Moran's Eigenvector Maps (MEM) and Asymmetric Eigenvector Maps (AEM) to model spatial patterns of heavy metals contamination in a polluted river basin in central Argentina

CUETO, GERARDO R. ; TRIPODI, MARIEL A.; SUAREZ, OLGA V.

Barcelona

Congreso; XXIX International Biometric Conference (IBC 2018); 2018

International Biometric Society?s (IBS)

The Matanza-Riachuelo River Basin, in the Northeast of the Buenos Aires Province, run through the most industrialized, populated and polluted region in Argentina. We utilized a variation partitioning approach in conjunction with Moran?s Eigenvector Maps (MEM) and Asymmetric Eigenvector Maps (AEM) to model spatial patterns of heavy metals contamination in Matanza Riachuelo basin. We used MEM for overland spatial eigenvectors (spatial structures on a plane) and AEM for watercourse spatial eigenvectors because AEM works better for modelling asymmetric spatial processes such as the directional effects of a river network. For AEM, we constructed the tree-like structure, based on connections (edges or river links) among sites. To represent the directional process in the spatial model, we created a table (edge table) that includes the information for both the connections among sampling sites and the direction of water flow. Log concentration of Pb, As, Cd, Cr, Ni and Cu in water determined by River Basin Authority (ACUMAR) during 2014 in 38 monitoring stations along the basin were used as response variables. Spatial eigenvectors were used as independent spatial variables. Forward selection of spatial variables was carried out using a cutoff level of α= 0.05. The significance of the fractions explained by the spatial eigenvectors (AEMs and MEMs) and local environmental variables were tested using 999 permutations at a significance level of 0.05. All statistical analyses were carried out in the R environment using the ?vegan? package. Heavy metals pollution showed a strong spatial pattern along the basin. The adjusted coefficient of determination (R2a) was 0.49 for RDA based on MEM and 0.59 for AEM ones. Asymetrics eigenvectors improved the model fit (the variance explained increased 20%), showing a unidirectional spatial gradient of pollutions from upstream to downstream along watercourses, which contributed more to explains heavy metals variations than overland. However, improve in variance explained changed among elements. Lead showed a low variance explained by AEM, possibly because it has low mobility in water. Thus, water flow is a strong force structuring spatially most of heavy metals pollution along the basin.