The complexity of antibiotic resistance dynamics in scarce surface water resources in northern Botswana

Antibiotic resistance (AR) is widely associated with intensive agricultural systems, pharmaceutical production, wastewater, and health facilities. However, little research has been conducted on AR gene (ARG) dynamics in natural environments lacking large-scale human inputs. In particular, we have a limited understanding of the complex dynamics influencing environmental AR in resource-limited dryland systems threatened by climate change. In northern Botswana, Escherichia coli isolates were obtained from river surface water (n = 426 samples; September 2017 – May 2018), sediments (n = 194; November 2017 – May 2018), and human fecal samples (n = 43 September 2017 and April 2018). A multiplex PCR assay was used to assess gene frequencies for sulfonamide (sul1 and sul2), tetracycline (tetA and tetB), and class 1 integron (intl1) resistance genes. The weighted frequency of sul1 in sediment E. coli isolates (µ= 0.07; SD = 0.39) was significantly higher than that observed in isolates obtained from surface water (µ= 0.03; SD = 0.15; p = 0.01). Weighted gene frequencies for sul1 and sul2 in human E. coli isolates from April 2018 were significantly higher than those in water (sul1 p = 0.01; sul2 p = 0.00) and sediment isolates (sul1 p = 0.01; sul2 p = 0.00) from the same time period. Significant differences for the five genes' weighted frequencies were observed between sampling months in water isolates (intl1 p = 3.318e-05; sul1 p = 3.217e-06; sul2 p = 4.392e-06; and tetA p = 2.477e-05), while only intl1 frequency differed significantly between months in sediment isolates (p = 0.05). While no significant spatial patterns of ARG frequencies were observed in E. coli isolates from water samples (p = 0.16), higher ARGs were observed in E. coli isolated from human-dominated land areas for intl1 (µ = 0.10; SD = 0.31) than in protected landscapes intl1 (µ = 0.03; SD = 0.13; p = 0.02). Land use also was associated with higher weighted frequencies for tetA in E. coli isolates from water in human-dominated land areas (µ = 0.10; SD = 0.30) compared to protected areas (µ = 0.04; SD = 0.23, p = 0.03). These results indicate that the interactions between land use and season-dependent hydrometeorological factors drive frequencies of some ARGs across this system, but do not fully explain the complexities observed. However, the lack of higher weighted gene frequencies for riverbed sediments suggests that they do not act as a reservoir for ARGs in the system, implicating humans as significant contributors to ARG persistence in the aquatic system.