Ecological processes and trophic status of two low-alpine patterned mires, south-central South Island, New Zealand
Abstract
Abstract Vegetation and environmental patterns, and associated ecological processes, were quantified from 42 sites on several transects in each of two extensive (5 and 220 ha) low-alpine patterned mires in the same region of south-central New Zealand. Plant communities, as derived from multivariate analyses, were correlated with 15 physical and chemical environmental factors. Various measures of water availability and chemistry were consistently the most significant factors in relation to vegetation patterns in both mires. In the smaller mire, plant cover adjacent to pools, which were partly or completely drained through underground tunnels, dominated the overall correlations. The nutrient status of surface water had a consistent negative relationship with water availability. No consistent spatial or temporal patterns were found in the concentrations of Ca, Mg, K or Na, nor pH or conductivity in pool water. Evaporative enrichment of cations on the surface of both mires was noted, with levels consistently higher in surface than in adjacent pool water. The somewhat higher nutrient status in the smaller mire may be a result of the size and/or the amounts of run-off from the surrounding slopes onto the mire surface or through the underground pipe system. Hydrogen (D) and oxygen (18O) isotopic compositions in water from pools, the mire surface and below ground from the smaller mire, suggested that there was negligible mixing of evaporation-enriched surface water with groundwater. Differences in overall nutrient levels in the two mires were relatively small and indicative of mesotrophic or marginally ombrotrophic status for these mires. Although of international significance, the wetland complex currently has inadequate formal protection. Possible options are assessed.
Introduction
Ecological processes in wetlands are both dynamic and complex because of their reliance on many interacting environmental factors, notably the availability, source and chemistry of water inputs, as well as climatic and topographic factors at various scales. Any or all of these factors may affect the rate of plant growth and decomposition, and hence peat accumulation (Moore & Bellamy 1974; Lindsay 1995). The state of these environmental factors, together with a wetland's historical development, can have major effects on its nutrient status (Eurola et al. 1984; Almquist-Jacobson & Foster 1995), and these factors in turn may affect the composition and importance of plant species on a mire (Moore & Bellamy 1974).
A considerable literature exists on the widespread Northern Hemisphere subarctic patterned wetland systems, their ecology and classification. These wetland types are not only generally rare in the Southern Hemisphere (see Mark et al. 1995 for review) but relatively little is known of their trophic status, ecological processes or the associated environmental gradients operating within them. In New Zealand, the only overview classification of mires has been made by Dobson (1979), in which patterned systems are mentioned only briefly. However, the study by Mark et al. (1995), which described a significant set of patterned, low-alpine wetlands in New Zealand, provides the basis for the comprehensive study reported in the present paper. More limited adjunct studies by Cullen (1996) and de Groot (1999) provided additional background information. Our aim in the present paper is to address two main questions. First, what are the hydrological processes operating in relation to the trophic status of these wetlands and, second, to what extent are their plant communities influenced by differences in water availability and chemistry?
Methods
Study area
Roaring Lion and Dome Mires (45°22′S), collectively referred to as the Nokomai mires, are extensive (approximately 220 and 5 ha, respectively), low-alpine patterned wetlands. They are located about 1.5 km apart at an elevation of approximately 1350 m a.s.l., and lie within a much larger mire complex. This complex occupies several square kilometres on the uplands of the southern Garvie Mountains (see Mark et al. 1995; 1, 2) in south-central South Island, New Zealand. The climate of the area is cool and perhumid with frequent fogs, with an estimated mean annual air temperature of approximately 3.5°C and estimated annual precipitation of approximately 1300 mm. The annual excess of precipitation over evapotranspiration for this location has been estimated at approximately 900 mm (McGlone et al. 1995).
Map of the southern end of Dome Mire drawn from an aerial photograph, showing (▪) the position of the data logger and (•) the 12 study sites along two transects bisecting the wetland (A–E, 0–6). The completely drained pool (see Mark et al. 1995; fig. 15) is located between Sites DB 0, 1 and A, whereas the pools north and south of site DB 2 are partly drained. (- - -) Contour lines at 1-m intervals. Site names are defined in the text.
Map of the upper part of Roaring Lion Mire drawn from an aerial photograph, showing (▪) the position of the data logger and (•) the 30 study sites arranged along one main and four lateral transects (1a,b; 2a–e, m–o; 3a–e, m–o; 4a,b; 5a–d; 6a–d, 7a,b). (- - -), Approximate contours at 1-m intervals are shown in relation to site RL 1a being at 1350 m a.s.l. Values for calcium concentration in 31 representative pools and in the stream bisecting the wetland, as measured on one or two (italics) occasions, are also shown. Islands within pools have not been shown at this scale.
Both mires occupy gently sloping (0–8°), broad interfluves between opposite draining catchments. The string-flark patterning of Dome Mire (note, however, that this wetland is not domed) is somewhat less distinctive than that of the nearby, larger Roaring Lion system. Dome Mire has fewer, less regular pools, some of which have drained or partially drained (see fig. 15 in Mark et al. 1995) through natural underground pipes in recent times.
The vegetation history of these wetland systems, as described by McGlone et al. (1995, 1997), indicates that the dominance of Sphagnum and a consequent rapid phase of peat development initiated these wetlands only approximately 700 years ago.
Sampling methods
Forty-two study sites were established in January 1996 along a series of transects traversing the two wetlands, both downslope and across several contours. Six sites were positioned 40–100 m apart across the long axis of the 5-ha Dome Mire (varying slightly in relation to pool location) and another six sites approximately 10 m apart on the short axis (again varying in relation to pool location: see Fig. 1). Seven sites were placed approximately 30 m apart along the central transect line studied by Mark et al. (1995) on the approximately 220-ha Roaring Lion mire with a second central series of sites established parallel and 10 m distant (Fig. 2). Secondary transects were established at right angles to these central ones, at four of the seven sites (2, 3, 5 and 6), with sites here also approximately 30 m apart (Fig. 2).
A piezometer was installed at each of the 42 study sites. These were constructed from 4-cm diameter PVC plastic pipes 1.2 m in length, with numerous fine transverse slits cut along the lower 50 cm. The pipes were capped at the base and were driven in full length or to the base of the peat, into the space left after the extraction of a peat core to be used in an associated study (Chagué-Goff et al., unpubl. data). The pipes were cut off close to ground level if necessary, then capped on top and labelled.
Water levels in all the piezometers were measured twice in January 1996, before and after a heavy rain event, and following periods of fine weather in December 1996, January 1998 and December 1998. Samples of pool water, where available within 5 m of the sites, were collected in December 1996 (at Dome Mire), in January 1998 (both mires) and December 1998 (Roaring Lion Mire). The same pools plus streams at the lower end of the study area in each mire were also sampled for colour in October 1999.
Samples of surface water were also collected (in duplicate) by depressing the surface of the mire at 40 of the 42 piezometer sites (one sample approximately 1 m on both sides of the tube). No surface samples could be obtained from two Dome Mire sites (DB 1 and DB A) adjacent to a drained pool. All water samples were acidified in the field with ultra-pure nitric acid, stored in acid-washed 125-mL polyethylene bottles and, on return to the lab, filtered using a 0.45-µm acetate membrane prior to analysis.
In order to assess possible sources of water and its pattern of movement through Dome Mire, samples were collected for stable isotope analyses of oxygen (18O) and hydrogen (D) from pools, surface water and groundwater near site DB 0 in December 1996, May 1997 and January 1998. The samples were stored in 20-mL screw-top glass bottles until analysed.
The elevation of each piezometer site was determined using a surveyor's automatic level and staff, and expressed relative to the highest study site on each mire (assumed to be 1350 m in both cases, based on the 1:50 000 Topomap F43 Garvie). Contours at 1-m intervals were sketched for both study areas from these data (see 1, 2).
Plant composition and cover were measured at each of the 42 study sites with four 0.5 m × 0.5 m quadrats, arranged in a line normal to the slope and centred 75 and 175 cm on each side of a piezometer. Each quadrat was divided into 25 10 cm × 10 cm squares to record frequency and allow a more accurate estimation of cover of each species, as well as surface water, bare ground and dead plants. Peat samples to a depth of 10 cm were collected in 1998; duplicate samples were taken adjacent to each piezometer, stored in sealed polythene bags and refrigerated within 5 days of collection until analysed for water and organic content.
Data loggers (Campbell model CR-10) were installed adjacent to a typical pool on Dome Mire (Fig. 1) and next to a large pool at the upper end of Roaring Lion Mire (Fig. 2). These recorded rainfall (tipping bucket gauge), humidity (screened Vaisala Humitter 50Y sensor at 1.4 m), shortwave radiation (350–3500 nm; Li-Cor pyranometer at 1.4 m), air temperature (shielded thermocouple at 1.4 m), plus peat temperatures at 5 cm, 50 cm and 100 cm depths (thermocouples). Water level (ISO Instrument Services and Development Type SS1) was measured in an adjacent pool. At both sites, records were obtained from mid-May 1997 until September 1997 at Roaring Lion Mire and October 1999 at Dome Mire and analysed on a daily basis (09.00 hours, apart from radiation).
Analyses of water and peat
All water samples were analysed for dissolved calcium, magnesium, potassium and sodium by Atomic Absorption Spectrometry (AAS). The colour of the pool and stream water was measured by a standard spectrophotometric method, which can detect small differences in dissolved organic matter (Cuthbert & Del Giorgio 1992).
The δ18O analyses were carried out by isotopically equilibrating the water with carbon dioxide gas at 29°C, and analysing the carbon dioxide using a mass spectrometer (Brenninkmeijer & Morrison 1987). The δD was determined by quantitatively converting the water to hydrogen gas by reaction with zinc at 500°C in individual reaction tubes (Stanley et al. 1984), then the hydrogen was isotopically analysed using a M602C mass spectrometer. The isotope ratios are reported using the usual delta (δ) notation relative to V-SMOW (the international water standard; Vienna Standard Mean Ocean Water; Gonfiantini 1978). The standard deviation of the measurements are ±0.1% for δ18O and ±1% for δD. The results were normalized assuming δ18O = −55.5?? and δD = −428?? for SLAP (the second international water standard; Standard Light Antarctic Precipitation) relative to V-SMOW. Comparative isotopic values for rainwater were obtained from the contemporary records of Ingraham and Mark (2000) for three upland sites in central and eastern Otago.
Duplicate peat samples were dried for 16 h at 90°C and water content expressed as a percentage of the field sample weight, then the organic content was determined by loss-on-ignition (2 g dried peat combusted at 550°C for 16 h), and expressed as a percentage of the dry weight.
Analyses of vegetation and environmental data
Vegetation data were entered into the ecological data handling package decoda (Minchin 1994). The percentage cover information from both mires was then classified using the polythetic, divisive, two-way indicator species analysis (twinspan) technique, which orders both the sample and species data into hierarchies or groups of similarity (Hill 1979a; Gauch 1982). Each division has an eigenvalue that indicates the distinction of the particular groupings. Default options in the package, including those for the creation of pseudospecies (1 = <2%; 2 = 2 − <5%; 3 = 5 − <10%; 4 = 10 − <20%; 5 = ≥20%) were used. The vegetation data were also ordinated using detrended correspondence analysis (DCA; Hill & Gauch 1980), using strict convergence criteria as recommended by Oksanen and Minchin (1997). Detrended correspondence analysis (available as decorana; Hill 1979b; Hill & Gauch 1980) produces ordinations of both species and samples on the basis of weighted averages. Eigenvalues show the amount of variance explained by each axis of the ordination diagram, although they are only a relative, not an absolute measure, as only the first four axes are calculated (Whittaker 1987). Separate analyses were made of the data for each of the two mires as well as those from all 42 sites embracing both mires. Each taxon was assigned a life form according to Raunkaier's (1934) system subdivided following Tivy (1993) and Mark et al. (2000). Nomenclature for indigenous taxa follows Johnson and Brooke (1989), Poole and Adams (1990), and Mark and Adams (1995) for all vascular plants, Beever et al. (1992) for mosses and Galloway (1985) for lichens. Voucher specimens are deposited in the Victoria University of Wellington Herbarium (WELTU).
The relationships associated with each of the 15 environmental factors measured at each of the 42 study sites were assessed using the vector-fitting technique, which determines the direction (vector) through the ordination that has the maximum correlation with each variable, plus the significance of each correlation (see Bowman & Minchin 1987; Kantvilas & Minchin 1989). The length of each arrow in a vector diagram indicates the rate of change in a certain direction, with each arrow directed towards its maximum change in relation to the community or species variations (Ter Braak 1987). Actual values for elevation were used in the vector fitting for the individual mires but, for the analysis of both mires, relative values for the elevation range were used. The 15 environmental variables used in the vector fitting with the sample and species ordinations, based on all 42 sites were: (i) elevation; (ii) two physical factors of the peat (water and organic contents); (iii) water table depth (as recorded on each of five occasions); and (iv) seven chemical attributes of the surface water at 40 (of the 42) sites (pH, conductivity, Ca, Mg, K, Na and Ca/Mg ratio).
To further characterize the environments of the two mires, the same chemical attributes were measured for the samples of pool water and the stream draining the centre of the Roaring Lion Mire. In addition, the data logger records were downloaded onto spreadsheets and manipulated to derive mean and extreme monthly values to characterize the environments of the two mires.
Results
Vegetation patterns
Seven communities or quadrat groups (QG 1–7) and eight species classes (SC A–H) were selected from the 42 sites, both at level three of the classification (Fig. 3). The list of taxa, with their particular species class and life form is shown in Table 1. Of the 56 taxa sampled in the two mires (21 dicotyledons, 20 monocotyledons, 12 mosses, three lichens), plus ‘other bryophytes’, there was only one exotic, sweet vernal grass, Anthoxanthum odoratum L., which was recorded only from Dome Mire. Among the life forms, hemicryptophytes predominated in all seven communities and species classes, with caespitose and rosette forms, as well as bryophytes, sharing importance (Table 1).
Two-way table from the twinspan classification, including dendrograms for the species and quadrats, each to level three (8 species classes and 7 communities) based on the cover data from 42 sites on both Roaring Lion and Dome Mires (the Nokomai mires). Indicator species and eigenvalues are shown for the quadrat classification. The upper values in each cell correspond to per cent presence (i.e. 100% implies all taxa in the species class occur in all quadrats in that quadrat group). The lower values (in italics) in each cell give the percentage of the total possible cover. The total sites out of the 42 in each community (quadrat group) are also shown, together with the number contributed from Dome Mire (DB) and Roaring Lion (RL) Mires. Number of taxa in each Species Class are shown in parentheses. Species names are abbreviated: see Table 1 for full names.
| Taxa | Species class | Life form | Presence | Community | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| DB | RL | 1 | 2 | 3 | 4 | 5 | 6 | 7 | |||
| Anthoxanthum odoratum * | A | Hc | + | – | – | – | – | 33 (0.30) | – | – | – |
| Calligeron sarmentosum | A | Hb | + | + | 33 (0.30) | 40 (0.80) | 67 (1.33) | 67 (0.67) | – | – | – |
| Campylopus sp. | A | Hb | + | – | – | – | 5 (0.05) | 33 (1.00) | – | – | – |
| Carpha alpina | A | Hc | + | – | – | – | – | 33 (0.67) | – | – | – |
| Dicranoloma obesifolium | A | Hb | + | – | – | – | 5 (0.05) | 67 (1.00) | – | – | – |
| Drepanocladus aduncus | A | Hb | – | + | – | – | 5 (0.05) | – | – | – | – |
| Epilobium atriplicifolium | A | Chh | – | + | – | – | 5 (0.05) | – | – | – | – |
| Poa kirkii | A | Hc | – | + | – | – | – | 33 (0.30) | – | – | – |
| Plantago uniflora | A | Hr | – | + | 33 (0.30) | – | 10 (0.19) | – | – | – | – |
| Sphagnum falcatulum | A | Hb | + | + | – | – | 52 (1.40) | 33 (0.30) | – | – | – |
| Sphagnum squarrosum | A | Hb | + | + | – | 10 (0.10) | 62 (1.10) | 67 (2.00) | – | – | – |
| Aulacomnium palustre | B | Hb | – | + | – | 10 (0.10) | 10 (0.10) | – | – | – | – |
| Anisotome flexuosa | B | Hr | – | + | – | 10 (0.10) | – | – | – | – | – |
| Carex sinclairii | B | Hc | – | + | – | 40 (1.50) | 5 (0.19) | – | – | – | – |
| Centrolepis pallida | B | Hm | + | – | 33 (0.33) | – | – | – | – | – | – |
| Cladonia sp. | B | Hl | – | + | – | 10 (0.10) | – | – | – | – | – |
| Dracophyllum prostratum | B | Cht | + | + | 67 (1.67) | 80 (1.50) | 14 (0.33) | – | – | – | – |
| Gaultheria parvula | B | Chh | – | + | 67 (1.33) | – | 5 (0.05) | – | – | – | – |
| Isolepis aucklandica | B | Hc | + | + | 67 (0.33) | 90 (1.70) | 90 (1.24) | – | – | – | – |
| Kelleria paludosa | B | Chh | + | + | 67 (1.33) | 60 (0.90) | 57 (0.67) | 100 (1.00) | – | – | 50 (0.50) |
| ‘Other bryophytes’ | B | Hb | + | + | 100 (2.00) | 90 (1.40) | 90 (2.57) | 100 (2.67) | 100 (2.00) | 100 (1.00) | – |
| Phyllachne colensoi | B | Hm | + | + | – | 90 (1.50) | 43 (0.57) | 67 (0.67) | – | – | – |
| Polytrichum longisetum | B | Hb | – | + | – | 30 (0.33) | – | – | – | – | – |
| Rostkovia magellanica | B | Hc | + | + | 33 (1.33) | 10 (0.10) | – | – | – | – | – |
| Dicranaloma billardierei | C | Hb | + | + | 67 (0.67) | 60 (1.30) | 62 (1.62) | 33 (0.67) | 50 (1.00) | – | 50 (0.50) |
| Gentiana amabilis | C | Hr | + | + | 100 (1.00) | 100 (1.10) | 95 (1.52) | 100 (1.00) | 100 (1.50) | 100 (1.00) | – |
| Sphagnum cristatum | C | Hb | + | + | 100 (4.33) | 100 (0.94) | 100 (4.67) | 100 (2.25) | 100 (4.00) | 100 (2.00) | 100 (1.00) |
| Abrotanella caespitosa | D | Hr | + | + | 100 (3.33) | 100 (2.00) | 86 (1.38) | 100 (2.33) | 100 (2.50) | – | 100 (1.00) |
| Carex gaudichaudiana (+ Carex lachenalii ssp. parkeri) | D | Hc | + | + | 100 (4.00) | 100 (4.20) | 100 (4.43) | 100 (1.33) | 100 (1.50) | 100 (2.00) | 50 (0.50) |
| Celmisia alpina | D | Hr | + | + | 100 (1.33) | 90 (0.90) | 95 (1.00) | 100 (1.33) | 100 (1.50) | 100 (2.00) | 50 (0.50) |
| Deschampsia novae-zelandiae | D | Hc | + | + | 33 (0.33) | 20 (0.20) | 33 (0.33) | 33 (0.33) | 100 (1.00) | 100 (1.00) | – |
| Epilobium tasmanicum | D | Chh | + | – | – | – | 5 (0.10) | – | – | – | 50 (0.05) |
| Euphrasia dyeri | D | Th | + | + | 67 (0.67) | 10 (0.10) | 24 (0.29) | – | 50 (0.50) | – | – |
| Luzula leptophylla | D | Hc | + | + | 100 (1.00) | 70 (0.80) | 76 (1.00) | 67 (0.67) | 50 (1.00) | 100 (2.00) | 50 (0.50) |
| Oreobolus pectinatus | D | Hm | + | + | 67 (1.67) | 90 (3.10) | 81 (2.24) | 100 (4.67) | 100 (4.00) | 100 (4.00) | 100 (2.50) |
| Coprosma perpusilla | E | Chh | + | – | 33 (0.33) | – | – | 100 (1.67) | 50 (0.50) | 100 (1.00) | 50 (0.50) |
| Drosera arcturi | E | Hr | + | + | 33 (0.33) | 30 (0.30) | 43 (0.52) | 100 (1.33) | 100 (1.00) | 100 (1.00) | 100 (2.00) |
| Bryum pseudotriquetrum | F | Hb | + | – | – | – | 10 (0.14) | – | – | – | 50 (1.00) |
| Oreomyrrhis sp. ‘bog’ | F | Hr | + | + | 67 (0.67) | 40 (0.40) | 71 (0.76) | – | 100 (1.00) | 100 (3.00) | 50 (0.50) |
| Polytrichum commune | F | Hb | + | + | 100 (1.00) | 50 (0.50) | 81 (0.95) | 67 (0.50) | 100 (2.00) | 100 (4.00) | 100 (3.50) |
| Agrostis muelleriana | G | Hc | + | – | – | – | – | – | – | – | 50 (0.50) |
| Carex echinata | G | Hc | + | – | – | – | 5 (0.05) | – | 50 (0.50) | 100 (1.00) | – |
| Chionochloa macra | G | Hc | + | + | – | 10 (0.10) | 5 (0.10) | – | 5 (0.50) | 100 (1.00) | – |
| Cladonia fimbriata | G | Chl | + | + | 33 (0.33) | – | 5 (0.05) | – | – | – | 100 (1.50) |
| Coprosma atropurpurea | G | Chh | + | + | – | – | 19 (0.19) | 30 (0.33) | 50 (0.50) | – | 100 (2.00) |
| Deyeuxia avenoides | G | Hc | + | – | – | – | 5 (0.05) | 33 (0.33) | 50 (1.00) | 100 (1.00) | – |
| Epilobium komarovianum | G | Chh | + | – | – | – | 5 (0.05) | – | – | 100 (1.00) | – |
| Gnaphalium mackayi | G | Hr | + | – | – | – | – | – | 100 (1.00) | 100 (2.00) | 100 (4.50) |
| Juncus antarcticus | G | Hc | + | – | – | – | – | – | – | – | 50 (0.50) |
| Lechanora broccha | G | Chl | + | – | – | – | – | – | – | – | 50 (0.50) |
| Plantago novae-zelandiae | G | Hr | + | – | 33 (0.67) | – | 10 (0.10) | 67 (1.00) | 50 (1.50) | 100 (1.00) | – |
| Poa incrassata | G | Hc | + | – | – | – | – | – | – | – | 50 (0.50) |
| Psychrophila obtusa | G | Hr | + | + | – | 20 (0.20) | 19 (0.29) | 67 (0.67) | 100 (3.00) | – | 100 (2.00) |
| Ranunculus gracilipes | G | Hr | + | + | 33 (0.30) | – | – | – | 100 (1.50) | – | – |
| Rytidosperma australe | G | Hc | + | – | – | – | – | – | – | 100 (2.00) | – |
| Poa colensoi | H | Hc | + | + | 100 (2.67) | 50 (0.50) | 52 (0.67) | 67 (1.67) | 100 (2.50) | 100 (1.00) | 100 (4.00) |
| Total taxa | 46 | 36 | 26 | 28 | 39 | 27 | 23 | 19 | 23 | ||
- Per cent frequency values and mean pseudospecies score in parentheses (see text for details) are given for each of the seven communities differentiated from the twinspan analysis of data from both wetlands. The life forms represented are: chamaephytes, plants with perennating buds up to 25 cm above ground level (Cht, trailing woody chamaephytes; Chh, herbaceous chamaephytes; Chl, chamaephyte lichen); hemicryptophytes, herbs or subshrubs with perennating buds at ground level (Hc, caespitose hemicryptophytes; Hr, rosette hemicryptophytes; Hm, mat hemicryptophytes; Hb, hemicryptophyte bryophyte); Th, annuals. *Exotic species.
Three of the communities (5, 6 and 7) were confined to Dome Mire, where they were located with increasing distances from one large, and two partially drained pools (see fig. 15 in Mark et al. 1995). Gnaphalium mackayi was restricted to these relatively dry sites and is the indicator species for these three communities (Fig. 3). Its importance increased with increasing dryness of the site, as indicated by its mean pseudospecies scores, which ranged from 1.00 to 4.50 (Table 1). Gnaphalium mackayi dominated community 7 (mean score ≥ 4.50) along with Poa colensoi (4.00). Polytrichum commune (3.50), Oreobolus pectinatus (2.50), Drosera arcturi, Psychrophila obtusa and Coprosma atropurpurea (2.00) were the only other taxa with a mean cover of =2% among the 23 recorded in this community (Table 1).
The mat-forming Oreobolus pectinatus and the moss Polytrichum commune dominated community 6 (pseudospecies mean scores = 4.00) with less of Oreomyrrhis sp. (3.00), Sphagnum cristatum, Gnaphalium mackayi, and the graminoids Carex gaudichaudiana/C. lachenalii, Luzula leptophylla and Rytidosperma australe (all 2.00) among the 19 taxa recorded in this community. Community 5, with 23 taxa, was dominated by Oreobolus pectinatus and Sphagnum cristatum (both 4.00) with Psychrophila obtusa (3.00), Abrotanella caespitosa, Poa colensoi (2.50) and Polytrichum commune having scores of ≥2.00.
The remaining four communities were all floristically richer than the three from the relatively dry sites, with community 4 (27 taxa) dominated by Oreobolus pectinatus (mean score = 4.67). Only three other species achieved scores of ≥2.00: Abrotanella caespitosa (2.33), Sphagnum cristatum and S. squarrosum (2.00), apart from ‘other bryophytes’ (2.67). The richest community, community 3, which had 39 taxa, was dominated by Sphagnum cristatum (4.67) and Carex gaudichaudiana/C. lachenallii (4.43), with only O. pectinatus (2.24) and ‘other bryophytes’ (2.57) achieving scores of ≥2.00. Community 2 had 28 taxa, with Carex gaudichaudiana/C. lachenalii (4.20), O. pectinatus (3.10) and Abrotanella caespitosa (2.00) being the only taxa with scores of ≥2.00. Community 1, with slightly fewer taxa (26), was also dominated by these same species of Carex (4.00) plus S. cristatum (4.33), with less of Abrotanella caespitosa (3.33), Poa colensoi (2.67) and ‘other bryophytes’ (2.00; see Table 1).
Results of the DCA analysis of all 42 sites (Fig. 4) confirmed the general distinction of five of the 12 Dome Mire sites (communities 5, 6 and 7). These all had relatively high scores on axis 1. There was considerable overlap in the axis 1 scores of the Roaring Lion Mire and the remaining Dome Mire sites, although the former tends toward lower scores on this axis. There are only weak patterns apparent along axis 2 (Fig. 4).
Two-dimensional ordination of the 42 sampling sites on Dome Mire (12 sites) and Roaring Lion Mire (30 sites) based on detrended correspondence analysis of the per cent cover values. The symbols refer to the seven communities from Fig. 3: (), 1; (), 2; (), 3; (×), 4; (), 5; (), 6; (), 7. The site numbers are shown adjacent to each symbol. Envelopes enclose each of the seven communities. Results of the vector fitting for the 14 significant environmental factors of the 15 measured at each sampling site (excluding pool water) are also included (top right). Cond, conductivity; orgmat, organic matter; moist, moisture content; w, water level for the month and year shown; *P < 0.05; **P < 0.01; ***P < 0.001. The relative length of the vector projections indicate the true direction of the vector in 3-dimensional space defined by the first three axes, of which only two (1 and 2) are shown.
Environmental factors
Elevation ranged over 9.73 m across the 30 study sites at the upper end of the Roaring Lion Mire but only 4.00 m over the 12 sites that embrace much of Dome Mire (1, 2). The water content of the near-surface peat (0–10 cm) across the 42 study sites varied from 85% of the field weight (community 7, DB 1) to 96.2% (community 3, DB 5), the organic contents varied from 87.8% (community 1, DB E) to 95.7% of dry weight (community 3, DB 5), and water table depths (measured on five occasions) ranged from +2.3 cm (community 3, RL 6a) to −45.7 cm (community 7, DB A).
The results of the chemical analyses of: (i) surface water from the two mires for the 40 sites where it could be obtained; (ii) water in each of the 31 pools that were considered to be ‘adjacent’ to the 42 study sites; plus (iii) the stream draining the centre of Roaring Lion Mire are summarized according to the relevant plant communities in Table 2. Nutrient concentrations of water in the small stream near the centre of Roaring Lion Mire were generally higher than the values for pool water on this mire, whereas values for pH, as expected, varied in relation to nutrient levels (Table 2).
| Community | n | pH | CD | Ca | Mg | K | Na | Ca/Mg | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2 | S | 1 | 2 | S | 1 | 2 | S | 1 | 2 | S | 1 | 2 | S | 1 | 2 | S | 1 | 2 | S | ||
| Dome Mire | ||||||||||||||||||||||
| 1 | 1 | 4.59 | 4.34 | 4.19 | 8.80 | 8.00 | 24.60 | 0.22 | 0.14 | 2.61 | 0.15 | 0.09 | 1.88 | 0.06 | 0.08 | 2.47 | 0.98 | 0.67 | 1.44 | 1.47 | 1.56 | 1.45 |
| – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | ||
| 3 | 3 | 4.85 | 4.47 | 4.17 | 7.43 | 9.35 | 21.07 | 0.30 | 0.13 | 2.33 | 0.17 | 0.09 | 1.79 | 0.06 | 0.06 | 1.26 | 0.82 | 0.72 | 1.50 | 1.92 | 1.75 | 1.32 |
| (0.17) | (0.42) | (0.17) | (0.88) | (0.29) | (3.70) | (0.04) | (0.02) | (0.02) | (0.04) | (0.02) | (0.14) | (0.01) | (0.00) | (0.06) | (0.06) | (0.02) | (0.55) | (0.27) | (0.61) | (0.10) | ||
| 4 | 3 | 4.93 | 4.57 | 4.07 | 8.00 | 7.90 | 36.07 | 0.26 | 0.14 | 4.85 | 0.17 | 0.01 | 9.36 | 0.07 | 0.06 | 3.18 | 0.89 | 0.72 | 1.87 | 1.59 | 1.53 | 1.47 |
| (0.21) | (0.41) | (0.09) | (0.00) | (1.07) | (11.48) | (0.03) | (0.02) | (1.42) | (0.02) | (0.01) | (7.75) | (0.01) | (0.02) | (1.44) | (0.02) | (0.02) | (0.58) | (0.02) | (0.27) | (0.47) | ||
| 5 | 2 | 5.35 | 5.07 | 4.55 | 8.15 | 8.00 | 14.90 | 0.43 | 0.19 | 15.30 | 0.21 | 0.12 | 4.60 | 0.05 | 0.03 | 1.68 | 0.88 | 0.70 | 1.14 | 2.02 | 1.64 | 3.33 |
| (0.18) | (0.25) | – | (0.05) | (1.00) | – | (0.02) | (0.01) | – | (0.00) | (0.02) | – | (0.01) | (0.02) | – | (0.11) | (0.09) | – | (0.12) | (0.26) | – | ||
| 6 | 1 | 4.83 | 4.53 | 4.37 | 11.30 | 10.00 | 65.70 | 0.61 | 0.31 | 40.20 | 0.28 | 0.21 | 11.40 | 0.05 | * | 16.50 | 1.22 | 0.82 | 6.30 | 2.18 | 1.48 | 3.59 |
| – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – | ||
| All Sites | 4.95 | 4.63 | 4.20 | 8.30 | 8.49 | 30.73 | 0.34 | 0.17 | 8.85 | 0.19 | 0.11 | 5.70 | 0.06 | 0.05 | 3.77 | 0.91 | 0.72 | 2.11 | 1.82 | 1.60 | 1.86 | |
| (0.11) | (0.17) | (0.07) | (0.45) | (0.43) | (5.86) | (0.04) | (0.02) | (3.95) | (0.02) | (0.01) | (2.45) | (0.01) | (0.01) | (1.58) | (0.04) | (0.02) | (0.55) | (0.10) | (0.15) | (0.32) | ||
| Roaring Lion Mire | ||||||||||||||||||||||
| 1 | 2 | 4.62 | 4.40 | 4.02 | 7.30 | 8.40 | 21.05 | 0.26 | 0.26 | 0.66 | 0.15 | 0.13 | 0.42 | 0.04 | 0.02 | 0.58 | 0.54 | 0.52 | 0.79 | 1.73 | 2.00 | 1.56 |
| – | – | (0.14) | – | – | (0.45) | – | – | (0.26) | – | – | (0.05) | – | – | (0.28) | – | – | (0.03) | – | – | (0.41) | ||
| 2 | 10 | 4.34 | 4.22 | 3.97 | 9.05 | 9.07 | 26.06 | 0.25 | 0.27 | 1.06 | 0.15 | 0.10 | 0.68 | 0.05 | 0.11 | 1.52 | 0.80 | 0.91 | 1.15 | 1.70 | 2.67 | 1.54 |
| (0.07) | (0.05) | (0.04) | (0.09) | (0.59) | (2.44) | (0.03) | (0.03) | (0.12) | (0.02) | (0.01) | (0.06) | (0.01) | (0.02) | (0.24) | (0.03) | (0.08) | (0.14) | (0.09) | (0.24) | (0.11) | ||
| 3 | 18 | 4.43 | 4.26 | 3.98 | 8.51 | 9.35 | 23.34 | 0.29 | 0.24 | 1.48 | 0.16 | 0.11 | 0.72 | 0.01 | 0.09 | 1.41 | 0.78 | 0.84 | 1.01 | 1.84 | 2.24 | 2.00 |
| (0.05) | (0.05) | (0.04) | (0.39) | (0.55) | (1.55) | (0.03) | (0.02) | (0.32) | (0.01) | (0.01) | (0.09) | (0.01) | (0.01) | (0.27) | (0.05) | (0.05) | (0.08) | (0.10) | (0.18) | (0.19) | ||
| All sites | 4.41 | 4.25 | 3.98 | 8.66 | 9.21 | 24.10 | 0.27 | 0.25 | 1.28 | 0.15 | 0.11 | 0.68 | 0.07 | 0.09 | 1.39 | 0.77 | 0.85 | 1.04 | 1.78 | 2.37 | 1.82 | |
| (0.05) | (0.04) | (0.04) | (0.51) | (0.49) | (1.60) | (0.02) | (0.02) | (0.26) | (0.01) | (0.01) | (0.08) | (0.01) | (0.02) | (0.23) | (0.04) | (0.05) | (0.09) | (0.08) | (0.18) | (0.17) | ||
| RL st. | – | 4.61 | – | – | 10.40 | – | – | 0.55 | – | – | 0.18 | – | – | 0.15 | – | – | 0.84 | – | – | 3.06 | – | |
- Values have been averaged on the basis of community type (1–7, except that no samples could be obtained for the relatively dry community 7). Values are also included for one sample from the stream adjacent to site RL 6 (community 3) in Roaring Lion Mire (RL st.). 1, sampled January 1998; 2, sampled December 1998 (Roaring Lion) and December 1996 (Dome Mire). Surface water (S) sampled December 1998 (Roaring Lion) and January 1998 (Dome Mire). Number of samples (n) for each community type are shown and, where appropriate, standard errors of means have been included (in parentheses). Conductivity (CD) is expressed as µS cm−3 and the four cations as mg L−1. *Value below level of detection; – for RL st. indicates that no water was collected.
Pool water had lower cation levels and conductivity values than water obtained from the mire surface. In contrast, pH values were consistently higher in the pool water. In general, nutrient concentrations and pH values were somewhat higher for surface water from Dome Mire than in that from Roaring Lion Mire, including water from the two communities (1 and 3) that were common to both wetlands. For Dome Mire, the pool adjacent to community 6 and closest to the partially and completely drained pools had the highest concentrations of Ca, Mg and Na, and the highest overall conductivity, consistent with the nutrient pattern in the surface water. Concentrations of K and the Ca/Mg ratio were both highest in the surface water at this same site. The Ca/Mg ratios, however, did not differ consistently between the surface and pool waters or between the two wetlands. Ionic concentrations of the surface water were inversely related to the moisture content of the peat at the same site, which, in turn, was related to the level of the water table (data not shown but see Fig. 4).
The results of all factors recorded with the data loggers in both mires from 17 May to 30 September 1997 were generally similar between the two mires. Only those data for the most recent 1-year period of records (November 1998 to October 1999 inclusive) on Dome Mire are presented here. Pool water level apparently varied by 480 mm over this period (Fig. 5) but the records for June to October, when the pool would have been covered with ice and snow, are likely to be anomalously high. Over the snow-free period, pool levels clearly responded to precipitation. The level rose 240 mm from its minimum recorded level over a 2-day period in early February 1998, and in response to two larger rainfall events in March 1998, when the inital pool water level was higher. The consequent pool level was similar in all three rainfall events (Fig. 5) and this was presumably the level of its sill. The presence of macrophyte (Myriophyllum sp.) deposits above the sill at the leeward (south-east) end of several pools attests to the periodic strength of north-west winds in the area.
Relationship between daily precipitation and water level in the pool adjacent to the rain gauge at Dome Mire for the period November 1998 – October 1999.
Recorded precipitation was higher in 1997–1998 (1757 mm) than in 1998–1999 (1160 mm), with the mean value being 1458 mm. Temperature values for air and three depths in the peat showed the expected seasonal progression, as well as cooling and heating effects that were both reduced and increasingly delayed with depth (Fig. 6). Freezing did not occur at the two lower depths (extreme minima 0.5°C at 50 cm; 1.7°C at 100 cm) and only rarely did temperatures drop below −1°C at 5 cm depth in the peat (extreme minimum −2.3°C). Air temperature extremes ranged from 22.7°C to −12.3°C at Dome Mire, whereas monthly means ranged from 10.5°C (January) to −1.6°C (July), with an annual mean of 4.0°C.
Temperature values (extreme maximum, mean maximum, mean, mean minimum, extreme minimum) for air (+1.4 m), and at three depths in peat (−5 cm; −50 cm; −100 cm) at Dome Mire for the period November 1998 – October 1999.
Shortwave radiation is shown for 1998–1999 in Fig. 7. The extreme and mean daily maxima, plus the mean daily values based on the daylight hours show the expected seasonal trends. Daily peaks reached 1063 W m−2 in midsummer, whereas mean daily maxima ranged from 890 to 295 W m−2 over the 12-month period. Mean daily values, however, were much lower (63–278 Wm−2) and probably reflect the high frequency of fog in the study area. Humidity values of less than 70% mean daily humidity prevailed for less than 20% of the days in each month (Fig. 8).
Values for incoming shortwave radiation recorded at Dome Mire for the period November 1998 – October 1999. Values shown for each month are the extreme maximum, mean daily maximum and mean daily values.
Mean monthly air humidity recorded at Dome Mire for the period November 1998 – October 1999. Separate results are given for five ranges of relative humidity as indicated.
Vector fitting
Vector fitting of each of 15 measured environmental variables (pool water values excluded) to the ordination of the sample and species information for both mires showed all but one elevation to be significantly related to the plant cover (Fig. 4). Water table levels, on all five occasions that they were measured, plus water and organic contents of the peat, were all significantly negatively correlated with axis 1 (for all but organic matter P < 0.001). The seven chemical factors of the surface water, however, had the opposite relationship with this axis, two being significant at P < 0.001 and four at P < 0.05 (Fig. 4). The pattern was generally similar when the vectors were fitted against axes 1 and 3 of the ordination, apart from the organic content of the peat being positively correlated, and Ca/Mg being negatively related with axis 3 (not shown). Elevation is probably not significant in relation to the vegetation pattern because the relatively dry sites adjacent to the completely and partially drained pools on Dome Mire were located within the mire system rather than on its upper or lower margin (Fig. 1).
The species ordination for the two mires (not shown) indicates that those groups associated with the driest sites (on Dome Mire: species classes F and G) occurred at the positive end of axis 1, whereas those associated with the wettest sites (B, C and D) had relatively low scores on this axis. Species class H contained only the generalist grass Poa colensoi, which made some contribution to all communities (Table 1; Fig. 4).
Stable isotope information
The stable hydrogen (D) and oxygen (18O) isotope values from water samples collected on three occasions from pools, the mire surface and groundwater at two depths from Dome Mire, and from rainwater collected from three upland sites in central and eastern Otago, are shown in Fig. 9. The values for rainwater plot close to the local meteoric water line for New Zealand (δD = 8 δ18O + 13; Stewart & Taylor 1981; Stewart & Williams 1981) but those from the mire all plot to the right of this line, along an evaporative fractionation line with a slope typical of temperate regions. Both the surface and pool water samples are consistently more enriched than the groundwater samples, indicating that there is probably little mixing of partially evaporated surface water with groundwater at either mid-depths (−40 cm) or near the base (−1.0 m) in these mires.
Stable isotope composition of water collected from () the mire surface, () the pools and both () shallow (−0.4 m) and () deep (−1.2 m) groundwater from sites on both Dome and Roaring Lion Mires on several occasions. These conform to an evaporation line where the equation is: δD = 5.2δ18O − 18.8, where D is stable hydrogen. Comparable values for rainwater collected from three upland sites in Otago [(), Swampy Summit, 736 m a.s.l.; (), Lammerlaw Range, 870 m a.s.l.; (×), Rock and Pillar Range, 1140 m a.s.l.] by Ingraham and Mark (2000) are also included. These values conform to the equation: δD = 8.3δ18O + 13.6, which is similar to the Local Meteoric Water Line (LMWL) for New Zealand of: δD = 8δ18O + 13.
Water colour
A limited amount of dissolved humic matter in both the pool and stream water was indicated from the spectrophotometric analyses. Absorption coefficients (g 440) for eight pools in Dome Mire ranged from 4.318 to 7.542 (mean 6.052 ± 0.408); the stream value (6.218) was slightly above the mean. Values for 10 pools in Roaring Lion Mire varied to a greater extent, but were not significantly different: range 3.915–8.751 (mean 6.586 ± 0.434), although the streamwater value (6.967) was somewhat higher.
Discussion
The subdued upland topography of the southern Central Otago Ecological Region, combined with its strongly perhumid low-alpine climate, are generally highly conducive to wetland development. Moreover, drainage is generally impeded either by the schist bedrock or by the silty colluvium that overlies it locally (Mark et al. 1995; McGlone et al. 1995, 1997). Our paper aimed to establish the trophic status of two of the most extensive wetlands of the Nokomai mire complex, in relation to the operative hydrological processes, as well as the extent to which their plant communities were influenced by differences in water availability and chemistry.
Although limited areas of wetland were present in the area from early Holocene times, a major expansion of Sphagnum moss and Sphagnum–sedge communities was associated with the rapid upward growth of peat from approximately 600 years ago. This peat now forms the distinctive 50–60-cm high walls of the many pools in the present string bog complex (McGlone et al. 1995). These authors offer two hypotheses to explain this relatively recent, rapid phase of Sphagnum growth: (i) changing climate altered both the amount and seasonal distribution of precipitation, thus increasing the water surplus conducive to the rapid growth of Sphagnum; or (ii) burning by the early Polynesian settlers from approximately 600 years ago destroyed much of the forest up to the natural treeline at approximately 1100 m, both here and elsewhere, as well as modifying extensive areas of tussock-shrubland and shrub-tussockland above the treeline (McGlone 1983). This burning resulted in conversion of these woody communities to those dominated by tall tussock (Chionochloa spp.) grasses. As a result, run-off patterns presumably were altered and water yield increased from the surrounding catchments. A combination of the two hypotheses was also proposed.
The stratigraphic pattern of peat described for these Nokomai mires, where approximately 50 cm of Sphagnum peat overlies a varying depth of sedge peat (McGlone et al. 1995), appears to have its counterpart in the development of several Northern Hemisphere raised bogs. Gross stratigraphic and certain other wetland features of several mires across Scandinavia and eastern North America have also been attributed primarily to changing moisture regimes (Almquist-Jacobson & Foster 1995).
A striking similarity with the Nokomai stratigraphic sequence of late Holocene events has been described by Seppälä and Koutaniemi (1985) from the Liippauso string-pool mire complex in Finland. At the Finnish site, the mire began to form the string-pool complex approximately 2000–3000 years ago, apparently in response to a long-term increase in groundwater levels. Meltwater flushing from the mire surface in spring apparently was the critical factor in promoting pool formation. McGlone et al. (1995) suggested that a global trend toward wetter and cooler late Holocene climates may be responsible for the similar temporal pattern in groundwater supply in these otherwise dissimilar mire complexes in New Zealand and Finland.
Mean annual precipitation at the Nokomai mires has been estimated from the records for 2 years, corrected for deviation from long-term normals at the closest official station (Kingston 310 m a.s.l., 20 km to the north-west). These values were 1476 mm and 1282 mm, respectively (mean 1380 mm). Precipitation at Dome Mire was generally 1.2 times that recorded at Kingston during the snow-free period but, as expected, was more erratic at other times when it was probably under-recorded at this high-elevation site because of snow. No seasonal pattern is apparent from the 24-month record (1997–1999) but the slight winter minimum in the long-term record from the nearest lowland site was not apparent in the Nokomai mire records and is unlikely. Rainfall affected pool water level below sill height while periodic strong winds depositied macrophytes above and beyond the sill margin. Air temperatures over 2 years at the Nokomai mires, when corrected for deviations from long-term norms in the region, based on records from the closest official station (Lumsden 187 m a.s.l.), 44 km to the south-west, indicate monthly means ranging from 8.3°C in midsummer to −1.1°C in midwinter, with a mean annual value of 2.6°C. The midsummer value, being 1.7°C below that prevailing at treelines in New Zealand (Wardle 1998; Mark et al. 2000), indicates that the mires are located in the low-alpine bioclimatic zone. Annual air temperature extremes, adjusted to the long-term values, are estimated to range from 19.3 to −12.8°C at the Nokomai mires. The peat at 5 cm depth was frozen continuously for 2 months and intermittently for 4 months but reached a minimum of only −2.3°C over the 2 years of recording. During the warmest month, the peat at this depth had a mean temperature of 14.0°C compared with 12.7°C at a depth of 50 cm and 10.7°C at 100 cm. Incoming shortwave radiation reached extreme monthly maxima and mean daily maxima that were indicative of clear skies, whereas mean daily values were considerably lower because, in part, of frequent cloud cover and fog. Frequent fog is also indicated by the generally high incidence of days (88%) when mean hourly air humidity exceeded 70%; 21% of days had a mean hourly humidity exceeding 95%.
In addressing our primary questions, we have established that at 42 sites selected to represent both wetlands, there is a strong relationship between the vegetation pattern and several environmental factors. Water availability expressed as either water table depth (measured on five occasions under both wet and dry conditions) or moisture content of the peat, were significantly related to the vegetation patterns on the mires. The organic content of the peat was also significantly related to the mire vegetation. The amount of dissolved humic matter in the water of the pools and central streams in both mires was indicated by its colour (range 3.915–8.751 g 440), which was generally greater than the range found in 20 Quebec lakes (0.092–4.744 g 440) by Cuthbert and Del Giorgio (1992). It was also greater than the range found in 11 other New Zealand wetlands (0.288–5.758 g 440) and was closer to the value for a small New Zealand lake that drains lowland indigenous coniferous–broadleaved forest (7.801 g 440; M. Schallenberg, pers. comm., 2000).
The range in elevation over the transects sampled was much greater for the Roaring Lion Mire (approximately 10 m) than for the Dome Mire (4 m). Elevation was positively related to the height of the water table at Roaring Lion Mire but no significant relationship was apparent at the much smaller Dome Mire (results not shown).
All seven chemical variables analysed for the water on the mire surface (pH, conductivity, Ca, Mg, K, Na and Ca/Mg ratio) were significantly positively correlated with axis 1 of the sample ordination: the reverse of the pattern shown for various aspects of water availability. Cation concentrations in surface water collected from the study sites on both mires were approximately an order of magnitude higher than those in the adjacent pool water and, moreover, were highest at the driest sites from which water was available near the drained and partly drained pools on Dome Mire. The surface water chemistry of Dome Mire was generally consistent with results from de Groot's (1999) more restricted study. Information on the stable isotopes of hydrogen (D) and oxygen (18O) for a limited range of water samples collected on three occasions from the surface, pool and groundwater (two depths) from Dome Mire indicated negligible mixing of surface water within the upper 50 cm of peat. Evaporation from both the mire surface and pools was indicated by isotopic enrichment from both such sites on the wetlands. The higher nutrient concentrations of the surface water can be attributed to greater evaporation on the mire surface than from the pools.
The inverse relationship between cation concentration and level of the water table probably reflects localized dilution, as described by Comeau and Bellamy (1986) and Giller and Wheeler (1986). However, the lower pH of water on the mire surface relative to the adjacent pool water is probably associated with the production of acids as a byproduct of organic matter degradation. The difference in the chemistry of the surface water between the two mires may indicate a variable trophic status.
In a global context, the water chemistry of the Nokomai mires is consistent with that described for several other patterned fens considered to be mesotrophic in Canada (Foster & King 1984) and Sweden (Foster & Fritz 1987). Ca/Mg ratios have been proposed as indicators of wetland trophic status but their reliability has been debated (Sjörs 1961; Waughman 1980; Foster et al. 1988). Either way, the differences in this ratio were relatively slight between the two wetlands studied. Values for calcium concentration of approximately 2–3 mg L−1 are considered to be indicative of mesotrophic status (or a ‘poor’ fen) by Glaser et al. (1981) and Glaser (1983), whereas ombrotrophic bogs have lower concentrations of Ca and generally an acute nutrient deficiency (Damman 1986). In addition, pH values between 4 and 5 are usually taken to be indicative of mesotrophic status and values less than this as ombrotrophic (Glaser 1983; Foster & King 1984; Karlin & Bliss 1984). Such criteria would confirm a mesotrophic status for Dome Mire, with Roaring Lion Mire being mesotrophic to marginally ombrotrophic. Because all our measurements of water chemistry in the Nokomai mires were made in midsummer, we are unable to assess seasonal differences.
Vegetation composition as well as chemical features have been used in wetland characterization and classification (Gore 1983). Among the 55 taxa sampled on the Nokomai mires, a few are shared with the Northern Hemisphere. The indigenous sedge Carex echinata and the exotic grass Anthoxanthum odoratum, plus Sphagnum cristatum, have been recorded in Finnish mires, where all have been used to indicate mesotrophic conditions, with A. odoratum also assumed to indicate some influence of groundwater (Eurola et al. 1984). The two graminoids are confined to the somewhat more nutrient-enriched Dome Mire and were closely placed on the positive side of axis 1 in the ordination of the mires (Fig. 4). Sphagnum cristatum is of widespread occurrence and was located at zero on this axis.
Comparisons of the main vegetation components of the Nokomai mires with studies of other New Zealand wetlands show both similarities and differences. The Dome and Roaring Lion Mires appear to be most similar in plant composition to subalpine mires in Canterbury (Dobson 1975) and to lowland mires in eastern Fiordland (Burrows & Dobson 1972; Mark et al. 1979), where the common species are Sphagnum cristatum, S. falcatulum and Oreobolus pectinatus. These and most other extensive wetlands in New Zealand are at a lower elevation than the Nokomai mires and have a significant cover of Empodisma minus. Furthermore, these lowland wetlands also support tall shrubs, particularly Leptospermum scoparium and Dracophyllum spp. (Mark & Smith 1975; Clarkson 1984; Williams et al. 1990; Agnew et al. 1993; Dickinson & Mark 1994; de Groot 1999). These lowland wetlands generally have no obvious patterning, although there are two notable exceptions in south-west South Island (Burrows & Dobson 1972; Dickinson & Mark 1994; Mark 1998). String bogs with similar life forms and generic composition (Abrotanella, Astelia, Carex, Carpha, Dracophyllum, Drosera, Euphrasia, Isolepis, Oreobolus) to those in the present study also occur in the low-alpine regions of Tasmania (Kirkpatrick & Gibson 1984). Ponds there drain by natural tunnelling through humified peat, as also occurs in the Nokomai mires.
As outlined in the earlier study of the Nokomai mires (Mark et al. 1995), the distinctive string-pool and island patterning is influenced by the interaction of several factors: topography, underlying substrate, hydrological and drainage properties, as well as differential peat formation associated with differences in plant cover. We have not further investigated the presence, extent and possible development mechanisms for the numerous islands within the pool system or for the underground drainage pipes. However, the application of fluoresceine Na dye to the outlet pipe of the completely drained pool on Dome Mire (Fig. 15 in Mark et al. 1995) confirmed its linkage with the adjacent pool approximately 15 m downslope. The outflow, however, appeared to be highly diffuse. We also noted that the water levels in the two pools closest to, and on opposite sides (north and south) of site DB 1 on Dome Mire, were often low and the southern one occasionally empty. We observed that water levels in both these pools fluctuated widely in response to rainfall.
The general presence of a subterranean pipe system in these mires accounts for the periodic partial or complete drainage of individual pools that presently occurs in Dome Mire. This system may also divert much of the considerable run-off water from the slopes surrounding the two mires to beneath the mire surface. Any difference in trophic status between the two mires may be the result of varying degrees of surface and subsurface flushing.
Conclusions
The Nokomai mires were included among the ‘recommended areas for protection’ and ranked as ‘internationally significant’ in ecological surveys for the New Zealand Protected Natural Areas Programme (Dickinson 1989; Dickinson et al. 1998). Held within a Crown pastoral farming lease, this wetland complex currently has very limited protection. Ideally, the entire catchment of the wetland complex, some 31 km2, should be formally protected to retain the integrity of the system. At present, the larger part of the complex, that within the Nevis catchment to the north, has been recognized and included within the Kawarau (River) Water Conservation Order, within the terms of the Resource Management Act, 1991. This Order, however, does not control the surrounding land use. Recently completed Conservation Management Strategies (CMSs) for the Conservation Department includes a commitment to seek formal recognition for the wetland complex under the Convention on Wetlands of International Importance (the ‘Ramsar’ Convention).
Our study of these patterned wetland systems has emphasized their complexity in relation to the landform, present and past environments, plant communities, hydrology and water chemistry. This is the first comprehensive study of the processes involved in these wetland types in the Southern Hemisphere and indicates, in this case, their mesotrophic to marginally ombrotrophic status. We have established the influence of the levels and nutrient status of water and we stress the significance of these Southern Hemisphere patterned mire systems in a global context, where wetlands rank among the World's most beleagured biomes.
Acknowledgements
We wish to thank the Hellaby Indigenous Grasslands Research Trust for providing much of the funding for this project, Marc Schallenberg for analyses and comments on water colour, Rob Daly for data logging support, and Brian Hore, lessee of Nokomai Station, for permission to conduct the study as well as assistance with helicopter transport. We also thank Stewart Bell, Robert Hofstede, Peter and Linda Kershaw, Peter Johnson, Chris Bycroft, Kelvin Lloyd and the late John Holloway for field assistance; Neil Gaudin for analysis of datalogger records; Ray Tangney for help with bryophyte identification; Richard Clayton and Gudrun Wells for graphics help; and two anonymous referees for their helpful comments.
