«Claudia Gray, Cambridge University, UK Evelien Jongepier, University of Groningen, The Netherlands Abstract The arthropod fauna of Madagascar is ...»
Spider abundance on the other hand showed significant responses to both disturbance (GLM, F=4.36, df=3,55, p=0.008) and its interaction with logging history (GLM, logging history: F=2.90, df=1,55, p=0.094; interaction: F=3.17, df=3,55, p=0.031). Post-hoc analysis shows that in the nongrid forest sites with less intensive logging history (FNG_low), spiders were more abundant than in non-grid forest site with more intensive logging history (FNG_high) (TukeyHSD, FNG_lowFNG_high: p=0.056). In addition, spiders were more abundant in the non-grid forest sites with less intensive logging history, compared to big paths in both low and high intensity logging areas (B_high and B_low respectively).
(TukeyHSB FNG_low-B_high: p=0.431; FNG_low-B_low:
Fig. 4 Number of ant morphospecies obtained at varying levels of disturbance. (mean ± SE; GLM, F=0.83, df=3,59, p=0.482B).
Ant species richness and distribution The number of ant morphospecies did not differ between the different levels of disturbance (Fig. 5a;
GLM, F=0.83, df=3,59, p=0.482) nor did logging history or its interaction with disturbance affect ant species richness (GLM, logging history: F=0.12, df=1,58, p=0.727; interaction: F=0.23, df=3,55, p=0.878).
The effect of disturbance and logging history on ant presence was evaluated for the three most abundant ant species; Aphaenogaster swammerdami, Monomorium sakalava and Pheidole spp. A.
swammerdami did not show a response to disturbance (Fig. 4a.; GLM, F=0.50, df=3,57, p=0.68), logging history (GLM, F=1.14, df=1,60, p=0.291) or their interaction (GLM, F=2.54, df=3,54, p=0.066).
M. sakalava did show a significant response to disturbance (GLM, F=5.93, df=3,59, p=0.001), where it was significantly more often present at the small and big paths compared to the forest in the grids (Fig 4b.; TukeyHSD, FG-S: p=0.005, FG-B: p=0.005). Logging history (GLM, F=0.80, df=1,62, p=0.375) and its interaction with disturbance did not affect M. sakalava presence in the community sample (GLM, F=0.21, df=1,55, p=0.892).
Pheidole spp. also showed a significant response to disturbance (GLM, F=4.90, df=3,59, p=0.004) although, in contrast to M. sakalava, it was significantly more often present in the forest in the grids compared to the big paths (Fig. 5c.; TukeyHSD, p=0.002). Pheidole spp. did not respond to logging history (GLM, F=0.13, df=1,58, p=0.721) or its interaction with disturbance (GLM, F=1.09, df=3,5, p=0.359).
Fig. 5 Pitfall trap contents for the 3 most abundant ant species for the four levels of disturbance. Log transformed abundance of A) Aphenogaster swammerdami (mean ± SE; GLM, F=0.50, df=3.58, p=0.680). B) Monomorium sakalava (mean ± SE; GLM, F=5.93, df=3.59, p=0.001) and C) Pheidole spp. (mean ± SE; GLM, F=4.90, df=3.59, p=0.004).
DISCUSSIONArthropod community diversity Arthropod diversity, as measured at the order level, decreases with an increase in human disturbance. This could be explained by a reduction in available habitat types. The reduction in plant cover with increasing disturbance will remove a range of ecological niches (e.g. habitat diversity or range of food resources for herbivores). Alternative niches are likely to be available in the more disturbed areas (e.g. open space for ground nesting hymenoptera), but overall the quantity of ecological specialisations will be lower, equating to a reduction in diversity.
We did not find a significant relationship between order richness and level of disturbance, which probably occurs because orders are such broad taxonomic groupings. Disturbance specialists are likely to occur in all orders, and so each order will be represented by species at all levels of disturbance. However, our data do suggest that there may be decrease in order richness with increasing disturbance. Given our limited sample size (we put out between 13 and 20 traps at sites of each level of disturbance) we suspect that increasing the number of replicates would be necessary to better establish the presence or absence of this relationship.
Our study indicates that human disturbance does not affect the absolute number of organisms within a limited forest area. This could be because the total number of organisms within a sampling area remains constant despite changes in niche availability and community composition. Simply measuring total number of individuals is not a sensitive measure of differences in community composition, as it assumes all animals are the same. Our decision to count all members of each ant morphospecies as members of the same colony, and count each colony as one individual also removes any variation due to differences in sheer number of ants.
As only the diversity measure (Shannon index) revealed an effect of disturbance, we think it is important to note that neither a simple abundance measure, nor a measure of richness at the order level should is sufficient to characterise an invertebrate community and its response to disturbance.
Abundance of specific taxonomic groups We assessed the affect of disturbance on cockroach abundance, as we expected this species to be adversely affected by the decrease in available leaf litter and damp conditions. However, we did not find any predicable variation in cockroach abundance, which may be due to the attraction of the water in the traps counteracting the less favourable conditions of the paths. Alternatively, the relatively high mobility of the cockroaches may mean that the paths do not form barriers; they are able to traverse the paths before experiencing any negative effects.
Disturbance only has an adverse affect on spider abundance in the less intensively logged forest, which could be explained by the intolerance of the spiders to the conditions on the paths. It is likely that the spider species collected in our traps were roaming, ground hunting species that forage along the forest floor. If the abundance of prey species on the paths was reduced by a lack of vegetation cover, these spiders would be less likely to forage in the more disturbed habitats. With respect to the non-grid forest sites, the higher spider abundance in the less intensively logged forest areas suggests that ecological effects have persisted for at least 18 years after the termination of logging activity.
The habitat quality of the non-grid forest sites in the intensively logged areas may still be lower than that elsewhere. This finding contrasts the opinion of Sorg et al., (2003) that there is well established regeneration of logged areas within 20 years of logging. This discrepancy highlights the importance of using a range of indicators to complement any vegetative assessment.
Out of the range of subgroups we identified, there were insufficient data on distribution to allow further statistical analysis. With an increase in sampling effort, it would be interesting to analysis the effect of disturbance on other groups such as non-desiccant resistant species (e.g. woodlice), detritivores (e.g. millipedes).
Ant species richness and distribution There is no relationship between the number of morphospecies found at each site and disturbance level, which may be due to the mobility and high sclerotization of the ant species. The paths are unlikely to form barriers to ants, since they can cross the paths quickly and do not suffer adverse effects of the drier conditions. However, it may also be the case that the ant community composition varies with human disturbance. As we only counted the number of morphospecies present in each trap, our results do not reveal whether the identity of the species present varies with disturbance level.
Further analysis on the three most abundant species, Aphaenogaster swammerdami, Monomorium sakalava and Pheidole spp, reveals that the ant community composition does change with the level of human disturbance. Monomorium sakalava was more abundant at higher levels of disturbance, which implies that this species is a disturbance specialist. The open areas of the paths may facilitate the foraging of this species and allow it to out-compete other ant species. Conversely, our results show that Pheidole spp. are more abundant at lower levels of disturbance. Thirdly, our results show that Aphaenogaster swammerdami does not vary in abundance with changes in disturbance. Clearly ant species vary in the extent to which they are affected by human disturbance. More information on the ecology of these species is needed to clarify the exact effects of disturbance and to distinguish the most important factors regulating species distribution (for example, whether disturbance alters food abundance or predation rate).
Conclusions Our results indicate that, at the scale of the path system within Kirindy Forest, an increase in human disturbance negatively impacts arthropod diversity. This has implications for the management of the reserve: the extent of the path system may need to be restricted in order to limit any negative impact on the ecology of the forest and its fauna. However, further study at the scale of the whole grid system is needed to assess the effect of disturbance on arthropod diversity at the level of the larger Kirindy community. Investigation into whether the path system actually increases habitat and arthropod diversity at a larger spatial scale would be valuable.
Analysis of the specific taxonomic groups shows that human disturbance affects some taxa but not others. Our results also indicate that the disturbance from the logging concession at Kirindy may have longer lasting effects than previously thought. To complement our study, and further extend our insight into the biological and ecological consequences of human disturbance, more information on a wider range of species is needed. We would recommend an emphasis on species that are not resistant to desiccation, or require a specific food resource or substrate. In addition, further characterising the changes in ant community composition would be very informative, especially given the ecological importance of this group.
It is clear that the study of arthropod communities can contribute to our understanding of the impact of human disturbance, and the recovery of logged or deforested areas. Given the shorter generation time of arthropods in comparison to larger vertebrate species, for example, we would expect arthropod communities to recover more rapidly, and therefore provide a more accurate indication of the current state of a forest region. It has been widely noted that insect species, and ants in particular, show (as yet unrealized) potential as a powerful tool for conservation planning (e.g.
Underwood & Fisher, 2006, New, 1999). The use of arthropod community assessment in analyzing forest quality is currently of particular significance for the Menabe region, where the recovery of the degraded corridor between Kirindy and the more northerly forest of Ambadira, is being initiated. Efficient and informative measures of forest regeneration will be important for monitoring protected areas. Our study shows that informative results can be achieved in a short time period and with a straightforward protocol. This reinforces the idea that arthropod assessments can provide a cost-efficient and simple insight into forest ecology, which is vital given the limited resources available to many conservation efforts, and an increasing involvement of local communities in the management of their local protected areas.
ACKNOWLEDGEMENTSWe would like to thank all members of the Tropical Biology Association staff, and the participants who helped us with fieldwork and processing samples. We would also like to thank Dimby Raharinjanahary and team, of the California Academy of Science, Madagascar, for assistance with identification of ant specimen.
REFERENCESAndriambelo L. H. (2005) Document de synthese des etudes de filieres de produits forestieres au sein du CFPF. Akon’ny Ala Bulletin du Department des Eaux et Forests de l’Essa 29:48-58.
Bohning-Gaese K. et al. (1996) Seed dispersal in the tree Commiphora guillaumini: a combination of ornithochory and myrmecochory in a dry tropical forest in western Madagascar. In J.P Ganzhorn and J. P. Sorg (ed.) Primate Report 46:305-310.
Fisher B. (2003) Formicidae, Ants. In S.M. Goodman and J. P. Benstead (ed.) The Natural History of Madagascar. University of Chicago Press, USA. p811-818.
Ingram et al. (2005). Tree structure and diversity in human-impacted littoral forests, Madagascar.
Environmental Management 35:779-798.
New (1999). Entomology and Nature Conservation. Eur. J. Entomology 96:11-17.