Introduction
Although the critical role of water inputs supporting the growth of plants and the flow of streams is indisputable, the way in which these two major fates of precipitation may compete with each other on a given territory is not always obvious and represents, in fact, a very active front of scientific inquiry. Moreover, this quest is perhaps one of the main attractors of the convergence of Ecology and Hydrology over the last decades. How any given  piece of land, and more operatively, each catchment, partitions the water that it receives from precipitation between evaporative losses (among which plant transpiration is usually dominant) and liquid losses (among which stream flow is generally the main protagonist), is something that has been quantified for more than a century using multiple approaches and that has sparked key integrative theoretical developments such as the Budyko curve (Budyko, 1974; Gentine et al., 2012). It is notable, however, that our knowledge about the mechanisms dictating this partition is still very fragmented (McDonnell et al., 2007). In which ways plants can ”capture” water that could eventually reach streams or, conversely, what are the mechanisms that allow streams to “capture” water that may be potentially usable by plants? Particularly relevant but disconnected are our representations of the role of saturated flows dictating a “push and pull” dialog between plants and groundwater (Fan, Miguez-Macho, Jobbágy, Jackson, & Otero-Casal, 2017) and eventually, streams. In this contribution we explore conceptually how the competition for water between plants and streams may operate in a dry basin, proposing a list of key mechanisms that involve groundwater-mediated interactions. Plant-stream competition should be particularly evident in dry regions (PET > PPT), where potentially all water inputs can get evaporated (no energy limitation) and the outcomes of the “push and pull” process should leave a strong imprint both on plants, which would increase their growth as their capture of limiting water resources becomes more exhaustive, and on streams, which can potentially have a total suppression of their flow. In this contribution, we illustrate the proposed mechanisms and evaluate their relevance both for land ecosystems and hydrology with data from a semi-arid sedimentary basin subject to strong vegetation changes and, mostly as their consequence, to an extraordinary rapid process of novel stream formation through sapping and surface erosion.
A simplified view of the atmosphere-soil-plant/stream continuum can assume that all the water consumed by vegetation comes from the unsaturated zone and that any by-pass or displacement flow escaping root access becomes free to rich the saturated zone and is no longer accessible to plants. This is still the prevalent representation of many models that can successfully describe the partition of evapotranspiration and streamflow in many catchments (Rahtjens, Oppelt, Bosch, Arnold, & Volk, 2015) based on what we can define as the single most relevant mechanism that plants have to take away water from streams (i.e. unsaturated uptake, mechanism 1, figure 1). In dry regions where evapotranspiration is water limited, flat and loose sedimentary substrates occupied by deep-rooted vegetation often display an exhaustive exploration of the wettable soil volume, leading to the total suppression of groundwater recharge through “mechanism 1”. In these settings any streamflow is limited to storm run-off episodes and there is no observable base flow (Scanlon et al. 2006; Santoni, Contreras, & Jobbágy, 2010). While simple and powerful enough to explain the behavior of many dry catchments, unsaturated uptake runs short to explain many biological and hydrological phenomena such as the presence of exceptionally green vegetation patches in deserts or the fluctuating daily or seasonal reduction of stream base flow during periods of high water demand and no rainfall inputs.  Plant roots access saturated water directly, as in the obvious case of riparian or wetland environments were specific vegetation types adapted to the anoxic conditions of the substrate thrive (Lowry, Loheide, Moore, & Lundquist, 2011), or indirectly by tapping the capillary fringe fed by the saturated zone, as seen (not as easily) with phreatophytic ecosystems (Naumburg, Mata-González, Hunter, & Martin, 2005; Jobbágy, Nosetto, Villagra, & Jackson, 2011). Therefore, saturated uptake needs to be included as a “second chance” that plants have to take away water from streams with the unique contribution of lateral connectivity allowing vegetation in one place to use water that plants did not use in another. The most acknowledged component of saturated uptake is the consumption of ground or stream water by wetland and riparian communities in the small fraction of landscapes occupied by topographic low positions and water body shores (i.e. focalized saturated uptake, mechanism 2, figure 1)(Bond et al., 2002). Yet, saturated uptake can be significant in the matrix of the landscape as well, when a more widespread contact between roots and the capillary fringe takes place (i.e. distributed saturated uptake, mechanism 3, figure 1), something that appears to be reciprocally and dynamically modulated by the vertical location of roots and water tables and highly sensitive to vegetation, climate and water managements shifts (Jobbágy & Jackson, 2004; Nosetto, Jobbágy, Jackson, & Sznaider, 2009, Fan et al., 2017). While being increasingly considered in catchment models (Immerzeel & Droogers, 2008, Doble & Crosbie, 2017), saturated water uptake by plants introduces a novel possibility for streams to take away water from plants.
Far from passively receiving the hydrological left-over of the three main water consumption mechanisms pathways highlighted above, streams, as active geomorphological agents, can take away water from plants when they remove and deposit sediments. By deepening their beds, streams can increase their groundwater capture into base flow, transiently enhancing hydraulic gradients but more permanently by impairing mechanisms 2 and 3 as water tables plunge to reach a new equilibrium and zones where plant roots can reach it shrink or disappear (i.e. streambed deepening, mechanism 4, figure 1) (Bravard et al., 1998; Wurster, Cooper, & Sanford, 2003; Schilling, Zhang, & Drobney, 2004). Another way in which streams can actively restrict plant water consumption is by burying land ecosystems with new sediment deposits, which are more likely to affect active riparian and wetland zones with higher water consumption than any other landscape component (i.e. sediment deposit, mechanism 5, figure 1) (Kui & Stella, 2016). The relative timing of ecological succession and geomorphological changes is critical defining the resulting balance of mechanisms 1,2 and 3 vs. 4 and 5. Ecological changes such as the establishment of riparian plants with deeper roots where stream incisions have gone deeper or the recovery of buried trees where streams have deposited fresh sediments (Rivaes et al., 2013; Kui & Stella, 2016), may outpace the slower work of streams, yet landscapes in which their erosive power get rapidly unleashed as result of climate or land use changes can match their forces offering an ideal observatory to explore their interaction.