Freshwater Biotogy (1986) 16, 509-520

The flora, fauna and water chemistry of Tagimaucia crater, a tropical highland lake and swamp in

W. SOUTHERN. J. ASH.* J. BRODlEt and P. RYANt Geography Depanment. Australian National University, Canberra, , and *School of Natural Resources, and tlnstitute of Natural Resources. University of the South Pacific, Suva, Fiji

SUMMARY. 1. Tagimaucia crater lake and swamp are located at an altitude of 820m In Taveuni, Fiji (lat. I6''S; long. 179°56'W). Rainfall exceeds evaporation in all seasons and residence time for water in the shallow lake (2.5-5.5 m deep) is only 4 days. 2. Dissolved ion concentrations are low (conductivity 14-18//Scm"' at 25X). and the water is slightly acidic (pH5.()-6.5 at 25°C). Median total phosphorus (0.5gm~"') and total nitrogen (3.3gm~"') are mod- erately high and probably represent dissolved and suspended organic matter. Median chlorophyll a concentrations (2.5 mg"'') were low and indicate low phytoplankton productivity. 3. The swamp vegetation is dominated by Leptronia ariiculata (Retz.) Domin and algae which form floating sedge peat islands, and Pandanns taveuniemis St John and other small trees where alluvium and colluvium arc infilling the margins of the crater. The surrounding slopes are forested. 4. The fauna of the lake and swamp are low in both diversity and abundance. The only aquatic vertebrates observed were Anguillti eeis and cane toads, Btifo niaritius L., and the only bird was the swamp harrier, Circus approximans approximans Peale. A variety of aquatic insects were observed but crustaceans and other invertebrate taxa were scarcely recorded. 5. Although the swamp and lake are otherwise undisturbed, several recently introduced and animals have been recorded.

Introduction the larger rivers, cut-off river meanders and channels form lakes which fill up after high Although there is high rainfall over most of the rainfall or flooding. Further upstream the Fiji islands, stream and river drainage is gener- rivers may be temporarily dammed by land- ally well developed so there are few natural slides, and there are legends and geological lakes or ponds, and freshwater swamps only evidence which indicate the sites of several occupy small areas. In the lowland valleys of former lakes (Rodda, 1976. 1984): none of these lakes exist at present, though there are Correspondence; Mr W. Southern, Geography Department. Australian National University, some small ponds created by the damming of G,P,O. Box 4. Canberra 2601, Australia. streams.

509 510 W. Southern et al.

km IOO 200

178

FIG, 1, Map of part of the Fiji archipelago showing the loeation of plaees mentioned in the text, D, Dreketi. valley; M, Monasavu dam; N. Nakelikoso; S, Salialevu; T. Tagimaucia erater; V. Vaturu dam.

Volcanic craters may form lake basins, and tion of peat swamps in Fiji. On areas of several lakes occur in recent craters on the impeded drainage, perhaps occasionally sub- island of Taveuni (Fig. 1). The largest of these, merged, there would naturally be forest an explosion crater, contains Lake Tagi- vegetation but. as a result of forest clearance, maucia. Several of the other volcanoes in gardening and burning during the past 3500 southern Taveuni have small craters, formed years, conditions were created which favoured by the eruption of lava, scoria and ash. and open-habitat species of sedges which form peat some of these contain lakes of 0.1-1 ha. Lava and thereby favour other peat swamp species: may block existing drainage patterns and there this type of peat swamp is both created and are a number of shallow ponds and swamps in maintained by man. In shallow or seasonal north-eastern Vanua Levu. especially in the lakes some sedges may establish on the lake Dreketi valley (Fig. 1), which seem to be the bed and gradually accumulate a layer of peat; product of Pleistocene lava flows and, perhaps, eventually infilling the lake. In deep lakes, tectonic tilting. some sedges may grow around the shore and Two major dams were completed in Viti form a floating peat mat which is both thick- Levu during 1983. the Monasavu and Vaturu ened and extended across the lake surface as dams (Fig. 1). and this has raised interest in the sedges grow. Parts of the floating peat mat the natural freshwater lakes of Fiji. may break loose, probably as a result of Freshwater peat swamps are associated with fluctuating water levels, and form a floating most of the lakes and other areas of impeded island: a characteristic feature of swamps of drainage. The freshwater swamps of the largest this type. There are two lakes with floating island. Viti Levu, have been described by Ash islands in Fiji, a small (lha) lowland basin & Ash (1984) who suggested that three diffe- near Nakelikoso in Vanua Levu (Fig. 1) and rent mechanisms are responsible for the crea- the Tagimaucia crater. The peat swamp and Tagimaucia crater 511 lake are closely inter-related and Tagimaucia by taking soundings with a weighted line on lake and swamp are treated together in this transects across the lake. Water transparency paper. was measured with a Secchi disc. Water sam- Tagimaucia crater is visited infrequently, ples for chemical analysis were taken at about and the 5 or 8h walk (from the west or east 9 a.m. from just below the lake surface and at coasts, respectively) is on steep and narrow a depth of about 2 m near the lake edge, on forest tracks. The lake and swamp have been each occasion that the crater was visited. little affected by human disturbance, though Water samples were frozen a few hours after sedges may have been harvested in the past collection and analysed at the Institute of and the footprints of more recent visitors are Natural Resources, University of the South forming a track across the swamp. The crater Pacific (USP). In general, methods for chemic- has been visited by various scientists, mostly al analysis of the water samples were those botanists, but little has been published except described in Standard Methods for the Ex- Seemann's (1862) description: '... a large amination of Water and Wastewater (American extinct crater filled with water, and on the Public Health Association, 1981). north-eastern part covered with a vegetable Alkalinity was measured by titration with mass, so much resembling in colour and lxlO"^moldm"' sulphuric acid using a mix- appearance the green fat of the turtle, as to ture of bromocresol green and methyl red have given rise to the popular belief that the indicator. Calcium, magnesium, sodium and fat of alt the turtles eaten in Fiji is transported potassium were measured by flame atomic hither by supernatural agency, which is the absorption spectrophotometry. Ammonia was reason why on the morning after a turtle-feast measured using the phenate spectropho- the natives always feel very hungry. This tometric method. Nitrate was measured using jelly-like mass is several feet thick, and entire- the cadmium reduction column method. Nit- ly composed of some microscopic cryptogams, rogen was analysed by Kjeldhal digestion fol- which, from specimens I submitted to the Rev. lowed by distillation of the ammonia and M. J. Berkeley, a weighty authority in these determination of the ammonia using the phen- matters, proved to be Hoomospora transversa- ate method. Phosphate was measured using Us of Brebisson. and the representative of the molybdate/tartrate/ascorbic acid method. quite a new genus, named Hoomonema Phosphorus was measured by oxidation with fluitans Berkl. A tall species of sedge was perchloric add followed by the phosphate growing among them, and gave some degree of determination (as above). Chlorophyll a was consistency to the singular body'. And Com- measured by tbe spectrophotometric method mander Goodenough's (1876) account of a after extraction with 90% acetone; no correc- visit to the crater during which he determined tion was made for phaeophytin a. Turbidity the depth of the lake as l.V16ft (4-5 m). More was measured using the nephelometric recently. Koyama (in Smith. 1979, pp. 231- method. pH and conductivity were measured 232) considered the swamp to be 'a fascinating in the laboratory using meters, at IS^C. and perhaps unique area in Fiji'. Plankton were sampled by towing plankton nets of 20 (6()//m) mesh at a depth of about 1 m in December 1982. and by examining the planktonic algae which settled from water Methods samples collected at the same time as those for chemical analysis. Algae were identified from Tagimaucia crater was visited periodically Prescott (1979) and Fritsch (1971), Vascular (September and December 1982, February, specimens were identified and lodged at July and December 1983. and November 1984) the SUVA herbarium, USP. Nomenclature and observations were made on the vegetation, follows Smith (1979, 1981). Brownlie (1977) structure of the peat swamp, lake bathymetry. and Parham (1972). Sedge biomass was esti- water chemistry and fauna. A palynological mated by oven-drying representative samples. study of the crater is being made by the first The vertebrate fauna was recorded by author, and cores were taken in several parts observation, including diving in the lake. One of the crater. Lake bathymetry was determined litre of Quinaldene was released in a small 33 512 VV. Southern et al.

(0) (b) 850 500 m

N

1050

950

FIG, 2. (a) Map of Tagimaucia crater catchment and take based upon aerial photography of 22 June 1%7; showing vegetation types (see text for details), (b) Depth conlourN of the main lake below the outHow level. emhayment of the lake but did not result in the specimens were sent for identification to the appearance of any aquatic vertebrates at the University of Canterbury. New Zealand. surface. Baited hooks were set to catch eels and fish but without success; terrestrial in- vertebrates were captured with sweep nets and Results by examination of the swamp vegetation. A fluorescent U.V. lamp, over an alcohol filled Tagimaucia crater and catchment collecting tray, was set up near the lake for Tagimaucia crater is situated on the eastern two nights each in December 1982. February side of the central mountain ridge of Taveuni 1983 and July 1983. Where necessary, animal at an altitude of 820m (Figs. 1 and 2). The Tagimaucia crater 513

mountains are formed of Pliocene-Quaternary range producing clouds and rain. TTiese condi- basic volcanics, and the crater is a late Quater- tions prevail on most days throughout the nary maar. The crater contains several small year, and Des Voeux Peak on the central volcanic cones, and is bounded on its eastern range has an average annual rainfall of margin by a fault scarp 60-80 m high. The 9800 mm (Public Works Department records. scarp continues to the south-west for a further 1976-83). Average rainfall in the catchment is 2 km and has diverted several eastern-flowing likely to be intermediate between that of Des rivers such that some now flow into the crater. Voeux Peak and sites on the east coast such as The catchment of the crater consists largely Salialevu which has an average annual rainfall of the eastern flanks of the central range, of 5800 mm (Fiji Meteorological Service re- which reach an altitude of 1100 m, while the cords, 1913-74). northern and eastern margins are formed by The cloudy climate greatly reduces both the short steep slopes of the maar and the incoming solar radiation and outgoing radia- scarp. The catchment has a total area of 619 ha tion such that air temperatures show little of which 213 ha (34.6%) is swamp and open diurnal variation. On the basis of short-term water and the remainder is forested. The observations and extrapolation from other sta- crater is drained by the eastern-flowing Waini- tions, the mean daily air temperature range at sairi Creek which has cut a steep-sided rocky the crater is likely to be between 15 and 2rC valley through the scarp; the outflow being in July and 18 and 24X in January, and relative about 8m wide. humidity is likely to be 94-98% for most of the The sedge peat swamp is contained within, year. and is infilling, the lake basin. Alluvium and It is unhkely that the catchment experiences colluvium are accumulating at the margins of many weeks in which evaporation exceeds the swamp, and are most extensive on the rainfall, and annual evapotranspiration is un- western and northern sides of the crater where likely to exceed 900 mm. Annual rainfall in the there are larger catchments. These inorganic catchment probably totals about 5 x 10' m^, deposits grade into sedge peat further from the evapotranspiration may be estimated as edge. In some places the transition to peat is 5xlO''m^. while the lake has a water volume gradual but around much of the edge there are (excluding peat) of about 5xlO''m\ By assum- narrow water-filled fissures 2-3 m deep and the ing that all rainwater not lost by evaporation central part of the peat swamp comprises a passes through the lake, the average residence floating peat mat 2-3 m thick (Figs 3 and 4). time for water in the lake (the time to fill the Tfie floating peat is dissected by fissures to lake) is estimated as only 4 days. The water form a number of interlocking islands, and reaching the lake may follow several routes, these enclose several open water bodies. The and the proportions may be approximated on water bodies have a total area of 16 ha and the the basis of habitat area in the catchment: largest, known as Lake Tagimaucia, has a direct rainfall (2.8%). run-off from the floating depth of 2.5-5.5 m below the outflow level sedge peat islands (14.5%), run-off from the (Figs 2 and 3). surrounding swamp (17.5%) and run-off from Much of the floating peat has a jelly-like forested slopes (65.4%). Rainfall is likely to be surface 10-25 cm thiek. composed of higher on the upper slopes of the catchment cyanophyta and algae, which has a low per- but this is offset by evapotranspiration, and the meability and it has been observed that proportion of water entering from the slopes is rainwater runs off the floating islands into the likely to be rather lower than the percentage fissures and lake. Water entering the swamp indicated above. and lake passes to the outflow by flowing beneath the peat islands and along fissures. Since the outflow is relatively narrow, the lake Water chemistry level may rise by 1 m or more after heavy Results of chemical analyses are summarized rainfall. in Table 1. They are presented in terms of the The climate of the catchment is strongly median observed value (or the average of the influenced by the prevailing humid south- two values closest to the median) and the easterly winds which rise over the central overall range of values. No significant differ- 514 W. Southern et at.

Alluvium Peal Mud ond Volcanics IOO m FIG. 3. Diagrammatic cross-section of Lake Tagimaucia under flood conditions, with a water level about 1 m above the outflow level. Vegetation types are indicated by number as in text and Fig. 2.

ences were apparent between lake surface rogen concentrations were moderately high, samples and deeper water satnples, and though however, and probably represented organic there was some variation between different forms of these nutrients, perhaps in peat or sampling dates this did not reveal any consis- humic materials. A Secchi disc was visible to a tent patterns. depth of 2-3 m, virtually the bottom of the The pH (5-6.5) and alkalinity (2.6g m"^ lake, indicating a moderate amount of dis- CaCO^) indicate slightly acidic conditions, solved or suspended matter. The water colour probably detertnined by CO2 and HCO3 ions. was brown, indicating organic matter. Chlor- Conductivity values (l4-18/iScm~') indicate a ophyll a concentrations were low and indicate low concentration of dissolved salts, and this low phytoplankton biomass. was apparent for individual cations, nitrate and Comparative chemical values for world ammonium ions. Total phosphorus and nit- mean rainwater (Garrels & Mackenzie, 1971)

FIG. 4. View looking south across a small lake in the northern part ol" Tagimuuciu crater, showing Ooatlng sedge peat with abundant Lepironia articulata. Freycinetia pritehardii along the lake margin, and forested cloud covered hills along the eastern side of the crater with Pandanus taveuniensis at their base. Tagimaucia crater 515

TABLE I. Chemical analyses of water from Lake Tagimaucia. World mean rainwater (Garrels & Mackenzie, 197!) and basaltic groundwater in Taveuni (pers. comm,. Mineral Resource Department, Fiji). All values are gm ""* unless stated otherwise. Median No. of value Range observations Rainwater Groundwater pH(pH units) 5.9 5.0-6.5 13 5.7 7.1 Bicarbonate alkalinity as CaCOj 2.6 2.1-6.0 13 Conductivity (^S cm "' at 25°C) 16 14-18 9 Ca^^ 2.6 0.5-4,1 9 0.08 17 0.5 0.3-0.9 9 0.28 10 Na* 5.9 1.7-17.0 9 1.9 9.5 K* 1.2 0.2-^.9 9 0.3 1.7 NHa <0.01

TABLE 2, Tagimaucia swamp vegetation (excluding epiphytes and algae). Introduced species are indicated by an asterisk (for further information see Ash & Ash. 1984). Pandanaceae Cyperaceae coiiimuni[y community

Species b a b Trees and shrubs *Ctidemia hirta (Linn.) D. Don Garcinia myrtifolia A, C. Smith Geniosioma vitiense Gilg. & Benedict Melastoma denticulatum Labill. Metrosideros collina (Forst,) A. Gray Pittosporum pickertngit A. Gray Rapanea myricifolia (A, Gray) Mez. Scaevola floribunda A, Gray Spiraeanthemum vitiense A. Gray Syzygium gracitipes (A. Gray) Merrill & Perry Weimnannia affinis A. Gray Pandanaceae and Arecaccae Freycinetia pritehardii Seem. Ereycinetia storckii Seem. Pandanus tavetmiensis St. John Physokenlia rosea H. E. Moore Cyperaceae Eleocharis dulcis (Burm. f.) Trin. ex Henschel Eleocharis ochrostachys Steudei Hypolytrum nemorum (C B. Clarke) Koyama Lepidosperma sp. Lepironia articulata (Reiz.) Domin Machaerina fakata (Nees) Koyama *Rhynchospora corymbosa (L,) Brilton -¥ + Herbs 'Centella asiatica Urban 'Mikania micrantha H.B. & K. 'Paspalum distichtim Linn. 'Paspalum orbiculare Forst. 'Polygonum dichotomum Bl. Spathoglottis pacifica Reichb. f. Fems and fem allies Culcita straminea (Labill.) Maxon Dicranopteris caudata (Copel) S. John Lycopodium cernuum Linn. Nephrolepis hirsutula (Forst.) Presl. Bryophytes Leucobryum sanctum (Brid,) Hampe Sphagnum cuspidatum Ehrh. ex Hoffm. species, is a zone dominated by Eleocharis fissures, and has a biomass of 2-4kgm ^, but ochrostachys. The substrate is more peaty and both the height and density decline towards may be flooded to depths of 0.5-1 m. This the centre of the islands where the biomass zone varies in width, from no more than a few was only 0.05-1 kgm"^. The moss Leuco- metres wide adjacent to steep slopes to 100 m bryum sanctum forms extensive patches and a wide adjacent to Pandanus communities. few herbs occur sporadically on the islands. There is often a water-filled fissure, 20-50cm The jelly-like surface of the islands is made up wide and 2-3 m deep, separating this commun- of live and dead plant material and supports a ity from the next, (c) The floating peat is diverse algal community including abundant characterized by the sedge Lepironia articulata filamentous forms. Prominent amongst the which grows to a height of 120-200 cm along algae are Cyanophyta, including Anabaena, Tagimaucia crater 517

Oscillatoria, Spirutina and various Chroococ- represented: Polycentropodidae, Philopotami- caceae; Chlorophyta, including Zygnema, Spir- dae and Leptoceridae. Large numbers of Poly- ogyra, Ulothrix, Chlamydomonas and various ptectropus fijianus Banks and a few Triaenodes Desmidiaceae including Tetmemorus, Cosmar- n.sp. (closest to but not T. dubia Mosely) were ium. Ctosterium and Pleurotaenium; and Bacil- collected in a light trap. Whirligig beetles lariophyta including Frustutia and Eunotia. (Gyrinidae) genus Dineutes were common Elongate shallow pools. 10-20cm deepand among the sedges around the lake edge. Large 20-40()m' in extent, occur in several areas of eels, genus Anguitta, were fleetingly seen but the swamp. These pools are virtually devoid of not caught on baited hooks. Based on size and vascular plants except for occasional stunted proportion these are likely to be A. marmorata Lepironia and a sedge which is identified Quoy & Gaimard, No other fish species were provisionally as Lepidosperma (this species is observed. not listed by Koyama in Smith. 1979). (d) The The sedge community around the lake sup- edges of the floating peat islands often are ports few vertebrates. Tadpoles and juveniles raised a few centimetres above water level. of the introduced cane toad, Bufo marintxs L., These and other raised areas, such as old were common, and coastal villagers reported Pandanus stumps, support small patches of the presence of the Fiji ground frog, Ptatyman- more diverse vegetation. A variety of mosses tis vitiantis Dumeril., though this was not are present, notably Sphagnum cuspidattim confirmed. One specimen of the Pacific islands which forms raised hummocks. Several epiphy- boa constrictor, Candoia bibroni Hombron & tic mosses grow on the hummocks and upon Jacquinot, was found in the swamp. Despite woody shrubs rooted in the hummocks. A apparently attractive conditions, no RalHdae variety of shrubs, stunted trees (l-2m high), (crakes, rails and swamphens) or Anatieae palms and ferns may be present. This com- (ducks) were seen or heard. A pair of swamp munity occupies a very small proportion of the harriers. Circus approximans approximans swamp surface but contributes greatly to its Peale, was seen patrolling the swamp. diversity. Plant debris is washed up around the edges of the lake by wave action and may contribute nutrients to these edge communi- Discussion ties. (iv) Aquatic algae. The lake contains a The climate of the Tagimaucia crater is of a sparse phytoplankton community dominated montane-oceanic type with high rainfall at all by Bacillariophyta, including Erustulia, Pinnu- seasons and little temperature variation either laria. Eunotia and Surirella; Desmidiaceae, daily or seasonally. In these respects the cli- including Ctosterium, Pletirotaenium, Euas- mate is similar to hyperhumid equatorial re- trum. Staurastrum and Tetmemorus; Zygnema- gions, and it differs from many less humid taceae including Zygnema and Spirogyra; and tropical regions which experience one or two Cyanophyta including Osciltatoria. Nitetta was drier seasons each year. The composition of found in the lake but is not abundant. the lake water reflects both the climate and the characteristics of the lake basin; including the relatively small ratio of the catchment area to Fauna lake area, the rapid flushing of the lake, and The non-aquatic invertebrate fauna in the the shallow deph of water. swamp appears to be limited to a few lepidop- Wetze! (1975) discussed the need to consider terans and the adults of aquatic insect larvae; the trophic status of lakes not only in terms of notably Odonata, Diptera and Trichoptera. A the phytoplankton in the water body, but also 2(X)m plankton tow in December 1982 col- with reference to the littoral of bog vegetation. lected only juvenile backswimmers, Anisops, Using Wetzel's classification, the Tagimaucia and bright orange hydrachnid mites. Benthic crater can be considered to be a Bog- samples contained larvae of a ceratopogonid Dystrophic lake system. species, and two species of Tanypodinae and The lake and swamp have similarities to Orthocladiinae (Chironomidae). Trichopteran equatorial lowland lakes, swamps and black- larvae were common and three families were waters that have been described in Malaya 518 W. Southern et al.

(Furtado & Mori, 1982; Johnson. 1967) and Pandanus taveuniensis indicate a richer nut- South America (Marlier. 1967; Sioli. 1968) but rient supply. These differences are also appa- differ in their water budget, ionic composition rent from the growth of Lepironia in different and productivity from most tropical African parts of the swamp. In the marginal lakes and swamps (Beadle, 1974; Burgis, 1978; Pandanaceae-rich swamp, Lepironia has a Tailing & Tailing, 1965; Vareschi, 1982; Car- biomass of 3-6kgm~^; on the edges of the ter, 1955; Gaudet & Melack, 1981; Ganf, islands the biomass is 2—4kgm"-, but away 1974) and Australian lakes (Williams, 1981). from the island edges the biomass is only In the neighbouring archipelagos of the south- 0.05 kg m~^. These differences in biomass west Pacific there are several islands with probably give a good indication of differences volcanic craters containing lakes and swamps in productivity and perhaps also the rate of but these are within a few metres of sea level peat formation. On the basis ot Ikusima's and contain slightly to strongly brackish water, (1982) obset^ations on Lepironia it may be e.g. Gaua in , Tikopia in the Solomon estimated that Lepironia gross production Islands, Aunu'u in Samoa, and Niuafo'ou, ranges from less than 0.5gm~^ day"' to more Tofua and Late in Tonga (Macioiek & Yama- than 8gm~^ day"' in different parts of the da, 1981). There are also a number of raised Tagimaucia swamp. This encompasses the limestone islands with shallow lagoons which range from very low to moderately high mac- may fluctuate in depth with rainfall or have rophyte production and is probably indicative marine influence, e.g. Nauru, and Nomuka of the total gross primary production in these and Vava'u in Tonga. communities. The comparison with Tasek Bera in Malaya One consequence of the difference between (Furtado & Mori, 1982) is of particular interest Lepironia growth on the surface and edges of since the swamp system has similar water the floating islands is that the islands are likely chemistry, algal productivity, and a swamp to extend across the lake surface much more flora dominated by Lepironia artictdata and rapidly than they will accumulate peat vertical- Pandanus (P. helicopus Kurz). There are, ly- however, many differences; the Malayan The plant species at Tagimaucia crater are swamp is in the equatorial lowlands, with higher mostly natives (Table 2) and in this respect the temperatures and more seasonal rainfall, and swamp differs from swamps in Viti Levu which has a richer and taxonomicaliy more diverse contain a preponderance of recently intro- flora and fauna. Much of the floristic and duced species (Ash & Ash, 1984). Eleocharis, faunistic difference including the lower plant Lepironia and Sptiagnum are probably long diversity may be attributed to biogeographic established and are virtually restricted to factors, notably the relative isolation of the swamp habitats. Most of the other vascular Fijian islands. The Lepironia swamp at Tasek plants also occur in forest or disturbed dryland Bera resembles the marginal swamps at Tagi- habitats. Recent introductions have colonized maueia but Tasek Bera has no equivalent to the marginal swamps but have not established the floating sedge peat islands. on the floating islands, e.g. Clidemia hirta, Most of the water entering the Tagimaucia Mikania micrantha, Paspalum orbicutare and crater passes over or through the marginal Rhynchospora corymbosa. swamps and these apparently function as fil- For most aquatic animals without aerial life ters, removing much of the particulate material history stages, access to Lake Tagimaucia is from the water before it enters the lake. The only possible through Wainisairi Creek. In floating islands appear to receive much of their many places the creek has cut through over- surface water and nutrients from rainwater, lying scoria and into lava which it has worn and only the vegetation along the edges of the smooth. The 820 m drop to the sea is precipi- islands is in close proximity to the lake water. tous and is punctuated by waterfalls. The creek Vegetation patterns observed can be corre- may present an insurmountable barrier to lated with these differences in water and upstream migration and subsequent coloniza- nutrient supply, and while species sueh as tion of the lake and catchment by some aquatic Sphagnum ctispidatum indicate a rain-fed and organisms including kuhliid fishes and atyid nutrient poor water supply, species such as shrimps which are abundant downstream. Tagimaucia crater 519

Other aquatic taxa such as molluscs, annelids, and benthic algae, the fauna is sparse and the various crustaceans and gobioid fishes oceur in only vertebrate carnivore, an eel, was upland parts of Fiji and could probably ascend observed in a flooded part of the swamp where the creek to the lake, but were not observed. it may have been feeding on tadpoles. The reasons for their apparent absence are On a time scale of 10'*-l(h'' years most of the unknown but it seems likely that the relatively Fijian swamps and lakes are temporary and the low primary productivity of both swamp plants biota is therefore unable to persist at a particu- and phytoplankton, the low diversity of vascu- lar locality. Dispersal from sites in other parts lar plants, low calcium concentrations and low of Fiji or from further away is therefore a pH of the lake water are contributing factors. characteristic of the swamp and lake biota. SicyoptertJs micrurus Kaumans, an algal graz- Those few plants restricted to swamp habitats, ing goby with a marine larval stage, occurs such as Sphagnum ctispidatum and the sedges above the Wainisavulevu falls near Monasavu Lepironia articutata and Eteocharis dulcis have in central Viti Levu and the young fish are ranges extending far from Fiji. Several ende- known to utilize their pelvic sucker to traverse mic Fijian plants occur in the Tagimaucia the falls (Ryan era/,, 1979). Sicyopterttsmicrur- swamp, e.g. Geniostoma vitiense, Spiraeanthe- us is present in the lower reaches of the mum vitiense and Pandanus taveuniensis, but Wainisairi Creek and could probably reaeh these also oecur in forest habitats. A third Lake Tagimaucia but have not been recorded component of the biota is neither restricted to there. On the other hand, young eels are swamp or lake habitats nor endemic to Fiji, evidently able to make the journey from the e.g. Metrosideros collina. Rapanea myricifolia sea since an eel was observed in the flooded and Lycopodium cernuum, and recently intro- margins of the swamp. duced species which mostly also occur in While this study was restricted in scope by habitats disturbed by man. e.g. Rhynchospora the inaccessibility of the crater, and further corymbosa, Mikania micrantha and Clidemia studies may reveal a more diverse aquatic hirta. biota, the low diversity of both the flora and During the past century many species have fauna are striking. Several factors may limit been introduced to Fiji either accidentally or the biota. deliberately, and there are a number of earlier The area of swamp and lake habitat at aboriginal introductions. Natural dispersal to Tagimaucia crater is relatively small (213 ha) Fiji is limited by its oceanic location and, while and these habitats account for only 0.3% of other archipelagos may function as 'stepping the total land surface of the Fiji Islands. The stones', there are few swamps on these islands apparently low primary productivity of the to assist in the spread of swamp or lake swamp, together with the relatively small area, species. are likely severely to limit the food availability for, and population of, any large carnivores. During a total of 53 man-days at the crater the only vertebrate carnivores observed were one Acknowledgments eel, one boa and (repeatedly) a pair of swamp We thank the following: Mr Saula Vodonaiva- harriers. Cane toad tadpoles were seen fre- lu for identifying many of the plant specimens; quently and it seems probable that juvenile cane the Public Works Department and the Fiji toads form a substantial part of the carnivores' Meteorological Service for providing climatic diet. Cane toads were only introduced to records: Professor George Kennedy, Dr Geof- Taveuni within the last eentury and may have frey Hope and Mr Michael Cecil for assistance displaced the Fiji ground frog from the crater; if with field work and advice; Dr M. J. Winter- not, then the food web may have been even bourn and Dr B. Cowie for identification of simpler in the past. While tadpoles feed on aquatic invertebrates; and Miss Aisha AM and algae, adult cane toads feed on insects and may Mrs Helena Wolf for typing the manuscript. have had a significant impact on the insect fauna; This study was partly funded by the Austra- tnost juvenile cane toads probably migrate from lian National University (W.S.), and the Insti- the swamp into the surrounding forest. tute of Natural Resources, University of the Although the lake contains phytoplankton South Pacific (J.B. and P.R.). 520 W. Southern et al.

References Macioiek J, & Yamada R. (1981) Vai Lahi and other lakes in Tonga, Verhandtungen Internationale Vereinigung der Limnologie. 21, 693-698. American Public Health Association (1981) Siandard Marlier G, (1967) Ecological studies of some lakes in Methods for the Examination of Water and Waste the Amazon valley. Amazonia, I, 91-115, Water. 15th edn, American Public Health Associa- Parham J.W, (1972) Ptants of the Fiji Islands. 2nd tion. American Water Works Association, Water edn. Government Printer. Suva. Pollution Control Federation, Washington, D.C. Prescott G,W, (1979) How to Know ihe Freshwater Ash J. & Ash W. (1984) Freshwater wetland vegeta- Algae. W. C Brown & Co., Iowa. tion of Viti Levu Fiji, New Zealand Journat of Rodda P, (1976) Geology of northern and central Botany. 22, 371-396. Viti Levu, Buttetin of the Mineral Resources Beadle L.C. (1974) The Inland Waters of Tropical Division Fiji. 3. Africa: an Introduction to Tropical Limnology. Rodda P. (1984) The lake at the source of the Rewa. Longman. London, Domodomo. 2, 15-22. Brownlie G. (1977) The Pteridophyie Flora of Fiji. J. Ryan P., Hassall D.C. Choy S.C, Penn N. & Cramer, Vaduz. Ryland J.S, (1979) Biological Studies Conducted Burgis W.J. (1978) Case studies of lake systems at in Wainisavulevu Creek. Viti Levu, Institute of different latitudes: ihc iropics. The I^ke George Natural Resources Environmental Studies Re- ecosystem, Verhandltmgen Internationale Vereini- port 3. University of the South Pacilic. Suva, gung der Limnologie. 20, 1175-1181. Fiji. Carter G.S. (1955) The Papyrtis Swamps of Uganda. Seemann B.C (1862) Viti: an account of a Govern- W. Heffer and Sons, Cambridge, ment Mission to the Vitian or Fijian Islands Fritsch F.E. (1971) The Structure and Reproduction IH60-I86L 1973 reprint. Colonial History Series, of the Algae. Vols. I and 2. Cambridge Universi- Dawsons, Folkestone, ty Press. Sioli H. (1968) Hydrochemistry and geology in the Furtado J.I. & Mori S. (Eds.) (1982) Tasek Bera: Brazilian Amazon region. Amazonia, 1, Iftl-ITI. the ecology of a freshwater swamp, Monog- Smith A.C, (1979) Ftora Vitiensis Nova I. Pacific raphiae Biologicae. 47. Tropical Botanic Garden, Lawai, Hawaii. Ganf G.G. (1974) Phytoplanklon biomass and dis- Smith A.C. (1981) Flora Viiiensis Nova 2. Pacific tribution in a shallow eutrophic lake (Lake Tropical Botanic Garden, Lawai, Hawaii, George, Uganda). Oecologia, 16, 9-29. Tailing J.F. & Tailing I.B. (196.5) The chemical Garrels R, & Mackenzie F.T. (1971) Evolation of composition of African lake waters. Interna- Sedimentary Rocks. W. W. Norton Press, New tionale Revue der Gesamten Hydrohiologie, 50, York. 421-163. Gaudet J.J. & Melack J,M. (1981) Major ion Vareschi E. (1982) The ecology of Lake Nakaru chemistr>' in a iropical African lake basin. Fresh- (Kenya). Ill, Abiotic factors and primary pro- water Biology, II, 309-333. duction. Oecotogia (Berlin), 55, 81-101. Goodenough J.G, (1876) Journal of Commander Wetzel R.G. (1975) Limnology. W. B. Saunders & Coodenough during his last command as senior Co.. Philadelphia. officer on the Austratian station. H. S. King. Williams W,D, (Ed.) (1981) Proceedings of the first London, International Symposium in Athalassic (Inland) Ikusima 1. (1982) Emergent macrophytes. In Furta- Salt Lakes. Adelaide 1979. Hydrobiologia, 81 do J.I. & Mori S. (Eds.) Tasek Bera: the ecology and 82. of a freshwater swamp. Monographiae Biologi- cae, 47. 267-269. Johnson D.S. (1967) On the chemistry of fresh- waiers in southern Malaya and Singapore. Archiv fur Hydrobiologie, 63, 477-496. (Manuscript accepted 21 October 1985}