Ecosystem services in the Amazon–Cerrado agricultural frontier: separating the wheat from the chaff in a functionally diverse riparian zone

The ecological services provided by protected riparian zones in human-altered landscapes are widely acknowledged, yet little is known about them. In this study, we assess ecosystem properties that a protected riparian zone maintains in contrast to environmental changes in its surroundings caused by agro-industrial activities in the northwestern fringe of the Brazilian Cerrado on the Amazon–Cerrado agricultural frontier. We assessed the plant biodiversity, soil hydro-physical properties, and water quality, to understand how the underlying ecological characteristics of a riparian zone withstand the effects of its neighboring cropland area on the stream water quality. We show that the riparian zone is fundamental in providing key ecosystem regulating services, including maintenance of plant biodiversity, soil properties, and water quality. Protection of plant biodiversity in the riparian zone sustains a synergy between soil, and functionally and phylogenetically diverse plant communities by promoting higher infiltration rates, higher soil porosity, and natural soil biogeochemistry conditions, which in turn have direct implications for stream water quality. Our study reaffirms that the conservation of riparian zones is crucial to buffer the negative impacts of agricultural practices on ecosystem services. Our results provide consistent evidence to support further studies and environmental policies for riparian environments, which are often the last fragment of natural vegetation remaining in the dominantly agricultural lands within the Cerrado and Amazon forests.

[This is a post-referring version, accepted in "Global Ecology & Conservation"] [Link to the published paper: https://doi.org/10.1016/j.gecco.2019.e00819] 6 with the former colonized by herbaceous and shrub vegetation, and presenting mostly woody savannah 149 species (De Oliveira-Filho, 1992;Eiten, 1972;Marimon et al., 2012;Ponce and Cunha, 1993;Resende 150 et al., 2004;Ribeiro and Walter, 2008). Within this catchment, the average width is approximately 250 151 m for the gallery forest and 175 m for the campo de murundus.  leave-one-out cross-validation method (Herbst et al., 2006). This procedure allowed the categorization 179 of the surface elevation in 5 equal intervals, binning clay content into quintiles, and the delineation of 180 transects from the catchment's crest to the stream valley passing over all elevation and clay content 181 categories.

182
For hydro-physical analysis, we used the regionalization of soil properties to select 2 points in the RZ 183 and 13 in the cropland area. At these approximately equally-spaced locations along the transects (Fig. 184 1B) we collected one disturbed sample and two undisturbed soil core samples (4.8 cm in diameter and 185 5.2 cm in height) at depth intervals of 0-10, 10-20, 20-40, and 40-60 cm for each sampling point.

186
Disturbed soil samples were analyzed for particle size distribution, and the undisturbed samples were 187 used to determine bulk density, saturated hydraulic conductivity (Ksat), total porosity, macroporosity, 188 microporosity, and field capacity. These procedures are in line with the soil geostatistical and hydro-

Soil hydro-physical and chemical properties 304
Soil hydro-physical properties of both RZ and cropland show a clay-loam texture ( Table 1). The 305 cropland area had greater clay content in the topsoil compared to the RZ. Bulk density in the RZ was 306 significantly lower than in the cropland area (p < 0.01). Ksat and field capacity did not show significant 307 differences between these areas, but upper layer (0-10 cm) total porosity was higher in the RZ. In both 308 areas total porosity was dominated by about 75% micropores due to high clay content (58 ± 7%, average 13 of both areas). Soil acidity at the 5-cm soil depth (Table 2) was significantly higher (p = 0.057) in the 310 RZ. The cropland area had higher Ca and P than the RZ at both 5 and 30-cm soil depths, and higher Mg 311 content at 5-cm soil depth.   The Kruskal-Wallis H test by ranks with the multiple comparison (Steel-Dwass-Critchlow-Fligner 329 method) exhibited the water quality varying from three to five groups with similar mean values (Fig. 4).

330
Mg was the parameter with less groups (total of three) and with the smallest variation (0-6 mg L -1 ). The  (Table S2). most abundant species Pleroma stenocarpum belongs to a genus that is well known for its ability to 365 colonize intensively degraded areas, thus contributing to their recovery (Lorenzo et al., 1994).

366
In the gallery forest, Tapirira obtusa was the most abundant, which is a pioneer species (Raaimakers and Cerradão. We attribute the lower acidity of the soil in the cropland area to the calcium carbonate 405 (CaCO3) commonly applied to the topsoil to reduce soil acidity and support nutrient availability to crops.

406
In our study area, the application of CaCO3 to croplands had implication on the soil Ca content, which 407 was significantly higher in the topsoil of the cropland area. As CaCO3 reacts with water, it produces

431
We found significantly higher P and Mg in the topsoil of the cropland area than in the RZ. This is likely 432 due to regular fertilizer application to croplands in this region while undisturbed Cerrado soils are highly 433 weathered and low in nutrients (Hunke et al., 2015). Other studies found nutrients, such as K, Mg or P, 434 higher in cropland areas than in native vegetation zones without direct agricultural influence (Cruz 435 Ruggiero et al., 2002;Haridasan, 2008;Silva et al., 2008;Tinker and Nye, 2000). However, we were 436 able to find a downward gradient of K, P, SO4, and Mg concentrations, which were highest in the 437 overland flow of the cropland area, exhibiting a gradual decrease in concentration from the cropland 438 area towards the stream (Fig. 4). On a farm in the USA, Lowrance and Sheridan (2005)