dipterocarps on Borneo
5.2 What is the importance of small fragments for conservation?
The species-area relationship (SAR) originally described for islands has been found in many taxa occupying habitat fragment ‘islands’, as demonstrated by studies on birds (e.g. Hill et al. 2011), ants (e.g. Brühl et al. 2003), dung beetles (Nichols et al. 2007), bats (e.g. Struebig et al. 2008), and vascular plants (Munguía-Rosas & Montiel 2014). These relationships call into question the value of small fragments for the conservation of biodiversity. I found that the number of tree seedling genera in plots in lowland rainforest fragments on Borneo increased with fragment area (Chapter 2), a pattern which was absent among mature trees or saplings, and which may indicate greater taxonomic impoverishment of seedlings in smaller fragments as a whole. Studies on tropical tree diversity are frequently conducted at the genus-level due to the difficulties of tropical tree identification (e.g. Ewers et al. 2017, Laurance et al. 2004). Although using this higher taxonomic level reduces the resolution of the data, it enables analyses of population trends that wouldn’t be possible at the species level due to the high proportion of rare species occurring at low densities in tropical rainforest. In addition, congeneric species are often ecologically similar (Laurance et al. 2004), and I found a strong correlation between the number of species and number of genera identified per plot (Figure S2.1). Thus, studies conducted on tropical trees at genus- and species- level are likely to reveal trends robust to the resolution of the analysis. My study was based on the assumption that the majority of seedlings (<1cm dbh and <1.5 m in height) and saplings (1 – 5 cm dbh) would have germinated after fragmentation occurred (about 20 years ago), and would therefore provide
an insight into the effects of fragmentation not yet visible in the mature tree cohort due to their longevity. The unexpected sapling results that led me to conclude that density dependent mortality and/or year-to-year variation in which genera are recruiting may offset losses, at the plot scale, in seedling richness. These ideas are discussed in detail in Chapter 2 (section 2.5), however my results for seedlings led me to conclude that small fragments are experiencing an overall decline in the genus richness of post-fragmentation seedling recruits.
The reduction I found in seedling alpha diversity in fragments may have important implications for future tree diversity, as well as for the diversity of other forest organisms, which often increase in parallel with plant diversity. The physical structure of the forest is largely determined by the tree community, so a more diverse tree community provides a greater diversity of niches for other species to occupy (Jones et al. 1994). Novotny et al. (2006) attribute the latitudinal gradient in insect species diversity from temperate to tropical regions to the equivalent gradient in plant diversity. They found that the host specificity and number of insect species per host plant were similar in temperate and tropical forests, but that the greater host plant diversity in the tropics provided the opportunities for greater insect diversity (Novotny et al. 2006). As a result, a reduction in the taxonomic richness of trees in fragments is likely to have similar consequences for insects dependent on these plants, as well as other taxa occupying specialised niches, if these niches disappear. Thus, small fragments with an impoverished tree community generally support lower levels of taxonomic richness than larger fragments or continuous forest.
However, in contrast to patterns of alpha diversity, studies considering patterns of beta diversity between fragments within a landscape reveal greater variation in responses to fragmentation. These may differ depending on geographic location, elapsed time since fragmentation, fragment disturbance history, study taxon, and the range of fragment sizes included in the study. I found that although the number of genera of tree seedlings recruited after fragment formation declined with fragment area (Chapter 2), different genera were recruiting in different fragments (Chapter 3). Thus, tree seedling alpha diversity declined with area, but beta diversity increased. Given that the diversity of seedlings offers insight into the diversity and trajectory of change of tree communities in future, this finding has important implications when considering the conservation of overall landscape-scale diversity of forest species within a human-modified landscape. If subsets of tree genera are able to persist long- term in small fragments by continuing to recruit seedlings, they may constitute valuable reservoirs of tree diversity, and hence retain considerable value for diversity conservation. This value is increased if different genera are recruiting in different fragments, as I found, such that
the network of fragments embedded in a human-altered landscape maintain a high level of diversity at this broader scale.
My finding, that small fragments can make significant contributions to landscape-scale tree diversity, corroborates those of other studies conducted in rainforest fragments in Central and South America (e.g. Sfair et al. 2016,Hernández-Ruedas et al. 2014, Arroyo-Rodríguez et al. 2013, Arroyo-Rodríguez et al. 2009, dos Santos et al. 2007). Seedling communities in small fragments are probably following contrasting pathways due to differences in fragment
attributes and disturbance history. However, my findings contrast with other studies that have reported taxonomic impoverishment accompanied by homogenization of tree communities in small fragments within a landscape (e.g. Lôbo et al. 2011, Pütz et al. 2011). Homogenization can occur if rare or specialist species decline and common, generalist or pioneer-type species spread and proliferate, thus increasing the similarity in composition between communities (Olden 2006). These studies, that question the conservation value of individual small fragments, are predominantly conducted in the Atlantic forest of northeast Brazil, a globally important biodiversity hotspot which is highly fragmented. Only around 16 % of the original forest remains, distributed amongst fragments of which 80% are smaller than 50 ha, and most forest fragments are composed of disturbed, second-growth forest (Ribeiro et al. 2009). The size of Atlantic forest fragments is in contrast to the forest fragments sampled in this study, of which only the two smallest were smaller than 50 ha (see Figure 2.1, Table 2.1). This difference in the size of ‘small’ fragments in different studies may suggest that there is a size and/or disturbance threshold below which fragments have a poor ability to sustain populations of rare or specialist species, but may also be due to regional differences in the flora and fauna, and the intensity of land use in the intervening matrix.
Carbon sequestration was not a focus of my study, but it is evident that many forest fragments continue to support a high number of large trees, and thus have some value for carbon
storage, as well as for conserving landscape diversity. I found that there was no significant difference in the density of large trees (>30 cm dbh) between plots in forest fragments and continuous forest, indicating the potential of fragments for carbon storage (Appendix 1B, Figure S2.3). Undisturbed dipterocarp forest on Borneo has some of the highest carbon stocks per hectare globally due to stem densities of large trees 2 – 4 times higher than forest in the neotropics, and 60% greater average above-ground biomass (Paoli et al. 2008; Slik et al. 2010). However, deforestation, degradation and forest fragmentation are significant sources of greenhouse gas emissions (Qie et al. 2017), and extensive logging and forest clearance over the last half-century has turned tropical lowlands from an important carbon sink to a significant carbon source: Malaysia and Indonesia have some of the highest annual country-
level estimates of carbon emissions from deforestation (calculated for 2000 – 2005; Harris et al. 2012). Not only do oil palm monocultures store about 80% less above-ground-carbon than primary forest (Ziegler et al. 2012), but cultivation exposes the soil, resulting in the oxidation of up to 30% of the organic matter held in the top metre of soil and releasing carbon dioxide (Houghton 2005). Forest fragments tend to store less carbon than the equivalent area of intact forest due to elevated rates of mortality of large trees near forest edges (Dantas de Paula et al. 2011), and forest within 0.5 km of an edge is more likely to be a carbon source than a carbon sink (Qie et al. 2017). In order to avoid the worst edge effects I located all plots at least 100m from fragment edges, as it was not a focus of my study to investigate edge effects. This may explain why I did not find significant reductions overall in large tree density in forest fragments (Appendix 1B, Figure S2.3). Changes in community composition towards disturbance-loving, fast-growing pioneer tree species with lower wood densities can also reduce above ground biomass (Bunker et al. 2005), but if heavily disturbed fragments recover their vegetation structure to be more similar to undisturbed forest they may re-gain above ground carbon (Powers & Marín-Spiotta 2017). Contrary to this expectation, I found no difference in the effects of fragment size on the alpha diversity of seedlings, saplings, and trees, when I tested low and high wood density genera separately (Appendix 1B, Table S2.8). Although larger blocks of forest generally experience fewer edge effects and less disturbance than the equivalent area separated into smaller blocks of forest, and thus have a greater capacity for storing carbon, existing small fragments may still have some value in this respect. Fragments are likely to retain substantially higher levels of above ground carbon than oil palm or other crop species, especially relatively undisturbed fragments like those in my study, and thus continue to represent important reservoirs of carbon that should be retained within the landscape. Turner and Corlett (1996) conclude that even fragments smaller than 100 ha are better than non-forest habitat for the conservation of forest specialist species. Lucey et al. (2016) suggest that a fragment with a core area of 200 ha can support up to 70% of the species richness of the same area of continuous forest. Considering that there are about 54 million patches of
fragmented forest in the tropics, that the mean fragment area is only 29 ha, and that they cover an area of more than 1.5 billion ha (Brinck et al. 2017), small fragments represent a substantial area of forest. Despite supporting impoverished assemblages of forest organisms and lower ecosystem functioning relative to primary forest, fragments smaller than 200 ha may sustain assemblages of trees which are no longer found anywhere else, as well as
supporting forest specialist species that cannot survive in an agricultural landscape. Fragments can act as stepping-stones, connecting larger areas of forest, and can provide food and shelter to migratory animals (Turner & Corlett 1996). Although fragment connectivity was
unimportant relative to area for tree seedling alpha diversity (Chapter 1), the distribution of fragments within the landscape is likely to have important consequences for the permeability of the intervening matrix for other forest species. Fragments can also act as reservoirs of species and genetic diversity, and may be seed sources for natural and assisted recolonization. Tree longevity provides longer time scales over which to act to save tree species, and in 50 or 100 years’ time, areas of forest may re-expand if trends for agricultural abandonment in Europe play out in other parts of the world (Navarro & Pereira 2015). Consequently, although continuous forest is clearly superior for the conservation of biodiversity, where fragmentation has already resulted in the formation of small forest fragments they can make a valuable contribution to landscape-scale conservation of biodiversity.