Groundwater dynamics
Within an overall context of raising groundwater levels, observations showed different water table behaviors depending on the temporal scales considered and the locations of observation points within the whole catchment and, more locally, regarding landscape features and vegetation types. Below we introduce long-term trends and their manifestation in a representative vertical section of the landscape, focusing afterwards on the more detailed dynamics captured after three years with periodic observations and, finally, selecting three illustrative cases of seasonal, daily and hourly level fluctuations.
Over the last four decades the El Morro catchment experienced a dramatic rise of groundwater levels leading to an expansion of the areas occupied by wetlands, lagoons and seepage zones. Water table levels raised on average 3.06 meters per decade across 27 wells where groundwater depth was measured in the past (most commonly in 1975) and measured or estimated again over the last decade (Figure 3). Noticeably, while the majority of those points showed level raises throughout one or two measuring intervals, those located within or very close (< 500m) to stream incisions (6 sites) showed absolute water table depth declines that were largest in the intermediate belt of the catchment. In the lower belt, a site that received the thickest deposit of fluvial sediments showed a relative decline of its water table level as the surface raised (Figure 3). The observed groundwater level raises were highest in the higher belt of the catchment where the most extreme case approached a 37 m level ascent and levels climbed on average 4.83 meters per decade (initial mean depth of 15.33 m)(Figure 3). That area was followed by the intermediate and lowest (terminal plain) belts where, ignoring the incision cases, decadal raises averaged 3.50 and 3.77 meters per decade, respectively (initial mean depths of 11.55 and 7.85 m, respectively)(Figure 3). The smallest changes were observed in the lower belt of the catchment, where the average ascent was 0.97 meters per decade, reflecting the fact that water tables were already shallow at the beginning of the study period there (initial mean depth of 3.6 m)(Figure 3).
Based on the higher availability of data and fast changes observed in the contact zone between the higher and intermediate belts of the catchment we reconstructed a transverse vertical cut depicting major landscape and groundwater shifts (Figure 4). By the year 2000 the water table in this section of the catchment (covering the full sub catchments of Quebrachal and Río Nuevo streams), approached the surface in multiple low positions that were disconnected from groundwater in the late 70s. Incisions formed in many of the local low positions in west to east temporal sequences (Figure 4). A relatively uniform lateral distribution of incisions or wetlands with permanently running surface wetlands can be appreciated along the transect with distances of 3.2, 3.2, 2.6, 1.9, and 2.1 km (2.6 km on average) among them. A similarly regular but larger spacing is observed in the intermediate and lower catchment, averaging 12.5 km between the main collectors of each sub catchment.
Water table depth observations in the network of monitoring wells reflected the dry conditions of the 3-year study period with an overall deepening trend (Figure 5). However, important contrasts were found between vegetation types and landscape positions. At site A, the wetland, which appeared in 2000 and has the crystalline basement close to the surface (see Figures 3 and S1), showed the most stable levels, always less than 15 cm below the surface, likely as a result of sustained seepage at that site (Figure 5a). The neighboring cropland instead, showed seasonal oscillations (summer declines, winter raises) over imposed to the general declining trend, suggesting that water consumption during the period of high vegetation activity and evapotranspirative demand, which is also the rainy season, outpaced recharge fluxes during the study years (Figure 5a, d, e). Wells at site B showed a more extreme, steady and parallel decline during the first year (Figure 5b) (they had to be deepened to continue measurements afterwards), revealing the likely effects of the recently carved or deepened incisions flanking the site (Upper Rio Nuevo and Uke tributaries), drying a wetland that emerged in the late 90s and interrupting its stream (site 7 in Figures 2b and 6).  A particular aspect of this level decline is its constant rate in the wet and dry season of the first year, which suggests drainage from below rather than consumption from above as the likely cause of this behavior. Located by the deepest incision of the catchment (enlarged from 9 to 20 m of depth and 28 to 60 m of width in 2015), site C represented the deepest water table situation showing a strong contrast for its three pairs of forest-cropland wells (Figure 5c). There, the forest, similarly to the cropland and wetland sites with shallower water tables, displayed seasonal oscillations suggesting a pulsed consumption of groundwater by the dominant winter-deciduous trees whose active roots were found growing around sensors at 9 m of depth. Croplands, only 600 m away at this site, showed a steady decline with very subtle seasonal rates of change. Differences along the transect at this site suggest that wells at position 1, ~100 m away from the deepened incision are closer to rich there a new equilibrium while those at positions 2 and 3 may still be yielding water towards the stream (Figure 5c).  Two additional monitoring wells not shown in the figure reflected similar patterns with Site F hosting a cropland with deep water tables close to the Quebrachal incision. That site followed croplands in site C until no saturated conditions could be detected above the crystalline bedrock at 14 m of depth. Site G, in a suburban area of the low catchment affected by periodic waterlogging followed the same pattern of the wetland at site E.
Continuous monitoring of groundwater depth with pressure transducers offered a more precise perspective on the recharge and discharge dynamics of the study sites (Figures 6 and 7). At Site A we highlight contrasting periods of net water gains (summer-fall 2018) and losses (spring-summer 2019-2020), showing the contrasting behavior of the wetland and the cropland. In the first of these periods, the wetland displayed high daily level fluctuations that amplified progressively following rainfall events (Figure 6, left panels), suggesting an increasing reliance on groundwater as the rainfall pulse was consumed. No fluctuations were observed in the cropland, that at that time was in fallow stage prior to the establishment of a pasture (see NDVI in figure 6, left panel). A single and intense rainfall event in this period (74 mm on April 1 measured at the precise site) switched the wetland to a flooded condition reducing but not eliminating fluctuations, likely as a result of the shifting groundwater yield-depth relationship under flooding. The same event and the following two triggered direct recharge episodes under the fallow with level rises of 2.7 – 3.8 mm per mm of rainfall (or a specific yield of 0.26-0.37)(Figure 6, left panels). In the second period, which followed three consecutive dry growing seasons, the wetland displayed wider diurnal fluctuations with the same pulsed reduction and gradual amplification in response to rainfall events (Figure 6, right panels). These fluctuations averaged 15.2 cm d-1 (daily maximum – minimum level) and, based on Loheide, Butler and Gorelick (2005) methodology, generated an average groundwater level depletion of 16.6 cm d-1. Noticeably, in spite of its apparent reliance on groundwater, the wetland greenness increased after rainfall events (see NDVI in figure 6, right panel). The cropland in this period was covered by an active (and typically deep-rooted) alfalfa pasture which was likely responsible for sustained water table fluctuations averaging 1.6 cm d-1 with an estimated mean depletion of 4.9 cm d-1. The largest rainfall pulse of this period (125 mm in November 17-20), triggered a slow 36 cm level raise of 10 days in the cropland which had a much lower velocity than the recharge pulses of the previous study period (Figure 6, right vs. left panel), suggesting that the level recovery resulted from the interruption of groundwater consumption.
Deep water table levels at site C, likely favored by its neighboring incision, provided an opportunity to explore vegetation-groundwater interactions under conditions that could have been more common for the catchment in the past (Figure 7). Sensor data provided here a more detailed description of the seasonal departure of water tables at forest stands with regard to their neighboring croplands (Figure 7 vs. Figure 5), showing that forest trees produce similar diurnal fluctuation, seasonal depression and slow level recoveries after rainfall events than those described for the alfalfa pasture above, yet they do so with water tables that are 8-10 m deep (Figure 7). Groundwater uptake, as indicated by daily water table fluctuations, started a few days after the deciduous tree species leafed out in mid-October (see NDVI cycles in Figure 7) and levels initiated their sustained and asymptotic recovery when the trees gradually lost their leaves in late May, approaching equilibrium with the surrounding cropland matrix throughout the dry (by ecologically inactive) season (Figure 7). Diurnal water table level fluctuations during the active period of the forest (October 15 - April 15) averaged 1.7 cm d-1 with an estimated mean depletion of 3.1 cm d-1. In the three consecutive years of observations, levels dropped/recovered -89/47, -71/55 and -79/57 cm in the forest site. The same net drop of -80 cm over the three-year measurement period was observed in the cropland but without seasonal fluctuations. No direct recharge episodes were documented at any of these wells (Figure 7), yet slow and partial level recoveries in response to rainfall pulses were observed in the forest, as depicted in detail for a series of events in March 2019 (Figure 7, bottom left panel) and recorded in November 2019 and January 2020 when of 15 cm-recoveries took place along 8 and 17 days following rainfall pulses of 90 and 77 mm, respectively (Figure 7).