Reciprocal Averaging Ordination Research Articles

Spatial and temporal variation in amphibian metacommunity structure in Chiapas, Mexico

Abstract:Amphibians are known to be sensitive to environmental change but their responses at the level of metacommunities to short-term environmental variability are poorly understood. We used field data from two protected areas, La Pera and Nahá (Chiapas, Mexico) to test for variation in metacommunity properties for two consecutive years (2009 and 2010). Amphibians and accompanying environmental data were recorded to a standardized protocol within each landscape, in four or five transects of 50 × 2-m per patch, for 30 and 31 patches, respectively. We found 23 species in La Pera and 30 in Nahá (21 species shared). Metacommunity structure was analysed using reciprocal averaging (RA) ordination by means of metrics for coherence, turnover and boundary clumping, with Spearman rank correlation used to examine relationships with environmental variables. The metacommunity structure varied differentially among the landscapes between years, being classed as quasi-Gleasonian in La Pera in both years, but Clementsian for Nahá in 2009 and Gleasonian for Nahá in 2010. In further illustration of variation between years, in 2009 the principal community gradient (RA axis 1) for La Pera was significantly positively correlated with altitude (r = 0.36), forest disturbance status (r = 0.78), mean canopy cover (r = 0.79) and mean litter depth (r = 0.67), while in 2010 it was correlated with latitude (r = 0.38), mean grass-layer height (r = 0.38), incidence of rainfall prior to sampling (r = 0.35) and presence of temporary ponds (r = 0.45). Our findings support the notion that amphibians respond to short-term environmental changes by individualistic movement within the landscape as well as via population dynamic responses.

The imprint of humans on landscape patterns and vegetation functioning in the dry subtropics

Dry subtropical regions (DST), originally hosting woodlands and savannas, are subject to contrasting human pressures and land uses and different degrees of water limitation. We quantified how this variable context influences landscape pattern and vegetation functioning, by exploring the associations between three groups of variables describing (i) human pressures (population density, poverty, and market isolation) and climate (water availability), (ii) landscape pattern (woody cover, infrastructure, paddock size, etc.), and (iii) vegetation functioning (magnitude and stability of primary productivity), in regions of Asia, Africa, Australia, and America. We collected data from global socioeconomic databases and remote sensing products for 4525 samples (representing uncultivated and cultivated conditions), located along 35 transects spanning semiarid to subhumid conditions. A Reciprocal Averaging ordination of uncultivated samples revealed a dominant gradient of declining woody cover accompanied by lower and less stable productivity. This gradient, likely capturing increasing vegetation degradation, had a negative relationship with poverty (characterized by infant mortality) and with market isolation (measured by travel time to large cities). With partial overlaps, regions displayed an increasing degradation ranking from Africa to South America, to Australia, to North America, and to Asia. A similar analysis of cultivated samples, showed a dominant gradient of increasing paddock size accompanied by decreasing primary productivity stability, which included all regions except Asia. This gradient was negatively associated with poverty and population density. A unique combination of small paddocks and high infrastructure differentiated Asian cultivated samples. While water availability gradients were related to productivity trends, they were unrelated to landscape pattern. Our comparative approach suggests that, in DST, human pressures have an overwhelming role driving landscape patterns and one shared with water availability shaping vegetation functioning.

Sediment transport and the distribution of shallow-water gorgonians

The objective of this study was to provide evidence supporting the contention that sediment transport (bedload) is a major environmental factor controlling the distribution of shallow-water gorgonian species. Gorgonians were surveyed annually from 2003–2008 in 3 replicated sets of transects at 2 shelf edge, 3 midshelf and 3 nearshore reefs on the southwest coast of Puerto Rico. Except for the shelf edge reefs, the transects were located in the fore reef terrace, break and base zones. A Reciprocal Averaging (RA) ordination analysis was used to identify the environmental factor underlying the distribution of gorgonian species. Substratum slope was used as a proxy of bedload levels to assess the overall RA results among and within reefs. Within reef RA scores decreased monotonically from the reef terrace to the slope to the base in all applicable cases. This pattern is consistent with sediment transport because increases in substratum slope from the terrace to the base of reefs should result in decreasing levels of bedload within reefs. RA scores were highly related to substratum slope (r2=0.79, p<0.01) supporting the premise that sediment transport controls the distribution of gorgonian species.

Vegetation of a Freshwater Dune Barrier Under High and Low Recreational Uses

BONANNO, S. E., D. J. LEOPOLD, AND L. R. ST. HILAIRE. Vegetation of a freshwater dune barrier under high and low recreational uses. J. Torrey Bot. Soc. 125:40-50. 1998.-The 27.4 km freshwater dune barrier on the eastern shore of Lake Ontario protects an extensive system of high quality wetlands. The vegetation was previously undescribed, and development pressures required management decisions for which data were needed. The objective of this study was to describe and compare the vegetation under high and low recreational use. We found four community types using hierarchical classification (TWINSPAN): (1) Ammophila breviligulata (beachgrass) (2) Toxicodendron radicans-Vitis riparia-Populus deltoides (brushland) (3) Quercus rubra-Acer rubrum (forest), and (4) Alnus incana (thicket). Reciprocal averaging ordination displayed communities along a first gradient of distance from the beach. A second gradient is likely related to disturbance. Species composition was similar in beachgrass and brushland communities under high and low recreational use. Under low use, forest and shrub thicket were present on the back of the barrier, but were absent under high use. Species richness and ground cover of vegetation, and density of colonizing species were lower on equivalent physiographic zones under high compared to low recreational use.

Niche segregation of terrestrial alpine beetles (Coleoptera) in relation to environmental gradients and phenology

At Finse, south Norway (60°36′N, 7°30′E), 247 pitfall traps were set in forty‐seven alpine habitats. More than 21,000 individuals were caught, including thirteen species of Carabidae, forty‐six Staphylinidae and twenty‐eight species from other families. The locations were selected to maximize variation in soil water contents between the habitats, but minimize variation in time of snowmelt and aspect. Soil water content of the sites was highly correlated with soil organic content, and showed little variation between dry and rainy periods. A reciprocal averaging (RA) ordination of the sites and coleoptera species gave results similar to a direct gradient analysis based on soil water content of the sample sites, clearly showing the importance of habitat humidity. Riverside and mire habitats had the most distinct species composition. Niche segregation among the species was studied based on guild, soil humidity preference, seasonal activity pattern and altitudinal range. Most species were well separated along one or more of these niche dimensions. No species from the wettest habitats were autumn active, otherwise all permutations of the described niche dimensions were used. Carabids and herbivores from various families were all spring active, while 38% of the staphylinids, mainly in the subfamily Omaliinae, were autumn active. In cases where species were not clearly separated, their behaviour, morphology, size, diet, etc. were considered.

Effects of wave energy, topographic relief and sediment transport on the distribution of shallow-water gorgonians of Puerto Rico

Environmental factors controlling the distribution of shallow-water gorgonians of Puerto Rico were inferred from a Reciprocal Averaging ordination analysis. The data set included several samples taken before and after the passage of Hurricane David and the mass mortality of the sea urchin Diadema antillarum. We could infer only a single environmental gradient associated with the distribution of gorgonians. Stations at opposite extremes of this gradient were characterized by combinations of high wave action with low topographic relief, or low wave action with high topographic relief. This gradient was also associated with sediment transport across the bottom (bedload). A detailed examination of ordination results in relation to Hurricane David and the Diadema mortality indicated that sediment transport, rather than water movement and topographic relief, is more directly related to the distribution of shallow-water gorgonians.

Environmental factors affecting physiognomic and floristic variation in an area of cerrado in central Brazil

ABSTRACTCerrado is a species-rich savanna-like vegetation that covers a large area in central Brazil. Soil, and particularly soil fertility, is considered to be the major factor determining the wide and diverse physiognomic-floristic gradients encountered within cerrado vegetation. To describe the vegetational variations within a cerrado in southern Mato Grosso state (15° 21'S, 55° 49'W), a 1 ha transect was located to pass through an interfluvial cerrado and climb a steep talus slope. The vegetational and environmental variations are described by means of field classification, direct gradient analysis and reciprocal averaging ordination. The effects of differences in ground water regime in the interfluvial cerrado and differences in inclination with associated rockiness in the talus cerrado are related to the vegetational variations. As the soils are very similar in texture and in the majority of chemical properties, the role of the above environmental factors was more easily distinguished. In the interfluvial cerrado, the vegetation variations are probably related to the seasonal fluctuation in water table level.

Parental Effects on Spatial Patterns of Plants: A Contingency Table Approach

We examined the effects of plant size and seed production on the seed shadows and mature offspring patterns of isolated parent plants of six biotypes of Panicum miliaceum. Spatial patterns of offspring were summarized in contingency tables that were classified by distance and orientation from the parent. These tables were analyzed with reciprocal averaging ordination and dispersions profile analysis. The first and second components of the ordination of a detailed data set reflected the distance and orientation, respectively, of the seed shadows from their parents. The mean infructescence height and the mean lateral spread of the parent plants were correlated with the first component of the ordination while the orientation of the parent plants was correlated with the second component. Little support was found for the hypothesis that biotypes with high numbers of seeds produced should have greater dispersal ability. However, when the individuals were grouped by seed production, irrespective of biotype, the more fecund groups had more distant cohorts of mature offspring. This latter trend was partly attributable to the increased density—dependent mortality near fecund parent plants and partly attributable to the larger sizes and hence larger seed shadows of these parents.

Composition and soil–environment analysis of prairies on Walpole Island, southwestern Ontario

Extensive prairie communities on Walpole Island Indian Reserve in Lake St. Clair, southwestern Ontario, were sampled for vegetation and soil profile characteristics. Two hundred and fifty-two species were found in the 20 stands. Dominant grasses included Calamagrostis canadensis, Andropogon gerardii, Sorghastrum nutans, Panicum virgatum, Poa pratensis, and Spartina pectinata. Dominant forbs were Pycnanthemum virginianum, Liatris spicata, Fragaria virginiana, Lysimachia quadriflora, and Viola papilionacea. Stands were ordered along a recognized moisture–substrate gradient divided into four segments; dry-mesic sandy, mesic sandy loam, wet-mesic sandy loam, and wet loam. Many species showed preferences for certain regions of the gradient. Species richness ranged between 44 and 91 species per stand. The composition of the vegetation was summarized for the four segments, using prevalent species lists. Effectiveness of the direct gradient analysis was assessed by ordination using reciprocal averaging and polar ordination. Percent sand, percent organic matter, and depth of A horizon were significant correlates of the primary axis. Frequent burning of the prairie by the Indians has effectively maintained these prairies. The composition of the Walpole Island prairies is similar to those in Michigan, Illinois, and Wisconsin, forming part of the eastern tallgrass prairie.

Ordination of vegetation change in Guadalupe Mountains, New Mexico, USA

Vegetation change over a nine-year period was studied in Guadalupe Mountains, New Mexico. Permanent transects in desert shrub vegetation were sampled in 1972 and 1980. Emphasis was given to shrubs because of their importance to big game diets. Univariate paired t-tests and reciprocal averaging ordination were used to detect and display coordinated changes in species composition over time. Despite apparently less browsing pressure in desert shrub vegetation in 1980 there were few significant changes in species composition. In addition, preferred forage species showed reduced reproduction while species of intermediate and poor forage value dis-played increased reproduction during this time. These data do not support traditional rangeland succession theory which states that enhanced reproduction of preferred species should follow grazing or browsing pressure reduction.

Successional Communities of Plants and Phytophagous Coleoptera

(1) A secondary succession on sandy soil in southern England is represented by seven experimental sites of known successional age, from the colonization of bare ground, through permanent pastureland to birch woodland. The development of the sites over 5 yr is described. (2) Communities of plants and phytophagous Coleoptera (weevils and leaf beetles) from sites of different successional age are defined using reciprocal averaging ordination. (3) The successional communities of phytophagous beetles are characterized by a sequence of dominant species, with the highest dominance occurring during very early colonization and mid-late succession; a pattern also seen in the herbs. (4) The species diversity of herbs and weevils is closely related throughout the succession. It is suggested that plant-species composition may be more important than plant structure in solely phytophagous groups of insects. (5) The host-plant associations of weevils and leaf beetles are discussed in terms of generalism-specialism and niche breadth. Measures for these are described. (6) The occurrence of a higher degree of generalism in weevils in late succession, than predicted in certain models of succession, is explained in terms of Rhoades' and Cates' theories of plant anti-herbivore chemistry.

Vegetation Patterns in and among Pannes (Calcareous Intradunal Ponds) at the Indiana Dunes National Lakeshore, Indiana

ABSTRACT: The relationships between plant species composition and dispersion, water chemistry and water depth/depth to water table were studied in five calcareous intradunal ponds (pannes) bordering the southern tip of Lake Michigan. The panne systems contained eight plant species threatened and endangered in Indiana. The aquatic zone was dominated by Chara, the pond edge by Rhynchospora capillacea, Juncus balticus and Utricularia cornuta, and the area surrounding the pond by Hypericum kal- mianum. The water chemistry was typical of hardwater ponds in the area, probably af- fecting species composition but not species dispersion within the pannes. A significant correlation between the first axis scores from a reciprocal-averaging ordination and water depth/depth to water was demonstrated. Panne species are fitted to a model based on hydrology proposed by van der Laan for dune-slack vegetation in the Neth- erlands. INTRODUCTION There is a series of intradunal ponds (pannes) just inland from the beach dunes at the Indiana Dunes National Lakeshore. These pannes differ in origin, age and plant species composition from numerous interdunal ponds found farther inland in older systems. The pannes now provide habitat for plant species found nowhere else in Indiana. Depressions which support the pannes were scoured out by wind around the turn of the century when Lake Michigan and groundwater levels were relatively low. When lake levels rose in the 1930s, the water table also rose and was exposed in the base of the depressions. These shallow ponds (< Im in depth) are normally perennial and the water is well-buffered with bicarbonates. Pannes are associated with systems along the E coast of Lake Michigan, along the southern shore of Lake Huron, and in one known location on Lake Superior. The number of plant species these have in common with the pannes at Indiana Dunes de- creases as distance from them increases (Chapman, 1982). Similar systems called dune slacks are associated with marine systems in the Netherlands (van der Laan, 1979; Londo, 1974), North Wales (Onyekwelu, 1972), North Devon (Willis

Multivariate Analysis of Protist Communities in Lentic Systems1

ABSTRACT. The purpose of this research was to examine the roles of physico‐chemical parameters in structuring protist communities that colonize artificial substrates. Polyurethane foam (PF) substrates were placed in five lentic systems in northern lower Michigan during summer 1983. These lentic habitats represented a range of trophic states and included three lakes, a bog, and a marsh. Triplicate PF substrates were sampled after 1, 3, 7, 14, 21, and 42 days of exposure. During this study, 90 living protist samples were examined for the number and kinds of species. Water samples were analyzed for pH, conductivity, alkalinity, chloride, silica, temperature, dissolved oxygen, ammonia, and total and ortho‐phosphate concurrent with artificial substrate collections. A total of 546 protist species was recorded. Only seven species were found in over 50% of the samples and 121 species were found in only one sample. The 96 most common species were examined in relation to environmental parameters using several multivariate statistical procedures. Factor analysis (principal components with varimax rotation) performed on the total environmental data set showed that three composite factors explained 85% of the variability of the data set. A reciprocal averaging ordination (RAO) was used to reduce species presence/absence data and to separate samples graphically by their species composition. Significant correlations, with RAO‐generated axes from all five systems, were found for pH, oxygen, and a nutrient factor to axis 1. Factor analysis on the physico‐chemical parameters of the three lakes showed that three factors explained 71% of the environmental data set variability. The RAO‐generated axis (axis 1) was correlated with silica, ortho‐phosphate, and Factor 2, which was primarily comprised of loadings from ortho‐phosphate. These techniques support the hypothesis that a limited number of environmental parameters strongly affect protist community composition.

Recognition of peat-forming plant communities from their peat deposits in two south Swedish bog complexes

Peat samples, 3 015 from 103 boring points, on two mires (Akhult mire, Store Mosse mire) south Sweden, have been subjected to macrofossil analysis. Based on plant remains, 9 peat groups were distinguished in the field. A further classification using phytosociological methods revealed 29 peat types. The affinities between the peat types were determined from TABORD classification and a Reciprocal Averaging ordination. The primary floristic differentiation is correlated with a gradient from treeless-to wooded stands, which coincides largely with the mire expanse-mire margin gradient. The poor-rich gradient seems to parallel the treeless-wooded gradient as well and may reflect the natural conditions in this mire before it was affected by man. The hummock-mud-bottom gradient is easy to distinguish in peat, formed by bog communities, but is not distinct in peat formed by fen communities and impossible to detect in peat dominated by wood remains. The amount of identifiable remains depends on the decomposition, which is determined by (1) the period of time the plant litter stays in the acrotelm, and (2) the nutrient status. The decomposition is greatest in fen-peat with abundant wood remains. This probably depends on a good supply of oxygen caused by greater horizontal water movements and better nutrient status.

Vegetation on Limestone in the Huachuca Mountains, Arizona

-Major plant communities on open limestone slopes (1600-2100 m) in the Huachuca Mountains of southeastern Arizona are pine-oak woodland, mesic phase Cercocarpus breviflorus scrub, and xeric phase Cercocarpus scrub. Other communities of more limited areal extent include pine forest, deciduous woodland, and desert grassland. Vegetation patterns as represented by the first axis of reciprocal averaging ordination are strongly correlated (Spearman's r = 0.86, p 2 mm diameter). Levels of organic matter and exchangeable calcium were high in these soils, whereas extractable iron and aluminum were generally undetectable. Data from Fort Huachuca, located at 1422 m near the northeastern base of the range, illustrate general climatic trends in the study area. The mean annual precipitation is 372 mm, of which 114 mm (31%) falls in the month of July, the start of the summer rainy season (Sellers and Hill, 1974). Wallmo (1955) estimated that precipitation at the highest elevations in the Huachucas exceeds 750 mm annually, and snow accumulations in winter may be a meter or more (Swarth, 1904). Mean annual temperature at Fort Huachuca is 16.8°C (7.9°C and 25.3°C in January and July, respectively). Regional lapse rates are -0.0073°C'm-' for mean July temperature and 0.347 mm.m-' for mean annual precipitation (Wentworth, 1976). Vegetation at the lowest elevations in the Huachucas (below 1370 m) is desert scrub, with a zone of desert grassland (1370-1520 m) above (Wallmo, 1955). Between 1520 m and 2130 m many sites support dense scrub dominated by evergreen oaks (encinal). Pine-oak woodlands are found above 1680 m and extend to 2290 m, while mixed coniferous forests occur as low as 1980 m and extend to the highest peaks. Occasional stands of quaking aspen (Populus tremuloides) are found on the highest, coolest sites. Superimposed on the elevational zonation of communities are shifts caused by topography and parent material. Exposed ridges and open south-facing slopes are xeric while north-facing slopes, draws, and canyons of similar elevation are relatively mesic. Vegetation on limestone usually differs in species composition, structure, and physiognomy from that found on soils derived from the more widespread acidic rocks, such as granite and quartzite (cf. Whittaker and Niering, 1968b; Wentworth, 1981). METHODS.-Topographic and geologic maps were employed to locate sample sites representing the variation in elevation, slope, aspect, and topographic position on limestone. During the summer and fall of 1973, vegetation at each of 77 sites was measured in a 0.1 ha (20 X 50 m) quadrat extending 10 m to either side of a 50 m tape (Whittaker and Niering, 1965). Site information included location, bedrock type, elevation, slope, aspect, topographic position, and exposure. Visual cover estimates for all vascular plant species (and miscellaneous classes such as litter and bedrock) were made in columns projected vertically upward from the soil surface in 25 m2 subquadrats arranged at 1 m intervals on alternate sides of the 50 m tape. Cover estimates were then averaged over the 25 subquadrats for each species or miscellaneous class. Each 0.1 ha quadrat was also searched for species not occurring in the subquadrats. The vegetation of each quadrat was classified by physiognomy and dominance into one of several community types. The classification was refined through use of reciprocal averaging ordination (Wentworth, 1976) and also confirmed by subsequent numerical classification of the quantitative data (Wentworth, unpublished). The various types and the numbers of quadrats allocated to these were: pine-oak woodland of open slopes (15), mesic phase Cercocarpus scrub of open slopes (23), xeric phase Cercocarpus scrub of open slopes (22), canyon deciduous woodland (2), canyon pine forest (3), scrub and woodland of draws (6), high elevation ( > 2100 m) scrub and woodland (2), low elevation ( < 1600 m) desert grassland (2), and vegetation of heterogeneous parent material (2). Nearly all of the 77 sites were judged to be free of disturbances. However, the 2 low elevation sites in desert grassland had burned recently (alternative sites were not available), and fire scars on large shrubs or trees in 6 higher elevation sites provided evidence of past fires. Only 1 site was judged to have been grazed by cattle in the recent past, while 2 others showed evidence of minor disturbance by native fauna. No evidence of logging or woodcutting was noted in any of the sites sampled. 386 vol.30, no.3 This content downloaded from 207.46.13.180 on Thu, 08 Sep 2016 06:03:11 UTC All use subject to http://about.jstor.org/terms Wentworth-Vegetation on Limestone 6700 I PINEOAK WOODLAND / w Pinus cembroides / 6500 o00 Quercus arizonica */ zu Cercocarpus ./ < u breviflorus

Relationships between Surface Sediment Diatom Assemblages and Lakewater Characteristics in Adirondack Lakes

Relationships between surface sediment diatom assemblages and lakewater characteristics were studied in 38 lakes in the Adirondack Mountains of northern New York. Most of the lakes are dilute, poorly buffered, and oligotrophic to mesotrophic. The diatom flora (240 taxa identified) is typical for circumneutral to acidic lakes. The purposes of this study were to identify the environmental factors most strongly related to the distributions of diatom taxa and the overall composition of diatom assemblages, and to derive equations to infer lakewater pH from diatom assemblage data. Relationships between diatom assemblages and environmental gradients were analyzed using reciprocal averaging ordination (RA). Correlations between RA axis 1 and pH—related factors were strong (pH, r2 = 0.90; alkalinity, r2 = 0.81; p(Ca positive charges), r2 = 0.67; pMg, r2 = 0.78; pNa, r2 = 0.48; p—total Al, r2 = 0.58; conductivity, r2 = 0.47). Correlations were weaker (but still statistically significant) with elevation, epilimnion temperature, and concentrations of SO4, Cl, and Si. Total P, chlorophyll a, water color, and mean depth were not important in explaining differences among assemblages. Cluster analysis of lakes and diatom taxa, multiple regression of taxon cluster groups with lakewater pH, and the occurrence patterns of individual taxa also indicated that pH—related factors are closely associated with diatom distributions. Euplanktonic taxa were present only rarely in surface sediments of lakes having a pH &lt;5.8, alkalinity &lt;20 μmol/L, and concentrations of Ca positive charges &lt;100 μmol/L and total Al &gt;60 mg/L. Predictive equations were derived for inferring lakewater pH from diatom assemblage data. Most were based on approaches of Hustedt (1939), Nygaard (1956), Merilainen (1967), and Renberg and Hellberg (1982). Agreement between predicted and measured pH was very good; r2 values ranged from 0.61 to 0.94 and se from 0.28 to 0.60 pH units. These predictive relationships can be used to interpret stratigraphic diatom assemblages to reconstruct lake pH histories. No significant relationships existed between Araphidinae: Centrales ratios and measures of trophic status.

Multivariate analysis of vegetational variation in different habitats at Omayed, Egypt

The application of indicator species analysis has proved useful in classifying stands into groupings coinciding with topographic variations. 41 indicator pseudo-species are identified in the vegetation along a phytosociological gradient: some are characteristic of specific habitats, and the others indicate transition between groups of habitats. The use of qualitative estimations (transformed density records) in multivariate analysis in the present study indicates that they may be more preferable than quantitative estimations, because of their easy and fast recording in the field, and the less comprehensive computations, while yielding precise results. The X-axis of the reciprocal averaging ordination is related to the salinity and fertility gradients, while the Y-axis reflects soil texture. The phytosociological gradients of the canonical variate axes reflect essentially contrasts between groups of species, each correlated with one or the other of the environmental gradients. Comparing the results obtained by the two ordination methods (reciprocal averaging and canonical variates) we find that the first method detects the overall phytosociological changes along strong environmental gradients which would be helpful in studying large surveys. On the other hand, the second technique, which is mainly predictive, is more sensitive to changes within both the phytosociological and environmental gradients and can detect the impact of these changes on the overall variance of each gradient.

Mallomonadacean (Chrysophyceae) assemblages and their relationships with limnological characteristics in 38 Adirondack (New York) lakes

The siliceous scales of 30 mallomonadacean taxa were identified and counted from the surficial sediments of 38 Adirondack lakes. The scales were usually abundant and were always well preserved. The Adirondack flora is typical of acidic to circumneutral oligotrophic lakes. Common taxa include Mallomonas acaroides, M. caudata, M. crassisquama, M. hamata, M. pseudocoronata, M. punctifera, Synura echinulata, S. petersenii, S. sphagnicola, S. spinosa, and Chrysosphaerella longispina. The abundance of certain species (e.g., Mallomonas hindonii, M. hamata) appears to be a reliable indication of acidic waters. Reciprocal averaging (RA) ordination was used to determine if distribution of the mallomonadacean assemblages corresponded with major environmental gradients. Correlations between the first RA axis scores and lakewater pH (r2 = 0.63) and related factors (log10 alkalinity, r2 = 0.66; pCa, r2 = 0.39; pMg, r2 = 0.43) were strongest. Relationships were weaker with average depth, summer epilimnion temperature, elevation, NO3, and total Al, and were not statistically significant with conductivity, color, Secchi disc transparency, total P, chlorophyll a, SO4, and Si. Cluster analysis grouped both lakes and taxa into categories best explained by their pH-related characteristics. Analysis of surficial sediment from lakes with known limnological characteristics is a useful approach for the study of the ecology and taxonomy of the Mallomonadaceae. Our data demonstrate that mallomonadacean distributions reflect lake conditions and, therefore, are potentially useful indicators in limnological and paleolimnological studies.

Forest Composition on the Lake Erie Islands

The Lake Erie Island archipelago includes 22 islands which are distinct from the neighboring Canadian and American shores in climate, topography, surface geology and soils. The forest communities of these islands were analyzed for species composition, relationships to environmental factors and similarity to mainland forests. When analyzed by reciprocal averaging ordination, forest stands split into two tight clusters. Upland forests, dominated by Acer saccharum and Celtis occidentalis, occurred where bedrock level> lake level. Lowland forests occurred on sites where bedrock level 65 km2. Dolomites and limestones of Silurian and Devonian age form the bedrock of the region (Core, 1948; Hamilton and Forsyth, 1972). The soils of the upland, exposed areas of the islands are shallow, generally less than 40 cm in depth, have many bedrock outcrops, and are low in available water (Redmond et al., 1971; Musgrave and Derringer, 1981). Most of these upland soils are currently classified as Castalia, very stony, silt loams (Musgrave and Derringer, 1981). Lowland areas and areas not exposed to direct lake action a-re covered by deeper soils (up to 175 cm) formed on limy glacial till (Redmond et al., 1971; Musgrave and Derringer, 1981). The climate of the Lake Erie Islands is temperate, humid-continental (Cooper and Herdendorf, 1977). It differs in several respects, however, from the neighboring Ohio, Michigan and Ontario shores. Due to the buffering effect of the lake, the average frostfree period is 205 days, 27% longer than Tiffin, Ohio, 50 km to the S (Verber, 1955). Significantly less precipitation falls on the islands than on the mainland. Annual

Geographic variation in habitat structure for the wood warblers in Maine and Minnesota.

The habitats occupied by species of wood warblers (Emberizidae) were compared at two study areas, Itasca State Park, Minnesota and Mount Blue State Park, Maine. Univariate comparisons of each variable of habitat structure show geographic differences for each species of warbler. Habitats available were also different because small trees were always more dense in Maine than in Minnesota. The Black-throated Green Warbler had the most dissimilar habitat with 9 of 11 variables different at the two sites.Cluster analysis identified four generalized habitat groups containing (1) species occupying territories with high percent shrub cover, (2) forest species from Maine, (3) forest species from Minnesota and (4) open country species. Reciprocal averaging ordination was used to identify habitat gradients at each site. The first axis of the Maine and Minnesota ordinations was a gradient from open country with dense ground cover to forest vegetation. The second axes differed, however. In Minnesota, the gradient separated medium deciduous trees from large conifers, whereas in Maine, vegetation graded from medium and large deciduous trees to coniferous habitats. Spearman rank correlation indicated that the warblers were similarly arranged along both habitat axes at each site despite differences in axis loadings of habitat variables.A combined reciprocal averaging ordination separated forest and shrub-forest edge species in Maine from the same two species groups in Minnesota along a smaller to larger tree axis. The results clearly demonstrate that habitat structure is not consistent throughout the range of many widely distributed species. It is suggested that the similar arrangement of species along the habitat axes probably results from an individualistic distribution of opportunistic bird species. Variation is probably induced at a site level by intraspecific competition for territories, small-scale vegetation dynamics, and resource fluctuation that occurs both within and between seasons.