Forest regeneration can positively contribute to local hydrological ecosystem services: Implications for forest landscape restoration

There is enormous global interest in forest restoration, with governments around the world making high profile and ambitious commitments (Fagan et al., 2020). Carbon sequestration an important driver of this policy interest, but it widely recognized that restoration needs to be considered as part a landscape approach which aims improve both ecological functioning livelihoods local people (Chazdon 2017). Passive through encouraging natural regeneration, can cost-effective successful mechanism for & Guariguata, 2016; Crouzeilles 2017; Lamb 2005; Molin 2018). In large areas tropics where shifting cultivation still loss (Curtis 2018), lengthening fallow periods or even phasing out system favour more productive agriculture on smaller land area could greatly increase tree cover. Ecologists have shown tropical regeneration following abandonment make substantial contribution carbon biodiversity conservation (Mukul Rozendaal 2019). However, remarkably little known about likely impacts ecohydrological processes underpinning locally hydrological ecosystem services. Such understanding critical ensuring benefits people. Hydrological services are accrue due terrestrial ecosystems freshwater (Brauman 2007). Forests play role large-scale water cycling, ultimately contributing precipitation downwind from forests (Ellison our focus paper services, such flood mitigation maintenance streamflow during dry (known baseflow). Ecology has strong influence hydrology therefore The risk flooding erosion strongly influenced by infiltration overland flow leaf litter decomposition enhances soil structure, faunal activity root channels form pathways move quickly into subsoil (Lozano-Baez 2019; Zimmermann 2006). Forest clearance reduce (Brookhuis Hein, Peña Recha 2012) while vegetation recovery reverse these (Germer 2010; empirical evidence knock-on mixed (Calder Aylward, 2006; van Dijk 2009) depends much than hillslope hydrology. Because intercept transpire other vegetation, all things being equal, will tend decrease total yield (Farley Liu 2016). effect increasing cover baseflow (the river drier periods) depend whether increased evapotranspiration greater less associated groundwater recharge improved precipitation. This ‘infiltration trade-off’ hypothesis (Bruijnzeel, 1989). systematic review, third 53 included studies suggested reforestation baseflows, although was extremely limited (Filoso Some started explore effects using data models (Krishnaswamy 2018; Lacombe Ribolzi 2018) how trade-off plays practice very because difficult collect. Field exploring terms rare, few quantified net (i.e. amount infiltrates not returned atmosphere rather flows underground towards streams). Most either Litt 2020) (Sommer 2002; Wolf 2011). result despite importance rainforests base remains limited. Madagascar committed reforest 4 million hectares under Bonn Challenge forms country's Individual Nationally Determined Contribution Paris Climate Agreement. Since taking office, new president underlined significant goal (Jones More 75% Malagasy live extreme poverty (The World Bank, irrigated rice rural economy. As such, vital understand impact affect production. These continuous streams ensures available irrigation between rainfall events (for indicator), peaks indicator) damage systems cultivation. We present one first link flow, cycle. combine detailed plot-scale measurements whole year semi-mature forest, degraded abandoned agricultural (Ghimire 2020), support inferences rates maximum depths across same spectrum types (Zwartendijk Over last 60 years, lost nearly 45% its (Vieilledent While forestry operations commercial played (Scales, 2014), major been small-scale Much remaining eastern rainforest mosaic patches regenerating (Styger Our study region Corridor Ankeniheny Zahamena, IUCN category VI protected (Figure 1). humid climate; average annual varies 1,500 3,500 mm/year, approximately three quarters falling November April (see Figures S1 S2). soils mainly highly weathered Ferralsols metamorphic igneous basement rock S3). majority 60,000 living 450 peripheral villages carry mixture hillslopes small valley bottoms (Poudyal Both lack at start wet season (November–December), (especially when ready harvest end season) limit production S4). took cycle (forest, land). aim construct budget including (relevant use erosion) evapotranspiration, estimate baseflows). Due heterogeneous area, possible identify catchments each single type compare dynamics. It logistically (and financially) impossible replicate instrumented plots locations. assess representativeness measurements, we determined sites were selected based presence indicator species 2007) interviews regarding history. total, 12 sites, 15 10 1; Table S1). See Figure S5 field work context Supporting Information justification space-for-time additional details site selection. 0.25-ha (one representing type; 1) chosen region, accessibility organization who provide daily monitoring equipment (Supporting Information). plot secondary had cleared memory, experienced selective manual logging until 1995. long history 1990. 2000, some enrichment planting project weeding, planted saplings died. used many cycles 2000. Regeneration poor, typical repeated short durations Although within 2.5 km 1), installed weather station obtain climate S6). Overland OF measured 3 × m bounded runoff collected metal troughs drained drums emptied daily. Two (values averaged). interception calculated difference throughfall plus stemflow. For plots, entire measurement period 66 funnel gauges daily, well V-shaped connected tipping bucket. Stemflow trees five stemflow collars containers also Ghimire (2017) details. Tree transpiration thermal dissipation probes 21 scaled multiplying sap flux density sapwood 2018 details). Litter evaporation water-holding capacity model simulations. plot, loss, simulations Soil moisture four different (5, 15, 40 75 cm). change zone storage ΔS 1-year stored upper cm beginning period. exact depth rooting known, roots observed below cm. Changes assumed negligible time-scale. R budget. residual fluxes (Equation includes errors. Differences among thus need interpreted care. Net represents leaves via lateral subsurface vertical drainage streams. may reach shallow contribute (particularly events) instead baseflow. assume feeds bodies beneath foot-slopes bottoms, concentrated. consider service communities further assumption). contrast, leads affects most useful quick response events. dis-service acknowledge re-infiltrate lower positions. complemented 37 wider At site, saturated hydraulic conductivity (Ksat) surface double ring infiltrometer 20–30 constant head permeameter likelihood generation deep infiltration. taken locations along upslope downslope transect; median value representative site. To determine statistical significance differences Ksat values per type, Kruskal–Wallis Dunn post hoc test (p < 0.05). subset (Table S1), investigated preferential spraying brilliant blue dye Zwartendijk similar fallow, higher plot. only 2% 11% 2a). wettest month period, ratio (22% precipitation; 2b). Total greatest (65% compared 42% 28% land; caused losses relatively minor component (<6% precipitation). (−1% plot). smallest (34% precipitation) (56% 63% respectively; month, highest lowest results corroborate findings unlikely land. 95th percentile 5-min intensity except 3a). means intense events, infiltrate soil, regardless type. depth, two orders magnitude 3b; common intensities, slower rate saturation occur above layer. If enough saturate layer, then occur. Visual inspection tracer experiments patterns differed 4). moved remained largely top soil. there no types; penetrated clay rich layer 30 S7). findings, dominates Madagascar, suggest perspective minimizing maximizing recharge, fallows perform well. mature better store (Andriamananjara 2016), supports (Lennox 2019), show growth further, come cost reduced recharge. 1989): (both forest) resulting finding (with underlying mechanism) reported Ilstedt (2016), found intermediate densities (in woodland semi-arid West Africa) optimal note young trees. Transpiration matures (cf. Giambelluca, 2002). reduction fairly 43 (mostly temperate) studied Bentley Coomes (2020), decreases after permanent, suggesting decline mature. An caveat partial fully rivers, information hydrogeological setting, example, provided riparian aquifers, partitioning fast would needed. non-land factors (soil texture, aspect, plots) fluxes. Of course, beyond those addressed paper. quality sediment contaminants, particularly replace (Herrera downstream (rather valleys paper) reservoirs setting regimes. users attenuation peak, capture peak flows. realized number, distribution (van Soesbergen Mulligan, When assessing variations (exposure solar radiation fog) important. Very could, least theory, positively 2017) though surrounding equivocal 70,000 km2 China did regional (Zhou 2010). (as elsewhere; Swanson 2011) so far tended underplaying potential contributed earlier stages succession. minimize just older stands, normal conditions. Decreasing (an threat farming), baseflows near farming often limited). Therefore, covered Given cost, success, active (Busch 2012), positive way deliver Madagascar's targets. valued acceptability interventions (Mansourian poorest dependent cultivation, efforts low input negatively their Conservation policies generally do costs, equitably shared (Gardner 2013). requires knowledge policies, impacts. real delivering benefits, modelling cases inform planning initiatives (e.g. Sun Trabucco 2008). However 2005), ignored infiltration). A recent infiltration) account (Peña-Arancibia 2019) showed spatially interplay seasonality, evaporative demand, condition. Models good mechanistic underpins them parameterize them. So far, lacking regions targeted restoration. works adds needed processes. biodiversity, already risk, reasonable maintain They they age. abundance deforested world, lengths, delivery. throughout help initiatives. authors thank Association Mitsinjo authorities permission practical help, M.W. Lubczynski L.A. Bruijnzeel valuable comments, p4ges team. artist Eric Andriantsialonina drawings 2. extended acknowledgements. funded alleviation program NE/K010220-1 NE/K010417/1. H.J.v.M., C.P.G., B.W.Z., M.M. J.P.G.J. conceived designed methodology; J.L. M.R. collection; H.J.v.M. led writing. All critically drafts gave final approval. https://doi.org/10.5285/5d080fef-613a-4f24-a613-b249ccdd12bf 2020); https://doi.org/10.5285/7987c6d4-973d-436d-a13b-c52997d0bce5 sapflow https://doi.org/10.5281/zenodo.3971689 (Poyatos Please note: publisher responsible content functionality any supporting supplied authors. Any queries (other missing content) should directed corresponding author article.