As mentioned above, O. humboldti sp. nov. is a brachypterous species. There are also another two brachypterous species of Onthophagus in Costa Rica: O. inediapterus Kohlmann & Solís, 2001 (Onthophagus dicranius Bates line) and O. micropterus Zunino & Halffter, 1981 (Onthopagus dicranius Bates line) (Kohlmann and Solís 2006). Other six brachypterous species are also known for Mexico: O. brachypterus Zunino & Halffter, 1997 (O. landolti Harold group); O. chilapensis Gasca-Álvarez, Zunino & Deloya, 2018 (O. chevrolati Harold group); O. gilli Delgado & Howden, 2000 (O. chevrolati Harold group); O. inflaticollis Bates, 1886–1889 (O. chevrolati Harold group); O. pedester Howden & Génier, 2004 (O. landolti Harold group); and O. zapotecus Zunino & Halffter, 1988 (O. landolti Harold group) (Kohlmann and Solís 2006; Gasca-Álvarez 2018). All the aforementioned species live in areas of old geological emergence: Cordillera de Talamanca and Sierra Madre del Sur (Kohlmann and Solís 2006). Outside the American continent there is only one brachypterous Onthophagus species known from Australia, living in vine scrub in arid areas, O. apterus Matthews, 1972 (Matthews 1972). All the American species have in common that they inhabit humid montane forests, a habitat considered as stable by most ecologists. It is interesting to note that Onthophagus brachyptery in Costa Rica is confined to the Onthophagus dicranius line, whereas in Mexico it is confined to the O. landolti and O. chevrolati species lines. This would suggest a line independent morphological convergence to similar ecological/historical conditions.

In relation to wing reduction, Zunino and Halffter (1988) described the most extreme case then known in Onthophagini with the example of O. zapotecus, where the wing does not show any trace of venation and the wing length to body length ratio has a value of 0.156. On the other hand, O. micropterus presents a ratio of 0.205 (Zunino and Halffter 1981), whereas O. humboldti sp. nov. has a ratio of 0.107. This ratio represents at present the most extreme case known so far of wing reduction in onthophagine scarab beetles. The wing does not show any trace of wing venation. The elytra are not fused together, but are strongly interlocked. This species shows also narrowed elytral humeri, as well as shortened elytra as had been already observed by Darlington (1936) for carabid beetles and Scholtz (2000) for scarab beetles. Contrary to the observation made by Scholtz (2000) and Scholtz et al. (2009) that flightless scarabs have reduced eyes with a smooth margin, O. humboldti has no such condition; however, as indicated by Scholtz et al. (2009), this species has a rounded body shape.

Accepted wisdom has proposed that in Scarabaeoidea the evolution of flightlessness is related to temperate highland forests in the tropics; arid environments, such as deserts; temperate forests at low latitudes in the southern hemisphere; islands; termite nests; and cold regions (Zunino and Halffter 1988; Génier 2000; Scholtz 2000). In montane environments, where Onthophagus, and other brachypterous genera, such as Ateuchus, Canthidium, and Cryptocanthon live in Costa Rica (Fig. 6), flight is apparently non-essential. According to Darwin´s classical explanation (Darwin 1859), known as “Darwin´s factor” (Darlington 1971), the presence of strong mountain winds could drag flying individuals towards unfavourable habitats for their survival (Zunino and Halffter 1981, 1988). This last explanation is strongly contested by Roff (1990), because it does not correctly take into account scale issues.

Figure 6. 

Known distribution of brachypterous Scarabaeinae in Costa Rica.

Scholtz et al. (2009) propose that flightlessness increases in scarab beetles with altitude in temperate forests in the tropics (which are considered to be a stable habitat), being rare in lowland tropical forests. A similar process has been observed in carabids, where brachypterous species also predominate in montane areas (Darlington 1971); as well with passalid beetles in tropical humid montane forests (Mac Vean and Schuster 1981). This seems to be the case for the present study, where all brachypterous species are montane, no species having been found in the lowlands so far. This would support Roff’s (1990) original hypothesis that flightlessness increases with altitude.

Wagner and Liebherr (1992) present an analysis of flightlessness in insects, indicating that around 10 % of temperate Coleoptera show this characteristic. Based on our current tally, 184 Scarabaeinae taxa have been listed for Costa Rica; of these, seven species are brachypterous (Fig. 6), thus representing 3.8 % of flightless scarab beetles. The low brachyptery percentage found in Costa Rica would support the hypothesis that insect and scarab-beetle flightlessness increases with latitude (Roff 1990; Scholtz et al. 2009).

At present, a very much accepted hypothesis that tries to explain the origin of this phenomenon is the one given by Lawrence et al. (1991), where they propose that wing brachyptery may have a selection value in insects that have adopted a sedentary, cryptic or a parasitic way of life, or that live in mountain, island or high latitude habitats. Another explanation for this situation, and the one we follow and expand here, is the one proposed by Kavanaugh (1985). Kavanaugh suggests that macroptery represents the ancestral (plesiotypic) condition among beetles and that brachyptery has evolved independently many times among Coleoptera and other pterygote insects. Such a widespread phenomenon requires explanation. Brachyptery is a major factor contributing to restricted distributions and it usually does not progress to a stage where the wing rudiment is actually absent (Kavanaugh 1985). However, Frolov (pers. comm., 2019) has observed that in Orphninae (Scarabaeidae) two genera are completely apterous. Brachyptery has also been suggested as a factor that promotes speciation (Mayr 1963; Hackman 1964).

It is clear that the distribution of brachypterous forms is not random, certain patterns are repeated. In North and Central America no brachypterous Scarabaeinae are known from the lowlands, alpine regions, or from rodent nests. They are only known from the mountains in Costa Rica, especially the Talamanca range and the Sierra Madre del Sur in Mexico. In all cases these flightless species live in humid montane forests, spanning an altitudinal distribution that goes from 1100 to 3000 m. Flightlessness in scarabaeines is confined so far in Mesoamerica to small-sized genera, like Onthophagus (9), Canthidium (4), Cryptocanthon (2), and Ateuchus (1), in descending order of known species number. On the other hand, brachyptery seems to be confined in South American Scarabaeinae to eight medium-sized species of the genus Dichotomius, out of 170 described taxa, where this condition has evolved independently, at least four times in this genus (Nunes and Vaz-de-Mello 2013, 2016). However, this genus does not show a clear attachment to a particular ecological environment, because the different brachypterous species have been collected ranging from tree sand dune habitats to riparian forests (Nunes and Vaz-de-Mello 2016). According to models developed by Roff (1994), a dominant brachyptery can spread if the cost of being macropterous and habitat stability are important. In other words, regarding the last point, habitat stability is a key factor favoring the loss of flight (Roff 1990, 1994).

If it is generally accepted that the occurrence of brachyptery reflects long-term stability of habitats (Roff 1990, 1994), then one could propose that the occurrence of such forms in particular geographical areas is an indication of long-term stability for these regions as well, especially if this pattern is repeated by different taxa in the same area. This train of thought has been used for recognizing areas of long-term occupation, such as glacial refugia in carabids by Lindroth (1979) in Scandinavia and by Kavanaugh (1985) in Canada.

If one would plot the geographical ranges of the fore mentioned brachypterous scarab-beetles on a map, coincident occurrence of such taxa is apparent. The pattern that emerges is one in which one particular area stands out, the Cordillera de Talamanca (Fig. 6). All four known brachypterous genera and six species are concentrated in this range. All other known areas boast one to two genera and species and would probably represent minor centers or areas of subsequent colonization. These data would strongly suggest that the Cordillera de Talamanca has served as a center for both long-term survival and differentiation in this group of beetles, acting as a stable area. The Cordillera de Talamanca is the highest mountainous area in Costa Rica, and the highest range in Central America, reaching 3820 m altitude, and thus acting as a possible built-in buffer for residents against sudden and dramatic climate change. If the climate were to change rapidly and drastically, montane species could be able to move a short distance up or down in elevation, tracking their required microclimate, whereas lowland organisms would have to move far greater distances north or south in order to achieve the same result.

Lachniet and Seltzer (2002) and Vázquez-Selem and Lachniet (2017) analyzed the effects of the last glaciation on the Cordillera de Talamanca and estimated that the last local glacial maximum (LLGM) for the Cerro Chirripó occurred at 21–18 ka with a depression of the equilibrium line altitude (ELA) or snow line of ~1500 m in relation to the modern regional ELA of 4900–5100 m, thus representing a LLGM temperature reduction on the order of ~8–9 °C. However, a more recent research reevaluation by Potter et al. (2019) estimates the age of the LLGM at 25–23 ka for the Cerro Chirripó with a reconstructed ELA depression of 1317–1536 m and an associated cooling of ~7–9 °C. Palinological studies have indicated that during the last glacier interval (50–15.6 ka) with temperatures 7–8 °C cooler than today the treeless páramo extended down to 2100 m altitude, whereas it is distributed from 3300 to 3819 m at present (Islebe and Hooghiemstra 1997; Horn 2007). At the end of the last deglaciation (15.6–13 ka), the upper forest limit rose to 2700–2800 m (3100–3 300 m present-day forest limit of subalpine tropical rain/elfin cloud forest), indicating a temperature increase of up to 4.6 °C (Islebe and Hooghiemstra 1997; Horn 2007). Subsequently, the upper forest limit dropped 300–400 m from 13.1 to 11.2 ka indicating a temperature decline of 2–3 °C (Horn 2007). From 12.3 to 11.2 ka the glaciers retreated above 3500 m and the subalpine tropical rain forest was gradually replaced by mountain rain forest as the forest limit and temperatures rose toward present-day values (Islebe and Hooghiemstra 1997; Horn 2007).

Figure 7 shows the present-day distribution of Onthophagus humboldti sp. nov. and O. micropterus and lines indicate proposed localities depressed by 1500 m (14 km in straight line) generated by the last glacial maximum, ~25–23 ka, in the Talamanca Cordillera. All mountain systems are depicted 150 m lower than present day height, considering a generalized continental uplift of 1 mm/year and an estimated sea level descent, due to glaciation, of 120 m. Interestingly, the glacially depressed localities are not only to be found at the base of the mountain system but also within the Valle de El General (Valley of the General or The General’s Valley), surrounded and probably climatically protected by the embracement of this very long valley (Kohlmann et al. 2002), that possibly dampened the cooling effect of the glaciation. Solano Quintero and Villalobos Flores (2001) did a climatic-geographic regionalization of Costa Rica and they found this intermontane valley to differ from the rest of the southern Pacific area in relation to having very homogeneous lower rainfall values (3050 m total annual precipitation) and a longer dry period (three months).

Figure 7. 

Present day distribution of Onthophagus humboldti sp. nov. (blue dot) and O. micropterus (red dot) and lines indicating proposed localities (rhombi) depressed by 1500 m (14 km in straight line) generated by the last glacial maximum, ~25–23 ka, in the Cordillera de Talamanca. All mountain systems are 150 m lower than present day height and an estimated sea level descent of 120 m is depicted. Dotted black lines represent present-day sea levels.

We propose here that the Valle de El General might have acted as a refugium for the brachypterous species during the last glacial maximum (LGM). This valley would then resemble what has been called a cryptic refugium (Stewart and Lister 2001; Stewart et al. 2010), a refugium situated at different latitudes or longitudes than would normally be expected, often resembling climatic islands in which conditions differ favourably from the surrounding areas. According to the classification proposed by Stewart et al. (2010), the Valle de El General would fit the description of being a glacial southern refugium, representing an accepted low-latitude refugium for temperate species during a glacial phase. Stewart et al. (2010) also indicate that one characteristic of cryptic northern refugia is that they are sheltered in habitats located in deeply incised valleys that provide microclimates for temperate species, which is precisely the case under study here. Stewart et al. (2010) do not present any examples in their study of cryptic glacial southern refugia (only cryptic interglacial southern refugia), so that the example of the Valle de El General could represent the first one reported of its kind. Finally, because the Valle de El General represents a small area (~1850 km²), and in accordance with island biogeography tenets (MacArthur and Wilson 1967) that indicate that because of low population size and a limited food base in small areas, this could explain that mostly small-sized (brachypterous and non-brachypterous) scarab species and not big-sized ones, could have had patches of suitable habitats to live in during glacial periods which seems to be the case for the present study, all reported brachypterous scarab beetle species in Costa Rica are small-sized. Coincidentally, six of the known seven brachypterous scarab beetle species are also distributed around the Valle de El General. This area is occupied by montane forests, representing incredibly biodiverse and more species-rich environments than lowland tropical forests in Costa Rica (Kohlmann et al. 2007, 2010; Kohlmann 2011). This high concentration of brachypterous species in these montane forests contradicts Scholtz (2000) and Scholtz et al. (2009) proposal that relatively species-poor, environmentally stable habitats, lacking complex biotic interactions, like temperate forests on tropical mountains, contribute toward flightlessness. Costa Rican montane forests are decidedly species-rich and thus most probably also having complex biotic interactions.

The Valle de El General must have been formed by the uplift of the Cordillera de Talamanca and the Fila Costeña (Costeña range), a process that began about 7 million years ago and accelerated during the last 4 million years, triggered by the arrival of the Coco submarine range (an extinct volcanic range, also known as Cocos Ridge) in the Pacific and by the compression of the microplate of Panama in the Caribbean. The Arenal depression (tectonic graben) must have originated less than 2 million years ago, although there is no better estimate. While the Valle Central (Central Valley) is of a more recent formation and its age goes back to less than half a million years (Alvarado et al. 2007; Alvarado and Gans 2012; Alvarado and Cárdenes 2016). The absence of endemic brachypterous species in the Valle Central can be a product of never having being present or that they were exterminated due to the cataclysmic volcanism that occurred several times in the last 800 ka, the last large one 322 ka ago, with the formation of pyroclastic flows (pyroclastic density currents), which destroyed everything in hundreds of square kilometers (burning ash clouds with a temperature > 600 °C, pyroclastic deposits with thickness between 10 and > 80 m); the last major event of this kind occurred some 322 ka, covered an extension of at least 820 km² in the Valle Central and neighboring areas, as well as a distance of up to 80 km from the eruptive source (Pérez et al. 2006; Alvarado and Gans 2012). The cataclysmic and paroxysmal volcanism of this type (pyroclastic density currents or ignimbrites) has been absent in the last few million years both in the Valle de El General and in the Arenal tectonic depression, where, coincidentally, the only other known endemic brachypterous species (Cryptocanthon lindemanae) that does not live around the Valle de El General Area is present.

Areas of endemism (AE) are fundamental areas in the analyses of biogeography and are defined as areas of non-random distributional congruence among taxa, whose biogeographical history probably shared common factors such as geological, ecological, or evolutionary processes (Harold and Mooi 1994; Morrone 2008). Important questions regarding the AE are its distribution and defining the major ecological/evolutionary factors (climatic/elevational gradients, geographic isolation, topographical heterogeneity) that affect the distribution of these areas. A quick glance at the distribution map of the new species and their closely related taxa (Fig. 5), as well as the map published by Kohlmann et al. (2007), showing the distribution of the endemic Costa Rican scarab beetle species (Fig. 8), shows a clear non-random distribution of the AE.

Figure 8. 

Map showing Costa Rican scarabaeine endemic numbers as distributed by life-zones (taken from Kohlmann et al. 2007).

The mapping of the AE’s (Fig. 8) clearly identified mountain ranges as important centres of endemism. Endemism seems to be higher in the tropical mountains than in the tropical lowlands of Costa Rica. Similar results have been found in other areas of the world. Noroozi et al. (2018) have identified in the mountains of Iran using Asteraceae that patterns of endemic richness and areas of endemism identify mountain ranges as main centers of endemism, likely due to high environmental heterogeneity and strong geographic isolation among and within mountain ranges. Noroozi et al. (2018) also found that endemic richness and degree of endemicity are positively related to topographic complexity and elevational range. The proportion of endemic taxa at a certain altitude (percent endemism) is not congruent with the proportion of total surface area at this elevation, but is higher in mountain ranges. While the distribution of endemic richness (i.e., number of endemic taxa) along an altitudinal gradient was hump-shaped peaking at mid-elevations, the percentage of endemism gradually increased with elevation. Sosa and Loera (2017) have shown that endemism of Mexican monocot geophytes was highest in montane regions (Mexican Trans-Volcanic Belt) and Millar et al. (2017) demonstrated that angiosperm endemism was highest in the mountains of New Zealand’s South Island. To very similar results came Buira et al. (2017) in relation to the vascular flora of the Iberian Peninsula. McDonald and Cowling (1995) found that endemic mountain fynbos flora (represented by shrubs with short-distance seed dispersal) are over-represented in high altitude wet habitats, where almost twice the number occur than expected on the basis of area occupied by these habitats. Finally, Scherson et al. (2014) found that coastal ranges in southern Chile have acted as glacial refugia for ancient flora during the Quaternary, showing a higher endemicity than expected by chance.

Noroozi et al. (2018) ask themselves the question, if endemics have higher numbers in the mountains than in the lowlands. This seems to be certainly the case in tropical Costa Rica for scarab-beetles (Scarabaeinae, Dynastinae) and monocot plants (Araceae, Arecaceae, Bromeliaceae) (Kohlmann et al. 2007, 2010; Kohlmann 2011), and as shown by other references, also around the world. High environmental heterogeneity and strong geographic isolation among and within mountain ranges seems to be a very plausible explanation. The effect of mountain systems as possible glacial refugia seems to also play an important role. So, as indicated in the previous section, because of simple geographical distance, mountains allow for small linear displacements that still maintain the same ecology, thus allowing a cenocron to stay concentrated in a small area; whereas, lowland species have to travel greater linear distances, hence presenting a much more extended distribution of endemic species.

An analysis of the cytochrome c-oxidase I (COI) for both new species was undertaken. The Bar Code Index Number (BIN) for each species is: Onthophagus humboldti, BOLD: ABA7524 and Uroxys bonplandi, BOLD: ABA3722. Results are clearly distinct, whereas the value registered for the Onthophagus pair gives an average Kimura–2–parameter [K2P] value of 6.35 % with a maximum distance of 10.6 %, the amount of DNA difference for the Uroxys pair is of only 3.3 %. Uroxys results stay in line with other similar ones calculated for a group of Caribbean-Pacific scarab-beetle sister-species pairs separated by the Cordillera de Talamanca which started its emergence around 7 million years ago; Phanaeus pyrois Bates, 1887 and P. malyi Arnaud, 2002 ([K2P]= 3.8 %) and Phanaeus beltianus Bates, 1887 and P. changdiazi Kohlmann and Solís 2001 ([K2P] = 3.0%), which show basically a similar amount of mitochondrial DNA difference (Solís and Kohlmann 2012). These average values are similar to the ones that Johns and Avise (1998) found (K2P difference) of 3.5 % in 47 pairs of bird sister species and divergences greater than 2% in 98% of vertebrate sister species.

However, the value shown for the Onthophagus sister-pair looks higher ([K2P] = 6.35 %). This is interesting if we consider that this pair is formed by flightless species and the geographical nearness between them, 52 km in a straight line (Fig. 5). It would seem therefore that a limited dispersal capacity tends to favour differentiation as Mayr had already suggested (1963). Another possible explanation is that this species pair represents an old clade, as is known that nucleotide substitutions accumulate through time (older clades tend to accumulate more substitutions).

This last explanation is concordant with the previous results shown in the brachyptery section, where it is suggested that the Cordillera de Talamanca has been an area of long-term stability, thus allowing the continuous and uninterrupted presence of clades. In general, areas with a preponderance of brachypterous populations represent areas of older populations (Lindroth 1979; Kavanaugh 1985). Volcanologically, the Cordillera de Talamanca and the Fila Costeña have been very stable during the last 9 and 3.5 million years, respectively. During these last 5 million years the Valle de El General starts to acquire its present geomorphological configuration due to the formation of the two aforementioned mountain systems (Alvarado and Gans 2012; Alvarado and Cárdenes 2016).

In a very interesting study of phylogenetics and biogeography of the genus Onthophagus inferred from mitochondrial genomes, Breeschoten et al. (2016) found that all New World species of Onthophagus form a monophyletic group. This study found an origin of the Onthophagini from an Afrotropical ancestral stock in the Eocene and a subsequent spread to the Americas via the oriental region at about 20–24 Ma. New World Onthophagini started diversifying around the Miocene (20 Ma).

Among the American Onthophagini that Breeschoten et al. (2016) studied, O. clypeatus Blanchard, 1843 from the tropical forests of Colombia to Bolivia (0–1000 m) and O. rhinolophus Harold, 1869 from tropical forests (0–1500 m) of Mexico to Guatemala, were included. They estimated a phylogenetic branching process between both of them at ca. 3 Ma (95 % HPD: 1.6–4.4 Ma). At that time and according to paleogeographic reconstructions of Alvarado and Cárdenes (2016), Costa Rica had already emerged. The Cordillera de Talamanca began to rise fast after 3.5 Ma with an estimated uplift rate of 1 mm/year (Miyamura 1975) and creating an uplifted area where montane species could start to evolve.

Onthophagus clypeatus Blanchard, 1846 and O. rhinolophus Harold, 1869, are part of the clypeatus species group and considered to be closely related to the dicranius species group (Zunino and Halffter 1997; Kohlmann and Solís 2001). These species are therefore related to O. humboldti sp. nov. and O. micropterus, and are also distributed in tropical mountain forests. This closeness might suggest a similar phylogenetic branching process of the Costa Rican species in the Cordillera de Talamanca around 3 Ma. One could suggest that after this relatively old speciation process the gradual evolution of flightlessness took place in a subsequent, much more recent time, where the orographic scenario was more akin to the present-day one. This scenario would be then in accordance with the aforementioned estimations of the rise of the Cordillera de Talamanca, as well as with the proposal of Sanmartín et al. (2001), where they found that in the western Nearctic animal diversification and animal species richness increased in the later part of the Neogene and early Quaternary (2.56 Ma), whereas they consider unlikely elevated speciation rates during the Pleistocene. However, Kohlmann et al. (2018) studying scarab beetles of the genus Geotrupes (Coleoptera: Geotrupidae), have found evidence that the last glacial maxima in the mountains of Oaxaca, Mexico (18–15 ka, but in Costa Rica around 24 ka), could explain very recent speciation processes, where speciation can be described as incipient, generating very closely related taxa with small taxonomic differences.

We present in Figure 9 a cytochrome c-oxidase I (COI–5P) mitochondrial DNA sequence-based Bold Taxon ID tree of the species group to which O. humboldti sp. nov. belongs. The Taxon ID Tree is a functionality of the BOLD System, that allows for the generation of dendrograms from sequencing using the neighbor-joining algorithm.

Figure 9. 

A cytochrome c-oxidase I (COI–5P) mitochondrial DNA sequence-based BOLD taxon ID tree of the nearest species of O. humboldti sp. nov.

BOLD uses neighbor-joining trees which group sequences together by the number of amino acid or nucleotide differences. The arrangement of the specimens in the tree is based on sequence similarities, with the sequences that are most similar placed closer together on the tree, and with the branch length indicating the degree of similarity. The percentage of similarity between sequences can be measured against the legend (line), where the longer the branch the more disparity between the sequences. It is often expected that specimens of the same species have more similar sequences and cluster closer together than specimens from different species.

This figure is part of a more general analysis done for the genus Onthophagus in Costa Rica. All four depicted taxa are mountain species distributed in the mountains of Costa Rica and Panama. O. humboldti sp. nov. seems to be a closely related species of O. dorsipilulus Howden & Gill, 1993, a species distributed in the Cordillera de Talamanca and the Cordillera de Chiriquí in Costa Rica and Panama from 1400 to 1800 m altitude and would seem to be its ecological equivalent at a slightly lower altitudinal belt. On the other hand, O. micropterus, also distributed in the Cordillera de Talamanca, seems to be the sister species of O. quetzalis Howden & Gill, 1993, a taxon distributed in the neighboring Cordillera de Tilarán and Guanacaste. The DNA mitochondrial analysis neatly recovers the formation of this cluster belonging to the O. dicranius species-group (Kohlmann and Solís 2001). The nearest species cluster to this last group is also included (Fig. 9), where all seven taxa belong to the O. clypeatus species-group (Zunino and Halffter 1997), as defined by Kohlmann and Solís (2001). The formation of these two well-defined, but closely related branches seems to support the proposal forwarded by Kohlmann and Solís (2001) that they are effectively two different groupings and not a single one, as proposed by Zunino and Halffter (1997).

Finally, Onthophagus having around 2200 valid species (Schoolmeesters 2016) and being a hyperdiverse and ecologically plastic genus, does not conform with Roff’s (1990) proposal that lineage size should favor the evolution of flightlessness. At present, only ten brachypterous species of Onthophagus are known: six in North America, three in Central America, and one in Australia. As a comparison, the genus Dichotomius Hope, 1838 has around 170 described species with eight brachypterous taxa (Nunes and Vaz-de-Mello 2013, 2016), all of them living in South America.