The purpose of this study was to analyze the effects of the application of multivariate criteria of principal components and hierarchical clustering as a mechanism for monitoring the performance of Tifton 85 grass (Cynodon spp.) planted in horizontal subsurface flow constructed wetland reactor (HSSF-CW) under different organic (OLR), nutritional and sodium loads of swine wastewater (SW). The HSSF-CW planted with Tifton 85 grass was used as a swine wastewater after treatment applying organic loading rates between 26.1 (1st cut) and 360.6 kg ha-1 day-1 COD (8th cut). The maximum performances of HSSF-CW consisted of 52.0 t ha-1 of productivity and 24.0% of crude protein, with the application of 59.7, 64.2, and 31.2 kg ha-1 day-1 of TKN, PT, and K+, respectively. The eleven original variables generated four new components, with PC4 accounting for 94.0% of total variance, a condition strengthened with four data groupings greater than 48% similarity and three data groupings greater than 95% similarity between the variables. There was a strong association between of nitrogen, phosphorus, and potassium concentration by the hierarchical grouping and the intermediate cuts and lower temperatures.
Keywords: Agroindustrial wastewater; Biomass production; Hierarchical cluster analysis; Nutrient removal; Organic load; Principal component analysis.