Populations and assemblages living on the edge: dung beetles responses to forests-pasture ecotones

Study area

The study was carried out in two localities with different temperate forest types: JF and POFs, in the state of Hidalgo, Mexico, in the MTZ (Fig. 1). In both localities, forest fragments are imbedded in pasture matrices where sheep, horses, and cattle graze. Mean annual precipitation in this region is ca. 2,000 mm, and the rainy season occurs from June to October, with a temperate climate (Pavón & Meza, 2009).

The JF locality (21°1′1.17″N, 99°12′2.35″W) is dominated by Juniperus flaccida Schltdl., J. deppeana Steud., and Cupressus sp. Mean elevation is 1,376 m a.s.l. and mean temperature during dung beetle sampling (see below) was 23.11 °C (31.52 °C maximum; 17.90 °C minimum) in native vegetation, and 24.74 °C (34.01 °C maximum; 17.52 °C minimum) at pastures. The POF locality (20°56′36.65″N, 99°12′29.23″W) is dominated by Pinus teocote Schiede ex Schltdl. & Cham., Pinus montezumae Lamb., Pinus cembroides Zucc., Quercus mexicana Bonpl. and Q. crassifolia, Bonpl. Mean elevation is 1,850 m a.s.l. and average temperature during the dung beetle sampling was 17.97 °C (27.12 °C maximum; 14.85 °C minimum) in native vegetation and 18.01 °C (25.17 °C maximum; 14.47 °C minimum) at pastures.

Sampling design

We selected two sampling sites at each locality. The two sites in JF were separated 480 m, and the two sites in POF were 530 m apart. Each site has a forest remnant (mean size of fragments: eight ha) surrounded by pastures. At each site, we located three linear transects, starting 90 m inside the native forest and ending 90 m into the pastureland, and each transect had seven baited pitfall traps (see below) separated 30 m (Fig. 1). Distances of traps from the ecotone toward the forest interior are indicated with negative values (−30, −60, −90 m), distances of traps from the ecotone toward the pasture with positive values (30, 60, 90 m), and the ecotone trap is considered as zero distance, set under the tree line (Fig. 1). A recent study on the dung beetle assemblage and species with the highest numbers of recaptured individuals in the Atlantic forest, Brazil, suggest a minimum distance of 100 m between pairs of traps to minimize interference between baited pitfall traps (Da-Silva & Medina-Hernández, 2015). However, a mark–recapture experiment in Venezuela found that 95% of the recaptured beetles were attracted from within 26.20 m of the trap (Larsen & Forsyth, 2005). Therefore, given the environmental contrast in the adjacent habitats studies, we assume that the beetles at each trap are reliable representatives of the distance to the ecotone.

Pitfall traps were baited with a mixture (3:1) of sheep and horse dung (ca. 250 g), as this method has been successfully used in the region for sampling dung beetles, given that the native mammal fauna was best represented by herbivorous like white-tailed deer, but currently wild mammals are scarce and the main source of dung is livestock (Barragán et al., 2014). Each pitfall trap was a plastic container (18.5 cm diameter, one L capacity) buried in the soil, containing ethylene-glycol to break the surface tension and preserve the beetles. Fresh dung was supported on a wire grid at the top of the container. To increase sampling completeness, we carried out this procedure twice (July and August 2008) and for each trap we summed the captures to obtain the cumulative number of beetles. This is the middle of the rainy season, when insects are most active and thus samplings in these months can maximize capture success of dung beetles in the region (Ortega-Martínez, Moreno & Escobar, 2016). Traps were active for 48 h each time.

Therefore, our analyses are based on the beetles that are attracted to the traps for food at close distances to the ecotone, and we assume that their identity and abundance at each sampling distance to the edge represents their foraging patterns near the ecotone. This sampling method is usual in dung beetle studies, although beetles may fly toward the trap but prefer different conditions for nesting in the surroundings. So, with this sampling method we do not aim to assess reproductive populations, but the flying beetles that can arrive for food near to the forest edge.

Dung beetles were preserved in alcohol, and some specimens were pinned and identified to species using taxonomic keys (Delgado, Pérez & Blackaller, 2000; Morón, 2003; Vaz-De-Mello et al., 2011) and comparing with voucher specimens from a local reference collection and with the help of Fernando Escobar (Instituto de Ecología, A.C., Xalapa, México).

Data analysis

Population responses to the edge

We analyzed the response to the edge of 18 species that had more than 30 individuals captured (24 cases: 15 species in JF and nine species in POF, Table 1; details of model fitting are in Table S3, the variation of abundance among transects of each site is in Table S4). From these, three species had a neutral response to the edge, that is, without a clear preference for any habitat condition: Copris incertus Say, 1835 in both forests, E. magnus Laporte De Castelnau, 1840 in POF, and Onthophagus sp. in JF (Fig. 2A). Canthon humectus and Phanaeus adonis responded otherwise, with a unimodal distribution of abundance in POF (Fig. 2B), although both species were more abundant in the pastures of JF.

As predicted, most of the species avoided ecotones: 87% of the species in JF and 56% in POF (Table 2). On one hand, six populations were more abundant toward the forest interior, five of them in the JF (Table 1): O. igualensis Bates, 1887 (with a complete response within our gradient, Fig. 2C), O. incensus, Sysiphus mexicanus Harold, 1863 (both with an incomplete response), Deltochilum scabriusculum Bates, 1887 and O. knulli (with undetermined avoidance response, Fig. 2D). In the POF only Onthophagus sp. showed an incomplete edge avoidance response and higher abundance within the forest interior.

On the other hand, in 12 cases populations avoided the edge and showed higher abundance toward pastures. In the JF locality eight species were more abundant in the pasture, four of them with a complete response: Canthon (Boreocanthon) puncticollis LeConte, 1866, Canthon humectus, Canthon imitator Brown, 1946, and Digitonthophagus gazella Fabricius, 1787 (Fig. 2E), and four with an incomplete response within our sampled gradient (Canthon cyanellus LeConte, 1859, Dichotomius colonicus Say, 1835, Glaphyrocanthon sp., and Phanaeus adonis; Fig. 2F). Interestingly, in the POF locality four Onthophagus species avoided the edge and had higher abundance toward pastures: O. mexicanus, O. incensus, O. knulli, and O. gibsoni Howden & Génier 2004 (the first with a complete response, the second with an incomplete response, and the last two with undetermined responses).

Overall, the proportion of species showing different types of response to the edge effect was similar between localities (G_test = 2.84, d.f. = 2, P = 0.17, Table 2). Only eight species (40%) exhibited a complete response within the −90 to 90 m sampled transects, while for other 12 species the extent of edge effects was longer than this (40% with incomplete responses, 20% with undetermined responses), and these proportions remained similar in both forest types (G_test = 0.29, d.f. = 1, P = 0.63, Table 2).

Assemblage responses to the edge

Contrary to our prediction, at the assemblage level dung beetle species richness, diversity, FRic, and FEve consistently had a neutral response to the edge, showing no clear preference for any habitat condition in both forest types (Table 1; Table S3). Abundance showed a neutral response in POF, but an avoidance response (incomplete) in the JF, where a higher number of individuals was captured toward pastures. On the contrary, FDiv responded neutrally in JF but had an avoidance incomplete response in POF, with higher values inside forest fragments (Table 1).

As expected, compositional dissimilarity showed a clear peak near the edge when species abundances were taken into account (1-Morisita index) in POF, while the Jaccard dissimilarity showed a neutral response to edge effects in the two forest types (Table 1).

Populations’ responses

Only Copris incertus showed a consistent pattern in both JF and POF (a neutral response to the edge). This big tunneler species is native from México, Central, and South America, and is common in tropical pasturelands (Cruz & Huerta, 1998), thus in our study area its abundance is not affected by the conversion of forests to pastures and can be considered as a habitat generalist.

Other five species had different responses to edge effects in the two studied forests, probably related to environmental changes due to their difference in vegetation structure, elevation (ca. 500 m), temperature (ca. 5 °C), and probably other variables. Canthon humectus and Phanaeus adonis have unimodal responses in POF, and an avoidance response with higher abundance in pastures in the JF locality. The biogeographic distribution pattern of these Neotropical species corresponds to the Plateau, and they have been reported in open and semi-open places usually with livestock (Morón, 2003; Halffter et al., 2011). O. incensus and O. knulli avoided the edge with higher abundance within forest in the Juniperus locality, but higher abundance in the pastures of POF; while Onthophagus sp. had a neutral response in JF but avoided the edge with higher abundance in the forest interior of POF. These Onthophagus species are widespread in the area, and near to our study area Barragán et al. (2014) found them more abundant on native vegetation than in pasturelands in a tropical forest landscape, however, on a pine-oak landscape, the same species are more abundant on pastures than in native vegetation. Several environmental variables associated with elevation and vegetation (such as soil hardness, humidity, temperature, slope, and tree cover) may have important effects on the responses of these species to environmental change. These interesting population responses to the environment should be further studied, probably at wider scales. The other 12 studied species had more than 30 individuals in only one of the two localities, thus we were not able to assess the consistency of their response patterns to edge effects.

In the same way as the habitat generalist Copris incertus, Onthophagus sp. in JF and E. magnus in POF also showed a neutral response to edge effects. Interestingly, E. magnus was likewise collected in both forest and savannah habitats in a sharp ecotone between open savannah and tall evergreen forest in Bolivia (Spector & Ayzama, 2003). Thus, this species behaves as a habitat generalist in these distant regions.

In six cases species avoided the edge and had greater abundance inside forest fragments, showing their sensibility to habitat conversion. Among them, only O. igualensis had a complete response within 90 m from the edge toward forest interior, while for the other species models showed that the extent of edge effects on abundance is larger than the sampled area (incomplete or undetermined responses). Our results indicate that all these species are characteristics of native forests in our studied area, so their habitat specificity becomes them vulnerable to habitat conversion and their ecological services might be compromised. For example, Deltochilum scabriusculum is one of the largest species in the area and as a result of its biomass it removes great amounts of dung in these forest ecosystems (Ortega-Martínez, Moreno & Escobar, 2016). Thus, forest conversion into open habitats may have strong effects on dung beetle functional groups and ecological services such as dung removal and seed post-dispersal (Hosaka et al., 2014).

In newly transformed pasturelands, native forest species are replaced by heat-tolerant species that endure the stressful conditions created in open, light filled agroecosystems (Halffter & Arellano, 2002). Thus, the low complexity of open-grazed areas, and the high availability of dung resources, may promote dominance of the species associated to open conditions habitats (Halffter, Favila & Arellano, 1995; Montes-De-Oca, 2001). In our study area eleven species avoided the edge and had higher abundances in pastures than in forest remnants. Heliophilous and thermophilous species such as Canthon imitator and C. (B.) puncticollis in JF, or O. mexicanus and O. incensus in POF, have abundant populations in pastures. The large Dichotomius colonicus was not very abundant but also avoided the edge and preferred the pasture. Similarly, in a mountain landscape of the MTZ Dichotomius colonicus was likewise abundant in pastures, and is supposed to use these agroecosystems as corridors to move from tropical landscapes to higher altitudes (Arellano & Halffter, 2003).

The exotic Digitonthophagus gazella is also abundant in pastures, especially in JF. This Asian–African beetle was introduced to America ca. 1972 and is now widespread in tropical and subtropical pastures (Montes-De-Oca & Halffter, 1998; Noriega, Moreno & Otavo, 2011). In the MTZ this species has invaded grazing habitats in lowlands and middle-mountain ecosystems (Barragán et al., 2014), probably because these pastures resemble its native habitats.

Regarding the extent of edge effects on populations, in almost half (11) of the 24 studied cases, species showed neutral, unimodal, or complete avoidance responses within the 90 m sampled at both sides of ecotones (Table 1). This coincides with the results of Spector & Ayzama (2003) in a forest-border-savannah system, where nearly half (24) of the species that were scored for habitat specificity were found in only one habitat, even when sampling was carried out only 50 m apart from the ecotones. Similarly, in forest–sun-grown coffee ecotones of Colombia Villada-Bedoya et al. (2017) found that the extent of some dung beetle species responses to edge effects occurred within 210 m from the ecotones. Furthermore, for ground beetles (Carabidae) clear edge effects have been detected in as near as 20 m from the ecotone (Boetzl, Schneider & Krauss, 2016; Jung & Lee, 2016). These examples show that many beetle populations might be sensitive to edge effects even at short distances, although this is not a generalized pattern. For example, Ewers & Didham (2008) found that the abundances of 20% of common beetle species caught in New Zealand were affected by edges at scales >250 m, and one in eight common species had edge effects that appeared to penetrate as far as one km into habitat patches. Therefore, large-scale edge effects may threaten forest interior species by promoting their local extinction in fragmented landscapes, and properly designed studies should been carried out to assess these processes.

Intrinsic characteristics of species, such as dispersal ability or tolerance to desiccation, determine dung beetle responses to habitat change (Montes-De-Oca, 2001; Vulinec et al., 2008). Thus, each species responds to edge effects in its own particular way, resulting in species-dependent filtration by edges (Ingham & Samways, 1996; Gaublomme, Eggermont & Hendrickx, 2014), or in spillover effects, that is, movements of species across the habitat edge due to contrasting resource availability in the adjacent habitats (Boetzl, Schneider & Krauss, 2016; Prass et al., 2017). These processes should be further studied at a landscape scale, accounting for the effects of broad scale factors that may influence beetles, such as the level of contrast between the native habitat and the matrix (Prass et al., 2017), overall landscape structure (Jung & Lee, 2016), or human activities in fragmented landscapes. For example, livestock produces different types of dung in variable quantities, thus resource availability may be a major driver of dung beetle abundance. In a recent study Bourg et al. (2016) found that tropical forest substitution by grasslands modifies the patterns of food selection by coprophagous beetles and affects the functioning of ecosystems. Therefore, we could expect that trophic generalist species will have neutral responses or will show unimodal preferences near to the edge, while food specialists will show avoidance to the edge responses (with species feeding on native mammal dung, rotten fruits or carrion being more abundant in forest interior, and species feeding in cattle dung being more abundant in pastures), as specialists are more sensitive to habitat fragmentation (Gaublomme, Eggermont & Hendrickx, 2014).

Our results with baited-pitfall traps show patterns on the abundance of species that fly to the food source. Some species may arrive from the surroundings to the ecotone for foraging, but their captures in traps do not indicate that they maintain viable populations on this area. Therefore, a future research line should explore which dung beetle species really nest at the ecotone. For example, in our study JF site, couples of Canthon humectus arrive to cattle dung, form a brood-ball and roll it 0.28–9.65 m away from the source until they bury it in pastureland (Ortega-Martínez et al., 2014). However, other species may not have a breeding population in pastures and our records in pitfall traps could be of individuals in transit (vagrants). Thus, the real habitat indicator character of species (forest of pasture preference) should be studied in deep with evidence of reproductive behavior.

Assemblages’ responses

Despite differential responses of dung beetle populations to edge effects, at the assemblage level these effects are masked, as species richness, species diversity, FRic, and FEve did not vary along forest-pasture ecotones. Only total abundance had an avoidance response (incomplete) to the edge in the JF, where more individuals were captured in pastures.

This neutral response in assemblage parameters (or even a positive effect of habitat conversion on abundance) may depend strongly on the interaction of ecological factors and the biogeographical context (Davis et al., 2010). In the MTZ, other studies have also found that grazing areas have a positive effect on dung beetle abundance and diversity at high elevations and in dry environments, specifically in pine–oak forest and semiarid scrubland (Barragán et al., 2014) and JF (Ortega-Martínez, Moreno & Escobar, 2016) of the Sierra Madre Oriental province. Similarly, dung beetle abundance is higher in human-created pastures than in native savannah-like vegetation in the Cerrado ecosystem in Brazil, and the influence of the edge is evident only for abundance when landscape matrix is pastureland (Martello et al., 2016). Thus, the presence of cattle may have a positive effect on dung beetle abundance, probably due to higher resource availability (dung), but very intensive grazing may reduce their diversity (Hutton & Giller, 2003).

In contrast, in tropical forests habitat conversion to pastures does usually have a negative effect on dung beetle diversity (Nichols et al., 2007; Beiroz et al., 2017), and several empirical studies have shown that tropical native forests harbor more species and more abundance than pastures because dung beetles prefer shaded habitats than open areas where heat limits their behavior (Halffter & Arellano, 2002; Escobar, 2004; Vidaurre, Gonzales & Ledezma, 2008; Lee et al., 2009; Noriega et al., 2012). Besides higher temperatures, pastures face wind variations, which may be a limiting factor for some sensitive species (Halffter, 1991; Hanski & Cambefort, 1991; Basto-Estrella et al., 2012; Frank et al., 2017). However, edge effects not always have a clear influence on dung beetle assemblage parameters in tropical forests. For example, Spector & Ayzama (2003) found a strong effect on dung beetles species richness and biomass through forest-border-savanna, and Villada-Bedoya et al. (2017) also found higher richness and abundance in the forest when assessing forest-coffee ecotones, although species diversity was higher at the edge. In land use change studies dung beetle functional diversity is also impoverished in cattle pastures (Barragán et al., 2011; Gómez-Cifuentes et al., 2017). On the contrary, as shown in this study, in some cases the species richness of dung beetle assemblages is not sensitive to forest edge effects (Durães, Martins & Vaz-De-Mello, 2005; Feer, 2008).

In a similar way, the neutral response of FRic and FEve to edge effects may result from the balance between individuals and species lost and gained from one habitat to the adjacent. This balanced variation may result in forest and pasture assemblages that do not differ in their functional trait space (FRic), or in the regularity of abundance distribution in traits (FEve), regardless on the identity of species. The lack of change in the FEve of dung beetle assemblages also occurs when tropical forests are converted into cattle pastures (Barragán et al., 2011), thus FEve might not be related to environmental variables related to habitat change, but with the assembly mechanisms of ecological communities, such as niche filtering (Mouchet et al., 2010).

Functional divergence, which measures how abundance is distributed within the volume of functional trait space occupied by all the species, had higher values inside POF forest fragments. Therefore, high values within forests indicate that abundant species are distant from the center of gravity in the assemblage functional trait space, relative to rare species, as a result of dissimilarity in functional traits among species. Thus, the inclusion of functional diversity in our assemblage level of analysis brings about important insights into the mechanisms behind edge responses. Likewise, Magura (2017) found significant edge effects on ground beetle assemblages when functional, phylogenetic and functional-phylogenetic diversities were used across forest-grassland gradients. Indeed, an integral approach encompassing different dimensions of biological diversity is highly desirable. And, although identifying the influence of edge effects on particular functional traits or ecological processes regulated by dung beetles was beyond the scope of this study, we believe that edge effects on the functionality of ecosystems deserve additional research (Ziter, Bennett & Gonzalez, 2014). For example, it would be interesting to link the functional diversity of beetle assemblages to particular functional processes such as dung removal, seed dispersal and nutrient cycling not only in contrasting habitats, but along forest edges. Also, the relationships of functional diversity with specific environmental variables of ecotones remain unknown.

As expected, dung beetle compositional dissimilarity increased abruptly at the edge, fitting the unimodal function, at least in one forest. Certainly, dissimilarity in species composition changed without an apparent loss of species, as richness and functional diversity had a neutral response. Likewise, Ewers, Thorpe & Didham (2007) and Villada-Bedoya et al. (2017) proposed that the coexistence of species on the edge is more evident when diversity does not differ significantly between the adjacent components of the ecotone. Compositional dissimilarity was more evident when species abundances were taken into account, which corresponds with the results of abundance avoidance to the edge in most of the populations studied.