Beyond these potential deleterious ecological effects of REDD are important potential ecological cobenefits that include the maintenance and restoration of hydrological functions, local climate regimes, soils, and native species assemblages through both direct and indirect effects (Fig. 1). Many of the cobenefits of REDD are best understood within the context of watersheds, the natural drainage units of the landscape. The output of water, energy, and minerals from a watershed is regulated by the ecosystems that occupy it (Bormann & Likens, 1979) and therefore strongly influenced by REDD interventions.
Figure 1. Summary of potential ecological cobenefits of the five principle interventions that tropical nations could make to reduce carbon emissions from deforestation and forest degradation. Reductions in deforestation, logging damage, forest fire, and increases in forest regeneration will generally reduce the occurrence of fire in non-forest environments, increase vapor release to the atmosphere (evapotranspiration), reduce soil compaction and diminish forest fragmentation. REDD could quickly achieve benefits for stream health if it fosters regrowth or restoration of riparian zone forests. These changes will generally lead to declines in the risk of regional rainfall inhibition (relevant primarily for large forest blocks), lower annual stream discharge and flood risk, less surface run-off and associated soil erosion, and improved habitat for terrestrial and aquatic biodiversity. The loss of nutrients and sediments to streams should decline, increasing the health of these aquatic ecosystems and improving water quality. The role of tree plantations in these cobenefits will depend upon the type of plantation management practices, and the ecosystems that it is replacing.
Download figure to PowerPoint
The long-term fate of many tropical forests and their carbon stocks – including many Amazon forests – will be determined by the speed with which near-term ‘forest dieback’ proceeds (Nepstad et al., 2008). Climatologists have presented evidence of a possible late-century shift in terrestrial biomes that is driven by the accumulation of heat-trapping gases in the atmosphere that could include the drying of a large portion of the Amazon Basin (Cox et al., 2000, 2008), described below. A more conspicuous forest dieback is already underway, however, driven by land use, fire, regional climate change, and their interactions. This dieback could lead to the substitution of large areas of tropical forest by fire-prone scrub vegetation and degraded forests by 2030, releasing large amounts of carbon to the atmosphere (Nepstad et al., 2008; Malhi et al., 2009, Fig. 1). Effective REDD programs could postpone near-term forest dieback, since the positive feedback cycles that drive it in its early stages can be broken through management practices that reduce the risk of ignition sources in flammable landscapes (Goldammer, 1990; Nepstad et al., 2001; Vayda, 2006; Bowman et al., 2008). All REDD interventions, with the exception of tree plantations, decrease fire risk in tropical landscapes (Holdsworth & Uhl, 1997; Nepstad et al., 2001; Cochrane, 2003; Ray et al., 2005) and diminish the risk of regional rainfall inhibition by maintaining or restoring evapotranspiration.
Hydrology and water resources
Deforestation, selective logging, and forest fires affect watersheds and the streams that drain them by increasing runoff, river discharge, erosion and sediment fluxes (Fig. 1). These effects occur at the local scale and are influenced by the type of ecosystems that replace the forest and the ways in which these ecosystems are managed.
Land-use change influences the quantity of surface water resources by altering the partitioning of incoming precipitation and radiation among sensible and latent heat fluxes, runoff, and river discharge (Costa & Foley, 1997; Bonan et al., 2004; Li et al., 2007). Observations in watersheds from small (<1 km2) to medium (1000s km2) spatial scales in the global tropics and extra-tropics show that in almost all cases deforestation reduces evapotranspiration and increases stream flow because of the reduced leaf area index, decreased root depth, and increased soil compaction that accompany forest replacement with less water demanding crops and pastures (Bosch & Hewlett, 1982; Bruijnzeel, 1990; Nepstad et al., 1994; Sahin & Hall, 1996; Moraes et al., 2006; Scanlon et al., 2007; Thanapakpawin et al., 2007; Chaves et al., 2008). The amount of increase depends on many local factors including the amount of rainfall, how much of the watershed is deforested, topography, soils, and the land use after deforestation, but observations indicate little effect with <20% of a basin deforested and a large increase in run-off (200–800 mm yr−1) with near complete forest removal. Tropical forest regeneration on abandoned lands restores evapotranspiration levels to that of mature forests (Hölscher et al., 1997; Jipp et al., 1998), indicating that tropical forest carbon enhancement through regeneration could help to restore hydrologic functions of the primary forest.
In addition to these water balance changes deforestation and conversion to agriculture alter the morphological and biogeochemical conditions of river systems through erosion and increased sediment flux, and can include the construction of dams that block species migration and water flow, the establishment of large cattle populations (Beaulac & Reckhow, 1982; Carpenter et al., 1998; McFarland & Hauck, 1999; Ballester et al., 2003), and the input of agrochemicals (pesticides, herbicides, fertilizers, and the chemical additives of the active ingredients) (reviewed in Nepstad et al., 2006). Agricultural and livestock expansion can provoke changes in vegetation and soil organic matter nutrient cycling (Neill et al., 1995, 2001; Markewitz et al., 2001; Biggs et al., 2004), and the development of urban populations (Vollenweider, 1971; Sonzogni et al., 1980; Howarth et al., 1996; Carpenter et al., 1998).
These local changes can have profound effects on the quantity, timing, and water quality of flows in even large rivers when integrated over entire watersheds. For example, analysis of discharge data in the 175 000 km2 Tocantins and 82 000 km2 Araguaia Rivers watersheds of eastern Amazonia suggest that land cover changes beginning after 1950 and culminating in the deforestation of about 50% of these basins by 2000, are associated with an approximately 25% increase in the annual mean discharge despite no significant change in precipitation (Costa et al., 2003; Coe et al., 2009). In the case of the Araguaia, a 28% increase in the sediment load was also observed and the geomorphology of the river has been fundamentally altered to more effectively transport the increased fluxes of water and sediments (Latrubesse et al., 2009).
Maintaining natural vegetation cover is one of the most secure ways of protecting soil resources. Soils not only store carbon (about 3000 Pg globally; Tarnocai et al., 2009), they also contain essential nutrients for plant growth, purify water, and serve as habitat for diverse flora, fauna, and microbial communities. Conversion of forest to agriculture can lead to varying degrees of soil erosion and degradation, depending on management practices and soil properties (Stocking, 2003). Deforestation need not always lead to the loss of soil and soil carbon if agricultural and pasture lands are properly managed (Neill & Davidson, 2000). Well-established soil conservation practices can minimize soil erosion in agriculture, but significant soil loss is common. We live today with the legacies of soil management and mismanagement, from ancient civilizations to recent times, with most examples of historical deforestation leading to soil degradation (Montgomery, 2007). Soil erosion is a global phenomenon, but some of the highest erosion rates have been observed in tropical regions, and the wet tropical climate is considered one of the most conducive for soil erosion (Lal, 1995). Hence, a likely cobenefit of REDD that goes beyond carbon stocks alone is the conservation of soil mass, nutrients, and habitat.
Perhaps a special case of soil loss with deforestation is that of deep organic soils in southeast Asia, where about 12 million hectares of peatlands have been drained for agriculture and for oil palm plantations (Hooijer et al., 2006). Drainage allows oxygen to enter previously inundated soils, thus promoting aerobic decomposition of soil organic matter. Drying of the organic soil layer also renders it more susceptible to fire. Hooijer et al. (2006) estimated that current emissions due to decomposition and fire in drained peatlands of southeast Asia are on the order of 0.5 Pg C yr−1; higher emissions have been estimated during years of extreme drought (Page et al., 2002).
Some essential plant nutrients are lost and others are redistributed within terrestrial ecosystems when forests are harvested for timber or cut and burned for conversion to agricultural uses (McGrath et al., 2001; Davidson et al., 2004). Losses from the terrestrial ecosystem include harvest products, transport of gases, aerosols, and ash following fire, trace gas emissions from soils, soil erosion, and leaching to surface and ground waters. Redistribution includes incorporation of slash and ash material into soils and sedimentation of eroded soil in toeslope positions and streams. By far the largest of these losses of nutrient capital occurs during the initial phase of biomass removal through a combination of logging and/or clearing and burning. Fire is used both for site preparation and for subsequent weed control, resulting in significant loss of nitrogen (N) and phosphorus (P) and sometimes potassium (K) through emissions of aerosols and wind-blown ash (Kauffman et al., 1995; Hölscher et al., 1997). Significant N loss also occurs through volatilization as nitrogen oxides. Mass balance studies have shown that losses of N from Amazonian forests caused by site-clearing fires are 51–62% and 7–32% the aboveground biomass N and P, respectively (Kauffman et al., 1995). The large fraction of biomass N that is often lost during fires depletes the pool of actively cycling ecosystem N and provokes an N limitation after repeated fire (Davidson et al., 2007). This loss of nutrients can slow rates of regrowth of secondary forests (Zarin et al., 2005; Davidson et al., 2007).
Although the largest nutrient losses occur with initial and repeated fire, additional modest losses of nutrients following disturbance can occur through inputs to groundwater and stream runoff and through gaseous emissions from soils. In Amazonia, increased hydrologic export of N and P has been measured in association with deforestation in small catchments (Williams & Melack, 1997; Neill et al., 2001) and in meso-scale watersheds (Ballester et al., 2003; Biggs et al., 2004). These effects of deforestation on water quality are also mediated by soil type (Biggs et al., 2004; Davidson et al., 2004), indicating that these responses are likely to vary across regions.
Local and regional climate
The primary goal of REDD is to maintain and potentially increase carbon in standing forests, thereby reducing the release of substantial amounts of CO2 to the atmosphere and slowing further climate change (Gullison et al., 2007; IPCC, 2007). However, Global Climate Model (GCM) simulations suggest that there may be an additional mechanism by which tropical forests directly influence regional climate in a way that is unrelated to CO2 emissions to the atmosphere. Therefore, a more immediate potential cobenefit of REDD may be its positive contribution to the protection of near-term regional climate.
GCM simulations with scenarios of future tropical deforestation indicate that the replacement of large areas of forest with other vegetation types such as grass or seasonal crops, which have greater reflectivity and lower water-demands, leads to reduced net surface radiation, decreased atmospheric moisture convergence, decreased water recycling, higher surface temperature and reduced precipitation (Dickinson & Henderson-Sellers, 1988; Nobre et al., 1991; Costa & Foley, 2000; Costa et al., 2007; Sampaio et al., 2007; Malhi et al., 2008; Coe et al., 2009). The fundamental changes to the energy and water cycles caused by large-scale deforestation may feed back to the atmospheric circulation and climate altering not just regional but continental-scale rainfall patterns (e.g., Nobre et al., 1991; Pielke et al., 1998; Delire et al., 2001) and are expected to propagate to other regions of the globe (Werth & Avissar, 2002, 2005a, b). It has been suggested that these predicted changes in climate, induced by large-scale deforestation, could produce a new climate equilibrium in many locations in the tropics that is out of balance with the current forest distribution (Malhi et al., 2009) and therefore threaten the existence of tropical forests in general, including those in protected areas.
The threshold at which tropical deforestation could provoke a continental-scale change in climate – or whether or not this change will take place at all – is not yet known. A large number of GCM studies of deforestation feedbacks to atmospheric circulation have been done, particularly in the Amazon Basin. Results depend on the model used and the assumptions made and have suggested that significant regional, deforestation-driven climate change can occur with 30–60% of the basin deforested (Oyama & Nobre, 2003; Costa et al., 2007; Sampaio et al., 2007). However, the deforestation threshold of climate change is easily underestimated. Simulations rarely include a full suite of feedbacks including atmospheric aerosols, dynamic vegetation, and forest fires. Any one of these feedbacks or a combination may be large enough to significantly affect forest extent and feedback negatively to regional climate (Fig. 1). For example, recent evidence from remote sensing and atmospheric studies (Andreae et al., 2004; Williams et al., 2002) indicate that dense aerosol loading in the atmosphere during periods of high biomass burning can inhibit rainfall for weeks at a time by creating an excessive concentration of condensation nuclei in the atmosphere and by reducing net solar radiation at the land surface.
Terrestrial and aquatic biodiversity
The conservation of biodiversity – defined here to mean the native assemblages of plant and animal species and their populations – is an important potential cobenefit of REDD. Habitat loss, alteration, and fragmentation are the leading causes of declines in species and populations around the world (MEA, 2005; Gallant et al., 2007; Sodhi et al., 2008; Cumberlidge et al., 2009). REDD interventions involving native forests should help to conserve biodiversity, although tree plantations could play an important role in restoring biodiversity in degraded lands if certain conditions in their establishment and management are met and if other, more permanent habitat is easily accessible (Parrotta et al., 1997; Barlow et al., 2007). As discussed previously, the greatest potential ecological cost of REDD with respect to biodiversity would be incurred if it provides incentives to clear or degrade lower biomass vegetation that contains high levels of biodiversity.
The greatest potential of REDD cobenefits for biodiversity conservation is through slowing deforestation. However, the same absolute reduction in deforestation rates could have dramatically different ecological cobenefits for biodiversity conservation depending on the level of fragmentation of the residual forests. In general, the ratio of forest edge to forest interior is highest in small remnants and in large remnants that have long, narrow shapes (Turner, 1989; Cook et al., 2002; Fischer & Lindenmayer, 2006). Species with large area requirements and those that are interior habitat specialists, avoiding modified habitats, are generally the first to disappear from fragmented terrestrial landscapes (Tilman et al. 1994; Woodroffe & Ginsberg, 1998; Laurance et al., 2001). In addition, small fragments can only support small species populations (Turner, 1989; Baguette & Schtickzelle, 2003). With greater distances between fragments, populations of species become fragmented, with fewer opportunities for genetic exchange (Cook et al., 2002).
Selective logging adversely affects forest structure (Putz, 1991, Costa & Magnusson, 2002) and food resource availability (Fimbel et al., 2001), contributing to forest fragmentation and edge effects (Gustafson & Crow, 1996; Laurance & Bierregaard, 1997) and creating barriers for the movement of arboreal organisms (Johns, 1986; Crome & Richards, 1988; Laurance & Laurance, 1996; Putz et al., 2001; White & Tutin, 2001). Canopy gaps caused by logging contribute to changes in the composition of understory vegetation, reducing habitat quality for understory-dependent species (Thiollay, 1992; Plumptre, 2001). Sustainable forest management, including reduced impact logging techniques, which can reduce the carbon emissions associated with logging (Putz et al., 2008), have also been found to significantly reduce the impacts of logging on insect and vertebrate populations (Azevedo-Ramos et al., 2004). Reduced impact logging can also diminish the damage to the soil, which can affect soil dwelling organisms and tree root systems. However, in some regions, notably in central Africa, animals may face increased threats from hunting as any logging regime opens access to forests to hunters (Poulsen et al., 2009). This will be an important issue for nations to consider as bushmeat represents an important source of protein in some forest regions.
Fires in standing forests lead to injuries and death for sedentary species, including plants, soil dwelling organisms, insects, birds, and other vertebrates. The most vulnerable species at the time of the fire are those with low mobility, poor climbing ability, and reliance on cavity nests in trees (Barlow et al., 2002; Peres et al., 2003); subsequently, understory birds (Barlow et al., 2002) and mid-canopy and canopy bird and monkey species (Peres et al., 2003) also show declines, presumably because of changes in habitat and resource (e.g., fruit, insect) availability. Thus, any interventions that reduce accidental (non-natural) fire as a way of reducing carbon emissions would also be likely to benefit biodiversity.
Finally, REDD could protect and restore landscape-level functions performed by species such as pollination (Ricketts et al. 2004) and seed dispersal. For example, in the tropics, up to 90% of all plant species are adapted for seed dispersal by vertebrates (Howe & Smallwood, 1982; Jansen & Zuidema, 2001). As a result, the ability of vertebrates to persist in and move around tropical forests is of great importance for natural regeneration processes, which themselves ultimately contribute to enhancing carbon in regenerating forests and maintaining carbon in native (primary or old-growth) forests.
REDD could promote dramatic cobenefits for aquatic biodiversity, especially where it leads to the maintenance or restoration of riparian zone forests and watershed function. Aquatic biodiversity may be the component of tropical biodiversity that is most vulnerable to land cover/land use change. The biodiversity of lower order streams is especially vulnerable due to its dependence on exogenous food sources and on environmental conditions created by the surrounding forest (Karr & Schlosser, 1978; Vannote et al., 1980; Gregory et al., 1991; Naiman & Décamps, 1997; Benstead & Pringle, 2004). Forest streams typically flow under the closed forest canopy and food chains develop from forest organic material (Goulding, 1980; Lowrance et al., 1997; Pusey & Arthington, 2003; Sweeney et al., 2004). Deforestation can increase solar loading, temperature, sedimentation, nutrient inputs, oxygen demand and turbidity of small streams, with important impacts on aquatic biodiversity, including sharp declines in fish diversity (Barton et al., 1985; Neill et al., 2001; Abell & Allan, 2002; Melo et al., 2003; Mendonça et al., 2005). In the tropics, vulnerability to local extinctions from forest clearing is heightened by the fact that there can be great variation in species composition between adjacent rainforest streams (Lorion & Kennedy, 2009a). The range of management interventions in streams, such as check dams, invasion by pasture grasses, direct cattle impacts, and agricultural chemicals, create physical and environmental barriers to the movement of aquatic species, altering species assemblages (Flecker, 1992; Pringle & Hamazaki, 1997). Even in larger streams and rivers, where external conditions no longer determine aquatic conditions and primary production within the aquatic ecosystem plays a more important role in the aquatic food-chain, removal of riparian vegetation can significantly reduce habitat quality for many fish species (Burcham, 1988; Bojsen & Barriga, 2002; Neill et al., 2006; Lorion & Kennedy, 2009b).