Effects of fragmentation and landscape variation on tree diversity in post-logging regrowth forests of the Southern Philippines

Abstract

The conservation value of forest fragments remains controversial. An extensive inventory of rainforest trees in post-logging regrowth forest in the southern Philippines provided a rare opportunity to compare stem density, species richness, diversity and biotic similarity between two types of post-logging forests: broken-canopy forest fragments and adjacent tracts of closed-canopy ‘contiguous’ forest. Tree density was much lower in the fragments, but rarefied species richness was higher. ‘Hill’ numbers, computed as the exponential of Shannon’s diversity index and the inverse of Simpson’s diversity index, indicated that fragments have higher numbers of typical and dominant species compared to contiguous forest. Beta diversity (based on species incidence) and the exponential of Shannon’s diversity index was higher in fragmented forest, indicating higher spatial species turnover than in contiguous forest samples. Lower mean values of the Chao-Jaccard index in fragmented forest compared to contiguous forest also indicated a lower probability of shared species across fragments. The high species richness of contiguous forest showed that an earlier single logging event had not caused biodiversity to be degraded leaving mostly generalist species. Fragmentation and further low-level utilisation by local farmers has also not caused acute degradation. Post-logging regrowth forest fragments present a window of opportunity for conservation that may disappear in a few years as edge effects become more apparent. For the conservation of trees in forests in south-east Asia generally, our findings also suggest that while conservation of remaining primary forest may be preferable, the conservation value of post-logging regrowth forests can also be high.

Appendices

Appendix 1: Background to estimating metrics of diversity

Metrics of species diversity have been extensively discussed in the recent literature, (e.g. see Jost 2006, 2007; Marcon et al. 2014; Chao et al. 2010; Magurran and McGill (2011); Chiu and Chao 2014). In particular, Jost et al. (2011) and Gotelli and Chao (2013) provided a comprehensive description of how diversity metrics are calculated. We present a brief description of the key metrics used in this research.

Species richness

Species richness is the total number of different species in an assemblage and is thus a key ecological attribute of an assemblage. However, unless the entire assemblage is inventoried, observed species richness is always an underestimate of true species richness, because increasing the number of samples increases the likelihood of adding new rare species. Species richness does not take into account levels of species abundance; thus rare and common species are not distinguished.

Within an assemblage, the observed numbers of species depend upon the area sampled and the number of individuals sampled. For sample-based data, species richness is often depicted in a species accumulation curve which shows the cumulative number of individuals or sampling units on the x axis and cumulative species richness on the y axis). Typical species accumulation curves rise rapidly at first, as common species are encountered and then rise more slowly as rare species are eventually inventoried, the curve eventually becoming asymptotic. The curves are often calculated using rarefaction, a resampling and interpolation technique, in which individuals or sampling units are selected at random from the entire sampling assemblage until all individuals and species have all been accumulated. The resampling process is repeated many times and mean species richness and other diversity indices are calculated for each level of accumulation. This results in a smoothed species accumulation curve which may be extrapolated. Extrapolation should not be extended more than double the reference sample’s size because variance may become very large (Colwell et al. 2012; Gotelli and Chao 2013). Alternatively, species accumulation curves of two (or more) datasets of different size, may be compared by scaling back the cumulative number of individuals in the larger dataset to the same number of individuals or sampling units of the smaller dataset.

Species diversity indices

Species diversity is a function of species richness and the relative abundance (evenness) of species in an assemblage (Maurer and McGill 2011). Two popular diversity indices are Shannon’s entropy (H) and the Gini-Simpson (D) diversity indices, calculated as:

$$H = - \sum\nolimits_{{i = 1}}^{s} {p_{i} (\ln p_{i} )\,{\text{and}}} \,D = 1/\sum\nolimits_{{i = 1}}^{s} {p_{i}^{2} }$$

where ‘s’ is the number of species in the assemblage and ‘p _i ’ is the population frequency of the ith species.

Shannon entropy quantifies the uncertainty of the species identity of a randomly chosen individual in the assemblage and the Gini–Simpson index measures the probability that two randomly chosen individuals (selected with replacement) belong to two different species. In their ‘raw’ form, these diversity indices calculate entropy i.e. the uncertainty of the species identity of a sample, not diversity itself and they are non-linear with respect to increasing diversity (Jost 2006; Chao et al. 2010). In certain circumstances, the indices can be misleading as they do not provide a measure based on numbers of species (e.g. see Chao et al. 2010). Hence, following the work of MacArthur (1965) and Hill (1973) these raw diversity metrics are often converted into ‘Hill’ numbers, which provide a measure of the equivalent number of equally abundant species.

Hill numbers (^qD) are calculated according to their order, ‘q’ which determines their sensitivity to species relative abundance. When q = 0, ⁰D is equal to species richness, when q = 1, the Hill number ¹D is the exponential form of Shannon entropy which weights species according to their frequency and can be approximately expressed as the number of typical species in the assemblage. When q = 2, the Hill number ²D is the inverse of the Gini-Simpson entropy and is heavily weighted towards very common or dominant species in the assemblage.

Decomposition into alpha (α), beta (β) and gamma (γ) diversity components

Spatially, tropical forests typically have high rates of species turnover, i.e. compositional similarity is low (Arroyo-Rodriguez et al. 2013). Hence, ecologists often wish to compare the diversity of a sub-assemblage to the diversity of the overall assemblage. If weighted appropriately, sample data may be pooled and diversity may be calculated as an overall assemblage diversity (gamma, γ), within or sub-assemblage diversity (alpha, α), and between-group (beta, β) diversity.

Recent debate concerning the most appropriate way to partition gamma diversity into alpha and beta diversity has resulted in consensus that when considering diversity in terms of Hill numbers, the relationship between gamma, alpha and beta diversity is multiplicative, rather than additive i.e., gamma = beta × alpha or D_γ = D_β × D_α (see Jost 2007; Jost et al. 2010; Marcon et al. 2014). Beta diversity is then calculated as the ratio of gamma to mean alpha diversity and may be interpreted as the effective number of distinct communities in an assemblage (Jost 2007). In general, high beta diversity implies that alpha is much less than gamma diversity and that species turnover between the sub-assemblage and the assemblage is high.

Alpha, gamma and hence beta diversity for q = 0 (species richness) and q = 1 (exponential form of Shannon entropy) may be calculated using equations presented by Jost et al. (2011), i.e.

$$^{\text{q}} {\text{D}}_{\upalpha } = \left( {\frac{1}{N}\sum\nolimits_{i = 1}^{s} p_{i1}^{q} + \frac{1}{N}\sum\nolimits_{i = 1}^{s} p_{i2}^{q}+ \cdots + \frac{1}{N}\sum\nolimits_{i = 1}^{s} p_{iN}^{q} } \right)^{{1/\left( {\left. {1 - q} \right)} \right.}}$$

$$^{\text{q}} {\text{D}}_{\upgamma } = \left\{ {\sum\nolimits_{i = 1}^{s} {\left[ {\frac{1}{N}\left( {p_{i1} + p_{i2} + \cdots + p_{iN} } \right)} \right]}^{q} } \right\}^{{1/\left( {1 - q} \right)}}$$

$$^{\text{q}}{\text{D}}_{\upbeta } = {^{\text{q}}}{\text{D}}_{\upgamma }/^{\text{q}} {\text{D}}_{\upalpha }$$

Biotic similarity

Whereas beta diversity describes the turnover of species from a sub-assemblage to the overall assemblage, indices of biotic similarity enable comparison of species shared by two assemblages. The comparison may be between two different assemblages or between a sub-assemblage and the combined assemblage. For this purpose the Jaccard index compares the number of shared species in an assemblage (i.e. a sample) to the total number of species in the combined assemblages. The Chao-Jaccard index considers the probability that two individuals (one each drawn from both assemblages) belong to any shared species, not necessarily the same shared species. The index has a maximum value of one when all species are shared and a minimum value of zero when no species are shared (Jost et al. 2011; Gotelli and Chao 2013).

Appendix 2

See Table 4.

Table 4 Common name, scientific name, guild (pioneer, secondary or climax species), inventory frequency, conservation status and commercial value of 8959 native trees inventoried