##############################################################################################################################################

Title: Data analysis of Caatinga Cristalino

Study authors: BRUNO COUTINHO KURTZ, DIEGO MEIRELES MONTEIRO, MARINEZ F. SIQUEIRA AND TAINÁ ROCHA

Codes and GitHub pages author: Tainá Rocha

We are investigating the phytogeographical patterns and species turnover/nestedness (beta diversity composition) in Caatinga dry tropics

In progress. For preliminary results click here

##############################################################################################################################################

Script folders structure as follow:

MATERIALS AND METHODS

Data source

We first compared the environment for 22 phytosociological surveys of Cristalino Caatinga as a test, considering five topographic data and 19 climatic data from INPE and wordclim v. 2.1 respectively. Next, we expanded sampling areas by data collection from online biodiversity databases, as follow:

• Survey of species list (and records) by Cristalino counties shapefile using Rocc pckg workflow. Code acess. Output: species list with occurrence (PS.:records occur beyond the Cristalino counties shapefile

• Select just records inside Cristalino counties shapefile, Qgis (Clip function of vector Geoprocessing tools)

• Create a spatial vector for the Caatinga Cristalino whole area with 10 km grid cells using Qgis(v.3.18). Each grid cell represents the sample site.

• Spatial join between occurrence records with spatial vector of 10km grid cells using Qgis(v.3.18).

• Extract enviromental values of occurrence records using raster . Code acess

Multivariate analysis

Generalized Dissimilarity Modeling (GDM)

• Inputs : two tables (spp_data and env_table)

• GDM models for 22 sites test. Code acess .

• GDM models for expanded sample

Principal Coordinates Analysis (PCoA): For florist matrix of 0 and 1

• Data_Prepare_for_PCA: script to prepare input data to run PCA analysis.

• PCoA . Ok for 22 sites test. Code acess .

Cluster

• Ok for 22 sites test using pvclust. Code acess .

Baselga metrics

• βSOR (overall beta diversity), 0k for 22 sites
• βsim (turnover), 0k for 22 sites
• βSNE (nestedness), 0k for 22 sites

Preliminary Results

The results of the Generalized dissimilarity model (GDM) showed a power of explanation higher than 50 percent (56.8%). A visualization of spatial pattern of dissimilarity is shown in Figure 1 , in which areas of similar colour are predicted to have similar floristic composition. Predictions for Petrolina sites (PE1 e PE2) show most difference compositions compared with others, i.e., major dissimilarities compared with other sites. Pedicted a greater ecological distance increasing observed compositional dissimilarity (Fig. 2 a) and higher compositional dissimilarity when the observed compositional dissimilarity increases (Fig. 2b).

Fig.1                                                                                  Fig.2