Across taxa, a great majority of elevational gradient studies in the Himalayan region returned a unimodal altitudinal pattern of species richness. This was typical for gradients covering a broad elevational range, whereas monotonously increasing or decreasing species richness applied for those covering short elevational ranges. In addition, monotonous increases were associated with low mean and maximum gradients’ elevations, whereas monotonous decreases were associated with those with high mean elevations. These observations support our original hypotheses that unimodal response of species richness to elevation prevail in Himalayan biota, and those studies reporting decreasing or increasing species richness with altitude covered subsets of the elevational range of the mountains. We agree with the observation of Nogués-Bravo et al.25 that studying only upper parts of elevational gradients results in apparently decreasing pattern species richness patterns, and with Kessler et al.35, who insisted on the necessity to cover entire elevational gradients in a global study of ferns.
Our Himalayan analysis supports the prevalence of unimodality for a broad range of taxa in a major mountain range. For the most frequently studied taxa, the species richness peaks were situated in 2000–3000 meters, i.e., in the altitudinal belt of deciduous broadleaf forests (southern Himalayan slopes oriented towards Oriental tropics) or coniferous forests (NE and N slopes, oriented towards Palearctic temperate zones). The high diversity of birds, insects, and many other groups in South Himalayan Mountain forests is a well-established fact36,37. Only for plants, some of the diversity peaks (n = 8) reached the subalpine to alpine vegetation (>≈ 3000 m). The two highest-elevation richness peaks were reported by Klimes38 and Bhattarai et al.39, who nevertheless covered rather short and primarily alpine gradients (elevational ranges 4180–5970 and 2800–4400 m a.s.l., respectively). The other authors reporting plant diversity peaks in alpine elevations covered substantially longer gradients, spanning >3000 altitudinal meters31–41. These observations suggest that at least in some parts of the mountains, diversity peaks of Himalayan plants may be located above those of animals. This may reflect the radiation of some plant groups at Himalayan (sub)alpine altitudes42,43, or high alpha-diversity of some plant groups in high altitude environments, resulting into highly situated plant diversity peaks. Alternatively, the apparently higher-elevated diversity peaks reported for plants may be due to considerably easier sampling of plants, which are immobile and non-cryptic, compared to difficulties with sampling mobile and/or cryptic animals in harsh terrains of high elevations.
Unimodal species richness patterns44,45 was also reported from other major mountain ranges, both temperate and tropical, the former temperate including, e.g., plants in Norway46, land snails in Europe47, mammals in the American Rocky Mountains48, or beetles and moths in Korea49; and the latter tropical including, e.g., leaf litter invertebrates in Panama9, ferns in Costa Rica50, moths in tropical mountains world-wide11, or mammals in the Philippines51.
Reversing the argument that sampling long elevational gradients results, almost invariably, in unimodal elevational species richness patterns, leads to the conjecture that the uniformly increasing or decreasing richness patterns are results of incomplete vertical sampling. If so, the monotonously decreasing or increasing gradients do not require additional biological explanation. Still, groups whose distribution towards elevational extremes is truncated by their biology likely represent exceptions to the rule. Towards the upper extreme, these most likely include trees, limited by physical limits to their growth52; fish, limited in high altitudes by absence of sufficiently large water bodies53; and perhaps other ectothermic vertebrates. Groups truncated towards lower elevational limits might include weakly competitive organisms, such as lichens or orchids.
Although the unimodal patterns fit the geometry-derived null hypothesis of mid-domain effect54, they deserve to be further analyzed regarding underlying physiological, ecological, or evolutionary mechanisms, which may vary among taxa, but also regions of the world. Hu et al.55 showed that biotas of various functional or climatic guilds and their turnover may effectively, together with climatic data, explain the unimodal pattern shape. Furthermore, high altitude species overlapping in mid-altitudes with lowland species could have originated by autochthonous high-altitude radiations56; dispersed to the mountains from higher latitudes, perhaps during periods of cooler climate (cf.57,58); or derived from lowland biotas by endemic speciation59. In the Himalayas, the diversity of high altitudes is often of Palearctic/Holarctic origin, whereas lowland species are Oriental60.
Cross-taxon analyses aiming on deciphering the mechanisms behind the unimodal patterns are highly desirable, but the data at hand do not allow them at this moment. The necessary conditions would be complete species lists for the attitudinal points surveyed, together with abundances. Such data would allow relating life history traits of species inhabiting different altitudes to their phylogeny and abiotic conditions. Only a small fraction (n = 3) of the 64 papers considered here reported original data. Without species-level data, it is impossible to understand the peculiarities of composition of individual species communities along the gradients and to explain how the general unimodal pattern of species richness is built.