Effects of land use and livestock watering on greenhouse gas emissions and water quality in rivers and water pans in tropical highland and semi-arid agroecosystems in Taita-Taveta County, Kenya

Economic development and food insecurity are an important drivers of land use change in tropical ecosystems.  In sub-Saharan Africa, forest and wetland conversion to agriculture and zero-grazing policies are common in highland systems, while livestock-dominated agricultural systems are more common in the drier, semi-arid low-land systems. Greenhouse gasses (GHG) from aquatic ecosystems is increasingly appreciated as missed sources of emissions from agricultural and forested landscapes, and rivers and artificial watering pans are used for livestock watering in these systems. They therefore receive manure and urea, making them potential hotspots for greenhouse gas (GHG) emissions through biogeochemical processing, but the role of livestock has not yet been examined. We performed 4 synoptic surveys for GHG concentration and fluxes in rivers in the Taita Hills and in water pans in the semi-arid region low-lands of Taita-Taveta County, Kenya in October–December 2019, which spanned the transition of short rainy season to the dry season. We also measured water-quality parameters and related them to GHG emissions in order to assess the biogeochemical processes likely responsible for the emissions in each system type. There were 9 agricultural streams (no livestock), 8 livestock streams, and 4 water pans. Results showed ten times higher CH₄ and N₂O  fluxes in the water pans compared to river systems (~4 vs. 40 mmol CH­₄ m^-2day^-1 and ~4 vs. 30 mmol N₂O m^-2day^-1)  while CO­₂ emissions were two times higher in the agricultural streams (~110 vs. 60 mmol CO₂ m^-2day^-1). Water pans also showed higher dissolved organic carbon (DOC) concentration and higher dissolved nitrogen and phosphorus (TDN and TDP) and lower dissolved oxygen (DO) concentrations than river systems. There was a significant positive relationship between pCO2 and fine benthic organic matter (FBOM) in livestock streams but no relationship with DOC, suggesting that increased sediment respiration from livestock may be responsible for CO­₂ emission. In river systems, there was also a positive relationship between CH₄ and CO₂, which indicated that methane production from CO₂ was a controlling mechanism. This contrasted with CH₄ production in water pans which was related to primary production and organic inputs from livestock. N₂O also showed different processes in riverine and water pan systems, with nitrification appearing to be more important in river systems, evidenced by the negative relationship of N₂O production with DOC and a positive relationship with CO₂. In water pans, N₂O was negatively related to NO_3­­, dissolved oxygen, and DOC. In addition, more negative fluxes of N₂O occurred in water pans than the other sites, which suggests complete denitrification of N₂O to N₂.  Diurnal measurements also indicated that fluxes were positively related to livestock density; however this effect was more pronounced in the drier season and under low discharge. Water pans were also hotspots in the landscape for CH₄ and N₂O emissions, showing 10 – 1000 times greater emission than the surrounding landscape. Further research should examine the water pans and riverine watering holes as distinct features with the potential to impact greenhouse gas emissions from agricultural landscapes.