Skip to main content
Dryad logo

Is a restricted niche the explanation for species vulnerability? Insights from a large field survey of Astragalus tragacantha L. (Fabaceae)

Citation

Baumel, Alex et al. (2021), Is a restricted niche the explanation for species vulnerability? Insights from a large field survey of Astragalus tragacantha L. (Fabaceae), Dryad, Dataset, https://doi.org/10.5061/dryad.bzkh1898p

Abstract

In this study, we examine the realized niche of A. tragacantha in Provence. We focus our fieldwork on habitat characteristics, floristic survey and perform an exhaustive demographic census. We base our analyses on total density and densities of different size stages of A. tragacantha. We delineate its realized ecological niche with the Outlying Mean Index (OMI) multivariate method and estimated the effects of environmental gradients on A. tragacantha populations with the Generalized Joint Attribute Modelling method (GJAM). Analyses based on occurrence data did not support the hypothesis that a too narrow niche by itself explains the vulnerability of A. tragacantha. However, niche modelling revealed differences among size stages supporting a restricted regeneration niche and a wider ecological range in the past.

This dataset is accompanying the manuscript published in Flora "Is a restricted niche the explanation for species vulnerability? Insights from a large field survey of Astragalus tragacantha L. (Fabaceae)"

Methods

The populations of A. tragacantha were described based on individual size stages and foliar necrosis, used as proxies of regeneration and health, respectively. Foliar necrosis is possibly due to the conjunction of water deficit stress and of sea-sprays polluted by maritime traffic and submarine outfall of discharges of wastewater. Five size stages were defined based on the mean diameter classes: inferior to 10 cm; between 10 and 50 cm; between 51 and 100 cm; between 101 and 200 cm; and superior to 200 cm. Four classes of foliar necrosis were considered: inferior to 10%; between 10 and 50%; more than 50%; and dead individual (i.e, 100%). A grid of 100 m² squares  was used to count the total number of individuals, then the number of individuals within each of the size stages and necrosis classes. Each square was used as a unit sample to summarize demographic data and assess densities within the A. tragacantha distribution.

The 499 relevés were situated within 2 km from seashore. Among them, 162 are occupied by A. tragacantha. For each of the 499 relevés, the following environmental variables were recorded within a surface of 100 m²: elevation above sea level (meters), distance from the seashore (meters), slope (degree), exposure (index from 0 to 4 corresponding to north exposure to south exposure, i.e. 0 for north, 1 north-east and north-west, 2 for east and west, 3 for south-east and south-west, and 4 for south), estimated cover rate of rocky substrate type such as rock (rocky outcrops), block i.e. Ø > 20 cm, stone i.e. Ø < 20 cm, estimated cover rate of bare soil, and estimated cover rate of herbaceous and woody plants. The vegetation and environmental cover variables were estimated on the basis of a semi-quantitative mode using 6 cover classes (0: cover = 0%, 1: 0% < cover < 10%, 2: 10% ≤ cover < 25%, 3: 25% ≤ cover < 50%, 4: 50% ≤ cover < 75%, 5: cover ≥ 75%). To analyze co-occurring plant communities with A. tragacantha, all plant species within the 100 m² plots were inventoried.

 To define a subset of companion species to A. tragacantha, we defined two groups of relevés according to the presence and absence of A. tragacantha and used the “multipatt” function of the indicspecies packages in R software which is based on specificity and fidelity criteria. A strongest value close to 1 indicates a strong association between A. tragacantha and the focal species (e.g. it is present and abundant only when A. tragacantha is present). To design a manageable subset for analyses, we selected the plant species having an indicator value superior or equal to 0.5.

Outlying Mean Index multivariate method (OMI)

 The ordination diagrams of OMI analysis  are based on the table of normalized environmental variables and the floristic table. In the OMI scatterplot, the distance of the centroid of a species to the origin of the diagram is an estimate of the environmental marginality of this species. A species with a high marginality is constrained by a narrower environmental requirement compared to other species, i.e. it has a more specialized niche in the context of the analysis. The OMIreleases also the species tolerance, which is a measurement of the species niche breadth. Marginality (omi) and tolerance (tol) are correlated when common species have the broader niche breadth, but the residual tolerance (rtol) is not correlated to omi . Niche parameters, omi and rtol values, were computed with the niche.params() function (R ade4) and compared between A. tragacantha, its companion species, species common in the study area and to mean values computed for all the species. To illustrate the ordination diagrams of OMI by A. tragacantha total density (number of individuals counted in each 100 m² plots) we used the coordinates of each relevé along the OMI axes to form a scatterplot of circles with diameter proportional to A. tragacantha density with the R function “point”. Spearman correlations between environmental variables and OMI axes were plotted as a heatmap  for inferences of the main ecological gradient revealed by OMI.

Generalized Joint Attribute Modelling method (GJAM)

The OMI axes were used as predictor variables for GJAM. The response variables were the presence/absence of all plant species present in at least 25 relevés (i.e. 5% of the data) and A. tragacantha densities. Two GJAM models were fitted with the main OMI axis as predictors: (i) A. tragacantha density of all living individuals (hereafter named “total density”) as first model and (ii) the five size stage densities (0-10 cm, 10-50 cm, 50-100 cm, 100-200 cm, only living individuals) as second model. Gjam was performed with the following model parameters: linear terms, non-informative prior for the coefficients and covariance matrix, 10,000 iterations with a burn-in of 2,500 iterations. The response variables concerning A. tragacantha were coded as “DA” (discrete counts) and the floristic data as “PA” (presence/absence). Astragalus tragacantha was removed from the list of plant species involved in the analysis. We controlled the convergence of the model and the model fitting to data by using the outputs of GJAM. We used build-in gjamplot functions to produce the matrix of beta coefficients and a clustering of the response variables based on sensitivity. Sensitivity to environmental predictors were displayed with boxplots. Beta coefficients were summarized in a table to show the response of A. tragacantha to ecological gradients.

Script and data of our analyses are deposed in ZENODO and DRYAD databases. We used the following packages: ade4 (Dray and Dufour, 2007), adegenet (Jombart, 2008), indicspecies ( De Caceres and Legendre (2009), labdsv (Roberts, 2013), gjam (Clark et al., 2017), ggcorrplot (Kassambara, 2016), and vegan ( Oksanen et al. 2020).

DATA : load dataset here

R script : https://doi.org/10.5281/zenodo.5388500

Outputs from Gjam : https://doi.org/10.5281/zenodo.5388504

 

References

Clark, J.S., Nemergut, D., Seyednasrollah, B., Turner, P.J., Zhang, S., 2017. Generalized joint attribute modeling for biodiversity analysis: Median-zero, multivariate, multifarious data. Ecol. Monogr. 87, 34–56. https://doi.org/10.1002/ecm.1241

De Caceres M., Legendre P., 2009. Associations between species and groups of sites: indices and statistical inference. http://sites.google.com/site/miqueldecaceres/.

Dolédec, S., Chessel, D., Gimaret-Carpentier, C., 2000. Niche separation in community analysis: A new method. Ecology 81, 2914–2927. https://doi.org/10.1890/0012-9658(2000)081[2914:NSICAA]2.0.CO;2.

Dray, S., Dufour, A.B., 2007. The ade4 package: Implementing the duality diagram for ecologists. J. Stat. Softw. 22, 1–20. http://www.jstatsoft.org.

Jombart, T., 2008. Adegenet: A R package for the multivariate analysis of genetic markers. Bioinformatics. 24, 1403–1405. https://doi.org/10.1093/bioinformatics/btn129.

Kassambara, A., 2016. ggcorrplot: Visualization of a Correlation Matrix using’ggplot2’. R Package. URL http://www.sthda.com/english/wiki/ggcorrplotversion 0.1, 1.

Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., ... & Wagner, H. 2020. vegan: Community Ecology Package. R package version 2.5-6. 2019

Roberts, D.W., 2013. Package ‘labdsv’: Ordination and Multivariate Analysis for Ecology. R Package. URL: http://ecology.msu.montana.edu/labdsv/R/

Usage Notes

Data files and R scripts to conduct OMI and GJAM methods are provided and described in the Readme file.

Funding

European Commission, Award: LIFE16 NAT / FR / 000593