Natural forest growth and human induced ecosystem disturbance influence water yield in forests

Forest complexity increases hydrological resistance to disturbances

In general, natural forests, old forests, forests with high coverage, and forests located in low aridity regions (P/PET ≥ 1) are characterized by higher ecosystem complexity than planted forests, young forests, forests with low coverage, and forests located in arid regions (P/PET < 1)^7,17,38. An ecosystem with a higher complexity is anticipated to be more resilient under the influence of external stresses and to maintain relatively stable ecosystem processes (including water cycles)³⁹. For example, a decrease in transpiration and interception due to tree removal by DECREASE treatment can be partially compensated by the additional evapotranspiration from other routes, including more open forestland, released trees formerly suppressed, reduction of wind barriers, and increases in canopy conductance, etc^40,41. In comparison, the increases in transpiration and interception from more trees introduced by INCREASE treatment can also be mitigated by the reduction of evapotranspiration resulted from more closed forestland, more suppressed trees, rising of wind barriers, and decreases in canopy conductance. These two distinct processes can be deemed as compensation and mitigation effects, respectively. Forests with higher ecosystem complexity possess stronger compensation or mitigation capabilities. The observed ΔR/Ps in PWE are, in fact, the results adjusted with compensation or mitigation effects, and thus the ΔR/Ps in PWE should be deemed as an “apparent effect” of forest changes on water yield.

The results presented in Figs. 1 and 2 confirmed the above perspective. For example, the absolute values of ΔR/P due to both DECREASE and INCREASE treatments decreased with the increases of forest coverage and age (Fig. 2), i.e., decreased with the increases of ecosystem complexity. Moreover, when comparing two contrasting combinations, we found that ΔR/Ps of the forests with lower ecosystem complexity were significantly more sensitive to both DECREASE and INCREASE treatments than the forests with higher ecosystem complexity (Fig. 1). This is consistent to a former study in which water yield was more sensitive to climate change in planted forests (lower complexity) than in natural forests (higher complexity)⁴². Similarly, evaluating whether ΔR/P differs significantly from zero, we found that DECREASE treatment significantly increased the R/Ps of seven combinations with relatively low ecosystem complexity, but this is not the case for the other one with relatively high ecosystem complexity (Fig. 1a–c). In parallel, INCREASE treatment significantly decreased the R/Ps of only three combinations with the lowest ecosystem complexity, but this did not apply to the other five ones with relatively high ecosystem complexity (Fig. 1d–f).

Generally, DECREASE treatment increases ΔR/Ps, whereas INCREASE treatment decreases ΔR/Ps (Fig. 1 and Supplementary Table 3). It is worth to note that R/P was more sensitive to DECREASE treatment than to INCREASE treatment (Fig. 1), which has been acknowledged since a long time ago³¹, but the underlying mechanism has not been clarified. The asymmetric effects are also closely related to ecosystem complexity changes caused by the two treatment events. Nevertheless, ecosystem complexity is a synthetic indicator depending on multiple factors and can barely be described by a single metric. Therefore, despite that DECREASE or INCREASE treatment is performed at the same magnitude, ecosystem complexity might not respond proportionally, which in turn exerts different hydrological effects on the system. This finding suggests that the impacts propagated through the chain from treatments, via the alteration of ecosystem complexity, to the changes of hydrological effect is highly nonlinear. Such a complexity, however, is also the primary cause of the high uncertainties of the forest–water relationship. A previous study has demonstrated theoretically that R/P changes nonlinearly with the variation of watershed characteristics¹⁷, but the nonlinear relationship between treatments and alteration of ecosystem complexity was not recognized.

Natural growth does not decrease forest R/Ps

Surprisingly, our results revealed no temporal downtrend for the mean R/Ps derived from the 278 watersheds under natural succession. This finding is robust as revealed by the comparisons of different combination categories, including the means of R/Ps derived from the two treatments, P/PET climate regions, forest origins, pre-treatment forest age categories and pre-treatment forest coverage categories (Fig. 3). Alternatively, significant increasing trends in the means of R/Ps were observed from the watersheds of lower forest coverages, various forest ages, and dryer regions (Fig. 3c, e, f). To sum up, our results suggest that natural growth of vegetation in forest ecosystems not only prevents water yield reduction in watersheds, but also increases the water yield to some extent depending on different climate conditions and underlying surface characteristics.

Ecological principles indicate that any surface on the Earth that is suitable for plant survival will be naturally covered by vegetation that is in harmony with the hydrothermal environment if given enough time²⁸. Moreover, even if external disturbance happens, the vegetation complexity (including biomass, greenness, coverage, etc) will increase continuously and naturally after interference until reaching the climax stage²⁸. This might explain the decrease and increase found in the periods before and after the 10 data-years (Fig. 3). Specifically, we suspect that forest natural succession process is artificially shortened by anthropogenic disturbance, such as tree-planting activities (e.g., INCREASE), but the ecosystem complexity has yet been fully developed in harmony with the environment (i.e., species composition, root structure, microbes, invertebrates) in the first decade after treatment. Thus, the water consumption increased due to asynchronous development of the disturbed ecosystem at first; however, as natural growth proceeds, co-evolving hydrological and other functional processes help reducing water consumption and increasing the runoff coefficient (Fig. 3). Therefore, in an ecosystem in progressive succession state over a long-term period, functional processes do not decline but are stable or increase²⁸, especially key processes such as water cycling. Hence, our results revealed no reduced water yields for the 278 watersheds experiencing a progressive succession process (natural growth without anthropic disturbances over 25 years).

One may argue that such a pattern could be explained partially by elevated atmospheric CO₂ concentration, which might promote stomatal closure^43,44, reduce water consumption, and boost photosynthetic activities in trees^40,45. Nonetheless, rising CO₂ concentration alone is insufficient to explain why the increase of R/Ps was only found in the forest ecosystems with low forest coverages. In fact, a recent study reported that rising CO₂ concentration would boost transpiration and reduce runoff in a region with limited vegetation cover since the effect of increases in leaf area index (LAI) on transpiration exceeds that of stomata closure in such ecosystems⁴⁶. Thus, the enriching atmospheric CO₂ concentration is unlikely the driver preventing forest ecosystem from water yield reduction during succession. Instead, a more plausible explanation may involve two contrasting processes that were sufficiently mutual-feedback during progressive succession. One process is the decreased water yield from enhanced transpiration and interception as vegetation complexity increases. The other process is reduced water demand for transpiration from coevolution of water cycling process and other ecological functional processes (see Supplementary Discussion). However, the enhancing effects of transpiration and interception by forest natural growth tend to diminish with the increasing “mitigation effect” as ecosystem succession progresses and ecosystem complexity develops. These mechanisms can also be explained by the changes of variables in potential evapotranspiration calculation using Penman equation⁴⁷ (see Eq. (1)):

where E_p is potential evapotranspiration, ∆ is the slope of the saturated vapor pressure curve, R_n is the net radiation, γ is the psychrometric constant, u₂ is wind speed, e_s and e_a are saturated and actual vapor pressure, respectively, and λ is the latent heat of vaporization. In Eq. (1), the values of u₂ and e_s– e_a will decrease definitely but that of R_n does not have determined change trends, which, thus, reduces E_p with forest ecosystem developing from pioneer stage to climax stage.

However, the coevolution among different ecological processes in ecosystems with progressive succession is omnipresent, even in the climax stage. With the coevolution of key functional processes such as water and carbon cycling processes, the species composition and population size, as well as seasonal and spatial community structures, are in adjustment to keep functional processes, including water cycling process, in the most effective state possible.

Perspective

Forests are not ephemeral and static entities, but exist for decades or centuries and grow constantly if no harsh external disturbances. Global PWE studies demonstrated well the effect of the abrupt forest change on water yields in an ephemeral time-span (Fig. 4). This study further highlights the effects of natural growth on water yields (Fig. 4). Our results indicate that global greening contributed from forest natural growth is unlike to decrease water resources. Even for the forestation in arid and semi-arid regions (P/PET < 1) or planting trees in the existing planted forests, the reduced water yield due to forestation will recover if no continuous artificial interference is exerted.

The present study addressed highly variable results from PWE worldwide and the underlying mechanisms, in which a convergence was identified based on ecological principles. Our study newly revealed that natural succession of forest ecosystems along with natural growth does not decrease or even increases water resource. The findings provide a generalized understanding about the integrated forest–water interactions with a global perspective. It should be pointed out that our analyses of the natural growth impacts on water yield focus on a time-span of 25 data-years, while further studies are required if more, longer measurement data become available in the future. Despite such a limitation, these results encourage us to advocate that nature-based forest restoration should be highlighted in global greening projects, especially in regions vulnerable to water shortage. Our study will benefit the sustainable development and management for promotion of the global forests. Moreover, it also provides a roadmap to the forest managers to adjust the forest management measures and policies when evaluating the balance of water supply and forest ecosystem services.

Source: Ecology - nature.com