Canopy fogging was used to sample the diversity of bark and ambrosia beetles (Coleoptera, Curculionidae, Scolytinae) at two western Amazonian rainforest sites in Ecuador. Sampling was conducted by Dr Terry Erwin and assistants from 1994–2006 and yielded 1158 samples containing 2500 scolytine specimens representing more than 400 morphospecies. Here, we analyze a subset of these data representing two ecological groups: true bark beetles (52 morphospecies) and ambrosia beetles (69 morphospecies). A high percentage of these taxa occurred as singletons and doubletons and their species accumulation curves did not reach an asymptote. Diversity estimates placed the total scolytine species richness for this taxon subset present at the two sites between 260 and 323 species. The α-diversity was remarkably high at each site, while the apparently high β-diversity was an artifact of undersampling, as shown by a Monte Carlo resampling analysis. This study demonstrates the utility of canopy fogging for the discovery of new scolytine taxa and for approximate diversity assessment, but a substantially greater sampling effort would be needed for conclusive alpha as well as beta diversity estimates.

ambrosia beetles, bark beetles, diversity, neotropical, rainforest, Terry Erwin

First pioneered by Southwood (1961) in temperate forests and then later adapted for tropical research by numerous others (Lowman and Wittman 1996), canopy fogging methods use insecticide to collect arthropods from the upper architecture of the forest habitat. These methods have been used in surveying arthropod diversity, particularly in lowland rain forests and at several neotropical localities (Erwin 1982; Basset 2001; Rice 2015). These data in part were used to make Erwin’s (1982) bold estimate of 30–50 million arthropod species which was a stark contrast to previous estimates of 1.5–10 million (Erwin 1982). Although this provocative publication produced immediate criticism, it initiated focused scientific inquiry for an improved estimate of the number of worldwide arthropod species (e.g., May 1988; Adis 1990; Gaston 1991; Erwin 1991; Ødegaard 2000; Hamilton et al. 2010).

For a more than a decade, Erwin and colleagues sampled arthropod diversity in the Ecuadorian Amazon rainforest canopy at two lowland sites with identified trees, separated by 21 km of contiguous primary forest, using a standardized insecticidal fogging protocol (Erwin 1983a, 1983b; Pitman et al. 2001; Erwin et al. 2005). Sampling occurred during dry, wet, and transitional seasons to measure temporal turnover in species composition. Over nine million specimens were collected from these fogging events (Rice 2015) and many new taxa have been described from the specimens (e.g., Erwin 2010).

The Scolytinae (Coleoptera, Curculionidae) is comprised of approximately 257 genera containing 6000 species worldwide (Hulcr et al. 2015). Scolytine species feed sub-cortically on a wide variety of woody and herbaceous plants. The typical life cycle of scolytines consists of a brief dispersal flight period after adult emergence, followed by colonization of new hosts. Once a suitable host has been located, adults bore galleries into the host tissue, where eggs are laid, and larvae complete their development into the next generation of adults (Kirkendall et al. 2015).

Scolytines are divided into two main ecological groups: bark and ambrosia beetles. The degree to which scolytines specialize on specific hosts varies considerably depending on these ecological groups. Bark beetles bore into the phloem of trees, feed on tree tissues, and tend to have more specialized host preferences. Ambrosia beetles bore into the xylem, feed on species of symbiotic fungi, which grow along the walls of their galleries, and tend to have more generalized tree host preferences (Beaver 1979; Hulcr et al. 2007; Kirkendall et al. 2015). The majority of scolytine diversity occurs in tropical regions of the world, and many tropical species remain undescribed (Wood 2007; Hulcr et al. 2015; Smith et al. 2017). The Ecuadorian canopy fogging samples collected in Erwin’s surveys offer a rich source of new scolytine specimens because of observed scolytine species specialization to tree taxa, tree parts, and other microclimate factors (e.g., Hulcr et al. 2007; Kirkendall et al. 2015). In addition, only fewer than 200 species have been recorded for Ecuador (Wood 2007; Martínez et al. 2019; Jordal and Smith 2020), but the Ecuadorian species richness is likely underestimated given the continuous discovery of new species (e.g., Petrov and Flechtmann 2013; Petrov and Mandelshtam 2018; Smith and Cognato, in prep.).

Remarkably few studies have addressed the spatial and temporal turnover of scolytine species (β-diversity) (Deyrup and Atkinson 1987; Thunes 1998; Peltonen et al. 1998; Hulcr et al. 2008a, b). The majority of studies have focused on the composition of the scolytine fauna of temperate regions and examined the distributions of a few economically important species (Deyrup and Atkinson 1987; Jakus 1998; Peltonen et al. 1998; Oliver and Mannion 2001; Gaylord et al. 2006). Several studies in the tropics have attempted to determine the effects of seasonal changes in rainfall and temperature on the composition of scolytine communities (Beaver and Löyttyniemi 1991; Madoffe and Bakke 1995; Morales et al. 2000; Flechtmann et al. 2001; Hulcr et al. 2008a; Sandoval Rodríguez et al. 2017; Martinez et al. 2019; Sanguansub et al. 2020). Studies in Malaysian forests have examined the spatial distribution of scolytines across horizontal (Maeto et al. 1999; Chung 2004) and vertical gradients (Maeto and Fukuyama 2003; Simon et al. 2003). Recent studies in Thailand (Hulcr et al. 2008a; Sanguansub et al. 2020) and Papua New Guinea (Hulcr et al. 2008b) have used quantitative sampling to examine scolytine community composition and attempt to determine the proximate causes of the distribution of species in tropical habitats. Scolytine communities in lowland homogeneous Papua New Guinean rainforests have a low β-diversity, thus contradicting the expected trend of high β-diversity in the tropics. Novotny et al. (2007) similarly demonstrated low species turnover for insect communities when examined on a large scale across Papua New Guinea. In Thailand, two study sites separated by only 5 km were found to have significantly different scolytine species composition (Hulcr et al. 2008a), however the differences were attributed to differences in forest type, elevation, seasonality, mean annual temperature and humidity (Hulcr et al. 2008a; Sanguansub et al. 2020). Given the scenarios above, it is uncertain whether Amazonian forests will produce the same trend of low β-diversity seen in Papua New Guinea or high β-diversity, as was observed between different Thai forests.

In tropical ecosystems, an increase in host specificity combined with an increase in plant diversity is often used to explain high levels of species diversity. However, bark and ambrosia beetles show a reverse trend with at least some groups exhibiting lower host specificity in tropical regions. This is largely due to the greater abundance of ambrosial feeding scolytines which, as discussed above, tend to be relative host plant generalists (Beaver 1979), with high symbiont specificity. Nevertheless, most scolytine groups have higher species diversity in tropical regions than they do in temperate ones (Beaver 1979; Deyrup and Atkinson 1987; Hulcr et al. 2015). It is therefore expected that canopy fogging will yield a high number of scolytine species, but that the β-diversity of scolytines across the Amazonian canopy may not necessarily be high.

In this study, we use a subset of scolytine specimens representing both the bark beetle and ambrosia beetle ecological groups to assess the diversity of scolytines at two western Amazonian forest study sites. We use these data to assess the value of canopy fogging as a source of scolytine specimens, determine the spatial turnover of scolytine species, and use the data collected from these fogging stations to predict the scolytine species richness at the two sites.

A total of 1158 canopy fogging bulk samples were analyzed from the Ecuadorian Amazon study transects; 965 from Onkone Gare and 293 from the Tiputini (Table 1). Scolytines were found in 60% of the Onkone Gare samples, 75% of the Tiputini samples, and 69% of the total samples from both sites. These samples contained a total of 2500 scolytines, representing more than 400 morphospecies including a large number of rare species occurring as singletons and doubletons. The subset of taxa examined here represented 688 individuals and 121 species (Tables 2, 3). The bark beetles (Bothrosternini, Phloeosinini, Phloeotribini, and Phrixosomini) totaled 183 specimens representing 9 genera and 52 morphospecies (Tables 2, 4). The ambrosia beetles (Xyleborini, and Premnobiina) totaled 504 specimens representing nine genera and 70 species (Tables 3, 4).

Species accumulation curves for both sites combined and for each site individually did not reach an asymptote (Fig. 1). The steady increase of the curves indicated a continued accumulation of rare species (singletons and doubletons). Estimates of species richness were very similar between abundance-based and incidence-based statistics (Table 1; Fig. 1). For the two sites combined, the abundance-based ACE and Chao1 estimated a total species richness of 301 and 308 species, respectively. The incidence-based ICE and Chao2 gave marginally higher estimates. Abundance-based and incidence-based estimates of species richness for the Onkone Gare overlapped slightly with 275–309 species. Abundance-based and incidence-based (Chao1) statistics arrived at similar estimates and the incidence-based statistics, ICE and Chao2 estimated a slightly greater species diversity for the Tiputini site. The second-order Jackknife estimated a species richness of 260 species for both study sites (Table 1; Fig. 1). For Onkone Gare, the second-order Jackknife estimated 217 species and for Tiputini it estimated 95 species.

Figure 1. 

A species accumulation curves for Onkone Gare B species accumulation curves for Tiputini C species richness estimators for Onkone Gare D species richness estimators for Tiputini E species accumulation curves for both sites combined F species richness estimators for both sites combined.

Simple Complementarity Indices suggested that the composition of the scolytine fauna was markedly different for the two study sites. Onkone Gare and Tiputini had a CI = 0.81 for the total analyzed subset of scolytine tribes (Table 1). Complementarity Indices for the tribes ranged from 0.73 to 1.00 (although a CI=1 for Premnobiina is meaningless because only one specimen was collected) (Table 5). In the samples, at least half of all the species occurring at either site was unique to that site: 76% of species at Onkone Gare and 50% or species at Tiputini (Table 1).

However, the correction for biases of incomplete sampling of the fauna using the Chao-Sørensen abundance-based estimator with the rarefied data estimated a faunal similarity of 0.79. Similarly, the Monte Carlo analysis returned a probability of 0.374 [median modeled similarity L’ = 0.77 (lower and upper 2.5% quantiles = 0.634 and 0.836)]. Both the Chao-Sørensen and the Monte Carlo analyses indicate that the apparent difference between the two sites was due to stochastic sampling error and the difference does not appear to be biologically significant.

The goals of this study were to assess the scolytine species richness, the use of canopy fogging to enhance the discovery of new taxa, and to estimate the faunal turnover (β-diversity) between a short distance in Amazonian rain forest. Species records for Ecuador range from 50 recorded in a monograph (Wood 2007) to 85 collecting at one site over a year (Martinez et al. 2019) and 248 have been recorded for Peru (Smith et al. 2017). Our entire collection of scolytines from the fogging represents 400 morphospecies out of 2500 specimens. This species richness is proportionally 30x greater than the species richness recorded for a western Ecuadorian forest (85 species out of ~18,000 specimens). Given the estimates of species richness suggest nearly twice the number of species in only these field sites (Tables 2, 3) and the diversity and long stability of Ecuadorian forest habitats (McKenna and Farrell 2006), it is likely the actual scolytine diversity exceeds 400 species.

Taxonomic study of these 400 morphospecies has yielded species descriptions of 21 Scolytodes species (84% of the known Ecuadorian fauna) (Jordal and Smith 2020), seven Camptocerus species (39% of the known Ecuadorian fauna) (Smith and Cognato 2010), and 15 Coptoborus (45% of the known Ecuadorian fauna) (Smith and Cognato, in prep.). In addition, faunal discoveries include descriptions of new genera with striking morphologies: Akrobothrus Dole & Cognato, 2007 and Callibora Cognato, 2018. These discoveries suggest that canopy fogging is not only a useful method for the collection of known scolytine fauna, but it is also a mean to access a poorly explored fauna.

Our analysis of the partial scolytine samples fogged from the Ecuadorian Amazonian canopy indicates a species-rich fauna but the estimate of a high β-diversity depended on the method’s sensitivity to undersampling. Specifically, the simple measure of complementarity suggested substantial faunal difference between the sites, but the Chao-Sørensen and Monte Carlo resampling showed that the differences were statistically inconclusive.

Our results indicate that even a large-scale, long-term sampling effort, such as this one, did not provide a reliable estimate of scolytine diversity for the two western Amazonian sites. Despite over 1100 individual fogging events (representing 14 fogging expeditions), which collected 688 individual beetles representing 121 species (used in this analysis), the species accumulation curves did not reach an asymptote (Fig. 1). Likewise, the accumulation of rare species in the form of singletons and doubletons did not decline. These phenomena were observed from the individual data collected for each study transect and for the collective species data for the two sites.

Despite this inconclusive result, we speculate on the scolytine diversity in the canopy. Given the relatively short distance (21 km) between Onkone Gare and Tiputini the real β-diversity is likely low because of the sites’ relatively proximity and very similar environment. The Onkone Gare and Tiputini sites differ in the occurrence of tree species but share the same tree families (Erwin et al. 2005). Bark and ambrosia beetle’s tree host specificity usually occurs at the family level (Beaver 1979) thus predicting low β-diversity between these sites. Conversely, previous canopy fogging studies have found significant differences in the beetle species composition of different forest types. In Manaus, Brazil, 83% of beetle species in canopy fogging samples were found in only one type of forest (Erwin 1983a). Of the four forest types characterized and studied by Erwin, “mixed-water” forests were found to be the most species rich. However, “terra firme” (non-floodplain) forests had the second highest number of species and the highest number of restricted species. Both sites sampled in this study were “terra firme” forest. Erwin also found that smaller insect species (1 mm class, as defined for his study) comprise the majority of insects in “terra firme” forests. This offers another predictor of scolytine diversity in this habitat, as most scolytine species are within the 1–3 mm range. If scolytines follow the same general patterns found for all Amazonian insects, we can predict that further sampling would uncover an even richer scolytine fauna with possibly many species not found in other forest types. The high percentage of singletons and doubletons at each site supports this prediction. Conversely, increased sampling may yield more specimens of rare species which would decrease β-diversity.

Large-scale sampling of tropical habitats offers a rich source of previously unknown species, but as this study demonstrates, even the most ambitious sampling schemes may not be enough to adequately assess the true species richness of hyper-diverse groups. Here we have uncovered a level of scolytine α-diversity that increases the known fauna of Ecuador nearly five-fold, only based on sampling a single Amazonian habitat. Future studies will need even more extensive sampling protocols to avoid erroneous conclusions and over-estimates of β-diversity based on stochastic sampling (Hulcr et al. 2007; Lewinsohn and Roslin 2008).

The authors are grateful to Dr Terry Erwin for his collaboration and generous sharing of samples that made this work possible. Throughout his distinguished career Dr Erwin was known for his warm and enthusiastic demeanor and his mentorship of students from around the world. SAD was one such graduate student and considers working beside Terry canopy fogging in the Ecuadorian Amazon to be one of the richest experiences of her career. Dr Erwin lives on in his substantial body of work and in the careers of the many scientists he mentored. We would like to thank Sarah Smith (Michigan State University) for her assistance and support with this research. Specimen collection funded by: Ecuambiente Consulting Group, Ecuador; Casey Fund, Department of Entomology, NMNH; NMNH Lowland Amazon Project. This research was supported by NSF-PEET grant (DEB-0328920) to Anthony I. Cognato (Michigan State University).