What are the mechanisms of regeneration post-disturbance in tropical dry forest? (systematic review)


Tropical or subtropical dry forests originally accounted for around 42% of total global forest cover (Murphy and Lugo, 1986), but a review by Miles et al. (2006) estimated the current extent of tropical dry forests to be just 1,050,000 km2. According to the FAO Global Forest Resources Assessment (2005), total global forest cover is currently around 40,000,000 km2. Thus, tropical dry forest now only accounts for around 2.5% of total global forest cover, and virtually all of the dry forests that remain are currently exposed to a variety of different threats, largely resulting from human activity. Their mode of regeneration and resilience to disturbance is not well understood, but moisture availability plays a critical role. Vegetative regeneration may be the primary regeneration mechanism in disturbed dry forest sites where resprouting offers considerable resilience when successful regeneration by seed is highly susceptible to rainfall seasonality. Selective pressure favouring resprouting may also be related to the intensity and frequency of natural disturbances, or any canopy-opening factor that might result in heavier drought stress and mortality of seedlings. The sensitivity to moisture availability renders the regeneration of the dry forests highly vulnerable to predicted climate change effects.


We conducted a systematic global review of the mechanisms of recovery of tropical dry forests after disturbance, in order to achieve an understanding of how tropical dry forests might best be restored.


Multiple electronic databases were searched using a priori search terms. In addition, electronic communications were made to individual experts, forest departments and research institutions. Studies were considered for inclusion if they contained data on modes of regeneration after disturbance, or provided evidence of facilitation. For each study considered, the study characteristics (location, climate, experimental design, nature of intervention and comparators and outcomes measured), sources of heterogeneity and main results were recorded. Two distinct sets of observation were recorded a posteriori; i) regeneration method post-disturbance (regeneration composition by seeds or vegetative sprouting ii) evidence of facilitation (seedling survival under forest canopy or in gaps).

Main Results

The initial systematic literature identified in excess of 10,000 articles of potential relevance. This was reduced to 534 on the basis of title. Reviewers’ comments on the draft protocol advised two additional search terms (drought and dry season length) which yielded a further 545 articles, of which 56 were duplicated in the initial list. Thus, 1023 articles were finally considered on the basis of title and abstract. Of these, the full text of 35 articles was assessed and 16 articles were retained for further analyses. Of these, 14 considered regeneration post-disturbance, and 2 considered facilitative effects of the forest canopy. The studies varied widely in the time since disturbance, and the intensity of disturbance. There was also a dearth of studies which considered seedling or sprout origins of individuals in the undisturbed forest against those in disturbed forest, and we were restricted to considering a gradient of intensity of disturbance as a comparator. This departed from the terms of the protocol and limited more detailed meta-analyses of the data. Only two studies considered the facilitative effects of the canopy and were not included in any analyses.


In the range of forests studied, and across the climates and location, the response of forests to disturbance was sufficiently consistent to allow for general conclusions about their mode of recovery. Along a gradient of disturbance, it seems to be the case that following recent disturbance events, resprouts are the most prevalent individuals. When disturbance has taken place over a longer interval, individuals of seedling origin become more important in forest composition. As rainfall increases, the proportion of regeneration accounted by seedlings increases, while the proportion of coppicing individuals decreases, following cutting. However, an equal contribution by seedlings and shoots to regeneration following cutting is expected to occur at sites with a total annual rainfall of 1400 mm or more and approximately 19 – 20 years following cutting. Burning, on average reduced the predicted time at which the contribution from both seedlings and coppice will be equal (7 – 10 yrs), and the total annual rainfall at which this will occur (approximately 1150 mm/year). The implications for dry forest restoration would appear to be that initial regeneration by sprouting is most important which will affect both species and genetic diversity in the short term, but providing that further disturbance is restricted, new seedling regeneration will occur. Burning will affect both species and genetic diversity, and future seed sources. The implications from a research perspective are that results such as these could be used to develop more generic models of dry forest regeneration patterns which will enhance our ability to restore these threatened ecosystems. We need more climate change and regeneration studies in dry forest ecosystems, so that a comprehensive theory and explanation of dry forest dynamics can be built to afford greater understanding of the process and mechanisms governing regeneration in the dry tropics, this has been largely achieved for temperate and tropical rainforest ecosystems, but is as yet under-developed for tropical dry forests.


It is estimated that of the total global extent of tropical forest, tropical or subtropical dry forest accounts for 42 % of total global forest cover (Murphy & Lugo 1986). They have been exposed to severe, large-scale changes, through the cutting of valuable trees, creation of pastures, accidental or intentional fires (Gerhardt & Hytteborn 1992) and as a source of fuel wood (Murphy & Lugo 1986) and virtually all of the tropical dry forests that remain are currently exposed to a variety of different threats, largely resulting from human activity (Miles et al., 2006). However, despite their over-exploitation there have been relatively few studies of tropical dry forest and even fewer studies done on their regeneration pathways, which can provide knowledge crucial to the restoration of these forests (Vieira & Scariot, 2006).

Tropical dry forests have particular natural regeneration attributes which are not currently well understood (McLaren & McDonald, 2003b; Vieiria & Scariot, 2006). Unlike tropical moist forests, regeneration in gaps is not the primary mechanism; and the availability of moisture plays a major role in successful establishment of seedlings. The timing and duration of rainfall is crucial as there is a high probability that a given seedling will be left stranded in a drying soil and there is therefore a low probability of successful regeneration from seeds (Ewel 1980). The pronounced seasonality affects patterns of seed production, germination, survival and seedling development (Khurana & Singh 2000). Seeds of a majority of dry tropical species mature in the dry season and they are dispersed at the beginning of the rainy season when sufficient moisture is available for germination and seedling growth (Singh & Singh 1992; McLaren & McDonald, 2005). The favourable growing period is restricted to short rainy seasons when seeds are expected to germinate and seedlings establish. The deciduous state of some or most tree species allows for an increase in irradiance that in the absence of moisture exacerbates desiccation in seedlings and hence higher rates of mortality (e.g. Gerhardt 1996a).

Vegetative regeneration may well be the primary regeneration mechanism in disturbed dry forest sites, where stem and roots remain in place (Ewel 1977; Murphy & Lugo 1986; Murphy et al. 1995; McLaren & McDonald, 2003b). In the wet tropics, where large-scale disturbance occurs as a result of clearing, burning and extensive storm damage, regeneration from stem coppice is also important (e.g. Byer & Weaver 1977; Ewel 1977; Stocker 1981; Uhl et al. 1981; Putz & Brokaw 1989; Kauffman 1991; Bellingham et al. 1994). It may however, be more important in dry forest sites where resprouting offers considerable resilience to disturbance in dry forests where successful regeneration by seed is highly susceptible to rainfall seasonality (Ky-Dembele et al., 2007; McLaren & McDonald, 2003c; Vieira et al., 2006; Vieira & Scariot, 2006). Also, because trunk bases are less subject to rapid decay in the dry tropics, trees in seasonally dry forests are much more prone to reproduce vegetatively through coppicing (Ewel 1980) in response to disturbance. However, the different abilities of species to produce shoots will affect long-term species diversity in disturbed forests.

The sensitivity to moisture availability renders the regeneration of the dry forests highly vulnerable to predicted climate change effects. Most predictions of the response of tropical forests to rising CO2 concentrations and temperatures and changing precipitation patterns have concluded that changes in precipitation will have the most impact, in that drying trends will remove drought-sensitive species from the forest (e.g. Condit, 1998). However, a recent assessment of neotropical rain forests over a 20 year period showed an increase in faster- growing species to the detriment of slower-growing sub-canopy species, apparently irrespective of their light or moisture requirements (Laurance et al., 2004). This may be due to the fertilising effect of rising CO2 concentrations. However, this is uncertain, and strong droughts have been shown to cause shifts in tree-community composition in Panama (Condit et al., 1996) and niche differentiation with respect to soil water availability has been shown to determine distributions of tropical trees at both local and regional scales (Engelbrecht et al., 2007). Körner (1998) postulated that changing plant water relations could become the most important of all elevated CO2 effects on tropical forests, as water is always a selective driver of plant growth. The length of dry periods is among the key determinants of the species structure of communities (Condit, 1998; Körner, 1998). 97% of the remaining global dry forest is estimated to be ‘at risk’, and in the Americas this is mostly from climate change (Miles et al., 2006). Most studies on the effects of climate change in the tropics have been centred on tropical rainforest with a noticeable absence of work in tropical dry forest (Körner, 1998). Given that seasonally dry forests are so heavily impacted in the neotropics, and largely surrounded by agricultural landscapes, it is unlikely that they will be able to spread their range in the face of hotter and drier climates because of the lack of habitat in anthropogenic landscapes (Pennington et al., 2004). Mayle et al., (2004) demonstrated that dry forest species merely shift their ranges rather than expand them in the face of drier climates in the Bolivian Chiqitano region. If this is generally true, then the future of the seasonally dry forests is very bleak as they will “die where they stand” (Pennington et al., 2004).

Drying trends in a dry forest environment may create conditions similar to those more usually found in xeric tropical environments where plants are more drought-tolerant and seedling survival is enhanced by both the direct effects of habitat amelioration by shade, and the indirect effects of shaded plants attaining greater size before the onset of extreme conditions (Hastwell & Facelli, 2003). This is regarded as facilitation, the positive effect of plants on the establishment or growth of other plants (Holmgren et al., 1997). In the last decade, plant ecologists have focused more on the occurrence of positive plant-plant interactions than ever before, especially in severe environments. These studies have shown that facilitative effects are stronger in these environments, leaving little doubt of their generality and importance and raising questions about the assumed ubiquity of competition as the dominant interaction between neighbouring plants. Postulated relationships between facilitation and stress (Bertness & Callaway, 1994; Callaway & Walker, 1997) propose that, as conditions for plant growth becomes increasingly adverse, facilitation becomes ‘usually common’ (Bertness & Callaway, 1994) or that the ‘importance’ and intensity of facilitation increases. Dormann and Brooker (2002) suggest a model of plant interactions whereby the importance of facilitation increases along a gradient of increasing environmental harshness, while the importance of competition decreases (Figure 1). They apply this model to harsh environments, such as the Arctic, where species removal experiments tend to find facilitative rather than competitive effects. This phenomenon has also been recorded in more mesic environments which encounter periods of environmental severity, particularly drought. Hastwell and Facelli (2003) found that the relationship between facilitation and environment severity is more complex than previously thought, as neither the intensity nor the importance of facilitation necessarily increases as conditions become severe. Maestre and Cortina (2004) found that competitive interactions dominated at both extremes of an environmental gradient and suggested that a shift from facilitation to competition under high abiotic stress conditions is likely to occur when the levels of the most limiting resource are so low that the benefits provided by the facilitator cannot overcome its own resource uptake.

Thus, the relationship between facilitation and environmental severity is more complex than previously recognised, and clarification of this relationship is central for further progress in plant facilitation research (Brooker et al., 2007). The relationship is likely to be particularly complex in seasonally dry forests in a changing climate where the balance between competition and facilitation is confounded by a decrease in stomatal conductance and increased water use efficiency in response to increasing CO2 concentrations (Lewis et al., 2004). McLaren and McDonald (2003a) observed that, in shaded plots, seedling density and survival were higher than in unshaded plots, also reflected in higher mortality rates in the unshaded plots. However, competition for moisture between adult trees and seedlings was indicated in the shaded plots by negative growth rates in the seedlings, and mortality in the shaded plots was observed to be higher in the wet season. Thus, it would appear that the facilitative effects of the canopy are stronger in drier periods, and that alleviation of the moisture stress switches the effect of the interaction from facilitative to competitive (Holmgren et al., 1997; McLaren & McDonald, 2003a). A number of other studies have confirmed the importance of shading on tree seedling survival in dry tropical forests where shading significantly improved dry season survival (Gerhardt, 1993, 1996a,b, 1998; Gerhardt & Fredriksson, 1995, Hammond, 1995; Ray & Brown, 1995; McLaren & McDonald, 2003b). However, the advantages of being below the canopy at a given location may change with seasonal or transient weather conditions (Greenlee & Callaway, 1996; Tielbörger & Kadmon, 2000; Hastwell & Facelli, 2003; McLaren & McDonald, 2003b). Smith and Huston (1989) hypothesized that the response of plants to the combined effects of irradiance (Photosynthetic Active Radiation (PAR)) and water is characterized by a trade-off between drought tolerance and shade tolerance. They proposed a trade-off model, which they used to predict the growth and survival of plants along gradients of PAR and water availability. While this model was supported by empirical studies, the results of field studies on one or a few species did not support the hypothesis (e.g. Hastwell & Facelli, 2003; Holmgren, 2000; Sack & Grubb, 2002; Tielbörger & Kadmon, 2000). Holmgren (2000) therefore expressed the need to test the model using a large set of species as an essential next step in understanding positive plant- plant interactions and these data could also be used to explain species’ responses to a drying environment. Hence, it is difficult to predict the impact of drying trends on species’ interactions in tropical dry forests. Species within these environments are already drought tolerant and facilitation does occur, consequently the importance of facilitation may increase at least up to a certain threshold level of drought (Maestre and Cortina, 2004). As well as reduced consumption and demand for water, differential responses of growth to CO2 fertilization between species can be predicted (Körner, 1998). How will this then modify or affect the interaction between plants? Different species have different climatic responses, so they will respond individualistically to climate change. Consequently, not only would one expect biome shifts (e.g. replacement of rainforest by seasonally dry forest or savannah) but also significant reassortment of species within plant communities in response to such changes (Mayle et al., 2004). Furthermore, relatively little is known about how different tropical forest types will respond to future climate changes (Enquist, 2002).