Hydrological observations
In this article we use two sources of field data, one derived from the compilation of previous studies of our own and other teams working in the region, and the other originated in a specific long term monitoring program in the El Morro catchment motivated by the last and most intense pulse of stream incisions in the area, that took place in February of 2015. We complemented these data sources with several remote sensing tools used to characterize structural and functional changes in ecosystems.
To characterize regional decadal changes in groundwater level we took advantage of an existing synthesis of well surveys (BRS, 2002) and additional records by local farmers. We identified 23 sites in the catchment with observations between 1966-1980 (most measured in 1975), and 4 additional sites measured in 1995. Some of these sites had more than one record through time. For these 27 sites we obtained a water table depth estimate for recent times (1999 to 2020, mean 2007) based on (i) direct measurements or records by farmers (11 sites) and (ii) signals of surface waterlogging or ponding by rising water tables obtained through visual inspection of high resolution imagery from Google Earth as described below in the section about ecological observation (16 sites). In the case of wetlands that were incised we consider water table depths before and after the incision considering the stream bed position below the surrounding surface as the new position.  We also took advantage of existing isolated base flow records obtained in the developing streams by our team between 2008 and 2011 (Contreras et al., 2013) and after the last incision episode in 2015 and by others in 2008 (Barbeito, 2008). Combining the earliest groundwater level measurements available with our more recent estimates, together with the wetland and incision emergence mapping described above and available vertical electrical soundings (Barbeito, 2008), indicating depth to the crystalline bedrock; we generated a 2D, 17 km-long vertical, WSW-ENE-oriented cut representing ecohydrological changes at the high-intermediate belt of the Quebrachal and Río Nuevo sub catchments over the last 45 years (Figure 2a), for this purpose we used all available observation within a range of 600 m at the sides of the transect line.
In April of 2017, 26 months after the last and most intense incision episode, we initiated a cycle of periodic water table depth and stream flow measurements mainly focused in the Río Nuevo sub catchment but complemented with observation in La Guardia and Quebrachal sub catchments. Aimed to follow the behavior of the groundwater and stream system to new sapping episodes, the period covered up to the present has been drier than the long term average for the meteorological station of Villa Mercedes (496 vs. 608 mm y-1), offering a good opportunity to explore the mechanisms highlighted above as ecosystems become more sensitive to groundwater contributions and streams spend most time under their base flow regime.
We installed a full meteorological station (Davis Instruments) at Site A to complement an existing network maintained by a provincial public agency which has three operating stations within or close to the catchment (Villa Mercedes, Coronel Alzogaray, La Esquina; Figure 2a). All these stations provided hourly precipitation, temperature, wind direction and velocity and relative humidity data. We obtained an integrated averaged daily precipitation series after accounting for gaps (<5% of the data). We established a network of 16 monitoring groundwater wells (Figure 2a), that included pairs of wetlands with their adjacent croplands in the higher (sites A and B) and intermediate belts (site D), the largest sediment deposit and its adjacent cropland (site E) in the lower belt, a transect running 1.2 km away from the deepest Río Nuevo incision along paired forest-cropland stands with three sites in each vegetation type (site C) and two additional wells sampling a cropland adjacent to the Quebrachal incision and a suburban area in the lowest segment of the Río Nuevo. All these wells were hand augered at least 1.5 m below the water table and cased with cribbed PVC pipes. The bottom of the wells ranged from 2 to 16 m of depth. These wells were monitored three times a year manually and sampled for chemical and stable isotopic composition (data not shown here) after purging. Several wells have to be deepened as the water table levels retreated. With different degrees of completeness, water table levels were also recorded hourly with pressure transducers connected to data loggers (Campbell Scientific Instruments).
Streamflow was measured monthly during the first year and three times per year thereafter at 6 locations (Figure 2b). These included the Quebrachal stream close to its terminal zone (Site 1), two tributaries of the Río Nuevo (Sites 2 and 3) in the higher catchment and the same stream at the end of the lower catchment (Site 4). At this same point we gaged the La Guardia stream just before its convergence with the Río Nuevo (Site 5) and then the two merged streams after the flow through a 5 km-long rectified tract (Site 6). We could complement base flow series with data obtained before (Sites 1 to 4 and 6) and shortly after the incision episode of 2015 (Sites 1,3 and 5). We also included in the analysis older data for other sites, such as an active tributary of the Río Nuevo that dried after 2015 (Site 7), the higher and oldest segment of Quebrachal (Site 8) and the oldest segment of the Río Nuevo before it was deepened in 2015 (Site 9). In all cases flow was gaged using an electromagnetic velocity sensor (Marsh-Mc Birney Flo-Mate 2000) at 10 to 25 positions across the section of the stream. At sites 5 and 6 we performed several attempts to obtain continuous flow gaging that were hampered by the unstable nature of the streambed and high sediment and plant debris transport. Stage records obtained with pressure transducers in the stream bed first, and with radar distance meters under bridges or culverts next, proved to be unreliable, yet they were useful to qualitatively identify peak flow events and qualitatively sort them into minor and major ones. These provided the context for manual peak flow measurements performed during two of these events and used to obtain a maximum boundary estimate of the contribution of peak flow to stream discharge during the study period.