Thin layers of water running over rocky surfaces are characteristic of madicolous habitats, which harbor a peculiar Chironomidae community. However, information on the identity, distribution, and ecology of madicolous chironomids in the Neotropical region are still sparse. The main purpose of this research is to reveal and contribute to the ecology of madicolous Chironomidae species, especially regarding their altitudinal distribution in the Atlantic Forest. Sampling was performed using our own designed emergence traps deployed from 0 to 2700 m a.s.l. in 70 sites in three mountains in southeastern Brazil. Sixty taxa of chironomids were collected and identified, of which only 22 are known to science. Most of the species showed a wider distribution than previously known, both in terms of geographic and altitudinal ranges, while others showed significant association with particular altitudinal bands (as evidenced by the indicator species analysis). Atlantic Forest mountainous regions are known to harbor one of the richest fauna in the world and have been suffering from several types of environmental impacts, including climate change, which will especially affect taxa living in specialized habitats. The narrow range of tolerance to environmental conditions verified for mountain species, and the fact that many of them are rare and endemic, make the conservation efforts in these areas indispensable.

hygropetric habitats, mountains, non-biting midges, semi-aquatic habitats, tropical forest

Madicolous habitats are characterized by a thin layer of water that frequently flows over rocky surfaces, and for this reason they are also known as hygropetric habitats. The first to use the term “hygropetrischen” was Thienemann in 1909, when studying the biology of trichopterans from Central Europe. Throughout the twentieth century, some catalogues of madicolous fauna were done in North America (Sinclair and Marshall 1987) and Europe (Bertrand 1948, Vaillant 1956). More recently, most of the progress done on the study of madicolous organisms came out of taxonomic works (Sinclair 1988, Cranston 1998, Roque and Trivinho-Strixino 2004, Short 2009, Short et al. 2013, Bilton 2015, Pinho and Andersen 2015, Trivinho-Strixino and Shimabukuro 2017, Shimabukuro et al. 2017a, b, Pinho and Shimabukuro 2018), emphasizing the potential of this habitat in harboring a rich and endemic overlooked fauna. In South America, madicolous habitats have recently provided remarkable discoveries on the occurrence of insects, from new records (Roque and Trivinho-Strixino 2004, Short et al. 2013, Pinho and Andersen 2015) to several new species (Pepinelli et al. 2009, Silva et al. 2012, Trivinho-Strixino et al. 2012, Miller and Montano 2014, Shimabukuro et al. 2017a, b, Pinho and Shimabukuro 2018).

In natural ecosystems, madicolous insects can live in a wide range of habitats, such as shoreline of streams or in isolated overflowing groundwater. Additionally, when robust water bodies like streams and lakes are scarce, for example on mountaintops, madicolous biotopes can be the only source of permanent water allowing aquatic and semi-aquatic insects to establish themselves and survive, contributing to the maintenance of biodiversity in natural systems. The true madicolous inhabitants (eumadicoles) present morphological and physiological adaptations favoring their survival in such a specific environmental condition, as seen by the presence of strong locomotor appendages to hang on the rocky substrate in larval stages (Trivinho-Strixino et al. 2012), presence of strong hooks on the pupal abdomen and the production of silk by the larvae (Boothroyd 2005) or living inside portable cases to avoid water carrying (Fittkau and Reiss 1998).

The Chironomidae family is one of the most diverse within Diptera. Species numbers reach an estimated 20,000 (Coffman 1995), though only 6,000 approximately have been described. This remarkable evolutionary success allowed them to occur in all zoogeographic regions, including Antarctica, tolerating even the harshest environmental conditions (Sugg et al. 1983, Linevich 1971, Watanabe et al. 2006, Andersen et al. 2016a). Although the immature stages of known species show high dependence on water (Ferrington 2008), some are semi-aquatic or terrestrial, and researchers have recorded some in artificial madicolous systems (Cranston 1984, Boothroyd 2005, Hamerlík et al. 2010).

A high diversity of chironomids is expected to occur in natural madicolous habitats from tropical regions, but this biotope has so far been neglected in freshwater researches, making it difficult to have an estimate on the diversity of insects living in such habitats. Furthermore, concerning the taxonomy of chironomids, most of the descriptions are based exclusively on adult males, making it difficult to obtain the information on the habitats, behavior, and other ecological information related to aquatic stages.

Despite significant progress on Chironomidae research in the last decade (Trivinho-Strixino 2011, Mendes et al. 2007a), most registered species are still concentrated in Nearctic and Palearctic regions, which emphasize the urgent need for studies in Tropical regions that present potentially higher diversity. In this research, the first checklist is provided of madicolous Chironomidae from the Atlantic Forest, which is one of the richest hotspots in the world, and still the most affected by habitat loss (Myers et al. 2000). In addition, notes on distribution in the altitudinal gradients and other ecological features are included.

A total of 60 species, including 22 known species and 38 morphospecies, was recorded, as follows:

Notes on altitudinal distribution

A summarized list of the species, morphospecies and the genera of immature found, along with respective ecological and geographical information, is presented in supplementary material (Table 1). In this study, the chironomid community was predominantly composed of species belonging to the subfamily Orthocladiinae (35 spp.), followed by Chironominae (21 spp), Podonominae and Tanypodinae (2 spp each). Among the 60 species recorded, a higher percentage has been found at APASM (45%), of which 85% were exclusive from this locality. Further, 31% of the possible new species occurred above 2100 m a.s.l. Only five from the 60 species recorded were significant indicators of specific altitudes, they are: Urubicimbera sp. 1, Cricotopus sp. 4, Pseudochironomus ruah, Lauterborniella sp. 1 and Podonomus mina (Figure 3). Urubicimbera sp. 1, and P. mina, represented the highest sites in this study (> 2600 m a.s.l.); P. ruah was a significant indicator of the 2500 m–altitudinal–band; Lauterborniella sp. 1 and Cricotopus sp. 4 were significant indicators of 1100 and 200 m–altitudinal–band, respectively (Figure 3). Furthermore, these five species were all unknown to science previous to this investigation in madicolous habitats of the Atlantic Forest.

Figure 3. 

Indicator species of altitudinal range. Indicator values in black and significance value (p) at 0.05 level in gray obtained for each species and morphospecies found in the present study.

Regarding the 22 recognized species, 17 of them had spread the altitudinal distribution (Figure 4). Even those species that have previously been found in mountain regions, such as Podonomus pepinellii, Lipurometriocnemus biancae, Urubicimbera montana, Pseudosmittia catarinense and Limnophyes guarani, were recorded at higher altitudes in this study, and, except for P. catarinense, the altitudinal distribution increased more than 1000 m for each of these species (Figure 4).

Figure 4. 

Altitudinal record of each species found in madicolous habitats in the present study. Previous altitudinal records (from literature) in gray and altitudinal records from this research in black.

For those species that have so far been verified at low altitudes, such as Tanytarsus giovannii, Polypedilum solimoes, Onconeura oncovolsella, Corynoneura hermanni and Bryophaenocladius carus, the amplitude of the altitudinal distribution was even more remarkable, adding more than 1500 m to the altitudinal range in some cases. The only exception was Hudsonimyia caissara, that have been firstly reported at the sea level, and here it was found at 200 m a.s.l., slightly increasing the altitudinal range of the species (Figure 4). Paratanytarsus silentii, Tanytarsus digitatus, Onconeura japi, Corynoneura septadentata, Corynoneura unicapsulata and Caaporangombera intervales also had the altitudinal distribution extended, while the remaining five species were recorded within their typical altitudinal ranges (Figure 4).

Compositional changing in chironomid assemblages along altitudinal gradients have been verified by many researchers worldwide (McKie et al. 2005, Tejerina and Malizia 2012, Henriques-Oliveira and Nessimian 2010, Scheibler et al. 2014, Robinson et al. 2016, Matthews-Bird et al. 2016). In mountain ecosystems the occurrence of chironomid species can be influenced by environmental changes related to altitude variation, such as temperature and oxygen availability (Oliver 1971, Pinder 1986, Eggermont and Heiri 2012), dispersal capacity (Ashe et al. 1987), historical events (McKie et al. 2005, Allegrucci et al. 2006, Krosch et al. 2011) or other regional particularities (Körner 2007). Mountains are therefore important objects to biogeography studies, revealing rich communities and many endemic species (Lods-Crozet et al. 2001, Garcia and Suaréz 2007; Brundin 1966). In our study, the locality with higher altitudes (APASM) yielded a higher number of species, most of them are unknown to science and were exclusively found in this place, especially above 2,100 m a.s.l.. These evidences are essential in view of the conservation perspective, once the majority of the species found are possibly endemic.

A clear gap on the taxonomic knowledge of mountain fauna can be observed. This gap is likely due to the low accessibility of these areas, thereby hampering sampling strategies. Studies in mountain regions are urgently needed, especially when dealing with one of the most threatened biomes in the world (Ribeiro et al. 2009) whose geomorphological characteristics are so heterogeneous. Mountains have been suffering from several types of environmental impacts, but the most alarming today is climate change (Burke 2003, Catalan et al., 2017). Current forecasts suggest that rainfall will be less constant and temperature will raise 2.0 to 6.0 °C by 2100 (Garcia et al. 2014), strongly affecting the flora and fauna. The climate changes will especially affect those living in small water bodies with high exposure to environmental pressure. The narrow range of tolerance to environmental conditions verified for mountain species, and the fact that many of them are rare and endemic, make the conservation efforts in these areas indispensable.

The indicator's analysis evidenced that all species significantly associated with their respective altitudinal band were previously unknown to science. All of them, except for Cricotopus sp. 4, recorded from 70–200 m a.s.l. were found exclusively at mountaintops. Lauterborniella sp. 1 was recorded at the highest sites in PESM and the remaining taxa were recorded at the summit of APASM mountains. Regarding the geophysical characteristics of mountaintops (shape, size, insulation value), also known as Inselbergs (Porembski 2007); some species, especially those with limited dispersal capacity, are more likely to deal with speciation process and local extinctions (MacArthur and Wilson, 1967). The narrow altitudinal range expressed by these unknown species, make us believe that they should present a high endemicity degree.

Our new records extend the altitudinal range of 17 known species. Most species seem to tolerate a wide altitudinal range, such as Tanytarsus giovannii and Limnophyes gercinoi, while others presented a narrow range, such as Hudsonimyia caissara. The altitudinal range is related to the extent of the geographical distribution of each species; species that are widely distributed are expected to occur in a wider range of altitudes compared to those that have limited distribution (Stevens 1992). Brundin (1966), analyzing the distributional patterns of Podonominae in South America concluded that species found in Patagonian region could be recorded at the highest sites of tropical Andean mountains. Similarly, in this study, species that have previously been recorded further south such as Lipurometriocnemus biancae, Urubicimbera montana, Pseudosmittia catarinense, and Limnophyes guarani, were found at higher altitudes, and may be related to temperature requirements.

Madicolous habitats have never been formally studied in Brazilian mountainous regions, in contrast to other Atlantic Forest water bodies in which the Chironomidae fauna have already been extensively investigated (Henriques-Oliveira et al. 2003, Roque et al. 2007,Silveira et al. 2015). Taxonomists, and especially ecologists, have paid little attention to semi-aquatic and terrestrial Chironomidae, and therefore, madicolous species were completely overlooked. In our current study, a remarkable diversity of Chironomidae living in madicolous habitats was revealed, and most of the species (about 64%) were probably new before this project. However, from the 38-unknown species collected, five have recently been described: T. alaidae, T. alienus, T. angelae (Trivinho-Strixino and Shimabukuro 2017), P. mina (Shimabukuro et al. 2017b), and P. ruah (Shimabukuro et al. 2017a), increasing the number of madicolous chironomid species.

Despite the low knowledge on semi-aquatic forms, evidences from chironomids fossils preserved in amber reveals that terrestrial life-styles have been common since the late Eocene (about 40 million years ago) (Zelentsov et al. 2012), raising the importance of madicolous and other semi-aquatic habitats on the evolutionary history of many Chironomidae taxa. Within them, Orthocladiinae harbors the majority of semi-aquatic species (Andersen et al. 2010), what might explain their notable richness in madicolous habitats. Sinclair and Marshall (1987) also noted the remarkable dominance of Orthocladiinae among madicolous chironomids in Southern Ontario, Canada. In their study, ten of 14 genera recorded were Orthocladiinae, including Parakiefferiella, Metriocnemus, Parametriocnemus, Thienemanniella, and Limnophyes, also verified in this study. One more evidence that these genera can adapt well to this habitat.

Only two of the species verified here were previously known to occur in madicolous habitats (Podonomus pepinellii and Limnophyes guarani). Although Podonomus larvae can be found in streams and other fast flowing running waters, they are also common on the edge of streams (Brundin 1966). Podonomus pepinellii and all Podonomus morphotypes in the Atlantic Forest highlands are associated with madicolous habitats (Trivinho-Strixino et al. 2012), and as such they occur in rocky outflows and stream shorelines. The remaining species identified in this study were previously considered stream-dweller, although some have been found in habitats close to madicolous ones, such as those from the Corynoneurini tribe and Hudsonimyia caissara.

The larvae of Hudsonimyia caissara were originally found in low abundance (two specimens) in leaf litter of a mountain stream (Silva et al. 2012), possibly an inhabitant of the stones in the stream's edge. Further, it is very plausible that some stream-dweller species can tolerate both conditions (Vaillant 1956, Sinclair and Marshall 1987). Thus, a richer fauna is expected to occur in marginal stream rocks compared to isolate seepages.

It is not surprising that members of Hudsonimyia, Bryophaenocladius, Metriocnemus, Limnophyes and Pseudosmittia have been found in madicolous habitats during this study. These genera are known to have larval instars associated with semi-aquatic and terrestrial conditions. Roback (1979) was the first to verify Hudsonimyia larvae living on a thin layer of current water with periphyton and moss. Metriocnemus species are adapted to an extremely broad range of habitats within Chironomidae (Cranston and Judd 1987), including madicolous, as exemplified by the species M. hygropetricus (Kieffer 1911), whose name was given after their type locality habitat – natural rock seepages and artificial madicoulous habitats. Most Limnophyes larvae are semi-aquatic (Saether 1990), and recently a new species of this genus, Limnophyes guarani, has been recorded on madicolous habitats in the south of Brazil (Pinho and Andersen 2015). Pseudosmittia and Bryophaenocladius larvae are largely terrestrial or semi-terrestrial; however, this was the first time that Neotropical species of these genera have been recorded in a madicolous habitat.

Although many genera were expected to occur, some were particularly intriguing, such as Stenochironomus and Oukuriella. Both are known to be highly habitat-specialized in larval phase. The first is a vegetal miner (Epler et al. 2013) while the second is typically associated with sponge or wood detritus, although the habitat of basal groups in the phylogeny of Oukuriella could not be defined yet (Fusari et al. 2014). Oukuriella sublettei recorded in this study was reported in association with submerged wood found in first order streams with bedrock (Bellodi et al. 2016). Their presence in the marginal rocks of the stream might have been accidental, considering that only one specimen was found. The same is expected for both Stenochironomus species. Their reduced size suggests that adults emerged from leaf detritus or small fragments of wood, possibly inside rock fissures. However, the emergence of these taxa in such conditions was interesting, since in this case they completed their development in a thin layer of water, a complement to previous observations of immature submerged in the streams (Fusari et al. 2014, Bellodi et al. 2016; Dantas et al. 2016).

Far from being semi-aquatic, most of the Rheotanytarsus species require flowing water conditions to survive and emerge (Coffman and Ferrington 1996). However, the capacity to live in madicolous habitats may not be disregarded as some species, such as R. gloveri, demonstrated tolerance to drying rock faces of streams and survived in thin layers of current water (Cranston 1997). The strict definition given by Vaillant, in 1956, considers hygropetric habitats all flowing water with less than 2 mm thick. However, this delimitation is hard to establish when dealing with microhabitats constantly susceptible to water flow oscillations due to climatic conditions. In some occasions, our sampling sites, especially those at the margins of the streams, had the water flux modified as a consequence of the contraction or expansion of the main channel. Probably, Rheotanytarsus species were favored when stronger currents rose, although the intense dark coloration of the cephalic capsule of the larvae may indicate that they are truly madicolous inhabitants (Brundin 1966, Sinclair 1988, Trivinho-Strixino et al. 2012).

The procedure of rearing immature specimens to obtain the adults is most of the time unsuccessful due to their environmental requirements (Ekrem et al. 2007). Therefore, descriptions are frequently based only on adults, whose sampling methods often preclude the knowledge of immature habitat and other aspects of their ecology. Some species recorded have only been known by the adults, previously sampled with malaise or light traps. This was the case of Lipurometriocnemus biancae, Urubicimbera montana, Pseudosmittia catarinense, Caaporangombera intervales, and Paratanytarsus silentii.

Using our modified emergence traps allowed us to assure that the immature organisms and the adults from the species sampled in this study were madicolous inhabitants. The association and description of the immature is a fundamental task when studying chironomids, best accomplished with the help of molecular tools, such as DNA barcode. The larva and the pupa of P. silentii have been successfully associated with adult males after this investigation (Trivinho-Strixino and Shimabukuro 2017). Furthermore, even for those species whose immature forms are known, the first record of them in madicolous habitats represents a remarkable note on their success in colonizing a wide range of habitats.

We thank CNPq for financial support: PhD scholarship and productivity grant (141031/2013–1 and 306402/2010–6, respectively). We also thank Dr. Luiz Carlos de Pinho, Dr. Sofia Wiedenbrug, and Dr. Livia Fusari for taxonomic assistance, and Gilmar Perbiche Neves and Victor Saito for their help with field work.