Original ArticlesDoes a correlation exist between environmental suitability models and plant population parameters? An experimental approach to measure the influence of disturbances and environmental changes
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)
- et al.
Predicting current and future disease outbreaks of Diplodia sapinea shoot blight in Italy: species distribution models as a tool for forest management planning
For. Ecol. and Manag.
(2017) - et al.
A practical method to speed up the discovery of unknown populations using Species Distribution Models
J. Nat. Conserv.
(2015) - et al.
The reliability of conservation status assessments at regional level: Past, present and future perspectives on Gentiana lutea L. ssp. lutea in Sardinia
J. Nat. Conserv.
(2016) - et al.
Evaluating the ability of habitat suitability models to predict species presences
Ecol. Model.
(2006) - et al.
Genetic variation and reproduction strategy of Gentiana pannonica in different habitats
Flora
(2009) - et al.
Revealing areas of high nature conservation importance in a seasonally dry tropical forest in Brazil: combination of modelled plant diversity hot spots and threat patterns
J. Nat. Conserv.
(2017) - et al.
Application of species distribution models for protected areas threatened by invasive plants
J. Nat. Conserv.
(2016) - et al.
Maximum entropy modeling of species geographic distributions
Ecol. Model.
(2006) - et al.
Improving the assessment and reporting on rare and endangered species through species distribution models
Glob. Ecol. Conser.
(2014) The coincidence of people and biodiversity in Europe
Glob. Ecol. Biogeogr.
(2003)