Study area

The study investigated microhabitat associations of individual amphibian taxa on the isolated lateritic plateaus in the NWG (Fig 1). These island-like plateaus are dominated by areas of exposed rock but contain a varied mixture of other habitats forming a heterogenous mosaic (Fig 2). The study focussed on 14 representative lateritic plateaus in the areas both above and below the North-South trending escarpment in the northern section of the Western Ghats/Sri Lanka Biodiversity Hotspot in western Maharashtra. The study area extends over 2° latitude (15.89°-17.92°N) and a 1112 m change in plateau elevation (67–1179 m above sea level [m]). Above the escarpment the plateaus are raised hilltop carapaces elevated from the ecologically contrasting countryside.

As temperature, rainfall seasonality and rainfall amount varies across the survey area, for comparative purposes the area was sub-divided into 2 Regions (High and Low), separated by the escarpment. Each region was further subdivided into three arbitrary latitudinal sections: North, Central and South. These are referred to as ‘eco-zones’ (similar to life-zones but in the absence of specific environmental data for the plateaus the term eco-zone is preferred [67]. Rainfall across this area ranges from <2000 mm per annum on low sites to >6000 mm on high sites peaking at >9000 mm on one high site [12, 68, 69].

These sites encompass a range of land-uses (Fig 1; Table 1). As anthropogenic disturbance within a patch is likely to change the availability of some microhabitats its type was recorded, and an arbitrary metric calculated by summing the number of disturbance factors observed on each site (Table 1). Although the figure is arbitrary, no relevant literature exists, and it allows for initial between patch comparisons. Disturbance factors recorded were; removal of loose rocks, surfaced road, unsurfaced road, built structures on the plateau, domesticated animal grazing, surfaced road within 200m of plateau, tourism, part conversion to plantation, adjacent built structures, importation of topsoil. Sites with 0–3 factors were considered to have low levels of disturbance, 4–7 Medium Disturbance, 8+ High Disturbance. Anthropogenic disturbance changed the availability of some classes of microhabitat, most notably the removal of loose rocks, reduction in woody plants in conversion for grazing and agriculture, creation of pools on some low-level sites and importation of soil at Panchgani (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Disturbance values and dominant land use for each site surveyed.

To facilitate spatial comment, the study area has been sub divided into three latitudinal zones each side of the escarpment.

https://doi.org/10.1371/journal.pone.0194810.t001

Field data collection methodology

Sampling of both microhabitat and amphibian presence was performed along the same belt transects concurrently. The rocky plateaus are relatively simple ecosystems dominated by areas of exposed rocks with varying amounts of other microhabitats. Their size varies by an order of magnitude but based upon the smaller sites within the survey it was determined that four belt transects each 100 m long and 6 m wide would adequately encompass all the microhabitat types available on an individual plateau. The direction and path of each transect was determined at each site to maximise sampling of all available microhabitats. The same methodology was applied on each survey on the same plateaus in each year but with different transect locations making a total surveyed area 4800m².

To maximise detection, both diurnal and nocturnal surveys were deployed during two temporally comparable survey seasons [70]. Surveying took place each year in the same weeks at the onset of the monsoons in late July to early August in 2013 and 2014 [71]. Survey timing was selected for the known range of amphibian autecology, encompassing taxa with both explosive and prolonged breeding strategies [36, 72].

To make samples comparable, standardised Visual Encounter Surveys (VES) with refugia searching [67] along the belt transects were performed [73, 74]. The identity of each amphibian taxa their abundance and their microhabitat associations were recorded for each section of the transect [75]. Where species identity was not immediately obvious in the field photographs were taken to permit later clarification.

Microhabitat variables recorded along the same transects as the VES surveys comprised; maximum refugia rock size (mm), number of loose rocks >50mm, woody plant cover (as % cover on transect), presence of soil depressions with vegetation, presence of flowing streams, presence of static pools, presence of surface flooding (vernal pools). Although some microhabitats co-occurred, e.g. surface flooding and stream presence, all were included in the analysis so that finer scale associations could be detected (Fig 2). As some NWG amphibians are semi-terrestrial humidity levels may be considered as a micro-habitat therefore Relative humidity included in the analysis, it was measured with a calibrated hygrometer (Hanna Instruments™ HI 9064; [76].

All amphibians were identified using the best available literature, and their nomenclature considered using the latest taxonomical authorities [6, 26, 69, 77–85]. The classification of several of the taxa found in this study is still evolving. While many herpetologists have adopted the new suggested taxonomies entirely, this study adhered to recommendations within [86] and [87] by presenting former nomenclature alongside more recent identifications to maintain the continuity of identification in years following taxonomic amendments. This system introduces new and unstable taxa with the formerly acknowledged genera first and the newly identified genera in parentheses. For example, although the changes proposed by Frost [88] for the genus Rana were made at the generic level, biologists wishing to recognize the subdivisions of this genus, but maintain the stability of familiar species names and still follow rules of the International Code of Zoological Nomenclature (ICZN), can recognise newly created subdivisions of these genera as subgenera [86, 87, 89]. Under ICZN rules, the subgenus category may follow the genus name in parentheses, e.g., Fejervarya (Minervarya) sahyadris or Rana (Lithobates) pipiens.

Statistical analytical methods

Primer-e and Permanova+, Primer-e v7 [90, 91] were used to investigate the relationships between taxa in the study area and their microhabitats. Biotic data were represented by a Bray-Curtis similarity matrix of square root transformed abundance. Environmental data were represented by a Euclidian Distance matrix which was normalised before analysis. Analyses were performed for all taxa combined and each individual taxon. Ordination and visualisation of the model was performed in distance-based redundancy analysis (dbRDA). To identify the microhabitats with significant taxa associations’ step-wise analysis was performed in distLM. The step-wise routine commences with a null model then adds each criterion before checking by tentative removal thus optimising the selection. As the sample and number of predictor variables were small the Akaike Information Criterion with second order correction (AICc), was used as it to accounts for the ratio of samples to predictor variables being lower than 40 and performed in distLM [91, 92]. The explanatory power of microhabitats for the distribution of the biota was assessed using LINKTREE, a form of constrained binary divisive clustering. The routine maximises the value of Rat each division in the biotic matrix in concordance with the underlying distribution of microhabitats within each patch (site) with the B% being the difference in each linkage [90, 93].

Ethics statement

Sampling was undertaken by kind permission of the Indian National Biodiversity Authority, Chenai, India under permit number: Maharashtra 2014 MC200621.

The advice from the representative of the University of Plymouth’s Animal Welfare and Ethics Committee was that no formal consent was required since the animals were only observed or received minimal handling on their site of origin. We followed strict handling and preventative measures for cross-contamination, following standard practice for working with amphibians as described on http://www.amphibiaweb.org. No endangered animals were specifically targeted in the study.

Spatial distribution of microhabitats

Sites could be spatially separated at the macroscale by the relative microhabitat composition with notable differences above and below the escarpment illustrated in Figs 3 and 4 (Fig 4; R = 0.53, B% = 85%; Fig 3). The two most distinctive sites, Amboli High (Fig 4; R = 0.89, B% = 91) and Zenda (Fig 4; R = 0.37, B% = 68), are low disturbance sites that have retained much of their loose rock cover and have taxa associated with rock refugia (Fig 3). Lanja, a low disturbance site, is the most charismatic of the low region sites (Fig 4; R = 0.55, B% = 43; Fig 2). The most diverse eco-zone was the High North as illustrated by the distribution of the data points in the dbRDA plot, reflecting the impact of three types of land use on microhabitat availability (Fig 3, Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. dbRDA analysis for microhabitats with sites illustrated within eco-zones to allow spatial comparison.

dbRDA1 explained 39.3% of fitted data and 25% of total variation with dbRDA2 explaining 19.3% of fitted data and 12.3% of total variation.

https://doi.org/10.1371/journal.pone.0194810.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Linktree analysis of plateau similarities based upon microhabitat explanations for the biotic distribution, Primer-e v7.

Annotated for Eco-zones; HS-High South; HC-High Central; HN-High North; LS-Low South; LC-Low Central; LN-Low North. A: R = 0.53; B% = 85; Woody plants<-0.117(>-0.0826). B: R = 0.89; B% = 91; Max loose rock size<0.832(>2.26) or Surface water<0.848(>1.67) or Woody plants>-0.999(<-1.23). C: R = 0.55; B% = 43; Surface water>-1.19(<-1.6) or Woody plants<-0.345(>-0.117). D: R = 0.37; B% = 29; Woody plants<-0.738(>-0.607). E: R = 0.54; B% = 20; Stream>0.743(<-1.1) or Max loose rock size<-0.871(>0.00354) or Surface water>-0.377(<-0.785) or Woody plants>-0.738(<-0.999). F: R = 0.00; B% = 11; Max loose rock size<0.00354(>0.832) or Soil Cover<-0.791(>0) or N. Rocks>-0.0943(<-0.7) or Surface water>-0.785(<-1.19). G: R = 1.00; B% = 26; N. Rocks<-1.09(>2.13) or Pools<-0.832(>1.23) or Surface water>0.848(<-1.19) or Soil Cover<0(>1.58) or Max loose rock size<-1.42(>-0.0425) or Stream>0.743(<-0.177) or Woody plants<-0.607(>-0.345). H: R = 0.37; B% = 68; N. Rocks<-0.067(>1.63) or Woody plants>0.376(<-0.0826). I: R = 0.54; B% = 67; Woody plants>2.31(<1.39) or N. Rocks<-1.03(>-0.997). J: R = 0.63; B% = 55; Soil Cover<0(>2.37) or N. Rocks<-0.206(>-0.067). K: R = 0.50; B% = 42; Surface water<-0.377(>0.44). L: R = 0.50; B% = 23; Stream<-1.1(>1.66) or Max loose rock size<-1.24(>0.786) or Woody plants<0.376(>1.39) or Surface water>1.26(<0.44) or Soil Cover<-0.791(>0) or N. Rocks<-0.997(>-0.206) or Pools<-0.832(>-0.317).

https://doi.org/10.1371/journal.pone.0194810.g004

Pools were more abundant below the escarpment where many are manmade; their hydroperiod is shorter on the northern sites and more consistent above the escarpment. The number of annual wet days declines south to north by 11% and there 12.7 times as many wet days above the escarpment (Fig 1; [69–96]). Rainfall amount peaks at Amboli where it exceeds 9000 mm per annum resulting in the microhabitat ‘surface water’ separating the two Amboli sites from the rest (Fig 3). Soil is scant on the plateaus but deepens where it has accumulated in shallow depressions in the ferricrete but it has also has been imported onto Panchgani plateau to assist in tourism related activity [12]. Loose rock abundance, important as refugia, breeding sites and mate advertising posts, was reduced by collection from accessible sites for construction resulting in a disturbed distribution pattern (Fig 1; Tables 2 & 3; [12, 13, 48, 97]). Larger loose rocks were most absent from plateaus below the escarpment and most abundant where human access was difficult for example in the remoter High Region plateaus for example Zenda and Amboli High and to a lesser extent Jagmin. (Fig 2; Fig 4B, 4H & 4L; [98–100]). That, combined with greater woody plant cover, and for some sites surface water, separated them from the low sites and explained much of the latitudinal divisions (Figs 3 & 4A, 4B, 4H, 4I, 4J and 4K).

Most sites were characterised by combinations of microhabitats and their associated taxa (Fig 3; Tables 2 & 3). Such combinations are key for some taxa for example soil and rocks used by caecilians as refugia and egg deposition sites associated with soil close to water. We found 56% of microhabitats to be impacted by anthropogenic activity Tables 2 & 3).

Spatial distribution of taxa and explanatory microhabitats

Pond presence on low elevation sites was the most significant abiotic variable separating their amphibian assemblages from those above the escarpment where woody plants, surface water and relative humidity were the principal characters (Figs 3 & 4). Woody plant abundance, maximum loose rock size, surface water and pond presence were significant factors defining the differences in the biota above and below the escarpment in both the dbRDA and LINKTREE analyses (Fig 3; Fig 4, 4A, 4B & 4D).

High-level sites had 9 taxa not found on low sites (Figs 2 & 3; Tables 2 & 3). Woody plants were significantly associated with 44% of exclusive high-level taxa with no such associations for low-level specialists (Tables 2 & 3). Of the 8-taxa found exclusively on low sites 63% had significant relationships with water bodies and 60% of those were associated with pools compared to only 22% on high-level sites. However, all 8 had a relationship with the co-occurring surface flooding, highlighting the need to carefully define the types of water body.

Twelve of the 21-study taxa had significant habitat associations with the remaining 9 having associations that, whilst not significant, were identifiable (Tables 2 & 3). Taxa in the study are described as generalists after the IUCN description where they lack habitat specificity. Generalists that are very widely distributed indicating broad climatic and habitat tolerances are described as ubiquitous [9]. The remaining taxa, Uperodon globulosus, characterised as a generalist, was only detected when it was raining and had a relationship with relative humidity perhaps explaining its limited detection (Table 3 [9]). The generalist taxa did not have a noticeably higher number of habitat associations than other taxa except Hoplobatrachus tigerinus, which whilst currently described as a generalist, should more appropriately be assessed as ubiquitous.

Some 52% taxa were found in habitats other than those recorded by the IUCN, with 91% of the taxa sampled not previously recorded from lateritic plateaus [9]. Just over 67% of taxa in the study were associated with water bodies. Surface Flooding was the most important form of water on the plateaus being significant for 48% of taxa, Pools for 33% and Streams for 24%. Of the pool specialists 50% were only found on low level sites where pools were more common. Loose rocks measured by both size and abundance were the next most important microhabitat being significant for 33% of taxa, 19% with small rock abundance and 14% associated with large rocks. Rock sizes and abundance were meaningful in defining the different regions where large rocks were important for 33% of high-level specialists (Tables 2 & 3). The abundance of small rocks (<50 mm) was essential for 33% of exclusively low-level taxa but only one high-level specialist (Tables 2 & 3).

There were 29% of taxa associated with soil-filled depressions. Only 19% of taxa were associated with woody plants despite their being one of the defining microhabitats (Fig 2, Tables 2 & 3). The IUCN lists just 2 of the 21 taxa found as being associated with lateritic plateaus [9]. Generalist taxa were associated with a higher number of microhabitats (mean 3.7) than the other taxa (1.7). Fejervarya (Zakerana) cf. cepfi, Raorchestes cf. ghatei have not been assessed yet and their association should be noted with their first assessment. Fejervarya (Zakerana) cf. caperata, Gegeneophis seshachari, Indotyphlus maharashtraensis and Indotyphlus cf. battersbyi are all data deficient, thus the data presented here will form part of their initial assessment.

Many amphibians were detected under lateritic rock refugia in diurnal surveys. Only 5 taxa, Hoplobatrachus tigerinus, Fejevarya cf. caperata, Fejevarya (Minevarya) cf. sahyadris, Indotyplus cf. beddomii and Xanthophryne tigerina were found across open areas during the day, and these were only encountered during rainy periods. The above open area taxa were often well camouflaged against the texture of the lateritic rock or among short grass growing on soil depressions. Nocturnal transects confirmed the presence of most of the diurnal anurans without adding new taxa to the sample.

Caecilian microhabitats on lateritic plateaus

Soil is important for many amphibians providing sites to aestivate but is critical the semi-fossorial caecilians [101–103]. Three of the 4 caecilian taxa were associated with rocks in addition to areas of soil or stream presence (Tables 2 & 3). The exception was Gegeneophis cf. ramaswamii, considered a generalist fossorial taxon, a view this study supports from results associating it with soil-filled depressions (Table 2; [9]). We observed that Indotyphlidae sp, were detected diurnally under lateritic rocks that were positioned on soil depressions indicating the importance of co-occurrence of some microhabitats. These soil depressions were often no deeper than 10 cm. Gegeneophis cf. ramaswamii, G. seshachari, Indotyphlus cf. battersbyi and I. maharashtraensis were all located between the rock and the soil substrate although not significantly for G. cf. ramaswamii. One single I. maharashtraensis at Jagmin was found emerging from a soil depression next to rain fed flowing run-off stream after nearby terrain was disturbed by searching activity. The rocks caecilians were detected under were all within a short distance (no more than 20 m) from surface run-off, stream, or wet seep areas supporting the view soil moisture was likely to be highly important to the group [104]. The Gegenophis sp are oviparous and use rocks to shelter their young, for example, Gegeneophis seshachari at Kudopi comprised a mixture of adult and juveniles all found under rocks within a single 50m stretch of wet run-off [80].

Seasonal changes in microhabitat use

Many of the plateau taxa breed close to the start of the monsoon and they may have been detected in association with their breeding microhabitats [48, 115]. The plateaus are all highly seasonal only receiving rainfall for around four months a year resulting in the need for seasonal movement to avoid desiccation and to access breeding sites [36, 106, 115]. Rainfall, hydroperiod and the associated relative humidity are important factors for taxa with terrestrial or semi-terrestrial larvae for example Xanthophryne tigerina which was found only in the very wet southern high sites [6, 36, 48].

Generalist taxa microhabitat associations

Generalist taxa in this study were associated with more than twice as many types of microhabitat than the mean for other taxa (Table 2). However, the IUCN definition may be spatially too coarse to adequately describe patch quality as it makes little reference to microhabitat associations. There were two non-generalist taxa, Gegeneophis seshachari and Microhyla ornata, with very similar number of associated microhabitats (4) to the generalist total (3.7) suggesting that they too were generalists. However, such a result can be explained by co-occurrence microhabitats necessary for some taxa. For example, the microhabitats for Gegeneophis seshachari encompass a range predictable for a caecilian; rock, soil and water (Table 2). Another, Gegeneophis cf. ramaswamii, was perhaps wrongly identified as a generalist as it appears to require specific combinations of microhabitats to persist but can also be found among a range of landscapes. Similarly, Microhyla ornata should be reclassified as a habitat generalist in the context of lateritic plateaus. The generalist taxa, Duttaphrynus melanostictus and Hoplobatrachus tigerinus each have associations with all three aquatic microhabitats. This was an unsurprising result as both are pond breeding taxa that are also associated with abundant woody plant cover (Table 2; [32, 33]).

Impact of elevation on microhabitat associations

Tropical site habitats are known to change with elevation a view supported by this study (Figs 2 & 3; [41, 42, 116]. The drivers of elevational differences in the amphibian assemblages on the plateaus of the NWG were microhabitats dependent upon rainfall increasing which increased in frequency and volume with increasing elevation and hydroperiod which decreases with latitude. Although not directly related to elevation the ease of access onto low elevation sites, and their agricultural land use, has increased man-made pool frequency and reduced the abundance of large rocks (Figs 2, 3 & 4). The combination of long periods of rainfall, the related high relative humidity and abundance of loose rocks on Amboli High and to a lesser extent Amboli Low creates a special habitat the critically endangered and declining Amboli Toad, Xanthophryne tigerina, The large rocks provide three major resources, refugia, breeding sites and mate advertisement sites [48]. All of these are highly important resources for not only X. tigerina, but as breeding sites for Caecilians [117]. Woody plant abundance was one of the main microhabitats to define the regional difference between the high and low-level sites (Figs 3 & 4). Together with its associated soil filled depressions woody plants were highly important in amphibian distribution on the NWG plateaus across all elevations but impacting different taxa (Tables 2 & 3).

The effect of anthropogenic disturbance on amphibian microhabitats

Microhabitat availability was changed by three forms of anthropogenic disturbance on the plateaus; removal e.g. loose rocks, damage e.g. trampling or cutting down of plants and alteration by addition of foreign material e.g. soil at Panchami. Anthropogenic disturbance was also evidenced by construction and pollution. We did not examine the impact of addition by construction, pollution or trampling and therefore cannot comment specifically on these, although the sites with wind turbines had some construction on them. All rural communities close to the plateaus carried out the common practice of harvesting loose rock and utilising it for construction of dwellings, walls and memorials [12]. Therefore, sites at which there were quantities of rocks >50 mm³ were often farther from human residences. Given that many of the amphibian taxa in this study were associated with, detected under, or proximate to cover provided by rocks >50 mm³ we suggest that the natural occurrence of rocks >50 mm³ on plateau sites is an essential microhabitat resource for all amphibians, and one that is a rapidly emerging conservation concern for all plateaus.

Disturbance by the addition of soil, together with tourist related activity; on the high-level site Panchgani has shifted the taxa assemblage towards one dominated by generalist or widely distributed taxa (Tables 2 & 3). The addition of soil has closed almost all the fissure refugia and all large loose rocks and most small ones have been removed, limiting the available types of refugia, breeding and mate advertisement sites. This site is popular with equine tourism and this local industry has resulted in infrastructure development (cafes, stables and roads), soil compaction and increasing levels of domestic refuse. The pools also have a high silt load from the imported soil and grazing. A total of 24 individual amphibians were recorded from this plateau. Although amphibian counts were relatively high in comparison to lower disturbance sites, several of the taxa recorded (D. melanostictus, E. cyanophlyctis and H. tigerinus) are considered widespread or generalist taxa, listed as ''least concern'' in the IUCN status reports (Tables 2 & 3). D. melanostictus, E. cyanophlyctis and H. tigerinus were anticipated as taxa known to associate with anthropocentrically disturbed or modified habitats (Daniel, 2002). However, the presence of Raorchestes cf. ghatei and Fejevarya cf. brevipalmata may be surprising as they had limited distribution and are data deficient taxa in need of more robust ecological and population studies (Tables 2 & 3). Panchgani has a number of large pools constructed for watering livestock and anthropogenic uses. The largest is likely to hold some water throughout an average year possibly shaping the community by offering aquatic refugia in the dry season not seen on many sites. That may be a significant factor structuring the assemblage as it would favour pond specialist taxa [118].

Surface topography on the low-level plateaus creates some pools but many additional ones have been created by farmers in association with their agricultural land use. At a landscape level pool frequency is important in maintaining population connectivity and persistence [118–120].

Impact of climate change on the amphibians of the northern Western Ghats

Two changes in climate are predicted to impact the amphibian microhabitat requirements in the NWG: increasing temperature and fragmentation of the monsoon rains [121]. Both will require microhabitat resources to mitigate their effects; as refugia from increased temperatures and desiccation [107, 122]. The woody plants and rocks in this study provide thermal refugia allowing behavioural temperature regulation and are therefore key microhabitats worthy of preservation [122]. Breaks in rainfall that occur when larvae are in pools or in hygropetric habitats are likely to cause significant losses. To offset these, maximum availabilities of both populations and microhabitats should be preserved.