Comparison of sampling efficiency for two widely used sample methods in detecting ant diversity in a high-elevation montane forest mosaic indicate the sample plot methods are more efficient and productive than pitfall trapping for collecting ant diversity. With data representing three sampling seasons each spanning two months, pitfall traps collected 20% fewer species than a single effort of sample plot sampling (sample plot = 25 species; pitfall trap = 21 species). Using rarefaction analysis, we found the sample plot method estimated a significantly higher site species richness and diversity than pitfall traps. To collect a similar richness using pitfall traps would have required a near doubling of the sampling effort. These findings are consistent with Gotelli et al. (2011) who identified standardised hand-sampling methods as the most efficient method for ant surveys. Our findings are also consistent with recent studies comparing pitfall trapping and quadrat methods for sampling ants carried out across rainforest habitats (Mbenoun et al. 2021) and along a gradients of increasing vegetation disturbance (Fotso Kuate et al. 2015) in equatorial Africa, as well as a comparison of pitfall trapping to standardised hand collecting in pine forests in Spain (Abril & Gómez 2013).
The sample-plot method is an active species-collection method that employs a set of standardised hand-sampling techniques (i.e., quadrat sampling (Bestelmeyer et al. 2000), nest excavation (Romero & Jaffe 1989), and foliage beating (Harris, Collis & Magar 1972)) that target available microhabitats such as decaying woody material, litter, surface soil and foliage within a defined space and time (Xu 2002). Different sampling methods often yield a distinctive set of species, and many authors advocate using complementary methods to gain the greatest coverage of species, including improved representation of species that are rare, occupy specialised microhabitats, or are patchily distributed (Gotelli et al. 2011; Antoniazzi et al. 2020). Indeed, we found the sample plot method was slightly more proficient than pitfall trapping at collecting rare ants. Of the 18 ‘rare’ species observed, seven were distinctive to the sample plot method in comparison to the four species unique to pitfall trapping. However, surprisingly, the representative range of ecological niches observed (Gotelli et al. 2011; Mark & Guenard 2017) appeared remarkably undifferentiated between the two sampling methods (Supplementary Information 3). Both methods captured species that nest and forage across a range of microhabitats soil, litter and canopy.
Pitfall traps target epigaeic (surface and litter foraging) ants; they are selective and spatially constrained, a problem that becomes more pronounced with increasingly complex habitats (Luff 1975; Majer 1997; Gotelli & Colwell 2001; Gotelli et al. 2011). Still, numerous studies suggest pitfall traps are a preferred method for comparing ant assemblages among some habitats (Steiner et al. 2005; Oliveira et al. 2009; Hoffmann & Pettit 2022) or are an effective complement to other methods as part of an integrated survey (Bestelmeyer et al. 2000; de Souza et al. 2012). Several authors have highlighted the need to identify added information gained from additional sampling methods to reduce sampling effort, costs, and the risk of sampling redundancy (Tista & Fiedler 2011; de Souza et al. 2012). We found pitfall traps were generally cheap and easy to install; they were effective in estimating ant species alpha and beta richness, capturing unique species, and identifying a distinctive ant assemblage, as determined by the presence of non-shared species (PERMANOVA, p < 0.05). However, if we increased the sampling effort of the sampling plot method to equal the sampling completeness of the pitfall trapping (95%), we would expect an increase of shared species, as well as further rare, cryptic or specialist species (Gotelli et al. 2011; Antoniazzi et al. 2020). Only with additional survey could pitfall trapping be confirmed redundant to sample plots in this habitat.
The pitfall trap method was, on a sampling unit basis, more time-consuming and labour intensive than the sample plot method, since traps had to be cleared and maintained regularly and samples were generally slower to process in the lab due to sample bycatch, litterfall and detritus in the sample. In contrast, sampling plot methods were completed in the field in two hours and collections returned to the lab ready for identification. Our experience highlighted the efficiency of methods is not independent of the site location and environment; travel time to the traps was an important constituent to efficiency.
The sample plot method does have limitations. Sample plot surveys can only be conducted in the daytime (i.e., adequate lighting) and in fair weather (Li et al. 2015b). There may also be long-term effects from destructive sampling, an important consideration for any monitoring in sensitive environments (Bowie & Frampton 2004; Zaller et al. 2015). Also, consideration needs to be given to its susceptibility to differences in collector efficiency or expertise (Longino, Coddington & Colwell 2002; Sørensen, Coddington & Scharff 2009; Gotelli et al. 2011; Antoniazzi et al. 2020). In this study, ant samples were collected by a student with limited expertise and training, indicating the method has wide applications for monitoring high-elevation forests. For remote sites, particularly in areas with prominent local or indigenous groups presence, such as Lasha Mountain, a test will be if long-term biological monitoring can be facilitated by interested locals. Lasha Mountain has communities representing multiple ethnic nationalities that depend on the local biodiversity and the sustainable management of grazing land, soil, and water for their livelihoods. Establishing robust methods that can involve indigenous or local people can benefit research as well as improve situations for local people.
This study demonstrates the sample plot method, a standardised hand-sampling methodology, is more efficient and productive than pitfall trapping for ants in this high-elevation forest system. Where resources limitations restrict the scope of sampling methods, the sample plot method is the most appropriate choice, providing a higher sampling efficiency on a sample unit basis. However, our results indicate pitfall trapping provides additional information on ant assemblages not detected in sample plot methods, indicating that where possible, adding complementary methods, such as pitfall trapping, to the sampling protocol improves sensitivity and accuracy of long-term biological monitoring.