4. Discussion
A growing body of literature attributes reductions in streamflow in South America to intensive plantation forestry using non-native species (Alvarez-Garreton, Lara, Boisier, & Galleguillos, 2019; Ferraz, Lima, & Rodrigues, 2013, 2019; Garcia, Salemi, Lima, & Ferraz, 2018; Huber, Iroumé, & Bathurst, 2008; Lara et al., 2009; Little, 2009, 2014). In North America, evidence is accumulating that intensively managed plantations of native species can reduce streamflow (Gronsdahl, Moore, Rosenfeld, McCleary, & Winkler, 2019; Perry & Jones, 2017; Segura et al., 2020). Drought and climate change can exacerbate these reductions (Crampe, Segura, & Jones, in review; Iroumé et al., in review).
Consistent with studies in other locations, post-clear-cutting increases in streamflow in absolute terms (mm) were highest and more consistent from year to year in fall and winter, and absolute increases were low and inconsistent in spring and summer. However, increases in streamflow in relative terms (%) were highest in the fall and summer, although summer streamflow was reduced in dry summers (2014-2015). Base flow increased gradually over the restoration period. The pre-treatment runoff ratios in the study catchments were on the high end of rates reported from Eucalyptus and other exotic forest plantations in south-central Chile and Brazil (e.g. Huber et al., 2008; Ferraz, Rodrigues, Garcia, Alvares, & Lima, 2019; Iroumé et al., 2020). If the study catchments had been replanted with Eucalyptus , we would have expected a reduction in streamflow within one to three years after the post-clear-cutting increase in streamflow, but under regenerating native forests, streamflow increased, and over the longer term, base flow also increased relative to conditions under the formerEucalyptus plantations. These findings indicate that the formerEucalyptus plantations had depleted soil moisture reservoirs in the study catchments. Clear-cutting of Eucalyptus , its replacement with Nothofagus native trees, and natural regeneration of other tree species appear to have reduced transpiration, increased soil moisture storage, and increased streamflow, except during some dry periods.
Multiple findings, including different runoff ratios and base flow responses among catchments, delayed responses of base flow to restoration, and increased streamflow in an untreated catchment which was adjacent to a clear-cut, indicate that the streams are fed in part by delayed subsurface flow. Groundwater flows as deep as 6-8 meters under saturated soil conditions in winter can be observed on road cuts in the study area. Dipping bedding planes of the underlying metamorphic rocks, which are steeper than the hillslope gradients, appear to convey subsurface flow among these small catchments. This delayed flow is an important contributor to base flow, especially during summer.
Relative increases in streamflow after clear-cutting were highest at the two catchments which had the lowest runoff ratios and where summer base flow fluctuated the most in response to seasonal and inter-annual variation in precipitation (RC5 and RC10, Table S2). Both absolute and relative responses were lower at the catchment which had the highest runoff ratio and where summer base flow was insensitive to variation in precipitation. In other words, catchments with limited streamflow contributions from delayed flow were more responsive to restoration than those with sustained base flow. These findings are consistent with other studies showing that catchment hydrology can moderate streamflow response to change in vegetation and climate (Spencer, Anderson, Silins, & Collins, 2020; Tague, Valentine, & Kotchen, 2008; Vose et al., 2016).
Streamflow increased following clear-cutting of Eucalyptusplantations and regrowth of planted native Nothofagus trees and naturally regenerated native forest species over the restoration period. The gradual recovery of annual base flow through the nine-year period of restoration, and the pronounced increase in base flow during the last three years of the study, despite low precipitation in the last two years of the study, imply that native forest restoration has the potential to restore deep soil moisture reservoirs that sustain base flow during dry periods, and therefore may enhance the resilience of restored catchments to drought.
The immediate post clear-cutting increase in streamflow in RC6 (aEucalyptus plantation that was not clear-cut), the delayed response of streamflow at the adjacent clear-cut catchment RC5, and the higher runoff ratio at RC6 compared to RC5 (Figures 3 and 4) indicate that flow is transferred from RC5 to RC6. Yet despite the streamflow subsidies from RC5, base flow at RC6 remained low from 2017 through 2019, whereas it increased in all the catchments under restoration (Figure 9). In addition, the highest correlation between summer streamflow and spring precipitation occurred at RC6. These findings indicate that high evapotranspiration rates of the priorEucalyptus plantations had depleted deep soil moisture reservoirs, which recovered, and restored base flow in restored catchments over the period of study. The Eucalyptus plantation aged from 7 to 20 years over the course of the study, whereas typical rotations in these plantations are 12-15 years. The plantation continued to grow and nearly doubled in basal area from 2010 to 2020 (Table 3). Thus, in contrast to the finding that water use by Eucalyptus is reduced in older plantations globally (Farley et al., 2005), this study indicates that aging Eucalyptus plantations continue to evapotranspire at high rates that prevent recharge of deep soil moisture reserves.
The relatively high and persistent streamflow increases in catchment RC10 (Figure 3, Figure 4, Figure 5, Figure 6) may in part be due to native forest restoration in this catchment, which had the highest survival of planted Nothofagus , highest density of other native saplings and seedlings, and no Eucalyptus saplings or trees (Table 3). However, despite similar forest restoration outcomes at RC10 and RC11, streamflow increased much less at RC11 than at RC10 during the period of restoration (Table 2, Table 3). Nevertheless, the substantially higher base flow at RC11 in 2016 to 2019 relative to before clear-cutting (Figure 7, Figure 9) implies that native forest restoration may have contributed to increased base flow even in RC11, which had the highest runoff ratio of all study catchments (Figure 3, Figure S4). Our results do not show a clear effect of the native forest riparian buffer (width and percentage of the catchment covered) on streamflow as reported earlier for this study site (Little et al., 2014).
The magnitude of streamflow response in this long-term experiment is roughly consistent with the streamflow responses modeled by Alvarez-Garretón et al. (2019), who estimated 40% reductions in streamflow in catchments with intensively managed fast-growing plantations, relative to native forest. In addition to restoring the hydrologic regime, forest restoration with native tree species also can contribute to carbon sequestration for climate change mitigation efforts (Bastin et al., 2019; Lewis, Wheeler, Mitchard, & Koch, 2019). Results from this study indicate that the restoration of native forests might increase carbon sequestration for climate regulation and at the same time improve water provision, two crucial ecosystem services. This study reinforces the tight coupling between forest management, water, and carbon, and their relationship to Sustainable Development Goals (e.g., Creed et al., 2019). The findings from this study may inform policy to address trade-offs between carbon sequestration and water yield, relevant to the National Determined Contributions of Chile within the United Nations Framework Convention on Climate Change.
This study also indicates that native forest restoration in areas of former exotic forest plantations may enhance the resilience of streamflow to climate drying in southern Chile. The study occurred during an unprecedented mega-drought (2010 to present), although precipitation records at the site do not show clear downward trends. Climate models project continued drying trends throughout the present century (Boisier, Rondanelli, Garreaud, & Muñoz, 2016; Bozkurt, Rojas, Boisier, & Valdivieso, 2018; Garreaud et al., 2017). Hence, there is a need for continued studies, such as this one, of forest management to enhance streamflow resilience to drought. A broader suite of experiments is needed to examine how native forest restoration affects water yield along a precipitation gradient in South America. Ideally, a long-term study program would consider the effects of different former forest plantation species (e.g. Eucalyptus, Pinus ), stand density and age, and soil type, as well as native forest composition, density and diversity along this climatic gradient.