Aquifer and Pond Relationship Under Potential In uence of Eucalyptus and Sugarcane

Rafael Terada (  rafael.terada@usp.br ) Universty of São Paulo https://orcid.org/0000-0001-6987-5938 Ricardo Hirata University of São Paulo Geosciences Institute Paulo Galvão Federal University of Minas Gerais: Universidade Federal de Minas Gerais Fernando Saraiva University of Sao Paulo: Universidade de Sao Paulo Norio Tasse University of Ibaraki Mariana Luiz Bernardino University of Sao Paulo: Universidade de Sao Paulo Bruno Conicelli Amazon Regional University IKIAM: Universidad Regional Amazonica IKIAM

in a different water ow system from the eucalyptus and sugarcane water sources? Is the pond hydraulically disconnected from the shallow aquifer? To solve these questions, stable isotopic (δ 2 H and δ 18 O) and hydrogeochemical analysis of groundwater and pond water were performed, and geophysical surveys and groundwater level measurements were carried out before and after the cutting of eucalyptus to understand the hydrodynamic relationships.

Site Description
The city of Rio Claro is located in the central part of the State of São Paulo, Brazil, about 170 km northwest of the city of São Paulo. The study is carried out in the western countryside of the urban area of Rio Claro, where there is an intensive agriculture activity. On the south of the study site, there is a circular shaped pond positioned at a lower topography, contrasting to the higher elevations on the left and the right sides, where the eucalyptus trees and sugar cane are planted (Fig. 1).
This area is over the Paraná Basin over the sediments of the Cenozoic Rio Claro Formation, which covers an extensive area in the state of São Paulo (Björnberg and Landin 1966). The Rio Claro Formation consists of sequences of sandy strata with minor intercalated clay lenses at its base and by argillaceous sediments with intraformational gaps and sandy lenses subject on the top (Fulfaro and Suguio 1968). The thickness of these sequences does not exceed 40 m (Freitas et al. 1979;Cottas 1983). The Rio Claro Formation overlaps older deposits of the Paraná Basin, named Pirambóia and Corumbataí formations. The rst one is featured by a succession of reddish sand layers that, in the surface, present thicknesses greater than 270 m. These sandstones are medium to ne-grained, having a higher clay fraction at the base of the formation, occurring locally coarse conglomeratic sandstones (IPT 1981). The Corumbataí Formation consists of ne sediments, purplish siltstones, and mudstones of marine origin. On the surface, there is an intense fracturing. These two formations, besides those associated with aeolian sandstones from the Botucatu Formation and to basic intrusive rocks from the Serra Geral, are considered the original material of the Rio Claro Formation (Zaine 1994).
Geomorphologically, the area is in the Paulista Peripheral Depression unit, an area with altitudes of 500-700 meters above sea level (m.a.s.l.). Scattered ponds or lakes in a circular to oval shapes with diameters of 100-500 m throughout the Rio Claro Formation landscape is a unique feature of that region ( Fig. 1). The local morphology consists predominantly of a thick and sandy soil in large and tabuliform hills, with a low number of drainages. These ponds are formed in shallows depressions, linked or not to drainage networks on the surface.
Fulfaro and Suguio (1968) interpreted the origin of the ponds as related to a uvial paleochannel, corresponding to a past Corumbataí river that was barred in downstream in the function of reactivation of local faults in Pitanga structure area. Another explanation could be the results of sedimentation processes of the Rio Claro Formation associated with abandoned meanders channels. The ponds also can be related to doliniform depressions on the plateau of Itapetininga, a result of the carving of the current drainage network associated with solubilization and the leaching of carbonate sediments (Irati Formation), or intrusive basic rocks. The ponds represent the indication of the current drainage network and that their alignment follows preferred structural NE-SW directions and, secondarily, NW-SE direction (Zaine 1994).
The climate, according to Köppen classi cation, is CWA type, characterized by a rainy tropical climate with rainy summer and dry winter. The average monthly temperatures are above 18°C, with the hottest months above 22°C. The precipitation of the wettest month is up to ten times greater than the driest. The Rio Claro region can be considered as tropical showing two distinct seasons: dry season, from April to September, with rainfall of 180-200 mm (17°C average); and rainy season, from October to March, with 55-60 days of rain, totalizing 1,200 mm (Troppmair 1992). Santos (1986) noted the existence of cycles in terms of dry and wet years in the city of Rio Claro. The driest year was 1,921 with 655 mm of rainfall and the wettest reached 2,144 mm, in 1976. The average rainfall in the last 20 years was 1,521 mm/yr. The Rio Claro Formation comprises an uncon ned aquifer that consists of clay materials with wells that produce pumping rates between 17 and 25 m³/h (DAEE 1981). The water table presents a wide variation of depth, prevailing the ones lower than 18 m (Oliva 2006). Because of its sedimentary composition with medium to high permeability, the water level ows according to the topography. Recharge zones comprise the entire outcrop area and discharge areas are rivers and drainages. The hydraulic conductivity values are between 10 − 2 and 10 − 4 cm/s, depending on their lithologic variation (Oliva 2002;2005;Zanetti 2012). According to the same author, these ponds represent aquifers with depths ranging from 1 to 2 m, where the ow of in ltrated surface water is blocked by low permeable materials, such as clay layers, common to a uvial depositional environment. The potentiometric surface follows the preferential orientation of the Corumbataí river, west of the area, and the Claro river course, to the east. The lowest part of the potentiometric surface is in the southern portion of the area, where the Rio Claro Formation presents thin thickness and it is near to the contact with underlying units (Oliva 2006).

Water balance, well drillings, and potentiometric surface
The water balance was estimated considering temperature and precipitation data for the year 2013 based on data from CEAPLA/UNESP -Meteorological Monitoring Center, located in the city of Rio Claro, São Paulo, Brazil (coordinates UTM: 22K 23'39" S, 47°32'53" W, elevation 629 m). The potential and actual evapotranspirations were estimated using the Thornthwaitt and Mather (1955) method.
A total of 27 monitoring wells were drilled and installed following the Brazilian Groundwater Monitoring Wells in Granular Aquifer Standard (ABNT-NBR 15495, 2007). The drillings were performed using manual and hollow stem auger techniques, reaching depths between 4 and 16 m and coated with 2" geomechanical tubes. Six monitoring wells were installed in the pond area (P1 -P6). The wells named S1 to S5 are in the sugarcane area, while wells named E1 to E16 are in the eucalyptus area (Fig. 2). During the drillings, soil samples were collected for lithologic and organic matter descriptions. After drillings, potentiometric surface maps equivalent of March and September (2013), respectively before and after eucalyptus cutting, were made using topographic survey and hydraulic head information to understand groundwater ow directions.

Geophysical survey
In order to determine variations over the horizontal geologic strata, including moisture and water level, as well as to identify possible resistivity differences before (June) and after (September) the eucalyptus cutting on 08/13/2013, two Electrical Resistivity Images (ERI) transects were performed, using a dipoledipole array with a SAS 300 (ABEM) instrument. The transects were taken in two different lines: the west-east line 1 surveyed both the eucalyptus and sugarcane areas using 48 steal electrodes with 5 m of distance and its multiples; the north-south line 2 surveyed the central area that divides the two plantation areas (where the road is located), using 48 steal electrodes with 2.5 m of distance and its multiples (Fig. 2). Before inserting into the ground, steel electrodes were wetted with a saline solution to improve the electrode/soil contact and then connected to multielectrode cables. The data obtained for the ERI transects were processed using the RES2DINV software.

Stable isotopic analysis
Pond water and groundwater were sampled in July 2012 for stable isotopes δ 18 O and δ 2 H analysis. Amber vials were lled with samples, avoiding air bubbles inside and stored in coolers maintaining the temperature to avoid post-sampling fractioning. The analyses were performed at the University of Tsukuba, Japan, and were normalized to internal laboratory water standards that were previously calibrated relative to the Vienna Standard Mean Ocean Water  For potential water mixing between groundwater and pond water, a simple linear algebra based on δ 18 O and δ 2 H values was used for quanti cation, according to the equation: δsample = χ·δA + (1-χ) δB (Clark and Fritz 1997). Thus, the proportion, in %, of a mixture of groundwater (one end member), and pond water (another end member) was estimated, relating directly to the samples' position over the evaporation water line.

Chemistry analysis
The same monitoring wells and pond were used for geochemistry analysis, also sampled in July 2012 ( Fig. 2 and Table 1). After pumping several well volumes, the stagnant water was removed to acquire fresh ones for analysis. Groundwater was sampled using a peristaltic pump (Geotech Geopump), according to the low-ow method (USEPA 1995). Before sampling, physicochemical parameters (pH, T, EC, ORP, DO) were measured until its stabilization using the multiparameter probe YSI Professional Plus. The alkalinity values were also measured in situ by titration with padronized H 2 SO 4 solution. Samples of via Diagrammes Software to separate waters from different sources (pond, eucalyptus, sugarcane).

GIS database
All the data sets were entered in a GIS database and georeferenced at ArcGIS 10.1 software. monthly with precipitation input. The budget for 2013 was: (1) water excess from January to March; (2) water de cit from April to September; (3) water replacement between October and December, a period of groundwater recharging; and (4) depletion, period when soil moisture is consumed (Fig. 3).
According to the potentiometric surface maps and groundwater ow directions, the pond's hydraulic head (~ 619 m) is higher than all piezometric measurements (< 619 m) and local drainages (springs and rivers). Groundwater ows from the pond, which is topographically lower (~ 620 m), to higher elevations (~ 630 m) frequently throughout the study period from SSE to NNW. It is also noted that the water coming from the pond does not pass through the sugarcane area, indicated by the groundwater ow directions over the eucalyptus area (Fig. 3).

Geophysical survey
The surveys before (June) and after (September) eucalyptus cutting on 08/13/2013 showed the differences between soil resistivities under eucalyptus and sugarcane plantations until 10 m of depth. Line 1 crosses line 2 at meter 100, while line 2 crosses line 1 at meter 80 (Fig. 4). The SSE-NNW transect represents the soil resistivity through the road that divides eucalyptus and sugarcane plantations (line 2, Fig. 4). In June, the resistivity values at uphill ranged from 800 to > 4,000 Ohm.m (prevailing values around 2,400 Ohm.m); the range remained on the same levels in September; however, prevailing resistivities around 2,800 Ohm.m. Near the pond, downhill, the range in June was detected between 400-1,200 Ohm.m., prevailing values around 800 Ohm.m. In September, the resistivity range was slightly higher, prevailing resistivities around 1,600 Ohm.m.
Comparing the values of the June and September transects (before and after cutting, respectively), the eucalyptus area showed the most signi cant increase in resistivity (76%), followed by downhill, sugarcane, and uphill, with 50%, 25%, and 16%, respectively (bottom table on Fig. 4). This indicates a disproportionate increase in soil resistivity from June to September, speci cally in the eucalyptus area. In other words, there was a strong decrease in the soil moisture after cutting in this area compared to those of the sugarcane, with no cutting, and with the road, with no cover.

Stable isotopes analysis
The  Table 1).
The isotopic values for waters in the pond were highly variable (red dots in For water mixing proportion estimates, the pond end member considered the monitoring well P4, representing the most evaporated water sampled (Fig. 5). The groundwater end member considered was from the E2 well due to its lowest isotopic values, indicating less evaporated water. Although wells E3 and E4 in the eucalyptus area are located farther from the pond, potentially suggesting that they are the most representative end members, their isotopic values indicated a possible mixture with waters isotopically less enriched from the sugarcane area, explaining their plots closer to the middle area of the gure. Analyzing the proportion mixing line of wells E1 and E7 located on the pond's edge on the gure indicates that around 45% of the water comes from the pond, and the remaining 55% corresponds to the aquifer. In well E6, 75% of the water comes from groundwater and 25% from the pond, while in E5 65% of the water is groundwater and 35% is from the pond (Fig. 5). As previously stated, the other wells may have mixtures of waters from the sugarcane, changing their mixing proportions.

Geochemistry analysis
The sample's ionic mass balances were checked for accuracy and deviation, and values between the range of ± 10% were ideally accepted, according to Freeze and Cherry (1979) and Custodio and Llamas (1983). Wells S4 and P3 showed values out of the range; however, they were considered in this work. In one hand, S4 presented elevated nitrate concentration, which corroborated with fertilizer applications in sugarcane. On the other hand, P3 showed a very low concentration for all the parameters, which leads to a large error in ionic balance ( Table 2).   Table 2).

Discussion
The occurrence of ponds in the Rio Claro Formation is a unique geomorphologic feature of the region, where large-scale agriculture dominates the landscape.
To understand the hydrogeology in rural areas with such conditions, an area containing a pond in a low topographic position surrounded by sugarcane and eucalyptus plantations at a higher topographic elevation was investigated. An initial potentiometric surface map was generated in March 2013 (Fig. 3a) indicating groundwater ow directions to the northwest, coming from the pond towards the high topographic area, preliminarily showing a pond recharging behavior. This hydraulic behaviour generated a series of research questions since a pond at lower topography, which is hydraulically connected to the local shallow aquifer, would typically be considered a potential discharge area.
The rst hypothesis describes that the behavior is related to the eucalyptus' high capacity to absorb water. Brazil in the same climatic conditions. They concluded that the change in crop type leads to an increase in evapotranspiration and a decrease in recharge.
Several species of eucalyptus are able to extract large amounts of water, known for drying out ooded areas, especially when precipitation reaches levels below 400 mm/yr (de Paula, 1999). It is also known that the faster the tree growth and the larger its leaf area, the higher the groundwater consumption (Poore and Fries 1985). In eucalyptus, the growth rate is considerably accelerated up to 7 years, where it approaches its optimal cutting age. It is estimated that a plantation at this age has an evapotranspiration rate of around 800 to 1,200 mm/yr (Foelkel 2005). With recharge being the process in which water in ltrates and reaches a saturated zone, the eucalyptus roots act as a barrier preventing this in ltration, absorbing part of the soil moisture. When soil moisture is no longer enough for the crop, the roots grow down into the soil to nd water and nutrients, sometimes reaching the saturated zone. So, analyzing the groundwater stable isotopes samples from the area, rstly, there was a clear linear regression trend indicating an evaporation signature or a mixture between groundwater and pond waters, especially in wells at the eucalyptus area (yellow and red dots plotted right below meteoric water lines, and proportion mixing line, Fig. 5). Secondly, samples of sugarcane (blue dots -more negative values) aligned with the meteoric water line, but not with the mixing line, due to its more negative values, explained by the groundwater ow direction, since the water from the pond does not pass through the sugarcane area ( Fig. 3). Lastly, the distribution of groundwater stable isotope signatures indicates that the greater the distance from the pond, the more isotopically depleted is the groundwater, suggesting that in ltrated water receives depleted rainwater isotopes while it moves away from the pond. In such transient conditions, it should be considered that surface water mixes with groundwater, resulting in distinct signatures along the route to the discharge area (the spring) ( Fig. 5 and Table 2). Therefore, groundwater stable isotope samples are aligned under global and local meteoric water lines (Fig. 5), suggesting that water from different monitoring wells contain a mixture of groundwater and evaporated pond water. These ndings corroborate the idea that the pond acts as a recharge area, and the discharge zone is located near the spring, in the northwest direction (Fig. 2). The aquifer is also recharged via excess water from the local rainfall and varies according to the land uses, occupied by the cultivation of sugarcane or eucalyptus, in this case.
Regarding the relative soil moisture level, it was determined by geophysical resistivity surveys at the eucalyptus and sugarcane plantations; and at the road to the pond (Fig. 4). Since the eucalyptus trees were removed in August of 2013, it was possible to compare soil moisture before (June) and after (September) this event. The fact that the rain has decreased since May 2013, situations were observed: (1) an overall increase in soil resistivity, more likely in humid areas, near the pond (50%), sugarcane plantation (25%), and eucalyptus (76%), suggesting that the lack of rain until September yield at least 50% of the resistivity increase in situations of no soil use alteration. (2) in the eucalyptus area, after the cutting, a disproportionate increase both in values of soil resistivity and impacted area was noted, more than those observed in sugarcane, uphill and downhill areas. On the other hand, (3) the uphill area showed a high resistivity range in June 2013, and changes caused by the dry period were not detected (16%), despite the increase of the affected area.
Regarding the geochemical data, eucalyptus groundwater samples showed overall lower ion concentrations than sugarcane, indicated by low values of EC (Table 2). Sugarcane samples presented higher ion concentrations in most groundwater samples, with calcium generally above 10 mg/L. The average sodium and potassium concentrations were relatively low compared to other ions, except in well S4, which had high potassium (up to 12.4 mg/L) and nitrate (up to 272.6 mg/L), with a low concentration of phosphate (< 0.04 mg/L), which could be related to the use of fertilizers NPK (ratio 20:05:20). Additionally, the shallow reach of sugarcane roots makes it di cult to take nutrients at depth. In other words, mobile ions, like nitrate, can escape from the root in uence and reach the aquifer. The pond presented bicarbonate-calcium-type waters with low nitrate and other ion content and intermediary values of EC. Interestingly, the pond water presented the lowest average EC comparing to eucalyptus and sugarcane groundwater. If the pond were in a discharge area context, EC's values would be higher, indicating the contribution of more mineralized water from the aquifer, as pointed by Oliva (2006) in the geochemical model of the Rio Claro Aquifer.
According to Tóth (1963), there is a hierarchical pattern of groundwater ow systems, classi ed as local, intermediate, and regional. A local ow system develops between high topographies (recharge areas) and low topographies (discharge areas), while an intermediate system consists of several low topographies intervening between recharge and discharge areas. A regional ow system has its recharge area at the highest part of the watershed and its discharge area at the lowest part of the basin. The pond is inserted within this local ow system as a recharge area, not a discharge area.
Based on these ndings, in corroboration to the potentiometric surface maps and the isotopic, geochemical and geophysical data, it is possible to a rm that: 1) the eucalyptus area does not have enough in uence to change the dynamic of the aquifer hydraulically; 2) the study area is inserted in an intermediate ow system, in uenced by the main drainage (spring located at NW of the map -Figs. 2 and 3) that commands the general hydraulic ow and with lower hydraulic heads, indicating that; 3) the pond is inserted within this system as a recharge area, not discharge (see the schematic hydrogeological conceptual model on Fig. 8).

Conclusion
The potentiometric surface map and the geophysical survey indicated that, even after cutting the eucalyptus, which is capable of changing groundwater ow directions via water absorption by roots, the groundwater ow remained virtually constant in SSE-NNW direction, where a main spring is located.
The pond water is isotopically enriched due to natural evaporation processes. As groundwater is isotopically depleted towards high topography elevations and the main spring, there is a mixture of enriched pond water with depleted uncon ned aquifer water, which re ects rainwater in ltration.
The analysis of major ions showed hydrogeochemical differences between groundwater by eucalyptus, sugarcane, and pond areas; groundwater under sugarcane cultivation is contaminated by nitrate, but not under eucalyptus cultivation. The observed increases of most ion concentrations from the pond towards higher elevations indicated that if the pond were in a discharge area context, ion concentration values would be higher in the pond, indicating more mineralized water from the aquifer to the pond.
Therefore, both eucalyptus plantation and topography elevation do not have a preponderant in uence in the area's hydrogeologic system, which is part of an intermediate ow system, in uenced by a main spring located at NW of the area that commands the general hydraulic heads.
Finally, it is not the case that all uncon ned aquifers tend to have groundwater ow directions according to terrain topography. In cases where "unusual hydrogeological behavior" is noted, as seen a priori in this paper, assessing the suitability, scale, and scope of research on the site is critical.    Piper's diagram of the major ion of waters collected in the eucalyptus (E), sugar cane (S) and pond (P) areas. On the map (right), the location of each one type of water is shown.