Evaluation of the importance of various processes determining community structure and function is an important topic in ecology. Unlike just a decade or so ago, few studies today question whether or not community assembly is strictly random, recognizing the role of both stochastic and deterministic processes . This change can be attributed to accumulating data on organisation in experimental and natural communities, and new perspectives gained from the fields of evolutionary population ecology and phylogenetics ( e.g., [3, 4]). It is worth noting that empirical tests for much of this theory have been attempted relatively recently in experimental microcosms research in particular, yielding valuable insights into the role of processes in driving long-term community dynamics .
This newfound view of community ecology is an excitingly realistic one, and has the potential to make valuable contributions to conservation biology as well [6, 7]. However, though studies on long term dynamics of communities are obviously important, they are extremely difficult to implement. A vast majority of field studies are restricted to exploring correlates and predictors of species diversity and other emergent community properties. This is in part due to the problems associated with studying complex natural communities. But to a great extent, difficulties also arise because community ecology studies have traditionally been skewed towards relatively long-lived vertebrate groups, or are restricted to short study periods due to logistical constraints, especially in the tropics [8, 9]. Although these studies yield valuable information, they are carried out at timescales that at best, give insight into short-term processes, providing limited information about community organization, persistence, or assembly.
Circumventing this problem is obviously very difficult. One possible approach is to study communities along gradients of habitat succession using space-for-time substitution (SFT) to obtain chronosequential communities . Thus, instead of studying changes in a single community over time, successional habitats of known ages that can be arranged on a temporal gradient are compared. This method can reveal changes in community structure, environmental predictors of these changes, and provide estimates of the rate of community change [11–14]. Although many studies have examined recovery of faunal communities with tropical forest regeneration, a vast majority have been restricted to one or two vertebrate (birds, small mammals) and invertebrate (ants, beetles) groups . For example, 19 of the 33 studies reviewed by Dunn , were on vertebrates, out of which only two were on amphibians and/or reptiles. Considering that amphibians and reptiles are ectothermic and have life history traits different from mammals and birds , more data on these taxonomic groups is important to test the generality of conclusions about effects of tropical habitat alteration on fauna.
This study takes an SFT approach to compare changes in frog and lizard community structure in two contrasting habitat succession gradients: (a) 1-yr jhum fallows giving way to mature forest, and (b) 1-yr jhum fallows planted over with teak, leading to monoculture stands. Slash-and-burn or shifting cultivation (jhum) agriculture involves clearing and burning of forest patches, so the original rainforest communities are effectively obliterated, and succession involves recovery of communities from scratch. The following questions were addressed in this study:
1. How much does frog and lizard community succession differ between the two gradients of habitat recovery?
2. Does composition of the entire community change in synchrony, or does the recovery pattern differ between subcommunities such as frogs vs. lizards and guilds?
3. What aspects of habitat change influence frog and lizard community recovery, and if habitat parameters are linked to niche axes, do they predict changes in guild composition?
4. Do successional changes in guilds also show trends in phylogenetic structure? This last question is expected to yield interesting insights into possible evolutionary mechanisms underlying changes in community composition , but has not explicitly been addressed in previous work on faunal recovery during tropical forest regeneration (cf. , and references therein).
In this paper, a chronosere is defined as a habitat that has recovered from perturbation for a known length of time, and can be assigned a place in the SFT. An assemblage is the set of all species of a taxonomic group in a landscape of interest. Ecological groups (EGs) are species' subsets of the assemblage with similar niche characteristics. Communities comprise species of the assemblage which share a habitat stratum (i.e., chronosere) in the landscape. Guilds are members of the EGs that actually coexist in the same chronosere i.e., belong to the same community, and are thus likely to have ecological and evolutionary interactions (cf. ).
To draw inferences about what aspects of habitat change determine sequential communities, habitat and frog-lizard community data were analysed hierarchically. As a first step, species richness and turnover of frog and lizard communities along habitat recovery gradients was summarised, and the entire assemblage classified into ecological groups (EGs) based on niche similarities. Guilds identified from this classification were then examined for phylogenetic structure. Using factor analysis, orthogonal combinations of variables that described biotic and abiotic aspects of habitat transition were extracted. We then tested for correspondence between these composite variables and composition of frogs and lizard communities and guilds. Based upon the relationships between different habitat factors and frog and lizard communities, variables were interpreted as composite adaptive zones, and we tested whether they predicted successional changes at different levels of community organization.