RESUMEN
This study examined the association of air, land, and water variables with the first historical occurrence and current distribution of toxic Prymnesium parvum blooms in reservoirs of the Brazos River and Colorado River, Texas (USA). One impacted and one reference reservoir were selected per basin. Land cover and use variables were estimated for the whole watershed (WW) and a 0.5-km zone on either side of streams (near field, NF). Variables were expressed in annual values. Principal component and trend analyses were used to determine (1) differences in environmental conditions before and after the 2001 onset of toxic blooms in impacted reservoirs (study period, 1992-2017), and (2) traits that uniquely discriminate impacted from reference reservoirs (2001-2017). Of thirty-three variables examined, two positively aligned with the reoccurring appearance of blooms in impacted reservoirs (air CO2 and herbicide Glyphosate) and another two negatively aligned (insecticides Terbufos and Malathion). Glyphosate use was observed throughout the study period but a turning point for an upward trend occurred near the year of first bloom occurrence. While the relevance of the decreased use of insecticides is uncertain, prior experimental studies reported that increasing concentrations of air CO2 and water Glyphosate can enhance P. parvum growth. Consistent with prior findings, impacted reservoirs were of higher salinity than reference reservoirs. In addition, their watersheds had far lower wetland cover at NF and WW scales. The value of wetlands in reducing harmful algal bloom incidence by reducing nutrient inputs has been previously recognized, but wetlands can also capture pesticides. Therefore, a diminished wetland cover could magnify Glyphosate loads flowing into impacted reservoirs. These observations are consistent with a scenario where rising levels of air CO2 and Glyphosate use contributed to the establishment of P. parvum blooms in reservoirs of relatively high salinity and minimal wetland cover over their watersheds.