An ecological analysis of wetlands in the high mountain jalca above 3700 m elevation in the Andes near Cajamarca, Peru, indicated that most wetlands are groundwater-supported peat-accumulating fens. The floristic composition of fen communities was controlled largely by groundwater chemistry, which was highly variable and influenced by watershed bedrock composition. Watersheds with highly mineralized rock discharged water as acidic as pH 3.7, which was high in CaSO4, while watersheds with limestone, marble, and skarn produced groundwater as basic as pH 8.2 and high in CaHCO3. Of the 125 plots sampled in 36 wetland complexes, >50% of plots had at least 3 m of peat, and 21 plots had peat thicker than 7 m. Most soil horizons analyzed had 18 to 35% organic carbon, indicating high C storage. A total of 102 vascular plants, 69 bryophytes, and 10 lichens were identified. Study plots were classified using TWINSPAN into 20 plant communities, which were grouped into four broad categories by dominant life form: (1) cushion plant communities, (2) sedge- and rush-dominated communities, (3) bryophyte and lichen communities, and (4) tussock grass communities. Direct gradient analysis using canonical correspondence analysis indicated that Axis 1 was largely a water chemistry gradient, while Axis 2 was a complex hydrology and peat thickness gradient. Bryophytes and lichens were more strongly separated in the ordination space than vascular plants and were better indicators of specific environmental characteristics.
In the world's tropical regions, the largest area of alpine vegetation above the limit of closed canopy forest and below the permanent snowline is in the Andes Mountains of northern South America (Smith and Young, 1994). This landscape includes broad continuous highlands and isolated peaks that rise to elevations of 3500 m to over 6000 m. The northern Andes in Venezuela, Colombia, and Ecuador receive abundant rainfall, and their alpine ecosystems are termed páramo (Troll, 1968; Walter, 1985). From central Peru south to central Chile the climate is drier, and the alpine ecosystems are termed puna (Smith and Young, 1994). The transition between páramo and puna occurs in northern Peru, between 4.5 and 8°S latitude, and is termed jalca (Rundel et al., 1994; Luteyn, 1992). Cloud forest forms the upper treeline in the east, while deep valleys on the Andes' western slopes are in rain shadows and support thorn scrub and desert below the montane vegetation. In the jalca, tree-sized Polylepis (Rosaceae) and Lupinus (Fabaceae) occur in scattered patches within a high-elevation steppe.
The alpine flora of South America is the most species-rich of all high mountain regions in the tropics (Smith and Cleef, 1988). It contains the smallest number of boreal and temperate zone species and the greatest proportion of endemics (Luteyn and Churchill, 2000). There have been analyses of regional biota, endemism, and historical and biogeographic relationships of the flora (Young and Reynel, 1997; Weigend, 2002; Young et al., 2002). However, relatively little is known about factors controlling vegetation composition (Squeo et al., 2006; Ginocchio et al., 2008).
Uplands in the jalca, páramo, and puna are dominated by grass species in the genera Calamagrostis, Poa, and Festuca. Taller rosette plants including species of Puya are common, along with woody species of Senecio and other genera, particularly in Asteraceae. Most areas are cultural landscapes (Young and Reynel, 1997) heavily grazed by domestic livestock and burned annually (Suárez and Medina, 2001).
Seasonally abundant precipitation has allowed the development of numerous wetlands in the jalca, which have been referred to as cushion mires (Bosman et al., 1993), highland bogs (Wilcox et al., 1986), and soligenous (formed on slopes) peatlands (Earle et al., 2003), and are locally and regionally known as bofedales (Squeo et al., 2006). Ombrogenous (rain-fed) bogs are known to occur in southern Chile (Kleinebecker et al., 2008) but are unlikely to occur in the seasonally wet jalca. Few wetlands in the jalca are known to be Sphagnum-dominated (Earle et al., 2003), and not all are soligenous or dominated by cushion plants (Squeo et al., 2006). It is unclear what proportion of wetlands have accumulated peat and should be termed mires, bogs, or peatlands. Little is known about the physical factors controlling wetland distribution, wetland types, and their floristic composition in the jalca. The goals of this study were to (1) characterize the types of wetlands and the plant communities occurring within them in a region of jalca in northern Peru, (2) compile a list of species occurring in the wetlands, (3) determine the proportion of wetlands that are peat-accumulating, and (4) identify the physical factors controlling the floristic composition of vegetation.
LANDSCAPES, LANDFORMS, AND HYDROLOGY
The study area is approximately 150 km2, located in Cajamarca Department, in the Andean Cordillera of northern Peru at elevations ranging from 3700 m to 4200 m (Fig. 1). Four rivers originate in the study area: Chirimayo to the east, Challhuagón to the south, Rio Grande to the north, and Mamacocha to the west. Most wetlands occur on slopes or at the toe of slopes and appear to be supported by groundwater discharging from relatively localized and small hillslope aquifers. Terminal moraines deposited perpendicular to valley bottoms retard runoff and have created the hydrologic template for wetland formation in many areas. Other wetlands occur up-gradient from and adjacent to lakes or pond basins that had filled with peat.
Bedrock in the study area is complex and highly variable, consisting of five major rock types: Cretaceous sedimentary rocks including limestone, Pliocene volcanic rocks, Eocene/Miocene intrusive igneous rocks, metamorphic rocks where the original parent material has been altered, and surficial Pleistocene sediments (Davies, 2002).
Cation and anion concentrations, and acidity, differ in groundwater originating in each bedrock type. Water emerging from highly mineralized metamorphic rock is acidic due to the formation of sulfuric acid from pyrite weathering (Cooper et al., 2002; Verplank et al., 2006), while the dissolution of calcium carbonate from limestone or marble (metamorphosed limestone) produces alkaline water (Macpherson et al., 2008). Because of the numerous lithologic contact zones, where two or more rock types are adjacent to each other, a single wetland may straddle bedrock types and receive runoff and groundwater from more than one chemically distinct water source (Fig. 2)
The rainy season in the jalca occurs from October to April, contrasting strongly with the dry season from May through September. Precipitation arrives from northeasterly winds that bring warm and humid air masses from the Amazon Basin. Mean annual precipitation is 1180 mm at Brillantana and 1400 mm at Minas Sipan, both privately operated weather stations located near the study area, and is highly variable from year to year. Sites that are saturated to the surface with water-filled pools during the rainy season may have dry pools and water tables >40 cm below the soil surface during the dry season. Monthly temperatures are relatively consistent year round, with January being the warmest month with a mean temperature of 7.8 °C.
FIELD METHODS AND SPECIES NOMENCLATURE
We sampled all 36 major wetland complexes in the study area (Fig. 1). In each complex we identified homogeneous stands of vegetation and collected data on the physical environment and floristic composition of each stand. The number of stands analyzed in each wetland depended upon its size and heterogeneity, and varied from one to seven. One 20 m2 plot was analyzed in each stand for plant species composition, and we visually estimated percent canopy coverage by species. A total of 125 plots were analyzed. All data were collected in September and October 2005. Collections of unknown vascular plants were identified at Herbario de la Facultad de Ciencias (MOL), Universidad Nacional Agraria La Molina, Lima, Peru, where all vouchers are housed. We used the APG II (Angiosperm Phylogeny Group) system of plant classification (Stevens, 2001; Angiosperm Phylogeny Group, 2003; http://www.mobot.org/MOBOT/research/APweb/) for species nomenclature. The geographic distribution of vascular plant taxa were investigated using the U.S. Department of Agriculture PLANTS database < http://plants.usda.gov> and The Missouri Botanical Garden's website database TROPICOS < http://www.tropicos.org>.
In each plot we measured the pH of groundwater that filled a 20- to 40-cm-deep soil pit using an Orion model 250A pH meter with combination electrode. We measured the wetland's slope using a Suunto clinometer, aspect with a compass, peat thickness with a tile probe pushed until rock or dense mineral soil was encountered, groundwater level in the open soil pit after it had equilibrated for 1 h, and soil temperature at 13 and 28 cm depth with thermometers.
Lichens were identified by Dr. S. Will-Wolf, Sphagnum by Dr. Richard Andrus, other mosses and liverworts were identified by Dr. W. Weber with the exception of Scorpidium and Drepanocladus (L. Hedenäs), Breutelia and Philonotis (Dana Griffin III), Campylopus (Jan-Peter Frahm), Leptodontium (Richard Zander), and Dicranum (Robert Ireland). Voucher specimens of all cryptogams are at the University of Colorado at Boulder (COLO).
LABORATORY ANALYSIS OF WATER AND SOIL SAMPLES
Water was collected from 32 wetland complexes for analysis of dissolved ions, metals, and nutrients. The lack of shallow groundwater made it impossible to collect water from 4 wetlands. Two 1 L sample bottles were filled from the 40-cm-deep soil pit after it filled with fresh clean groundwater. Samples were stored on ice in the field and for transport, and in a refrigerator until analyzed by ALS Environmental in Lima, Peru. One sample bottle from each site was acidified in the field to stabilize the metals. All samples were filtered in the laboratory. We used standard analytical procedures recommended by the American Public Health Association (APHA, 2005) for analysis: Ca, Cu, Mg, and Fe using direct nitrous oxide-acetylene flame method (APHA 3111-D), Na and K using flame photometric method (APHA 3500-K-B), HCO3 and CO3 using titration (APHA 2320-B), SO4 using turbidimetric method (APHA 4500-SO4 E), and total P using persulfate digestion method (APHA 4500-P-E). Total N was determined using the Kjeldahl method, following ASTM method D5176-91 (standard method for total chemically bound N, http://amaec.kicet.re.kr/cd_astm/PAGES/D5176.htm). Percent organic carbon (OC) for one soil sample in each stand was collected at ∼30–40 cm depth from a soil pit and analyzed using a LECO soil analyzer.
Soils in the upper 40 cm horizon were analyzed for % OC and organic horizon thickness. Soils containing <18% OC were classified as mineral horizons, while those with >18% OC were classed as organic horizons following USDA Soil Taxonomy (USDA, 1999). If organic horizons were >40 cm thick, the soils were considered to be peat, and wetlands with peat soils to be peatland ecosystems. Sloping wetlands that lacked peat soils were classified as wet meadows, while those on lake shores that flooded deeply were classified as marshes.
Divisive cluster analysis using Two Way Indicator Species Analysis (TWINSPAN) was performed with the computer program PC Ord (McCune and Mefford, 1999) to sort plots into groups based upon their floristic composition. Braun-Blanquet table sorting methods (Mueller-Dombois and Ellenberg, 1975) were used to further refine the plot groupings and produce a vegetation classification.
Direct gradient analysis was used to identify the major environmental gradients controlling the floristic composition of study stands. Canonical correspondence analysis (CCA) was performed using the default settings in the computer program CANOCO (ter Braak, 1986, 1992; ter Braak and Prentice, 1988). Forward selection was used to test the main environmental variables. Variables were tested for their significance using a Monte Carlo permutation test (499 permutations), and significant variables (P < 0.05, except Cu which was accepted at P < 0.01) were included in the analysis and are shown on the canonical ordination diagrams. A Monte Carlo permutation test (499 runs) was performed to evaluate the significance of the environmental variables and the first ordination axis in explaining the vegetation composition (P < 0.05). In addition, a Monte Carlo permutation test (499 runs) was used to test the significance of all canonical eigenvalues. Raw data were not transformed prior to analysis. Two CCA diagrams are presented, one for vegetation composition of plots, and a second for plant species.
Wetland groundwater pH ranged from 3.7 to 8.2. For the 125 stands analyzed, 35 had pH readings from 3.7 to 5.0 indicating highly acid conditions, 32 from 5.0 to 6.0 indicating moderately acid conditions, and 58 from 6.0 to 8.2 indicating slightly acid, neutral to alkaline conditions. Using the geochemical classification of peatlands developed by Sjörs (1950), the three groups of pH ranges listed above would be classified as bogs or poor fens, intermediate fens, and rich to extreme rich fens, respectively. Most stands had pH values from 4.5 to 7.0, but many stands had more acid or basic values (Fig. 3) indicating the broad range of geochemical environments affecting wetland water chemistry in the study area.
The most abundant anions in sampled groundwater were HCO3, SO4, and Ca (Table 1). HCO3 ranged from undetectable in the most acidic sites to >200 mg L−1 for sites in limestone watersheds. Bicarbonate concentration was positively correlated with pH (r = 0.667). SO4 varied from <10 to >90 (with one extreme value of 590) mg L−1 and was highest in watersheds with the most acid waters. Calcium was the most abundant cation in most samples, and its concentration was positively correlated with both pH (r = 0.486) and HCO3 (r = 0.580). The relatively low correlation of Ca with pH was due to Ca concentration having a bimodal distribution, being high in both basic and highly acid waters, and lower in waters with moderate acidity. The concentration of Mg was low at most sites, and Na concentration was high in acidic waters.
Ion and metal concentrations, in mg L−1, for samples from the study wetlands listed by site. Ca = calcium, Mg = magnesium, Na = sodium, K = potassium, HCO3 = bicarbonate, Cl = chloride, Cu = copper, Fe = iron, SO4 = sulfate. Sum of cations, anions, and sum of ions are presented as well as the water type based upon the major cations and anions present. Cells that are left blank are below detection levels (0.5 mg L−1 for Cl, 0.005 mg L−1 for Cu, and 0.01 mg L−1 for Fe).
A Piper plot of anions and cations (Fig. 4) illustrates the relative abundance of ions in the waters sampled. On a proportional basis, Ca was the dominant cation in most samples; however, at seven sites >50% of cations were Na and/or K. Waters were classified based upon their most abundant ions, and a relatively equal number of sites had CaHCO3-dominated and CaSO4-dominated type water, indicating that the influence of carbonate and sulfide rocks on the study wetlands were dominant and of equal importance. Three sites had NaSO4-dominated water (Table 1). Our water samples were collected during the dry season after months without significant rains and reflect the chemistry of local and regional groundwater discharge.
SOIL ORGANIC CARBON CONTENT AND PEAT THICKNESS
Soils were grouped into seven categories to represent their percent (%) organic carbon content (Fig. 5). Soils with <18% OC were considered mineral soils. Soils with ≥18% OC were organic. Most soils analyzed had 18 to 35% OC, and all but 17% of the soils analyzed had organic horizons at least 40 cm thick (Fig. 6) and are classified as peat soils. Ten of the 16 soil samples that had <12% OC were from one wetland complex, Cocañes, indicating that the soil in most other wetland complexes was organic. Mineral soils also occurred where highly mineralized groundwater discharge led to high rates of mineral deposition, for example in pools within acidic wetland complexes.
Using USDA (1999) criteria for organic soil thickness and organic carbon content, 65% of sample stands had organic soils of sufficient thickness to be classified as peatlands, while the remaining sites are classified as wet meadows and marshes. Eleven plots had organic layers <40 cm thick, 17% had <18% OC, and 7% had both <40 cm soil and <18% OC. At least one sampled stand within 31 of the 36 study wetland complexes had peat soils indicating that peat accumulating communities occurred throughout the study area.
Peat bodies in many wetlands were very thick. In 22 plots the peat was >7 m thick, >25% of plots had peat >5 m thick, and nearly 50% had peat >3 m thick (Fig. 6). Thick and rapidly growing peat bodies have been identified in the Andes Mountains in Chile (Earle et al., 2003) where peat accumulated at a rate of approximately 2 m per 1000 years, which is 2–10 times faster than accumulation rates in boreal or mountain peatlands in the northern hemisphere (Chimner et al., 2002). The rapid peat accumulation is likely due to the 12 month growing season, and the very dense carbon-rich peat formed by cushion plants, Carex spp., and bryophytes.
A total of 102 vascular plants (Appendix A), 69 bryophyte (Appendix B), and 10 lichen taxa (Appendix C) were identified in the study wetlands. Most Sphagnum spp. are unidentified. Vascular plants comprised 56% of the flora, bryophytes 38%, and lichens 6%, indicating that non-vascular plant species richness was nearly equal to vascular plant species richness. A total of 49% of vascular taxa are in just three families, Asteraceae with 12 taxa, Cyperaceae 14, and Poaceae 24 (Table 2). The geographic distribution of 54 vascular taxa (53%) was limited to the Andes. A number of taxa are widely distributed along the Cordillera from the Antarctic to the Arctic, including Carex microglochin, Carex praegracilis, Eleocharis acicularis, Juncus arcticus, Poa annua, and Veronica serpyllifolia. Because our collections occurred at the end of the dry season, it is possible that some herbaceous species were not collected.
Number of vascular plant taxa and proportion of flora by family.
Carex microglochin is a calciphile in the northern hemisphere, and in the study area was found only in wetlands with high pH waters. However, the very similar Carex camptoglochin is endemic to South America and restricted to acidic Sphagnum magellanicum–dominated peatlands in the study area and elsewhere in South America (Wheeler and Guaglianone, 2003). Carex camptoglochin has not been previously reported for Peru, although it occurs from Ecuador and Colombia, as well as southern Chile and Argentina (Wheeler and Guaglianone, 2003). The Amotape-Huancabamba zone of northern Peru is known to have many endemic taxa (Weigend, 2002, 2004); however, no species new to science have been identified from our collections to date.
Bryophytes covered the ground in many wetland communities. In particular, Sphagnum magellanicum, Breutelia chrysea, B. polygastria, B. subarcuata, B. tomentosa, Campylopus argyrocaulon, C. cucullatifolius, C. sharpii, Cratoneuron filicinum, Drepanocladus longifolius, Hamatocaulis vernicosus, Polytrichum juniperinum, Scorpidium cossonii, S. scorpioides, and Warnstorfia exannulata had high canopy coverage in many stands and characterized some wetland community types. Bryophytes, because they lack roots, are highly sensitive indicators of environmental conditions at the soil surface, particularly water pH and chemical content, and water table depth and duration of soil saturation (Vitt and Chee, 1990; Cooper and Andrus, 1994).
Because bryophytes and lichens comprise >40% of the flora and dominate the vegetation at many sites, they are essential to include in wetland classification and characterization studies. This is particularly important in the study area where many vascular plant species—for example Carex pichinchensis, C. bonplandii, Werneria nubigena, and some species of Calamagrostis—occurred in both acid and alkaline wetlands, making them poor indicators of specific environmental conditions. Classifications based solely upon vascular plant taxa could miss even the most important environmental differences among communities.
The bryophytes Straminergon stramineum, Scorpidium cossonii, Scorpidium scorpioides, Warnstorfia exannulata, Drepanocladus longifolius, Drepanocladus polygamus, Drepanocladus sordidus, and Meesia uliginosa are characteristic of basic to slightly acid peatlands throughout the northern hemisphere, and occupy similar environments in the study area. Species of Sphagnum, Polytrichium, and many Campylopus and Breutelia were indicative of acidic waters in the study area. Lichens were common only in acid peatlands, and the fruticose species Cladina confusa, C. arbuscula, and Cladia aggregata formed dense patches.
Plant communities of the study area were placed into four broad categories based upon the dominant life form: (1) cushion plant, (2) sedge and rush, (3) bryophyte and lichen, and (4) tussock grasses. Within these broad categories 20 plant communities were identified from the 125 plots using TWINSPAN, and are briefly described below.
Cushion Plant Communities
Wetlands with cushion-forming vascular plants have been reported only for the southern hemisphere including the Andes, New Zealand (Wardle, 1991), and Africa (Hedberg, 1964, 1979, 1992). Typically termed cushion bogs (Bosman et al., 1993), these wetlands are fens supported by groundwater discharge, not solely by direct precipitation. The most common cushion-forming species were Plantago tubulosa, Oreobolus obtusangulus, Werneria pygmaea, Distichia acicularis, Aciachne pulvinata, and D. muscoides, with the first two species being most abundant. Cushion communities dominated by species of Oreobolus occur in many other regions; for example O. pectinatus in New Zealand (Wardle, 1991) and O. cleefii in Ecuador (Bosman et al., 1993). Distichia muscoides, Plantago rigida, and Oreobolus cleefii dominate cushion wetlands in the Andes of Colombia (Cleef, 1981) and D. muscoides dominates similar wetlands in Chile (Squeo et al., 2006). Donatia fascicularis and Astelia pumila dominate cushion plant communities in southern Chile (Kleinebecker et al., 2008). Cushion community soils had the highest organic carbon content of any community type, ranging from 30 to 40%. The high organic content, and nearly year-round growth of the dense cushions has produced peat deposits >7 m thick in many sites. These communities are heavily used for livestock forage, are deeply hummocked by cattle trampling, and sod has been cut from many sites for use as a building material and for fuel, and in some areas all of the vegetation and peat has been removed. Three cushion plant communities were identified:
(1) Plantago tubulosa–Oreobolus obtusangulus–Werneria pygmaea–Distichia acicularis: This cushion community is common and widespread in the study area. The vegetation can be dominated by any species for which the community is named, but all stands have prostrate forms and most have thick peat accumulations (Fig. 7A).
(2) Distichia muscoides–Breutelia polygastria: This cushion-forming community is rare in the study area and characterized by dense turfs of Distichia muscoides.
(3) Werneria nubigena–Campylopus spp.: This is one of the most characteristic communities in many wetlands because the leaves and flowers of Werneria nubigena are distinctive.
Sedge- and Rush-dominated Communities
Species of Cyperaceae and Juncaceae dominate wetlands on the margins of lakes and ponds where seasonal standing water occurs, and in some sloping wetlands. Stands of Schoenoplectus californicus and Juncus arcticus form dense stands bordering many lakes and ponds (Fig. 8). Stands of short stature Carex crinalis dominated pool margins in acid wetlands. The most common sedge was Carex pichinchensis but Carex bonplandii, C. hebetata, C. praegracilis, C. camptoglochin, Uncinia hamata, and Eleocharis albibracteata were also common. Fourteen sedge and rush community types are described, indicating the high diversity of this type of wetland.
(4) Carex pichinchensis–Scorpidium scorpioides–Cratoneuron filicinum: This community is dominated by the tall sedge Carex pichinchensis and forms dense vegetation in shallow flooded basins with water that has a pH > 6.0. The mosses Scorpidium scorpioides and/or Cratoneuron filicinum are common in this alkaline wetland type (Fig. 7B).
(5) Carex pichinchensis–Werneria nubigena: Sloping wetlands dominated by the tall sedge Carex pichinchensis are common in acid sites and have an understory of Werneria nubigena.
(6) Schoenoplectus californicus: This tall bulrush marsh community occurred on the margins of lakes in the study area.
(7) Juncus arcticus–Scorpidium scorpioides–Brachythecium stereopoma: This community dominates flooded marshes on pond margins in alkaline environments where it has an understory of Scorpidium scorpioides and other bryophytes.
(8) Juncus arcticus–Campylopus nivalis: This community occurs in seasonally or perennially flooded marsh basins with an understory of Campylopus nivalis and other bryophytes.
(9) Carex hebetata–Cratoneuron filicinum. This tall sedge community occurred in sloping alkaline fens.
(10) Carex praegracilis–Cratoneuron filicinum: This community occupies seasonally flooded sloping lake shorelines. Carex praegracilis is common in wet meadows and fens in North America.
(11) Hypsela reniformis–Drepanocladus longifolius: This community was found in seasonally flooded but sloping lake shorelines.
(12) Carex camptoglochin–Jensenia erythropus: This community occurred on sloping acid sites, and has an open canopy of Carex camptoglochin and an understory of the liverwort Jensenia erythropus and other mosses.
(13) Carex crinalis–Sphagnum pylaesii: This community occurred in and around pools in the larger acidic wetlands, with Sphagnum pylaesii submerged in pools.
(14) Carex bonplandii: Stands dominated by Carex bonplandii occurred in several acid wetlands.
(15) Carex bonplandii–Drepanocladus longifolius: This highly productive sedge community occurs in areas with a plentiful supply of alkaline water, and nearly constant saturation to the soil surface.
(16) Uncinia hamata–Puya fastuosa: A community dominated by the distinctive sedge Uncinia hamata was sampled once, but may be regionally common.
(17) Eleocharis albibracteata–Scorpidium cossonii: This is the characteristic community of sloping fens in limestone watersheds. The spike rush Eleocharis albibracteata has high fidelity to this community type, and the ground layer is covered by mosses that occur only in alkaline substrates, particularly Scorpidium cossonii.
Bryophyte- and Lichen-dominated Community
Sphagnum spp. and Cladina spp. are characteristic of acidic fens throughout the study area. These wetlands have complex topography with pools and hummocks, although submerged aquatic vascular plants were not seen.
(18) Sphagnum magellanicum–Cladina confusa–Loricaria lycopodinea: This is the common and distinctive fen community in acid wetlands. The ground layer is dominated by Sphagnum spp., most commonly S. magellanicum, several unidentified Sphagnum taxa, and the fruticose lichens Cladina confusa, C. arbuscula ssp. Boliviana, and Cladia aggregata. Loricaria lycopodinea is also distinctive (Fig. 7C).
Tussock Grass Communities
Tussock grass-dominated communities were present in every wetland complex investigated, and are dominated by two genera, Calamagrostis and Cortadaria. Calamagrostis tarmensis, C. rigida, and C. recta were abundant, as were Cortaderia hapalotricha and C. sericantha. In the Ecuadorian páramo C. sericantha dominates wetland communities (Bosman et al., 1993).
(19) Cortaderia hapalotricha–Cortaderia sericantha: Stands dominated by Cortaderia were typical of wetlands with acid soils. This community occurred on sites with deeper water tables, some of which lacked peat soils (Fig. 7D).
(20) Calamagrostis tarmensis–Campylopus cucullatifolius–Scorpidium cossonii: Bunch grass communities dominated by Calamagrostis tarmensis, and other species of Calamagrostis, are characteristic of many wetlands, with acid and alkaline soils.
DIRECT GRADIENT ANALYSIS OF THE VEGETATION AND SPECIES DATA
The eigenvalue of the first canonical axis is 0.507 (Table 3), and it was statistically significant at P < 0.01 using a Monte Carlo permutation test. The sum of all canonical eigenvalues was also significant at P < 0.01, demonstrating that the relation between the species and the environmental variables is highly significant (P < 0.01). The canonical coefficients and intra-set correlations of environmental variables from the CCA analyses (Table 3) indicate that Axis 1 is driven by water chemistry, particularly HCO3 content and pH. Sites with high HCO3 have low SO4 concentrations, high pH, and are plotted on the right side of the canonical ordination diagram (Fig. 8), while those with low HCO3 and low pH are on the left. Acidic wetlands form a relatively tight cluster along Axis 1 between 0 and −1, indicating that these communities are abundant and have relatively similar floristic composition. Most acid waters had detectable concentrations of copper and iron, and other metals are likely present as well.
Results of the canonical correspondence analysis showing eigenvalues for the first four axes, as well as the species environment correlations, and cumulative variance in species and species environment relation data and the total inertia in the data set. The lower portion of the table includes the canonical coefficients, and intraset correlation coefficients of environmental variables with the first two ordination axes are also shown. Axis variables are pH, WT (water table depth), peat thickness (Peat), % organic carbon (%OC), soil temperature at 13 and 28 cm depth (13 cm and 28 cm), wetland slope, aspect, and dissolved HCO3, total N, total P, Cl, SO4, Cu, Fe, Mg, K, and Na. Variables in bold explained a statistically significant proportion of the variance as determined using a Monte Carlo permutation test (499 runs) (P < 0.05).
CCA axis 2 is a complex hydrologic, soil temperature, and peat thickness gradient. Sites with deeper water tables during 2005 had warmer soil temperatures and thinner peat deposits, and plotted at the bottom of the canonical ordination diagram. Sites near the top of the diagram have thicker peat, cooler soil temperatures at 28 cm, and a water table closer to the soil surface.
On the right side of the species canonical ordination diagram (Fig. 9), the mosses Cratoneuron filicinum, Scorpidium scorpioides, S. cossonii, and Drepanocladus longifolius along with the vascular plant species Eleocharis albibracteata and Carex praegracilis are diagnostic of alkaline waters. On the acidic (left) side of the gradient are species of Sphagnum, Polytrichum juniperinum, and the fruticose lichens Cladina confusa, C. arbuscula, and Cladia aggregata. Among the vascular plants, Loricaria lycopodinea, Cortaderia hapalotricha, C. sericantha, Carex crinalis, and Werneria nubigena are most abundant in acid wetlands. Juncus arcticus and Carex pichinchensis are plotted near the center of the ordination space indicating their occurrence in acid and basic wetlands, and those with thick and thin peat deposits. Bryophyte and lichen species were more closely tied to specific hydrological and geochemical environments than vascular plants.
Geologically complex mountain systems, such as the northern Andes, are composed of a range of bedrock types, including highly mineralized igneous and metamorphic rocks as well as limestone and other sedimentary rocks. Groundwater contacting each rock type is geochemically distinct, and along with landforms produced by glaciers and hillslope processes form heterogeneous environments that support a diversity of wetlands. Direct gradient analysis indicated that wetland floristic composition is controlled primarily by groundwater geochemistry, which explained more of the total variation in species composition than all other environmental variables combined. Similar controlling gradients occur in boreal (Vitt and Chee, 1990; Nicholson et al., 1996) and mountain wetlands (Gignac et al., 1991; Cooper and Andrus, 1994) in Canada and the U.S.A., where biogeochemical variables, more than climate, explain the most floristic variation. In the study area, acid and alkaline sloping wetlands have few species in common. This acidic to alkaline gradient is typically referred to as the “rich to poor” gradient (Sjörs, 1950; Malmer, 1986) and provides a conceptual understanding of peatland vegetation variation throughout the world.
Poor fens and bogs are typically dominated by species of Sphagnum whose ion exchange capacity acidifies water and soil. However, acidic waters in the study area are produced by the oxidation of mineral sulfides to form sulfuric acid. This water discharges from hillslopes and its pH is little altered during its flow path through the wetland or apparently by ion exchange by Sphagnum species. The acid water may leach cations and metals from watershed rock producing acidic and ion-rich water which have been termed iron fens or acid geothermal fens in the Rocky Mountains of the U.S.A. (Cooper et al., 2002; Lemly, 2007). Iron fens occur in watersheds with iron pyrite, which results in precipitation of iron and the formation of iron rich peat, mineral terraces, and bog iron ore formation, similar to those in the study area. Acid geothermal fens occur in volcanic landscapes where sulfur vents produce mineral sulfides. These fens are distinct from poor fens or bogs, which are saturated largely by precipitation and have mineral ion poor waters. The acid fens in our study area were dominated by S. magellanicum, which also dominates ombrotrophic (rain fed) bogs in hypermaritime regions of south Patagonia (Kleinebecker et al., 2008) and southeastern Alaska (Klinger, 1996), indicating that water acidity more than climate variables control the distribution and abundance of many peatland species. The alkaline portion of the rich-poor gradient supports Scorpidium scorpioides and other brown mosses in the family Amblystegiaceae which dominate rich and extreme rich fens around the world.
Axis 2 of our CCA analysis (Fig. 9) represents a complex hydrologic gradient. Wetlands with perennial discharge of cool groundwater maintained the lowest temperatures, highest water tables, and formed the thickest peat bodies. The hydrologic characteristics of wetlands that formed in basins vs. slopes also produce floristic differences. Basin wetlands dominated by Carex pichinchensis, Juncus balticus, and Schoenoplectus californicus were floristically similar in both alkaline and acid waters because deep flooding limited the species that could occur in these marsh communities, while slope wetland communities were distinctly different in each water type.
Most wetland complexes had at least one area with organic soils of sufficient thickness to be classed as peatlands. Because of the strongly seasonal precipitation regime, peatlands formed only where perennial groundwater discharge occurs, and all peatlands sampled were fens. Previous researchers have termed alpine Andean wetlands cushion mires (Bosman et al., 1993), highland bogs (Wilcox et al., 1986), and soligenous peatlands (Earle et al., 2003). However, communities in the study area are dominated by a wide range of vascular plants, not solely cushion plants. Bogs are ombrogenous (rain-fed) peatland ecosystems (Rydin and Jeglum, 2006) and do not exist in the study area. Peatlands were not all soligenous and many basins with nearly level surfaces supported well developed peat. Our study sites included cushion plant communities, a range of soligenous peatland community types, and topogenous peatlands that had formed in basins. While communities dominated by Sphagnum spp. and cushion plants were common, most communities sampled were dominated by species in Cyperaceae.
Biogeographically 53% of vascular plant taxa in the study area are Andean endemics, although no narrow endemics were found. Several species of Carex that occur along the Cordillera from the Arctic to Antarctic, such as Carex microglochin (Hultén, 1968), were present. In addition, several ruderal taxa, such as Poa annua, were found. More than 20% of the bryophyte species are common in the southern and northern hemisphere, and many are worldwide dominants in fens, for example Scorpidium scorpioides and Warnstorfia exannulata. Of the Sphagnum taxa, three occur in the northern hemisphere, and S. magellanicum is common in oceanic regions throughout the world.
A wide diversity of wetland plant communities occur in the study area supported by complex geological and geochemical gradients that produce widely varied groundwater pH and ion concentrations and hydrogeologic settings. Most study wetlands were peat-accumulating fens, and many had peat bodies >5 m thick. Rapid peat accumulation rates appear to be common in high elevation tropical peatlands in South America, and may not indicate great antiquity. For example, Chimner and Karberg (2008) measured peat accumulation rates of 1.3 mm yr−1 in the Ecuadorian páramo while in the puna of northern Chile, Earle et al. (2003) reported that peat bodies had a long-term mean peat accumulation rate of ∼2 mm yr−1, and up to 3.5 m of peat had accumulated in no more than 1700 years. In contrast, high elevation peatlands in the Rocky Mountains of Colorado have averaged 0.25 mm yr−1 (Chimner et al., 2002), and subarctic and boreal peatlands in Canada average 0.375 mm yr−1 and 0.635 mm yr−1 (Tarnocai and Stolbovoy, 2006). Thus, tropical montane peatlands appear to support the highest rates of peat accumulation known for mountain wetlands. However, these rates are slower than lowland tropical peatlands, where peat accumulation rates can exceed 10 mm yr−1 (Page et al., 2006).
Plant communities in the study area are dominated by cushion plants in basins, sedges and rushes on pond and lake margins and sloping alkaline sites, and bunch grasses on the margins of all fens. Acid fens are widespread in the region, supported by groundwater flow systems discharging from natural acid generating rocks. The acid fens support distinctive Sphagnum spp.-dominated communities similar to boreal and austral ombrogenous bogs and geochemically acid fens reported elsewhere in the western hemisphere (Cooper et al., 2002; Lemly, 2007).
This research was funded by Minera Yanacocha S.R.L., Cajamarca, Peru. We are grateful to Dr. William A. Weber for moss identification, numerous other experts for moss and lichen identification, José Alegría for his generous assistance in plant determination, especially Poaceae and Cyperaceae, and his valuable comments. Thanks to Daniel Stančík for identifying or confirming Festuca species. Two anonymous reviewers provided comments that helped to improve this manuscript.