow gentian and compare relationships among results with different parameter sets and demographic traits.

Specifically, our aims were 1) to test whether population parameters are correlated with environmental suitability derived from SDMs, 2) to test whether SDMs including disturbances and environmental changes indices show higher correlations with demographic traits than SDMs based only on environmental parameters, and 3) to assess threats by measuring the degree of influence of natural and human disturbance-induced demographic variation by comparing strengths of relationships among the considered SDM results.

Section snippets

Plant species and study area

Gentiana lutea L. subsp. lutea (hereafter G. lutea) is a rhizomatous perennial plant distributed throughout central and southern Europe (Fig. 1), mainly living in mountainous habitats (approximately from 700 to 2000 m asl) (Tutin et al., 1972, Rossi et al., 2015). G. lutea multiplies through vegetative propagation: it develops into a basal rosette during first-year spring, and may further grow some lateral rosettes in the following years (Hesse et al., 2007). Accordingly, we relate to G. lutea

Species distribution models

According to AUC values (0.845 − 0.921), all models performed with good accuracy (AUC > 0.8) and captured patterns far from random (AUC ≤ 0.5) for each of the six replicate runs (Appendix B). In particular, the best-fitting model involved all the natural and disturbance parameters (AUC = 0.921 ± 0.02) while the AUC decreased from the model with only natural parameters (AUC = 0.855 ± 0.04) when ClimChange (AUC = 0.845 ± 0.04) and Trail indices (AUC = 0.850 ± 0.05) were added (see Appendix B for

Discussion

The ecological theory underpinning SDMs assumes a positive relationship between patterns of occupancy and environmental suitability; however, such ‘abstract constructions’ need to be connected with the history of species (Pellissier et al., 2013, Hereford et al., 2017). While presence and absence predictions were in many cases corroborated by subsequent field investigations (e.g. Pearson et al., 2007, Fois et al., 2015, Brichieri-Colombi et al., 2016, Rhoden et al., 2017), few have linked them

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

References (69)

  • G. Bacchetta et al.

    Floristic traits and biogeographic characterization of the Gennargentu massif (Sardinia)

    Candollea

    (2013)
  • C.M. Beale et al.

    Regression analysis of spatial data

    Ecol. Lett.

    (2010)
  • W.T. Bean et al.

    Species distribution models of an endangered rodent offer conflicting measures of habitat quality at multiple scales

    J. Appl. Ecol.

    (2014)
  • L. Bosso et al.

    Potential distribution of Xylella fastidiosa in Italy: a maximum entropy model

    Phytopathol. Mediterr.

    (2016)
  • M. Brambilla et al.

    Species distribution models as a tool to estimate reproductive parameters: a case study with a passerine bird species

    J. Anim. Ecol.

    (2012)
  • T.A. Brichieri-Colombi et al.

    In aid of (re) discovered species: maximizing conservation insights from minimal data

    Anim. Conserv.

    (2016)
  • C.P. Carmona et al.

    Taxonomical and functional diversity turnover in Mediterranean grasslands: interactions between grazing, habitat type and rainfall

    J. Appl. Ecol.

    (2012)
  • O. Cotto et al.

    A dynamic eco-evolutionary model predicts slow response of alpine plants to climate warming

    Nat. Commun.

    (2017)
  • A. Cuena-Lombraña et al.

    Gentiana lutea L. subsp. lutea seed germination: natural versus controlled conditions

    Botany

    (2016)
  • A. Cuena-Lombraña et al.

    Discovering the type of seed dormancy and temperature requirements for seed germination of Gentiana lutea L. subsp. lutea (Gentianaceae)

    J. Plant Ecol.

    (2017)
  • V. Di Cola et al.

    Ecospat: an R package to support spatial analyses and modeling of species niches and distributions

    Ecography

    (2017)
  • I. Dullinger et al.

    Climate change will increase the naturalization risk from garden plants in Europe

    Glob. Ecol. Biogeogr.

    (2017)
  • J. Elith et al.

    Do they? How do they? WHY do they differ? On finding reasons for differing performances of species distribution models

    Ecography

    (2009)
  • J. Elith et al.

    A statistical explanation of MaxEnt for ecologists

    Divers. Distrib.

    (2011)
  • G. Fenu et al.

    Conserving plant diversity in Europe: outcomes, criticisms and perspectives of the Habitats Directive application in Italy

    Biodivers. Conserv.

    (2017)
  • M. Fois et al.

    Using extinctions in species distribution models to evaluate and predict threats: a contribution to plant conservation planning on the island of Sardinia

    Environ. Conserv

    (2017)
  • D.A. Fordham et al.

    Plant extinction risk under climate change: are forecast range shifts alone a good indicator of species vulnerability to global warming? Glob

    Change Biol.

    (2012)
  • J. Fox et al.

    An R Companion to Applied Regression

    (2011)
  • A. Guisan et al.

    Predicting species distribution: offering more than simple habitat models

    Ecol. Lett.

    (2005)
  • J.A. Hanley et al.

    The meaning and use of the area under a receiver operating characteristic (ROC) curve

    Radiology

    (1982)
  • J. Hereford et al.

    The seasonal climate niche predicts phenology and distribution of an ephemeral annual plant, Mollugo verticillata

    J. Ecol.

    (2017)
  • E. Hesse et al.

    Seed bank persistence of clonal weeds in contrasting habitats: implications for control

    Plant Ecol.

    (2007)
  • B. Jaeger

    r2glmm: Computes R Squared for Mixed (multilevel) Models (LMMs and GLMMs)

    (2016)
  • J. López-Tirado et al.

    Predictive modelling of climax oak trees in southern Spain : insights in a scenario of global change

    Plant Ecol.

    (2016)
  • Cited by (0)

    View full text