New Zealand Journal of Marine and Freshwater Research ISSN: 0028-8330 (Print) 1175-8805 (Online) Journal homepage: https://www.tandfonline.com/loi/tnzm20 Bryophyte distribution patterns in relation to macro‐, meso‐, and micro‐scale variables in South Island, New Zealand streams Alastair M. Suren To cite this article: Alastair M. Suren (1996) Bryophyte distribution patterns in relation to macro‐, meso‐, and micro‐scale variables in South Island, New Zealand streams, New Zealand Journal of Marine and Freshwater Research, 30:4, 501-523, DOI: 10.1080/00288330.1996.9516738 To link to this article: https://doi.org/10.1080/00288330.1996.9516738 Published online: 30 Mar 2010. Submit your article to this journal Article views: 629 View related articles Citing articles: 46 View citing articles Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=tnzm20 New Zealand Journal of Marine and Freshwater Research, 1996: Vol 30: 501-523 0028-8330/96/3004-0501 $2.50/0 ©The Royal Society of New Zealand 1996 501 Bryophyte distribution patterns in relation to macro-, meso-, and micro-scale variables in South Island, New Zealand streams ALASTAIR M. SUREN National Institute of Water & Atmospheric Research Ltd P. O. Box 8602 Riccarton, Christchurch, New Zealand Abstract A broadscale survey of 118 streams was made throughout New Zealand's South Island to establish relationships between aquatic bryophytes and selected environmental variables. Bryophytes were found in 95 of the streams examined, and covered large areas of the substratum (mean cover of 17%, maximum cover of 86%). TWINSPAN analysis revealed the existence of five sample groupings, of which one supported no bryophytes. Streams without bryophytes typically flowed through developed catchments (pasture or pine) composed of easily eroded rocks. Nutrient levels were high in these streams, and low-flow events were also common. Streambed stability was low, reflecting lack of stable bedrock and boulder substrata. TWINSPAN identified two major groups of streams with bryophytes: those supporting mosses and those supporting liverworts. These streams differed from those without bryophytes by having a higher streambed stability, and fewer lowflow events. These two variables thus appear fundamental in controlling aquatic bryophyte distribution patterns. Flood events had no significant impact on these plants once they grew on stable substrata. No differences in stream stability, or substrate composition, were found between streams dominated by mosses and those dominated by liverworts. Catchment geology, land use, water quality, and the number of high-flow events differed between these streams, suggesting that these factors influenced what type of bryophyte community occurred. M96010 Received 1 March 1996; accepted 3 September 1996 Keywords aquatic bryophytes; distribution; hydrology; substrate stability; geology; land use; water chemistry INTRODUCTION Stream ecosystems are controlled by a wide range of environmental variables, many of which are strongly interrelated. In an attempt to simplify our understanding of stream ecosystem functioning, it is useful to categorise these variables on the basis of their spatial and temporal characteristics (Minshall 1978; Biggs et al. 1990; Naiman et al. 1992; Biggs & Gerbeaux 1993). Macroscale variables such as climate, altitude, geology, and catchment vegetation affect large geographic areas (> 1 km2) and are influential over a long time period (e.g., > 100 years). These in turn dictate the nature of mesoscale variables, such as a stream's hydrological or nutrient regime. Such variables influence only small geographical areas (< 1 km2) within a catchment, and operate over generally short time periods (< 1 year). At the smallest spatial and temporal scale are microscale variables such as substrate size, bed stability, current velocity, and stream slope. These variables reflect small (c. < 100 m) localised changes to the stream bed. The importance of the above variables in regulating periphyton biomass and community composition is becoming apparent. Research directed at periphyton communities reflects their importance as primary producers in many open streams (e.g., Minshall 1978), food for invertebrates (e.g., Lamberti & Resh 1983), and potential for developing aesthetically displeasing blooms (e.g., Biggs & Price 1987). An extensive survey of periphyton in 101 New Zealand streams during summer low flows illustrated the importance of stream conductivity in influencing their biomass and community composition (Biggs 1990). Conductivity, however, was regarded as a surrogate determinant for the 502 New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 percentage of soft sedimentary rocks in a catchment, and approximated relative nutrient levels within a stream. A more intensive study of algal dynamics in the Motueka and Riwaka Rivers in the northern South Island found that periphyton communities here were controlled primarily by geology, particularly by the amount of marble in the catchment. This was thought to have influenced the long-term supply of nitrogen to the algae (Biggs & Gerbeaux 1993). Recently, however, the influence of flood events in structuring periphyton communities has been highlighted in a long-term study of 16 sites throughout New Zealand (Biggs 1995). This study differed from previous periphyton studies here, in that it was not based on a one-off lowflow sampling occasion (e.g., Biggs 1990), and was not confined to one small geographical area (e.g., Biggs & Gerbeaux 1993). It clearly implicated the dominant role that flood events have in regulating periphyton biomass, and suggested that such events represented the principal axis of the habitat template for periphyton in New Zealand streams. Nutrient resources (controlled by geology and land use) were implicated as operating within this principal axis, and were responsible for controlling the rate of biomass accrual during interflood periods. Little, however, is known about the importance of macro-, meso-, and micro-scale variables to aquatic bryophytes. This is surprising, especially considering the ecological importance of these plants to invertebrates (Percival & Whitehead 1929; Hynes 1961 ; Brusven et al. 1990; Nolte 1991 ; Suren 1991) and periphyton (Suren 1992), and their contribution to stream energetics (Dawson 1973; Naiman 1983; Bowden et al. 1994). This lack of fundamental knowledge about factors that regulate bryophyte communities represents a major deficiency in our understanding of stream ecosystems. The aim of the present study was thus to examine bryophyte communities in several locations throughout New Zealand's South Island to establish relationships between species occurrence and environmental variables. METHODS Stream survey Seventy-eight sites were selected from Walter (1989), based on the criteria of being streams from small catchments (< 20 km2) which had good (> 5 yr) hydrological data. Few suitable small streams on the west coast had been gauged, especially in areas of high rainfall such as Fiordland and northwest Nelson. Forty streams in these regions were thus randomly surveyed in as many locations as possible. The location of all 118 surveyed streams was recorded on 1:63 000 or 1:50 000 maps (NZMS Series 1 or 2). Gauged streams were surveyed either at the gauging station if this was suitable, or at smaller upstream sites if the stream was too big. Ungauged streams were simply surveyed at a suitable location within each catchment. A 40 m long transect was placed up the centre of each stream and measurements made every metre of stream width and maximum depth. Substrate composition was estimated visually in a 20 cm wide strip across the stream at these intervals. Bedrock outcrops were common in many streams, so all streambed materials were categorised into five qualitative classes (Jowett et al. 1991): • bedrock (solid rock surface that was obviously deeply embedded and would rarely move in a flood event), • boulders (> 256 mm), • cobbles (64-255 mm), • gravel (10-63 mm), and • sand (1-9 mm). All substrate measurements were converted to a single substrate index by summing the weighted substrate percentages such that: Substrate index = 0.08 bedrock % + 0.07 boulder % + 0.06 cobble % + 0.05 gravel % + 0.04 sand % Percentage bryophyte cover was estimated visually to the nearest 5% within each 20 cm wide strip across the stream. Representative samples of any plants present were collected from each site, and notes made on their microhabitat. Only those bryophytes which were completely submerged, or those that were continually wetted by splashing water were collected. Bryophytes were kept frozen (—18°C) before analysis. Upon thawing, they were teased apart using water pressure, and the species identified using standard keys (Appendix 1 ). Flow characteristics of the stream were assessed every metre according to one of six qualitative categories: waterfall, chute, riffle, run, pool, or seepage. Such categories considered the relative water velocity and depth at each point, whether it was super-critical (white water) or sub-critical (no white water), and the localised streambed slope (Table 1). Stream slope was measured every 5 m using survey poles and an Abney level. Suren—Aquatic bryophyte distribution patterns Streambed stability was assessed using a modification of the "streambed" component of the Pfankuch (1975) survey (see Suren 1993; Death & Winterbourn 1995). This is a method of assessing a stream's overall stability on the basis of the surface texture, roughness, and embeddedness of substrate particles. Water samples were collected from a subset of the 118 sites, filtered (Whatman GFC), and stored in acid-washed polycarbonate bottles before being frozen (—18°C) pending analysis. Upon thawing, all samples were analysed by a Technicon II autoanalyser for dissolved reactive phosphorus (DRP), ammonium-N, nitrate-N, total dissolved phosphorus (TDP), total dissolved nitrogen (TDN), dissolved organic phosphorus (DOP), and dissolved organic nitrogen (DON) using standard techniques (Downes 1978). Measurement of physical and hydrological variables Underlying geology, dominant land use category of the catchment up stream of each transect, and average catchment elevation were quantified by use of the New Zealand Land Resources Inventory data base. All catchments were digitised, and the percentage of catchment area of a particular geology or land use calculated by the GIS system Arc-Info. For ease of analysis, parent geological rock types were classified into broad categories. Each of these was then assigned an arbitrary "erosion index" (Table 2). Land use categories were assigned to one of nine classes: developed pasture, tussock grassland, scrub, beech, podocarp, and exotic forest, alpine vegetation, non-vegetated areas, and miscellaneous areas. The inclusion of gauged streams in this survey enabled the influence of flow regime on aquatic bryophytes to be assessed. In particular, variables 503 that reflected high- and low-flow periods, as well as flow variability, were selected for study. Flood events represent the primary axis of the habitat template for periphyton (Biggs 1995), whereby mean annual biomass is negatively correlated with flood frequency. This reflects physical tearing of periphyton communities from the substrate when shear stress exceeds the attachment and tensile strength of algal cells. Complicating this simple interaction between high shear stress and periphyton sloughing is the preflood incubation period. This has a major effect on the resistance of an algal community to high-flow events, as water velocity often alters both the taxonomic composition of the community, as well as its physiognomy (e.g., Reiter 1986; Biggs & Hickey 1994). Such a close interaction between shear stress and the community's inherent and conditional {sensu Biggs & Thomsen 1995) properties was observed by Biggs & Close (1989), who found that periphyton biomass always declined following a flood of six times the preceding 7-day mean flow. Bryophytes are not as likely to be scoured from the substrate during high-flow events as periphyton. Bryophytes are also slower growing, and so are unlikely to become "conditioned" to a particular velocity regime. Their flood resistance will therefore reflect better their inherent tensile and attachment strengths, and not be as dependent on the preceding flow regime before a spate as are periphyton communities. A slightly different approach was thus adopted in defining a flood event for this study than was used by Biggs & Close (1989). A flood was arbitrarily defined as one which exceeded the longterm median annual flow by six times, and not one which exceeded the preceding 7-day mean flow by six times. This slightly more conservative Table 1 Field work involved an assessment of the dominant flow category of streams every metre up a 40 m transect. Six arbitrary flow categories were thus defined, based on water velocity, depth, slope, and turbulence characteristics. No quantitative means are given for these variables, reflecting their large inter-site variability as a result of different stream sizes. Flow category Velocity Depth Slope Turbulence 1. Fall 2. Chute 3. Riffle 4. Run 5. Pool 6. Seep Fast Fast Fast Moderate Slow Fast Shallow Moderate Shallow-moderate Shallow-moderate Deep Very shallow Very steep Steep Moderately steep Shallow Flat Steep Super-critical Not super-critical Super-critical Not super-critical Not super-critical Super-critical New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 504 estimate of floods does, however, allow for some comparison with the study by Biggs & Close (1989). The mean number of floods per year in each gauged catchment was determined from site hydrographs. The mean annual total number of days each stream was in flood was also calculated. Aquatic bryophytes are intolerant to desiccation (Glime 1971; Kimmerer & Allan 1982), and are generally absent from rocks that are not continually submerged. The influence of low-flow events on their distribution was thus also assessed. A lowflow episode was arbitrarily defined as one where flows fell below 0.5 times the long-term median, as it was assumed that many of the larger boulders in these small streams would become exposed at such low flows. The frequency that each gauged stream fell below this level each year was determined from site hydrographs, as was the mean number of days at low flow. The annual maximum duration of a low-flow event was also calculated. Because the definitions of high and low flow events were arbitrary, an assessment was also made as to whether selection of different cut-off points for a flood or low flow event had any major effect on the observed results. Ten streams were selected at random, and the frequency of flood events at 4, 6, 8, and 10 times the median flow was calculated. Additionally, the frequency of low flow events of 0.9, 0.7, 0.5, and 0.2 times the median flow was calculated. The slopes of the resultant regression lines showing flood or drought frequency against the cut-off points for each of the ten sites were analysed by ANCOVA to see whether they differed. Flow variability often has profound effects on biota (e.g., Biggs et al. 1990), and the coefficient of variation of flow has been shown to be the best single discriminator of flow variability in New Zealand streams (Jowett & Duncan 1990). This coefficient was calculated for each gauged stream. Statistical analysis The data set of all 118 streams was first analysed by TWINSPAN to assess whether discrete bryophyte communities occurred in any of the surveyed streams. Streams without bryophytes were Table 2 Twenty-three rock types are found in the South Island of New Zealand, from four major rock groups. For the purposes of this study, these were assigned to one of five "erosion index" classes, based on their resistance to fluvial erosion. Group Erosion Index Rock types Surficial 1. Very high erosion Uncemented gravels Windblown sand Loess Peat Alluvium Sedimentary (Weakly indurated) 2. High erosion (Strongly indurated) 3. Moderately high erosion Igneous 4. Moderate erosion Metamorphic 5. Low erosion Mudstone Siltstone Interbedded sandstone Conglomerate Argillite Sandstone Greywacke Conglomerate Limestone Pyroclastics Lavas Ancient volcanics Plutonics (Granites) Ultramafics Semi-schist Schist Gneiss Marble Suren—Aquatic bryophyte distribution patterns included in this analysis to assess differences in environmental variables between streams with and without bryophytes in the resultant TWINSPAN groups (see below). This analysis was conducted on species presence-absence data only, as it was impossible to accurately quantify cover estimates for individual taxa growing in mixed colonies. Many bryophyte taxa were collected only once. These were taxa normally regarded as being only facultatively aquatic, and so were removed from the data set. TWINSPAN is generally unaffected by rare taxa (Gauch 1982) and so removal of these was unlikely to have greatly influenced the resultant classification. This strategy was also more relevant to a study of aquatic bryophytes, and would have reduced "noise" from the resultant classification caused by non-aquatic species. TWINSPAN hierarchically classifies samples into groups of similar species composition. Differences in the observed biological communities (e.g., percentage bryophyte cover, taxonomic richness) between each TWINSPAN group were first assessed by ANOVA with the TWINSPAN group as the a posteriori grouping variable. Following this, differences in measured macro-scale (e.g., underlying geology, land use) and meso-scale variables (e.g., hydrological parameters, nutrient status), as well as the stream's inherent physical attributes (e.g., size, substrate index, slope) between TWINSPAN groups (including the group representing streams without bryophytes) were also assessed by ANOVA. All data were first examined for normality, and transformed appropriately pending analysis. Where significant differences were observed, Tukey's test corrected for unequal sample sizes (Wilkinson 1990) was done to determine which samples differed. Aquatic bryophytes are regarded as being restricted to areas of fast water flow such as waterfalls and chutes, and less common in slowflowing areas (Suren 1991). The generality of this was assessed by ANOVA to determine whether percentage bryophyte cover differed according to the dominant flow characteristics from where plants were collected. Again, a post-hoc Tukey's test was used to see where observed differences occurred. Relationships between bryophyte cover and hydrological and physical parameters as indicated by the initial TWINSPAN analysis were further explored by backward stepwise multiple regression analysis (Zar 1984). All variables were first examined for collinearity, and highly correlated variables removed. There was no evidence of 505 curvilinear relationships between any of the remaining independent and dependent variables so a linear model was thought appropriate for this analysis. All residuals were examined to determine the adequacy of the resultant regression models. Two regressions were done. The first investigated relationships between physical variables (e.g., stream size, substrate index, slope, and stability) on bryophyte cover and taxon richness, and included data from all sites. The second examined the influence of hydrological variables on bryophyte cover and taxon richness using data from gauged sites. RESULTS The bryophyte flora The survey examined mostly headwater streams (mean width 2.4 m, mean depth 20 cm) which drained small catchments (mean catchment size: 7.3 km2). A total of 83 taxa were collected from these streams (Appendix 1). Bryophytes were present in 95 of the 118 streams, and the mean cover was 17% of the stream bed. Eleven streams had at least 50% of their stream bed covered by bryophytes, and the highest cover was 86%. Streams in the Doubtful Sound region of Fiordland had the most diverse bryophyte flora, with one stream supporting 21 taxa. Five other streams in this region contained 10 or more taxa. Streams in the North-West Nelson region also supported a rich bryophyte flora; one site supported 12 taxa. These streams were all from areas of high rainfall, with erosion-resistant geological parent material and largely unmodified catchments of native tussock or forest. Aquatic bryophytes were absent from 23 sites: areas on the Canterbury Plains, and catchments near Nelson, Dunedin, and Invercargill. These locations receive less rainfall and have an easily eroded geological parent material. Dominant land use in these catchments was mostly either exotic plantation or pasture. The most commonly collected taxa were the mosses Fissidens rigidulus (57 sites), Cratoneuropsis relaxa (26 sites), and Bryum blandum (24 sites). The most common liverworts were Anthoceros laevis (21 sites) and Hepatostolonophora paucistipula (18 sites). Eighteen taxa were collected only once. Some of these were facultative aquatic species, which grew submerged at some points up the transects. New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 506 Group 1 Group 2 Group 3 Group 4 Group 5 42 Streams 27 streams 13 streams 13 streams 23 streams Mosses Liverworts No bryophytes Fig. 1 Location of samples assigned to the five TWINSPAN groups, based on species composition data. Samples from Groups 1 and 2 were distributed throughout the South Island and were dominated by mosses (open circles). Samples from Groups 3 and 4 were mostly from the west coast and Stewart Island, and were dominated by liverworts (shaded circles). Bryophytes were absent from streams in TWINSPAN group 5 (filled circles). Community composition TWINSPAN analysis of species presence-absence data from the 118 transects was stopped after three divisions. Five discrete TWINSPAN groups were thus identified, including one group (Group 5) which contained 23 streams without bryophytes. TWINSPAN Groups 1 and 2 were spread throughout the South Island, and samples were common from Stewart Island, Central Otago, South Canterbury, Banks Peninsula, the central Southern Alps, the Kaikouras, and North-West Nelson (Fig. 1). Streams allocated to TWINSPAN Groups 3 and 4 were, however, predominantly from either Fiordland, or the north-west of the west coast. Samples from TWINSPAN Group 5 were all restricted to eastern, southern, and northern areas of the South Island. Examination of the species classifications for the TWINSPAN groups revealed distinctive patterns (Table 3). Streams in TWINSPAN Groups 1 and 2 were dominated by mosses. Only three of the 14 plants characterising Group 1 were liverworts, and only two of ten plants in Group 2 were liverworts. TWINSPAN Groups 3 and 4, however, were dominated mainly by liverwort taxa (Table 3). Bryophyte cover differed significantly between TWINSPAN groups (Table 4). Of the streams where bryophytes were present, streams in Group 1 had the least cover, whereas streams in Groups 3 and 4 had the most. Taxonomic richness also differed between TWINSPAN groups, with the mossdominated streams supporting fewer taxa than the liverwort-dominated streams (Table 4). Relationships with macroscale variables Catchment bedrock geology differed markedly between streams in each of the TWINSPAN groups (Fig. 2). The percentage of catchment area composed of weak sedimentary rocks was highest in areas containing streams without bryophytes (Group 5), whereas the percentage of catchment area composed of strong sedimentary and igneous rocks was highest in areas containing streams supporting mosses (Groups 1 and 2). Catchments dominated by metamorphic rocks were common in areas where streams supported liverworts (Groups 3 and 4; Fig. 2). Dominant catchment land use differed greatly among streams in each of the TWINSPAN groups (Fig. 3). The percentage of catchment area composed of improved pasture or exotic pine plantations was highest in areas containing streams without bryophytes (Group 5), and relatively high in areas containing streams with mosses (Groups 1 and 2). This suggests that certain taxa in these TWINSPAN groups are somewhat tolerant of catchment modification. The percentage of catchment area composed of beech forests was highest in areas containing streams supporting liverworts (Groups 3 and 4). Catchments dominated by tussock grasslands were Suren—Aquatic bryophyte distribution patterns common in areas containing streams with either mosses (Groups 1 and 2) or liverworts (Groups 3 and 4; Fig. 3) Significant differences in mean catchment elevation were evident between TWINSPAN groups (Table 4). Mean catchment elevation was lower in sites without any bryophytes ( x = 200 m a.s.l.) than sites supporting bryophytes. There was, however, no significant difference in elevation between sites supporting mosses ( x = 514 m) or those supporting liverworts (x= 640 m). 507 Relationships with mesoscalc variables Analysis of data pertaining to the frequency of flood events at different arbitrary cut-off points showed that there was no difference in the resultant regressions between sites. This suggests that similar results would have been obtained in the following analyses if a different definition of a flood event had been made. The same conclusion was also drawn for analysis of the low-flow data. Analysis of hydrological variables of sites in each TWINSPAN group clearly implicated the Table 3 Species preferences for the defined TWINSPAN site groupings as defined by TWINSPAN analysis of species presence-absence data from 118 sites. Also shown are the number of mosses (Musci) and liverworts (Hepaticae) characteristic of each group. Group 1 Musci Achrophyllum quadrifarium Amblystegium sp. Bartramiu papillata Breutelia pendula Bryum blandum Biyum laevigatum Cratoneuropsis relaxa Fissidens rigidulus Coniobryum subasilare Tridontium tasmcinicum Thuidium furfurosum Hepaticae Aneura alterniloba Clasmatocolea vermicularis Plagiochila banksiana 11 Musci 3 Hepaticeae Group 2 Group 3 Cryphaea tasmanica Cyathophorum bulbosum Echinodium hispidum Eurhynchium austrinum Plagiomnium novae-zealandiae Thamnobryum pandum Thamnobryum pumilum Thuidium laeviusculum Group 4 Blindia immersa Blindia lewinskyae Blindia magellanica Blindia robusta Camptochaete gracilis Distichophyllum crispulum Distichophyllum pulchellum Drepanocladus aduncus Eriopus cristatu.s Hypnodendron spininervium Pyrrhobryum mnioides Racomitrium crispulum Sematophyllum uncinatum Lejenuea sp. Plagiochila circinalis Lophocolea cf. semiieres Anthoceros laevis Balantiopsis convexiscula Metzgeria decipiens Balantiopsis tumida Monoclea forsteri Clasmatocolea humilis Crypotchila grandiflora Hepatostolonophora paucistipula Hygrolembidium sp. Hymenophyton flabellatum Lophocolea gunniana Lophocolea planiuscula Marchantia berteroana Metezgeria fauriana Metzgeria rigida Plagiochila circinalis Plagiochila deltoides Plagiochila fruticella Plagiochila fusceila Plagiochila retrospectans Radula buccinifera Riccardia sp. A Riccardia sp. B Shistochila lehmanniana Teleranea cf. dispar 8 Musci 2 Hepaticeae 0 Musci 3 Hepaticeae 13 Musci 23 Hepaticeae New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 508 Fig. 2 Percentage of different bedrock material in all catchments from each TWINSPAN group ( x + 2 SE, « given at the top of each bar; shading as in Fig. 1): weak sedimentary rocks (F4 113 = 28.68); strong sedimentary rocks (^4, 113 = 8.27); igneous rocks (^4, 113 = 5.97); metamorphic rocks (F4< H3 = 7.32; al| P < 0.001). Letters denote which TWINSPAN groups are similar to each other (Tukey 's test, P > 0.05). 100-, 1OOn o o « 80 H o 80 2 60 10 3 a. S 40 I 40 O) k 4> X 20- E 20 1 2 3 4 5 Twinspan group 1 2 3 4 5 Twinspan group importance of low-flow events in ultimately controlling bryophyte occurrence (Fig. 4). The longest duration of low-flow events, the highest coefficient of variation of flow, and the lowest catchment yield all occurred in streams that contained no bryophytes (TWINSPAN Group 5). These streams also had the highest number of days at low flow, although this was not significantly different from streams supporting liverworts in TWINSPAN Group 4 (Fig. 4). The highest number of low-flow events each year occurred in streams in Groups 4 and 5. The mean number of floods in each stream also differed between groups, but this showed little consistent pattern between streams with, and without bryophytes. The highest number of floods were from streams in Group 4 (containing liverworts), but streams supporting mosses, as well as those streams without bryophytes, had fewer floods (Fig. 4). The number of days each stream was in flood for was similar in all TWINSPAN groups, suggesting that this variable was of little importance in structuring bryophyte communities. Table 4 Mean bryophyte cover and taxonomic richness ( x + 2 SE) of the five TWINSPAN groups identifed from the 118 streams sampled throughout the South Island. The mean altitude of streams in each TWINSPAN group is also given. Data were analysed by ANOVA to determine whether variables differed between TWINSPAN groups: means with the same superscripts are not significantly different from each other (P > 0.05) as determined by a post-hoc Tukey's test. TWINSPAN group 4i 1 2 3 4 5 H3) ratio Percentage cover 1 .8 ± 2.1" 23 .5 + 3.5a'b 35 .8±6.5 b 37 .0 ± 6.4b 0c 17.95 Taxonomic richness a 3.7 +0.2 5.5+0.8 3 8.8 ± 1.7b 7.0 ± 1.2a-b 0c 18.89 Mean altitude 604 ± 54 a ' b 387 + 48 b 510 ± 55a-b 687± 110a 200 ± 36C 9.88 509 Suren—Aquatic bryophyte distribution patterns Fig. 3 Percentage of different landuse categories in all catchments from each TWINSPAN group ( x + 2 SE, n and shading as in Fig. 1): improved pasture (F 4 , i n = 10.57); tussock grassland (F 4 , u 3 = 8.02); beech forest (F 4 t1I3 = 26.61); pine plantations (F 4 113 = 6.61 ; all P < 0.001). Letters denote which TWINSPAN groups are similar to each other (Tukey's test, P > 0.05). 1OO-i 1OOT 1 2 3 4 5 Twinspan group Water chemistry differed between streams in each TWINSPAN group (Fig. 5). Highest nitrogen levels (as ammonia-N, total dissolved N, and dissolved organic N) came from streams without bryophytes, reflecting the generally modified land they drained. These developed streams, however, had a lower phosphorus content than streams in Group 2 which supported mosses. The lowest nutrient concentrations, except for dissolved organic nitrogen, were found in streams supporting liverworts (i.e., TWINSPAN Groups 3 and 4; Fig. 5). Relationships with microscale variables Calculation of the substrate index enabled a single value to be assigned to streams based on their sediment sizes. Index values near 8 came from streams where bedrock was dominant, whereas values near 4 came from streams whose substrata were dominated by sand (Fig. 6). Streams with an intermediate index (6) contained intermediate proportions of gravel, cobble, and boulder. Analysis of physical variables characteristic of sites in each TWINSPAN group clearly implicated the importance of substrate size and streambed stability to bryophytes. The lowest substrate index, 1 2 3 4 5 Twinspan group and highest Pfankuch streambed instability ratings were from streams without bryophytes (TWINSPAN Group 5; Fig. 7). Such streams were thus dominated by smaller substrate particles, and were more unstable than streams supporting bryophytes. Substrate index and streambed stability, however, were similar between streams supporting mosses (TWINSPAN Groups 1 and 2) and those supporting liverworts (TWINSPAN Groups 3 and 4; Fig. 7). Small-scale flow characteristics had a significant effect on bryophyte distributions within streams. Bryophyte cover was greatest in waterfalls (Fig. 8), despite the obviously increased water velocities in these habitats which would have scoured away periphyton growths. Bryophyte cover was considerably lower in riffles, runs, and pools, despite the lower water velocities in these habitats. Stepwise regressions: physical variables (Table 5) Stepwise multiple regression analysis enabled the effects of measured physical parameters in influencing bryophyte distributions to be assessed. Resultant stepwise regression models explained a high fraction of the variance in sample spread, with New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 510 Fig. 4 Hydrological variables from streams in each TWINSPAN group ( x + 2 SE, n given at the top of each bar; shading as in Fig. 1): maximum duration of a lowflow event (F 4 , 73 = 3.65); coefficient of variation of flow (FA, 73 = 6.28); catchment yield (F4^ 73 = 12.07); number of days at low flow (F 4 73 = 6.46); number of low flow events (F4 73 = 5.28); number of flood events per year (F 4 73 = 6.04; allP<0.01). Letters denote which TWINSPAN groups are similar to each other (Tukey's test, P > 0.05). 1 2 3 4 5 Twinspan group 1 2 3 4 5 Twinspan group Table 5 Results of backward stepwise multiple regression analysis of A, bryophyte cover and B, bryophyte taxonomic richness against physical parameters in the 118 sites. Coeff., coefficient of determination (r2) of the resultant model statement. P - value Coeff. Variable 2 A. Bryophyte cover (n = 110, r = 0.63) Pfankuch stability score Substrate index Slope -0.86 33.71 4.18 -5.79 3.21 2.79 0.001 0.002 0.006 B. Taxonomic richness (n = 110, r2 = 0.605) Substrate index 3.99 S l o p e U7 ( 10.09 2^29 0.001 01)04 Suren—Aquatic bryophyte distribution patterns 511 Fig. 5 Nutrient analysis of selected streams in each TWINSPAN group (x ± 2 SE, n given at the top of each bar; shading as in Fig. 1): DRP (F 4 56 = 10.57); N H 4 - N ( F 4 56 = 2.94); TDN (F 4 56 = 3.04); DON (F 4 56 = 3.82; all P < 0.05). Letters denote which TWINSPAN groups are similar to each other (Tukey's test, P> 0.05). 1 2 3 4 5 Twinspan group 63% of the variation in bryophyte cover explained, and 60.5% of the variation in taxonomic richness explained. Bryophyte cover was negatively related to Pfankuch stability score, suggesting that these plants are less common in streams with high scores (i.e., unstable streams). Cover was positively related to substrate index and streambed slope, supporting the notion that bryophytes are abundant in steep streams with large, stable substrata. Taxonomic richness was also positively related to substrate index and streambed slope. 1 2 3 4 5 Twinspan group Stepwise regressions: hydrological variables (Table 6) Stepwise multiple regression models explained only 34.2% of the variance in sample spread for estimating bryophyte cover from hydrological variables. Unexpectedly, a significant positive relationship existed between the number of flood events and bryophyte cover, suggesting that frequent flood events per se are not detrimental to these plants. Cover was also positively related to catchment yield, but significantly and negatively Table 6 Results of backward stepwise multiple regression analysis of A, bryophyte cover and B, bryophyte taxonomic richness against measured hydrological parameters in the 78 gauged catchments surveyed in the study. Coeff., coefficient of determination (r2) for the resultant model statement. P - value Coeff. Variable 2 A. Bryophyte cover (n = 74, t = 0.342) Catchment yield 7.49 Number of floods 12.55 Number of low flows -10.99 4.03 3.12 -2.32 0.001 0.003 0.023 B. Taxonomic richness (n = 74, r2 = 0.364) Catchment yield 1.61 Number of low flows -1.87 4.07 -1.86 0.001 0.06 New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 512 7 6 SCO 3 3 50 £40 !5 100 n 80 9 (0 60 «30 •o a> o o 40 20 |20 Q a> S 10 100i 80 1 2 3 4 5 Twinspan group > 60 1 » 40 Fig. 7 Physical parameters of the substrate index and streambed stability of the streams in each TWINSPAN group ( x ± 2 SE, n and shading as in Fig. I): Substrate index (F4 <m = 38.61); Pfankuch streambed stability (F 4 H3 = 14.20) all P < 0.05). Letters denote which TWINSPAN groups are similar to each other (Tukey's test, P > 0.05). 20 100 i 5 6 7 Substrate index Fig. 6 Substrate index showing the relative composition of bedrock, boulders, cobbles, gravels, and sand in streams with different index values. related to the number of days a catchment experienced low flows for. Stepwise regression models for taxonomic richness explained a similarly low fraction of the variance in sample spread (36.4%). Here, taxonomic richness was positively related to catchment yield, but negatively related to the number of days a catchment experienced low flows. Suren—Aquatic bryophyte distribution patterns 100 80 60 a. 2 40 20 Water nature Fig. 8 Percentage bryophyte cover growing under different flow characteristics of the streams surveyed ( x ± 2 SE, n = 118). Letters denote which TWINSPAN groups are similar to each other (Tukey's test, P > 0.05). DISCUSSION This work represents the first detailed study of aquatic bryophyte distributions in New Zealand, and sought to identify significant relationships between bryophytes and environmental variables. As such it is based upon similar studies to the water quality, periphyton, invertebrate, and fish classifications of the New Zealand 100 Rivers programme (see Biggs et al. 1990). These studies all used a similar methodology, whereby regional classifications were developed based upon species assemblage patterns. Differences in physical environments within these regions were then assessed to determine what environmental parameters were responsible for the species assemblage patterns within each region. As found for other communities in the 100 Rivers survey, bryophyte distribution patterns in the present study were regulated by both macroscale and mesoscale variables, although the nature of these relationships often appeared fundamentally different from those regulating periphyton communities (see below). This study also showed how bryophyte distributions were controlled by microscale variables, which had not been assessed in the 100 Rivers study. The importance of substrate stability in regulating bryophyte distributions was clearly 513 evident in this study. Sites with the highest bryophyte diversity were from Fiordland and northwest Nelson. Streams in these regions flowed through catchments dominated by erosion-resistant igneous or metamorphic rocks. Bryophytes were also common in streams flowing through catchments dominated by strongly indurated sedimentary rocks. Streams flowing through catchments of weak sedimentary rocks, however, commonly did not support bryophytes. This suggests that underlying catchment geology may regulate bryophyte distributions by directly influencing the degree of weathering of parent bedrock, and thus the relative degree of potential substrate movement within each stream. Substrate movement is dependent upon substrate size and stream hydrology. It is well known that bryophytes are restricted to large, stable substrata (e.g., McAuliffe 1983; Sheath et al. 1986; Englund 1991; Steinman& Boston 1993): indeed it is almost axiomatic that "the rolling stone never gathereth moss" (J. Heywood 1363, quoted in Stevenson 1947). Similar results were evident in this study, where bryophytes were more abundant on larger substrata, irrespective of stream location. The importance of stable substrata to bryophyte occurrence was also implicated by negative relationships between both bryophyte cover and taxonomic richness to Pfankuch stability scores in the present as well as an earlier study (Suren 1993). Both studies have shown how bryophyte cover and diversity was higher in streams with a high streambed stability rating (i.e., low Pfankuch score). Many of the categorical variables used in this rating scheme reflect how often bed moving spates occur. Thus in highly unstable streams where rock movement is common, rocks will be well rounded, easily moved, and have no visible epilithon on them. Bedrock areas will usually be absent with the stream channel in a continual state of flux. Under such conditions, bryophytes are unlikely to occur, as confirmed in this study. Restriction of bryophytes to large stable substrata reflects the length of time it takes for plants to recolonise a stone after it has overturned. Englund (1991) found that the mosses Hygrohypnum ochreacum and Fontinalis dalecarlica took at least 2 months to recolonise upper surfaces of recently overturned stones. He estimated that it would take several years before biomass reached a predisturbance level. Moreover, although many aquatic mosses reproduce by asexual fragmentation (Glime 1984), studies on Fontinalis in North 514 New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 America have shown that such fragments need at least 2 months in contact with the substrate to attach (Glime et al. 1979). Such a long period in intimate contact with a cobble may seem unlikely in many high-gradient New Zealand streams, and may explain the observation that the percentage of streams supporting bryophytes in small alpine subcatchments was directly related to the average streambed stability of that subcatchment (Suren 1993). Transportation of suspended sediments during spates may cause significant abrasion to moss plants. This was suggested by Lewis (1973), who showed how apical tips ofEurhynchium riparioides became damaged from waterborne coal-dust particles. Lateral branches, which contained reproductive organs, were suppressed. This reduced the plants' ability for sexual reproduction, and may have affected their subsequent dispersal. Abrasion by suspended sediments has been implicated by other workers (e.g., Glime 1970; Conboy & Glime 1971) to explain absence of bryophytes in many streams, and is undoubtedly a major controlling factor in many New Zealand streams where suspended sediment yields are high (Duncan 1987; Hicks & Griffiths 1992). Although streambed stability is of fundamental importance to bryophyte distribution patterns, there was no difference in Pfankuch stability scores of any of the streams in the TWINSPAN sample groups, despite the fact that they supported distinctive species assemblages. This suggests that a particular stream must be of a "minimal stability" before bryophytes can occur there. Once this stability threshold has been met, however, the resultant species assemblage appears independent of measured physical or hydrological parameters. This is in sharp contrast to findings by Muotka & Virtanen (1995) who found that bryophyte species assemblages in 25 rivers and streams in northeastern Finland differed according to disturbance regime. Small-statured mosses such as Blindia acuta and Hygrohypnum luridum were characteristic of disturbed environments, whereas large, canopy-forming species (e.g., Fontinalis spp. and Hygrohypnum ochreacum) were characteristic of more stable environments. In contrast to the present study, Muotka & Virtanen (1995) quantified stability by using the "instability index" (Cobb & Flannigan 1990) as a relative index of substrate movement, defined as the tractive force at bankfull depth divided by the median substrate diameter. Low values of this index suggest a stable substrate at bankfull discharge. This instability index should give a similar result as the Pfankuch assessment, which is based largely on evidence of substrate movement during spates: the more common these bed-moving spates are, the higher the Pfankuch score. If the relative trends between species composition and stream bed instability can be compared between this study and that by Muotka & Virtanen (1995), then no similar clear demarcation between species composition or growth forms and streambed stability was evident in these New Zealand streams. This may reflect the more seasonal, and predictable nature of floods in Finland, in contrast to the generally unpredictable flow regimes characteristic of New Zealand streams. This unpredictability has often been used to explain the occurrence of ubiquitous "weedy" invertebrate taxa that can rapidly recolonise areas after flood events (e.g., Winterbourn et al. 1981; Mackay 1992). A similar ability to colonise a wide variety of unpredictable stream environments may also be displayed by some bryophyte taxa. The moss Fissidens rigidulus, for example, was found in almost half the streams surveyed. It has a flexible morphology, ranging from small clumps growing in tiny cracks on some boulders, to dominant, luxuriant "turfs" that cover much of the underlying substratum. Most of the c. 550 specimens collected in this and other survey work around the South Island have been sterile, despite the fact that collections have been made year-round, and that moss sporophytes are usually a persistent feature. This absence of sporophytes suggests a heavy reliance on vegetative propagation. Such a reproductive strategy would be particularly suited to unpredictable conditions, where both the timing of sporophyte production and spore release, as well as the survival and subsequent development of the protonema and new gametophyte would be reduced. Once the minimal stability threshold for bryophyte development to occur is met, flood frequency within that particular stream appears to have little detrimental effect on either bryophyte percentage cover, or on taxonomic richness. This contrasts sharply with studies on river periphyton, which are greatly affected by flood events (e.g., Homer et al. 1990; Biggs & Gerbeaux 1993; Biggs 1995). Indeed, periphyton biomass is usually negatively correlated with the duration of flood in a river (Biggs 1995). Such negative relationships between flooding and bryophyte cover were not observed in the Suren—Aquatic bryophyte distribution patterns -\ No Periphyton Bryophytes ) ( X i O ^- 1 ?- -" -v No Bryophytes Periphyton ( Low -High Streambed stability Fig. 9 Conceptual relationships of the theoretical responses of periphyton and bryophyte communities to streambed stability and the number of low flow days experienced by that stream. While stable substrata are beneficial for both periphyton and bryophytes, the predicted responses of low flow events to these plants are fundamentally different. present study. Indeed bryophyte growths were often most luxuriant in streams that flooded regularly, such as those in Fiordland and North-West Nelson. Unlike periphyton, bryophytes appear unaffected by high water velocities and are able to withstand extremely high shear stress. This is exemplified by their common occurrence in waterfalls characterised by fast, turbulent water. It is therefore unlikely that these plants would be easily washed away by high flows. Interactions between spates and a plant's inherent morphology can result in plants behaving differently under high-flow events. Thus Steinman & Boston (1993) reported a dramatic decline in percentage cover of the leafy liverwort Porella pinnata following an almost 200-fold increase in discharge in a small woodland stream in Tennessee, USA. This was thought to reflect the lower tensile strengths of this liverwort when compared with the co-occurring mosses Brachythecium cf. campestra and Ambiystegium sp., which actually increased in biomass after this flood. No such differences in tensile strength between mosses and liverworts were evident in this study, as liverworts often covered very large areas of the stream bed, despite occurring 515 in streams which had the highest flood frequency. More detailed studies are needed to address this interesting question of whether different bryophyte taxa, or growth forms have different flood resistances. Such differences are well established between different periphyton taxa, whereby filamentous communities are less resistant to high shear stress than diatomaceous communities (e.g., Reiter 1986; Uehlinger 1991), and where older communities are sloughed at lower velocities than newer communities (e.g., Biggs & Thomsen 1995). Similar interactions between a community's inherent nature and shear stress may exist for bryophytes, and help explain observed distribution patterns. Although spates had no consistent effect on established bryophyte communities, low-flow events appeared to adversely influence aquatic bryophytes, reflecting their intolerance of desiccation. This negative impact of low flows for bryophytes contrasts sharply with periphyton, where maximal biomass is only observed after extended periods of low flow (e.g., Biggs & Price 1987; Lohman et al. 1992). Indeed, flow perturbations which reduce periphyton biomass can make it hard to conduct surveys of these communities which are often best sampled during periods of summer low flow (Biggs & Hickey 1994). Fundamental differences thus exist with the habitat template of bryophytes and periphyton, and with their interactions with streambed stability and duration of low-flow events (Fig. 9). At low streambed stability, neither algae nor bryophytes will be found (e.g., Biggs 1995), reflecting high abrasion rates and substrate instability during high flows. Low-flow events, however, are often necessary for periphyton biomass to significantly accrue (e.g., Biggs & Price 1987), whereas bryophytes are generally absent from streams where low flows are common. This may explain the lack of bryophytes from streams in the Canterbury Plains and in the regions of Nelson, Dunedin, and Invercargill, where low flows are common (Pearson 1995). Catchment land use patterns had a dramatic influence on bryophyte communities, particularly liverworts. These plants were generally absent from catchments of pasture or pine. Catchments dominated by tussock grasslands, however, commonly had streams which supported mosses and liverworts, and liverworts were especially common in catchments dominated by undisturbed native forests. Mosses appeared slightly more tolerant of 516 New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 catchment modification, as pasture and pine contributed almost 25% and 10% of the total catchment area in which these plants grew. Land use is often correlated with underlying geology, so some of the observed differences in land use and the TW1NSPAN groups may reflect this more fundamental relationship. Observations of many streams around the South Island have, however, shown that bryophyte cover, especially that of liverworts, does decrease even within the same stream as soon as it flows from undisturbed native forest into open pasture. This is in sharp contrast to periphyton communities, which often attain highest biomass in catchments with developed pasture (Biggs 1990, 1995). Land use practices influence in-stream light regimes, and streams flowing through developed catchments often receive more light than streams flowing through undeveloped catchments (Beschta & Taylor 1988; Collier et al. 1995). The impact of altered light regimes on bryophytes is, however, inconclusive at present. Bryophytes are well known from their low light compensation points (e.g., Martin & Churchill 1982; Longton 1988) and are common in even shaded forested streams where autotrophic production is otherwise low (Naiman 1983; Suren 1992, 1993). They can, however, also grow under high light intensity regimes, where some species can produce secondary pigments to minimise damage to chlorophyll and proteins (Glime 1984; Glime & Vitt 1984). This ability to grow well under different light regimes was highlighted by Suren (1993) who observed similar bryophyte biomass in shaded and unshaded streams in the central southern alps. Land use changes greatly influence stream nutrient levels (e.g., Cooper et al. 1987; Cooper & Thompsen 1988), with generally elevated nutrient export in streams draining developed catchments. This was also found in this study, where bryophytes were generally absent from streams rich in nutrients, especially nitrogen. Relationships between bryophytes and stream nutrient status, however, are unclear. Elevated stream nutrients may be beneficial to some species, as increased organic pollution in the River Wear in England caused large increases in biomass of the moss Amblystegium riparium (Birch et al. 1988). Similarly, long-term experimental additions of N and P to an Alaskan river dramatically increased the cover of the mosses Acrocarpa sp. and Grimmia sp. (Bowden et al. 1994). Moreover, bryophytes have grown to nuisance proportions in highly eutrophic environments such as brewery effluent channels (Kelly & Huntley 1987) and in trickling filter beds at sewage treatment works (Hussey 1982). Conversely, a study by Miller (1990) showed a decline in the percentage cover of mosses below a sewage treatment outflow in the Waikato River. This was attributable to physical smothering of plants by luxuriant growths of filamentous algae which had greatly increased in biomass below the outflow. Bryophytes are known to provide a beneficial habitat for periphyton even in small oligotrophic alpine streams (Suren 1992), and may thus become even more smothered by periphyton in pasture streams. These are generally warmer and have a higher nutrient load than alpine streams, so periphyton development would be higher. This physical displacement of bryophytes by periphyton may explain why they were not common in nutrientrich streams flowing through pasture. Highest percentages of bryophyte cover occurred on bedrock areas within waterfalls and chutes. Bryophytes were, however, not common on bedrock areas within slower-flowing runs and pools. This observed preference of bryophytes to highly turbulent, fast-flowing habitats may reflect their need for atmospheric CO2 for photosynthesis (e.g., Bain & Proctor 1980). Boundary layers in these highly turbulent habitats are also likely to be thinner. Lacking roots, bryophytes must additionally rely on absorbing nutrients from the water column and so face similar problems as periphyton in obtaining enough nutrients by diffusion alone. Many of the upland streams where bryophytes grow are low in nutrients (e.g., Suren 1991), and so a thin boundary layer is advantageous in assisting them to obtain enough nutrients. A summary can thus be made to predict bryophyte occurrence in streams, and the resultant community type. This shows how bryophyte distributions are strongly related to macro-, meso-, and micro-scale variables (Table 7), some of which are highly intercorrelated. Thus substrate stability may be a direct consequence of substrate size and streambed slope, as well as catchment hydrology and underlying geology. Water quality may be a consequence of underlying geology, as well as of catchment land use. Bryophytes are generally absent from streams flowing through modified catchments of easily eroded geology and where low flows are common. These streams may also have relatively high nutrient levels (e.g., Close & Davies-Colley 1990). They will be dominated by small substrate sizes, and be of shallow gradient. As such, they will Suren—Aquatic bryophyte distribution patterns have few turbulent areas, but many runs and pools (Table 7). More high-gradient streams, flowing through catchments of more erosion-resistant rocks, and with fewer low flow events, however, will be expected to support bryophyte communities. Given stable substrates, flood events in these streams will be of little consequence, and bryophytes are expected to flourish. Microscale variables appear unimportant in regulating which biyophyte communities occur in a stream. Streams of sufficient stability will be colonised by bryophytes, but the resultant community appears influenced by meso- and macro-scale factors (Table 7). TWINSPAN analysis revealed that mosses and liverworts formed 517 distinctive communities, which differed on the basis of land use, geology, hydrology, and water quality. Liverworts had narrower niches than mosses, and were absent from many streams colonised by mosses. They also appeared more sensitive to changes in catchment land use, and to elevated stream nutrient levels. The summary presented here identifies major causative factors regulating aquatic bryophyte distributions, and influencing the type of resultant communities that develop in suitable sites. Substrate stability, and lack of low flows appear to be the variables influencing bryophyte distributions in the small streams studied, but more detailed studies are required to correctly predict the resultant Table 7 Summary of the interaction between macro, meso, and micro-scale variables and the presence or absence of aquatic bryophytes. Note how microscale variables control only the presence of bryophytes within streams, but have little influence on the resultant communities. Meso and macroscale variables, however, appear to control the type of bryophyte community. Presence of bryophytes Variable Resultant community MACROSCALE Modified Land use Unmodified High erodibility Geology Low erodibility •No Yes Some modification - Mosses Natural catchments Liverworts Strong sedimentary, igneous Mosses Metamorphic, igneous Liverworts Infrequent flooding Mosses Frequent flooding — Liverworts High phosphorus - Mosses Low phosphorus - Liverworts MESOSCALE Many low flows - -No Hydrology Water quality Few low flows — Yes High nutrients No Low nutrients Y Yes Small sizes No Large sizes Yes Smooth water No Turbulent water Yes Unstable No Stable Yes Shallow gradient No Steep gradient Yes MICROSCALE Substrate size Water nature Stream stability Stream slope < 518 New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 communities that eventually colonise these stable substrata. Little is known of nutrient or light requirements of bryophytes (but see Arts 1990), the influence (if any) that grazing has on bryophyte community composition (Suren & Winterbourn 1991 ), and the minimum duration a given substrate needs to be stable for it to be colonised by bryophytes (e.g., Glime et al. 1979). Further work is currently underway to determine how these and other factors affect bryophyte communities once they become established within streams. ACKNOWLEDGMENTS Thanks to Alex McFarlane, Brian Smith, Kim Dennison, and Cathy Kilroy for assistance with field work; and to Kathy Walter and Stephanie Brown for assistance with measuring hydrologica! parameters. Barry Biggs and Ian Hawes (NI WA) made helpful comments on earlier drafts, and two anonymous referees are thanked for their constructive criticism of the manuscript. Financial assistance from the New Zealand Lottery Board is greatly appreciated for provision of essential equipment. Funding came from the New Zealand Foundation for Research, Science and Technology, contract number CO1210. REFERENCES Allison, K. W.; Child, J. 1971: The mosses of New Zealand. Dunedin, University of Otago Press. 155 p. Allison, K. W.; Child, J. 1975: The liverworts of New Zealand. Dunedin, University of Otago Press. 300 p. Arts, G. H. P. 1990: Aquatic Bryophyta as indicators of water quality in shallow pools and lakes in The Netherlands. Annales botanici fennici 27: 19-32. Bain, J. T.; Proctor, M. C. F. 1980: The requirement of aquatic bryophytes for free CO2 as an inorganic carbon source: some experimental evidence. New phytologist 86: 393-400. Beever, J. E.; Allison. K. W.; Child, J. 1992: The mosses of New Zealand. 2nd edition, Dunedin, University of Otago Press. 214 p. Beschta, R. L.; Taylor, R. L. 1988: Stream temperature increases and land use in a forested Oregon watershed. Water resources bulletin 24: 19-25. Biggs, B. J. F. 1990: Periphyton communities and their environments in New Zealand rivers. New Zealand journal of marine and freshwater research 24: 367-386. Biggs, B. J. F. 1995: The contribution of flood disturbance, catchment geology and land use to the habitat template of periphyton in stream ecosystems. Freshwater biology 33: 101-120. Biggs, B. J. F.; Close, M. E. 1989: Periphyton biomass dynamics in gravel bed rivers: the relative effects of flows and nutrients. Freshwater biology 22: 209-231. Biggs, B. J. F.; Gerbeaux, P. 1993: Periphyton development in relation to macro-scale (geology) and micro-scale (velocity) limiters in two gravel rivers, New Zealand. New Zealand journal of marine and freshwater research 27: 39-53. Biggs, B. J. F.; Hickey, C. W. 1994: Periphyton responses to a hydraulic gradient in a regulated river in New Zealand. Freshwater biology 32: 49-59. Biggs, B. J. F.; Price, G. M. 1987: A survey of filamentous algal proliferations in New Zealand rivers. New Zealand journal of marine and freshwater research 21: 175-191. Biggs, B. J. F.; Thomsen, H. A. 1995: Disturbance of stream periphyton by perturbations in shear stress: time to structural failure and differences in community resistance. Journal of phycology 31: 233-241 Biggs, B. J. F.; Duncan, M. J.; Jowett, I. G.; Quinn, J. M.; Hickey, C. W.; Davies-Colley, R. J.; Close, M. E. 1990: Ecological characterisation, classification, and modelling of New Zealand rivers: an introduction and synthesis. New Zealand journal of marine and freshwater research 24: 277-304. Birch, S. P.; Kelly, M. G.; Whitton, B. A. 1988: Macrophytes of the River Wear: 1966, 1976, 1986. Transactions of the Botany Society of Edinburgh 45: 203-212. Bowden, W. B.; Finlay, J. C ; Maloney, P. E. 1994: Longterm effects of PO4 fertilization on the distribution of bryophytes in an arctic river. Freshwater biology 32: 445-454. Brusven, M. A.; Meehan, W. R.; Biggam, R. C. 1990: The role of aquatic moss on community composition and drift of fish-food organisms. Hydrobiologia 196: 39-50. Close, M. E.; Davies-Colley, R. J. 1990: Base flow water chemistry in New Zealand rivers 2. Influence of environmental factors. New Zealand journal of marine and freshwater research 24: 343-356. Cobb, D. G.; Flannigan, J. F. 1990: Trichoptera and substrate stability in the Ochre River, Manitoba. Hydrobiologia 206: 29-38. Collier, K. J.; Cooper, A. B.; Davies-Colley, R. J.; Rutherford, J. C ; Smith, C. M.; Williamson, R. B. 1995: Managing riparian zones: a contribution to protecting New Zealand's rivers and streams. Wellington, Department of Conservation. 39 p. Conboy, D. A.; Glime, J. M. 1971: Effects of drift abrasives on Fontinalis novae-angliae Sull. Castanea 36: 111-114. Suren—Aquatic bryophyte distribution patterns Cooper, A. B.; Thomsen, C. E. 1988: Nitrogen and phosphorus in streamwaters from adjacent pasture, pine, and native forest catchments. New Zealand journal of marine and freshwater research 22: 279-291. Cooper, A. B.; Hewitt, J. E.; Cooke, J. G. 1987: Land use impacts on streamwater nitrogen and phosphorous. New Zealand journal of forestry science 17: 179-192. Dawson, F. H. 1973: Notes on the production of stream bryophytes in the high Pyrenees (France). Annales de limnologie 9: 231-240. Death, R. G.; Winterbourn, M. J. 1995: Diversity patterns in stream benthic invertebrate communities: the influence of habitat stability. Ecology 76: 14461460. Downes, M. T. 1978: Taupo Research Laboratory chemical methods manual. Taupo Research Laboratory, DSIR, Report 102. Duncan, M. J. 1987: River hydrology and sediment transport. In: Viner, A. B. ed. Inland waters of New Zealand, pp. 113-137. Wellington, DSIR. Engel, J. J. 1980: A monograph of Clasmatocolea (Hepaticae). Fieldiana: Botany 3: 1-229. Englund, G. 1991 : Effects of disturbance on stream moss and invertebrate community structure. Journal of the North American Benthological Society 10: 143-153. Gauch, H. G. J. 1982: Multivariate analysis in community ecology. Cambridge, Cambridge University Press. 298 p. Glime, J. M. 1970: Zonation of bryophytes in the headwaters of a New Hampshire stream. Rhodora 72: 276-279. Glime, J. M. 1971: Response of two species of Fontinalis to field isolation from stream water. Bryologist 74: 383-386. Glime, J. M. 1984: Physico-ecological factors relating to reproduction and phenology in Fontinalis dalecarlica. Bryologist 87: 17-23. Glime, J. M.; Vitt, D. H. 1984: The physiological adaptations of aquatic musci. Lindhergia 10: 41-52. Glime, J. M.; Nissila, P. C ; Trynoski, S. E.; Fornwall, M. D. 1979: A model for attachment of aquatic moss. Journal of bryology 10: 313-320. Hicks, D. M ; Griffiths, G. A. 1992: Sediment load. In: Mosley, M. P. ed. Waters of New Zealand, pp. 229-248. Wellington, New Zealand Hydrological Society. Homer, R. R.; Welch, E. B.; Seeley, M. R.; Jacoby, J. M. 1990: Responses of periphyton to changes in current velocity, suspended sediment and phosphorus concentration. Freshwater biology 24:215-232. 519 Hussey, B. 1982: Moss growth on filter beds. Water research 16: 391-398. Hynes, H. B. N. 1961 : The invertebrate fauna of a Welsh mountain stream. Archiv fur Hydrobiologie 57: 344-388. Inoue, H.; Schuster, R. M. 1971: A monograph of the New Zealand and Tasmanian Plagiochilaceae. Journal of the Hattori Botanical Laboratory 34: 1-225. Jowett, I. G.; Duncan, M. J. 1990: Flow variability in New Zealand rivers and its relationship to instream habitat and biota. New Zealand journal of marine and freshwater research 24: 305-317. Jowett, I. G.; Richardson, J.; Biggs, B. J. F.; Hickey, C. W.; Quinn, J. M. 1991 : Microhabitat preferences of benthic invertebrates and the development of generalised Deleatidium spp. habitat suitability curves, applied to four New Zealand rivers. New Zealand journal of marine and freshwater research 25: 187-199. Kelly, M. G.; Huntley, B. 1987: Amblystegium riparium in brewery effluent channels. Journal of bryology 14: 792. Kimmerer, R. W.; Allen, T. F. H. 1982: The role of disturbance in the pattern of a riparian bryophyte community. American Midland naturalist 107: 370-383. Lamberti, G. A.; Resh, V. H. 1983: Stream periphyton and insect herbivores: an experimental study of grazing by a caddisfly population. Ecology 64: 1124-1135. Lewis, K. 1973: The effect of suspended coal particles on the life forms of the aquatic moss Eurhynchium riparioides (Hedw.). I. The gametophyte plant. Freshwater biologv 3: 251-257. Lohman, K.; Jones, J. R.; Perkins, D. 1992: Effects of nutrient enrichment and flood frequency on periphyton biomass in northern Ozark streams. Canadian journal of fisheries and aquaticsciences 49: 1198-1205. Longton, R. E. 1988: The biology of polar bryophytes and lichens. Cambridge, Cambridge University Press. 391 p. Mackay, R. J. 1992: Colonization by lotic macroinvertebrates: a review of processes and patterns. Freshwater biology 49: 617-628. Martin, C. E.; Churchill, S. P. 1982: Chlorophyll concentrations and a/b ratios in mosses collected from exposed and shaded habitats in Kansas. Journal ofbiyology 12: 297-304. McAuliffe, J. R. 1983: Competition, colonization patterns, and disturbance in stream benthic communities. In: Barnes, J. R.; Minshall, G. W. ed. Stream ecology: application and testing of general ecological theory, pp. 137-156. New York, Plenum Press. 520 New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 Miller, S. T. 1990: Investigation into the benthic vegetation of the upper Waikato River, New Zealand. Unpublished MSc thesis, University of Waikato, Hamilton, New Zealand. 336 p. Minshall, G. W. 1978: Autotrophy in stream ecosystems. Bioscience 28: 767-771. Minshall, G. W. 1988: Stream ecosystem theory: a global perspective. Journal of the North American Benthological Society 7: 263-288. Muotka, T.; Virtanen, R. 1995: The stream as a habitat templet for bryophytes: species' distributions along gradients in disturbance and substratum heterogeneity. Freshwater biology 33: 141-160. Naiman, R. J. 1983: The annual pattern and spatial distribution of aquatic metabolism in boreal forest watersheds. Ecological monographs 53: 73-94. Naiman, R. J.; Longarch, D. G.; Beechie, T. J.; Ralph, S. C. 1992: General principles of classification and the assessment of conservation potential in rivers. In: Boon, P. J.; Calow, P.; Petts, G. E. ed. River conservation and management, pp. 93-123. Chichester, John Wiley & Sons. Nolte, U. 1991: Seasonal dynamics of moss-dwelling chironomid communities. Hydrobiologia 222: 197-211. Pearson, C. P. 1995: Regional frequency analysis of low flows in New Zealand rivers. Journal of hydrology (New Zealand) 33: 94-122. Percival, E.; Whitehead, H. 1929: A quantitative study of the fauna of some types of stream-bed. Journal of ecology 17: 282-314. Pfankuch, D. J. 1975: Stream reach inventory and channel stability evaluation. U.S.D.A. Forest Service, Region 1, Missoula, Montana, U.S.A. 26 p. Reiter, M. A. 1986: Interactions between the hydrodynamics of flowing water and the development of a benthic algal community. Journal of freshwater ecology 3: 511-517 Scott, G. A. M. 1985: Southern Australian liverworts. Canberra, Australian Government Publishing Service. 216 p. Scott, G. A. M.; Stone, I. G.; Rosser, C. 1976: The mosses of Southern Australia. London, Academic Press. 495 p. Sheath, R. G.; Burkholder, J. M.; Hambrook, J. A.; Hogeland, A. M.; Hoy, E.; Kane, M. E.; Morison, M. O.; Steinman, A. D.; Van Alstyne, K. L. 1986: Characteristics of softwater streams in Rhode Island. III. Distribution of macrophytic vegetation in a small drainage basin. Hydrobiologia 140: 183-191. Steinman, A. D.; Boston, H. L. 1993: The ecological role of aquatic bryophytes in a woodland stream. Journal of the North American Benthological Society 12: 17-26. Stevenson, B. 1947: The home book of quotations, classical and modern. 5th, Revised edition, New York, Dodd, Mead and Company. 2812 p. Suren, A. M. 1991: Bryophytes as invertebrate habitat in two New Zealand alpine streams. Freshwater biology 26: 399-418. Suren, A. M. 1992: Enhancement of invertebrate food resources by bryophytes in New Zealand alpine headwater streams. New Zealand journal of marine and freshwater research 26: 229-239. Suren, A. M. 1993: Bryophytes and associated invertebrates in first-order alpine streams in Arthur's Pass, New Zealand. New Zealand journal of marine and freshwater research 27: 479-494. Suren, A. M.; Winterbourn, M. J. 1991: Consumption of aquatic bryophytes by alpine stream invertebrates in New Zealand. New Zealand journal of marine and freshwater research 25: 327-339. Uehlinger, U. 1991: Spatial and temporal variability of the periphyton biomass in a prealpine river (Necker, Switzerland). Archiv für Hydrobiologie 123:219-231. Walter, K. M. 1989: Index to hydrological recording sites in New Zealand. Christchurch, Hydrology Centre, Division of Water Sciences, Department of Scientific and Industrial Research. 181 p. Wilkinson, L. 1990: SYSTAT: The system for statistics. Evanston, IL. SYSTAT Inc. Winterbourn, M. J.; Rounick, J. S.; Cowie, B. 1981: Are New Zealand stream ecosystems really different? New Zealand journal of marine and j'reshwater research 15: 321-328. Zar, J. H. 1984: Biostatistical analysis. Second edition, London, Prentice-Hall International, Inc. 718 p. Suren—Aquatic bryophyte distribution patterns Appendix 1 Taxonomic list of all bryophytes (both obligately and facultatively aquatic) collected from 118 streams throughout the South Island. Nomenclature follows Allison & Child (1971), Beever et al. (1992), and Scott et al. (1976) for mosses; and Allsion & Child (1975), Engel (1980), Inoue & Schuster (1971), and Scott (1985) for liverworts. CLASS MUSCI SUBCLASS Family ANDREAEALES Andreaeaceae Andreaea subulata Harv. SUBCLASS Family BRYALES Amblystegiaceae Acrodadium cuspidatum (Hedw.) Lindb. Amblystegium sp. Cratoneuropsis relaxa (Hook. f. & Wils.) Fleisch. ex Broth. Drepanocladus aduncus (Hedw.) Warnst. Family Bartramiaceae Bartramia papillata Hook. f. & Wils. Breutelia pendula (Sm.) Mitt. Philonotispyriformis (R. Br. ter.) Wijk & Marg. Family Brachytheciaceae Brachythecium rutabulum (Hedw.) B. S. G. Eurhynchium austrinum (Hook. f. & Wils.) Jaeg. Family Bryaceae Bryum blandum Hook. f. & Wils. Bryum erythrocarpoides C. Muell. & Hampe Bryum laevigatum Hook. f. & Wils. Family Cryphaceae Ciyphaea tasmanica Mitt. Family Echinodiaceae Echinodium hispidum (Hook. f. & Wils.) Reichdt. Family Fissidentaceae Fissidens rigidulus Hook. f. & Wils. Fissidens sp. Family Grimmiaceae Racomitrium crispulum (Hook. f. & Wils.) Hook. f. & Wils. Family Hookeriaceae Calyptrochaeta cristata (Hedw.) Desv. Distichophyllum crispulum (Hook. f. & Wils.) Mitt. Distichophyllum pulchellum (Hampe ex C. Muell.) Mitt. Achrophyllum dentatum (Hook. f. & Wils.) Vitt & Crosby Achrophyllum quadrifarium (Sm.) Vitt & Crosby Family Hypnodendraceae Hypnodendron spininervium (Hook.) Jaeg. Family Hypopterygiaceae Cyathophorum butbosum (Hedw.) C. Muell. Family Lembophyliaceae Camptochaete gracilis (Hook. f. & Wils.) Par. 521 New Zealand Journal of Marine and Freshwater Research, 1996, Vol. 30 522 Family Mniaceae Plagiomnium novae-zealandiae (Col.) T. Kop. Family Neckeraceae Pendulothecium sp. Thamnobryum pandum (Hook. f. & Wils.) Stone & Scott Thamnobryum pumilum (Hook. f. & Wils.) Nieuwl. Family Pottiaceae Tridontium tasmanicum Hook. f. Family Ptychomniaceae Ptychomnion sp. Family Rhizogoniaceae Goniobryum subbasilare (Hook.) Lindb. Pyrrhobryum mnioides (Hook.) Manuel Family Seligeriaceae Blindia immersa (Bartr. & Dix.) Sainsb. Blindia lewinskyae Bartlett & Vitt Blindia magellanica Schimp. Blindia robusta Hampe Family Sematophyllaceae Sematophyllum uncinatum Stone & Scott Family Thuidiaceae Thuidium furfurosum (Hook. f. & Wils.) Reichdt. Thuidium laeviusculum (Mitt.) Jaeg. CLASS HEPATICAE SUBCLASS Family ANTHOCEROTALES Anthocerotaceae Anthoceros laevis L. SUBCLASS Family METZGERIALES Aneuraceae Aneura sp. Aneura prehensilis (Hook. f. & Tayl.) Hook. f. Aneura alterniloba (Hook. f. & Tayl.) Tayl. Riccardia (thin) Riccardia (thick) Family Hymenophytaceae Hymenophyton flabellatum (Labill.) Dum. ex Trev. Family Metzgeriaceae Metzgeria decipiens (Massal.) Schiffn. & Gott. Metzgeria fauriana Stephani Metzgeria rigida Lindb. Family Pallaviciniaceae Symphogyna prolifera Col. SUBCLASS Family MARCHANTIALES Marchantiaceae Marchantia berteroana Lehm. & Lindenb. Suren—Aquatic bryophyte distribution patterns SUBCLASS Family MONOCLEALES Monocleaceae Monoclea forsteri Hook. SUBCLASS Family JUNGERMANNIALES Balantiopsidaceae Balantiopsis cf. convexiscula Berggr. Balantiopsis tumida Berggr. Family Herbertaceae Herberta alpina (Steph.) Hodgs. Family Jungermanniaceae Cryptochila grandiflora (Lindenb. & Go Jungermannia inundata Hook. f. & Tayl. 523 Grolle Family Lejeuneaceae Lejeunea sp. Family Lepidoziaceae Hygrolembidium sp. Teleranea cf. dispar (Mont.) A. Hodgson Family Lophocoleaceae Chiloscyphus coalitus (Hook.) Nees Clasmatocolea humilis (Hook. f. & Tayl.) Grolle, var. humilis Clasmatocolea strongylophylla (Hook. f. Tayl.) Grolle Clasmatocolea vermicularis (Lehm.) Grolle. Hepatostolonophora paucistipula (Rodway) Engel Lophocolea bispinosa (Hook. f. & Tayl.) Tayl. Lophocolea gunniana Nees Lophocolea lenta (Hook. f. & Tayl.) Tayl. Lophocolea planiuscula (Hook. f. & Tayl.) Tayl. Lophocolea cf. semiteres (Lehm. & Lindenb.) Mitten Family Plagiochilaceae Plagiochila banksiana Gott. Plagiochila circinalis (Lehm.) Lehm. & Lindenbg. Plagiochila circumdentata Steph. Plagiochila deltoidea Lindenbg. Plagiochila fruticella (Tayl.) Hook. & Tayl. Plagiochila fuscella (Tayl.) Tayl. & Hook. Plagiochila lyallii Mitt. Plagiochila retrospectans (Nees) Nees Plagiochila stephensoniana Mitt. Family Radulaceae Radula buccinifera (Hook. f. & Tayl.) Tayl. Family Schistochilaceae Schistochila lehmanniana (Lindenb.) Stephani