Capital letters indicate significant differences according to range tests. .... and Mike Bassett for help with fieldwork, and to our guide Deshek Kafab Sherpa.
Fres/iii-fl/er Biolosy (1994) 32, 309-322
Altitudinal trends in the diatoms, bryophytes, macroinvertebrates and fish of a Nepalese river system S.J. ORMEROD,* S.D. R U N D L E / S.M. WILKINSON,* G.P. D A L Y / K.M. DALE* AND I. JUTTNER^ 'Catchment Research Group, School of Pure and Applied Biology, University of Wales College ofCartiijf, PO Box 915, Cardiff CFl 3TU, U.K. ''National Rivers Authority, Aztec West, Bristol, U.K. ^Waterland Environmental Consultancy, Doncaster, U.K. ^GSF — Forschung:^ntrum fur Unwelt und Gesundheit, Institut fUr Okologische Chemie, Neuherberg, 85764, Oberschleisseim, Germany
SUMMARY 1. Hydrobiological changes were assessed along an altitudinal transect of eighteen to twenty-three tributaries from 6(X) to 3750 m in two adjacent river systems in east-central Nepal. The transect incorporated catchments under terraced agriculture at the lowest altitudes in the Likhu Khola, through streams in forest, alpine scrub and tundra at higher altitudes in Langtang. 2. Diatoms, bryophytes, macroinvertebrates and fish all showed pronounced altitudinal changes in assemblage composition as shown by TWINSPAN and DECORANA. A few taxa were restricted to streams at high altitude, but many more occurred only at lower altitudes where taxon richness increased substantially despite catchment disturbance by terraced agriculture. 3. Diatoms characteristic of lower altitude streams were mostly motile, epipelic or episammic Navicula and Nitzschia spp., which occur typically at greater electrolyte and nutrient concentrations. Those characteristic of higher and steeper sites included attached Fragilaria spp. and prostrate Achnanthes spp., tolerant of turbulent flow. 4. Cover by bryophytes varied within catchment type; high altitude springs supported dense mats, unlike streams fed by ice and glaciers. Taxa confined to low altitudes included those characteristic of humid subtropical conditions. 5. Invertebrate families occurring only at lower altitudes included a range of burrowers and pool dwellers. Numerically, filter feeding Hydropsychidae and Simuliidae dominated streams in terraced and forested catchments, whereas grazing baetid mayflies dominated higher altitude streams in scrub and tundra. 6. The combined density and biomass of at least six fish species in the Likhu Khola were 23-250 (per 100 m"^), and 86-1282 g wet mass (per 100 m"^), respectively. No fish were found in Langtang streams, probably because torrential headwaters prevented colonization. 7. Our data confirm that altitudinal transitions in stream biota are pronounced in the Himalaya of Nepal, but are likely to reflect a wide array of potential influences. • -
in the distributions of freshwater oreanisms are undescribed. Polar, high altitude, tropical and subtropical regions are among those known inadequately, and together they cover a large proportion of the earth's land surface. As a result, much ecological
Although freshwater ecology is maturing into an experimental and process-orientated science, there are still large areas of the world where basic patterns 309
310
S.J. Ormerod et al
theory for freshwaters has been developed and tested in the absence of potentially important data. A further corollary is an incomplete understanding of the management problems that affect river catchments and river conservation in such regions, even though examples of catchment-scale disturbance are widespread (e.g. for high altitude and subtropical regions see Ives & Messerli, 1989; Dudgeon, 1992; Onnerod & Daly, unpubl. data). Situated on the subtropical and biologically diverse borders of the Oriental and Palearctic regions, the Himalayan mountains represent both the highest altitudes and largest altitudinal range on earth. There is great interest in their ecology, but there have been few hydrobiological studies of the streams which drain them (Suren, 1994; Ward, 1994). This is unfortunate because of the opportunity that the Himalaya provides to assess the ecological character of high altitude streams in the context of wider altitudinal change. In Nepal, such changes include altitudinal trends in slope, temperature and climax catchment vegetation, shifts from a monsoonal to an alpine climate, changes in the intensity of agriculture, and changes in stream chemistry (Rundle, Jenkins & Ormerod, 1993). Deforestation, uncontrolled shrub clearance, and agricultural intensification have been particularly implicated as sources of catchment degradation (cf. Ives & Messerli, 1989), but impacts on stream ecology are largely undescribed. In part, this reflects the lack of baseline hydrobiological data for Nepal as a whole. In a previous paper (Rundle et al., 1993), we described the chemistry and benthic invertebrates of streams in the Everest, Annapuma and Langtang Himalaya, reporting differences between regions, altitudes and catchments with different land uses. Here, we expand this previous work by assessing the altitudinal distributions of a wider range of plant and animal taxa. We increase the altitudinal range of our earlier study to include more lowland streams in catchments dominated by terraced agriculture, and more streams at higher altitude draining mountain springs and glaciers. We concentrate on one major river system, where the upper catchment is partly protected as a national park. Study area All our studies were conducted on tributaries in
two subsections of the Langtang/Trisuli river system in east central Nepal (Fig. 1). Together, these rivers flow south and west from the Himalaya to join the Kali Kandaki, and ultimately the Ganges. The whole area is underlain by acid crystalline rocks, with sediments of Proterozoic, Palaeozoic and Mesozoic age, metamorphosed by mountain building to low grade schists. The water is circumneutral (pH c. 7.3; calcium c. 4—10gm~^), and there are pronounced downstream increases in sodium (c. 2—4gm~ ), silica {c. 2-15gm~'') and nitrate concentrations (0.06-0.2gm^^ Rundle et al., 1993). Land in the Likhu Khola Valley (LK on Fig. 1), at 600—1200m a.s.I., is predominantly given over to terraced agriculture for rice, com, wheat and other crops, with areas also of degraded Sal Shorea robusta scrub (all higher plant nomenclature follows Miehe, 1990). Land of these types is widespread throughout the middle hills of Nepal, and has involved substantial modification to rivers through irrigation, clearance of some bankside vegetation, the release of sediment, and the application of nutrients and pesticides. Nevertheless, structurally complex rivers with steep gradients (>100-200mkm"'), fast flows (>50-100ans"'), and pronounced pool-riffle sequences, where large boulders and cobbles are interspersed with sand and gravel, still remain. The climate is predominantly monsoonal, with up to 4000 mm of rainfall annually, mostly between May and September. Midsummer air temperature range is 19—34''C, and midwinter temperatures are 8—23°C. Studies were undertaken on eleven different second to fourth order streams; fish, bryophytes and diatoms were sampled in six to ten of them (Fig. 1). In the Langtang Valley, studies were undertaken on fourteen second to fourth order streams, at altitudes between 1200 and 3750 m a.s.I. All drain into the Langtang Khola, which is torrential down to its confluence with the Trisuli. The land has been maintained in more of a semi-natural state than the Ltkhu Khola, partly through the presence of the Langtang National Park. There has, nevertheless, been some change in vegetation as a result of burning and grazing for at least three centuries. At lower altitudes (1200— 2500 m), streams Tl, T2, C13 and C16 flow through mixed evergreen and deciduous forests of Quercus, Lithocarpus, Michelia, Betula, Alnus, Rhododendron,
Pinus and Tsuga spp. (Miehe, 1990). At middle altitudes (2500-3250m), streams C19, T7, C23, TIO, C33
Altitudinal trends in Himalayan streams b
\
311
u /-—(^ J
V? ^
J Lk •K N
10 km
Fig. 1 Location of the study sites in Nepal. In (a) A, Lg and E represent the Annapuma, Langtang and Everest regions sampled by Rundle et al. (1993); K is Kathmandu. (b) shows the relative position of the Langtang and Likhu Khola catchments, (c) and (d) shovw the location and altitudinal arrangement of the Langtang sites in forest (TI, T2, C13, C16), alpine scrub (C19, T7. C23, TIO, C33, C26) and tundra (C27, C29, C31, CC). (e) shows the location of sites in the Likhu Khola (Bh, Bhondare Khola; Jo, Jogi; Ma, Mahadev; Bo, Bore; Ch, Chinniya; D, Dee; Ka, Kahare; Gh, Ghyambe; Sy, Syalping; Cs, Csalese).
and C26 flow through Juniperus recurva and /. indica C29 rose directly from a spring. Mid-winter temscrub with tussocks, rosettes and grassy vegetation. perature ranges at Langtang Village (3307 m) are —11 Higher still (3250-4(X)0m), streams C27, C29, C31 to 8''C, while in midsummer they are 9-19°C; annual and CC flow through alpine tundra and boulder rainfall in the inner Langtang valley is about 20% of fields with dwarf shrubs, lichens, Kobriesi mats and that in the Likhu Khola. Like the Likhu Khola, all Carex sedges. Some streams were sampled within the streams have steep gradients, and consist pre0.5-1km of the bases of hanging glaciers, which dominantly of riffles or torrents with substrata of drain by waterfalls at the valley sides. Stream CC was cobbles and boulders. As in the Likhu Khola, not all partly ice covered at the time of sampling, whereas investigations were undertaken at all sites.
312
S.J, Ormerod et al
Materials and methods
Bryophytes
Most biological fieldwork was carried out in November and December in 1991 and 1992. Some comparative data on inveiiebrate communities in Langtang were available for March 1992 (from Rundle et al, 1993), while visits to the Langtang region for fisheries work were made during July and August 1993, when upstream migrants were most likely to be present.
Macrophytes, particularly angiosperms, were scarce in all the rivers studied. However, assemblages of bryophytes (mosses and liverworts) were assessed over sample reaches of 20 m by hand collection of coded specimens in the field, and by estimates of relative and absolute cover in the channel, wetted perimeter and splash zone of each stream. In contrast to diatoms, invertebrates and fish, this approach treated bryophytes as a focus for ecological effects from the whole range of factors impinging on the river corridor, rather than solely aquatic factors. It also circumvented the problem of discriminating between truly aquatic species, and those that were inundated only occasionally (e.g. Slack & Glime, 1985). Coded specimens were subsequently identified, as far as taxonomically and practically possible. In two streams where bryophyte cover was pronounced (C13 and C29), species abundances were assessed by subsampling from five 1 m^ quadrats.
Physicochemistry
In the Likhu Khola, chemical data were collected from weekly spot samples from the Bore, Dee, Chinniya, Bhondare and Jogi Kholas; other sites were sampled once or twice in March and/or December 1992. Conductivity and pH were measured on site using portable meters. Concentrations of anions and cations were detennined by ion chromatography and inductively coupled plasma spectrophotometry, respectively. The physical structure of each stream was recorded on site using the method of Holmes (1983). For this, the percentage occurrence of different depths (lm), widths (20m), substratum characteristics (bedrock, boulders, cobbles, pebbles, gravel, sand, silt, mud clay), habitats (pool, slack, riffle, rapid, run, cascade, waterfall, exposed rock) and fringe marginal vegetation (2m) were estimated by eye over 20 m stretches. Altitudes and slopes at each site were recorded from 1:50000 maps; catchment land use was recorded on site. Diatoms
Diatoms were scraped from five stones in riffles at each location using a soft nylon brush, and preserved on site in 5% formaldehyde. Organic matter was oxidized from each sample using hydrogen peroxide and potassium dichromate before each sample was mounted and scanned at xlCKX) magnification under oil immersion and interference contrast. Nominally, 400 vaivae from each sample were identified to spedes or subspecies using Krammer & Lange-Bertalot (1986, 1988, 1991) and Krammer (1992) to provide data on percentage abundance. The few species not matching those described were given type labels for use in subsequent data analysis.
Macroinvertebrates
Macroinvertebrates were sampled in two ways. First, a 2min kick sample (net mesh size 250|j,m) was taken from arifflein each stream to give a semi-quantitative assessment of invertebrate abundance for use in assemblage classification and ordination (see below). In the Likhu Khola, animals were elutriated from mineral matter and individuals from each morphofamily (from here on called families) were placed into one of three abundance categories (1—10, 11-100 or >100 per sample). In Langtang, animals were preserved in the field in 70% alcohol and sorted, identified and counted in the laboratory. A second sampling procedure was used at a subset of sites in the Likhu Khola to generate more precise data on relative abundances. Quantitative data from six replicate Surber samples (area 0.1 m^, mesh 250 |im) were taken from i4entical habitats in five of the streams (Bore, Chinniya, Dee, Bhondare and Jogi Kholas), and used alongside exact counts from the Langtang kick samples. All data were converted to relative abundances, with counts for each group expressed as a percentage of the overall total. For the eight Langtang streams that had been sampled for invertebrates in both March and December 1992 using identical methods, the relative abundance of each
Altitudinal trends in Himalayan streams 313 taxon was compared between sampling intervals. We also assessed differences between sites in the composition of macroinvertebrate feeding guilds. Families were placed into one of five categories based on information from the literature and from gut content analyses. Although such a procedure has been criticized (see Palmer, O'Keefe & Palmer, 1993), we believe it is instructive in helping to understand the trophic structure of streams along the pronounced ecological gradient of the Himalaya. Moreover, recent studies support the view that feeding guild analyses are valid when differences in mouthpart structure and behaviour between groups are taken into account (Palmer et al, 1993). Fish
Fish in six streams in the Likhu Khola and eight in Langtang were sampled with safari backpack electric fishing gear, powered by batteries recharged on site from solar panels. Population estimates were made by standard catch depletion methods over two or three separate reaches of 20 m. No stop nets were available to close off each reach, so all estimates are conservative. Fish were anaesthetized in benzocaine and identified on site using Shrestha (1981). All but a few voucher specimens were returned to the stream after weighing and measuring. Data analysis
Except for fish, most biological data analysis involved an a posteriori approach, in which the macroinvertebrate, bryophyte and diatom data were separately ordinated by detrended correspondence analysis (DECORANA; Hill 1979a) and classified by two way indicator spedes analysis (TWINSPAN; Hill, 1979b). Pseudospecies cut levels varied according to the nature of the data in each group. For invertebrates, we used log abundance categories of 1, 10 and 100 individuals per sample; for bryophytes, percentage cover (1 =
