Water quality in catchments under mechanised agriculture

A main finding of my dissertion was that fertilisation at agricultural sites caused nutrient accumulation in soil and water resources with seasonal effects for water quality.

The meta-analysis revealed significantly increased available P and K contents in the upper soil layer (up to ~ 40 cm) for croplands and pastures (Hunke et al. 2015 Ecohydrology). Likewise, we found notably high P concentrations in the topsoil of soybean sites which were manifold higher (~6-fold) than the soil nutrient recommendations of EMBRAPA (2003) and de Morais et al., (2009), thus indicating over-fertilisation and a potential risk of P displacement via surface runoff and subsurface flow (Hunke et al. 2015 Geoderma Regional). However, regardless of the land use, P was not accumulated in deeper horizons.

P absence in surface waters of agricultural sites

Surprisingly, P pollution was widely absent in surface waters of the reviewed water quality studies, likewise in our own wet seasonal water quality sampling. Especially for the rainy season, these unambiguous results were unexpected in an area which is characterised by active erosion processes and over-fertilised soils. We conclude from these results that very high-fixing capacities of deeply weathered soils as well as the filtering capacity of riparian vegetation (e.g. Boechat et al., 2013) seems to play an important role in preventing surface waters from pollution. Contrarily, we recorded in dry seasonal increases of phosphate across the entire catchment which is likely to origin from animal waste and fertiliser application for the second crop. However, there are still a lot of open questions as we could not give further evidence to explain these findings with the data of our field study.

Elevated (inorganic) N in surface waters but not in the soils of agricultural sites

Soil total N did not differ between land uses because most of the cropland sites in the review study were cultivated with nitrogen-fixing soybeans, which are not artificially fertilized with N. Likewise, we could not find altered concentrations of total N for the soybean sites. However, the role that nitrogen leaching plays in agricultural Cerrado is inconclusive in this thesis. The results of the dissertation showed nitrogen enrichment in agricultural catchments, indicating fertilizer impacts and potential susceptibility to eutrophication from other crops (maize, cotton) after the soybean havest. Direct NO₃^– leaching appears to play a minor role; however, water quality is affected by agricultural non-point sources, due to topsoil fertiliser inputs affecting the entire catchment, from small low order streams to the larger rivers of the modified catchment. Additionally, nitrate fertilizer is also applied directly to crops, and higher levels of inorganic nitrogen forms such as nitrate and nitrite we identified by own measurements  in surface waters and in the review study may originate from granular fertilizer applications on bare soil between the seedling rows.

In absolute terms, nitrate concentrations in surface water were low compared with concentrations measured in European rivers; in relative terms, however, the nitrate values of the rivers in catchments under land use change are orders of magnitude greater than those found in natural Cerrado streams. Additionally, field sampling of Chapter 3 intended also to explore if there are any seasonal or spatial patterns of water quality evident. In the snapshot sampling, we could identify a strong seasonality with higher temperature, oxi-reduction potential (ORP), NO₂⁻, and very low oxygen concentrations (<5 mg·l ⁻¹) and saturation (<60 %) in the rainy season.

Our assumption that water quality parameters are spatially correlated to stream order, catchment size, up-stream land use, and the land use in a 300 m buffer was generally not supported by the snapshot data, because no significant differences were detected.

However, from the own observations like on the photographs (right stream cristal clear water from a forested sub-catchment and left a stream with brownish water caused by suspended sediment transport in an agricultural area) it is likely that there is a correlation with the percentage of different land use types with water quality indicators. For that there is a need of proper samplings, more high resoluted data etc.

From the literature several effects of pesticide contamination of the water cycle in the Cerrado isevident. There are only a few studies available across the entire Cerrado but it is likely that the described effects of the case studies are transferable to other catchments. 

It turned out that the main pesticide pathways to surface and ground water are (1) wind drift from plane applications (Laabs et al., 2002), (2) direct runoff from agricultural fields (Casara et al., 2012) at sites with a non-intact riprian vegetation and (3) leaching through macropores (Dores et al., 2009). In several case studies, extremely high-peak concentrations exceeded the considerably less strict Brazilian and EU water quality limits, which were potentially accompanied by serious health implications by the consumption of untreated water (Nogueira et al., 2012, Brando et al., 2013). Due to higher volatilisation rates in tropical areas, the wet and dry deposition rates of pesticides appears to be much more relevant and complex than in temperate regions (Laabs et al., 2002), thus contributing to a much larger distribution of pesticidal effects across the region. Overall, we concluded that even with low-detection frequencies the reviewed studies found pesticide pollution in the water cycle which failed to comply with the stricter EU limits