Clay mineralogical evidence of a bioclimatically-affected soil, Rouge River basin, South-Central Ontario, Canada

Geomorphology 228 (2015) 189–199

Contents lists available at ScienceDirect

Geomorphology journal homepage: www.elsevier.com/locate/geomorph

Clay mineralogical evidence of a bioclimatically-affected soil, Rouge River basin, South-Central Ontario, Canada W.C. Mahaney Quaternary Surveys, 26 Thornhill Ave., Thornhill, Ontario L4J1J4, Canada Department of Geography, York University, 4700 Keele St., N. York, Ontario M3J1P3, Canada

a r t i c l e

i n f o

Article history: Received 17 December 2013 Received in revised form 18 August 2014 Accepted 19 August 2014 Available online 16 September 2014 Keywords: Bioclimatic factors Soil morphogenesis Adiantum pedantum Soil evapotranspiration Clay minerals Bioclimatic/pedologic anomalies

a b s t r a c t Holocene soils in drainage basins of South-Central Ontario, Canada, are generally Fluvisols (Entisols) in floodplains transitioning to Brunisols (Inceptisols), Luvisols (Alfisols) and Podzols (Spodosols) in older terraces and in the glaciated tableland. A single landslide sourced from the highest fluvial terrace in the Rouge basin, with a rubble drop of ~12 m emplaced a lobe-shaped mass of reworked stream gravel, glaciolacustrine sediment and till, emplaced approximately 6 m above mean water level at a height roughly equivalent to previously dated mid-Holocene terraces and soils. Clay mineralogy of the soil formed in this transported regolith produced the usual semi-detrital/pedogenic distribution of 1:1 (Si:Al = 1:1), 2:1 and 2:1:1 clay minerals as well as primary minerals consisting of plagioclase feldspar, quartz, mica and calcite. Unexpectedly, the presence of moderate amounts of Ca-smectite in the Bk and Ck horizons, relative to a clay-mineral depleted parent material (Cuk), argues for a soil hydrological change affecting the wetting depth in the deposit. The presence of the uncommon ‘maidenhair fern’ (Adiantum pedantum) in the mass wasted deposit, a plant capable of high evapotranspiration, is interpreted as producing a bioclimatic disruption limiting soil water penetration to near root depth (wetting depth), thus producing a clay mineral anomaly. © 2014 Elsevier B.V. All rights reserved.

1. Introduction In local catchments of South-Central Ontario (Fig. 1A, B) the evolution of floodplain soils to terrace soils follows a progression from Ck/Cuk to Ahk/Ck/Cuk (Regosol) profiles in the floodplain to more complex soils in higher and older terraces developing either Ah/Bm/Bk/Ck/ Cuk (Brunisol), Ah/Bt/Ck/Cuk (Luvisol), and occasionally under pine forest L/E/Bh/Ck/Cuk (Podzol) profiles which deepen with time. In the US taxonomic soil system (NSSC, 1995), Regosols in the Canadian system (CSSC, 1998) key out as Entisols, Brunisols as Inceptisols, Luvisols as Alfisols and Podzols as Spodosols. Entire profiles in the Rouge catchment reach ~ 1.0 m depth developing over ~ 11 ka which, given well-dated fluctuations of Glacial Lake Iroquois (Jackson et al., 2000; Mahaney et al., 2014), is the maximum age for any riverine profiles in the area. Aside from occasional 14C dates on younger floodplain soils, the entire soil evolutionary sequence has been dated by relative dating (RD) methods including pebble/sand weathering characteristics, topographic position and soil stratigraphy (Mahaney and Sanmugadas, 1986; Mahaney and Hancock, 1993a,b; Mahaney et al., 2014). Changing profile morphology over time brings staged removal of calcite from soil epipedons over time, slow increase in concentrations of extractable cations and C/N ratios, increases in Fed/Fet (dithionite extractable/total Fe)

E-mail address: [email protected].

http://dx.doi.org/10.1016/j.geomorph.2014.08.033 0169-555X/© 2014 Elsevier B.V. All rights reserved.

(Mahaney and Sanmugadas, 1986; Mahaney and Hancock, 1993a) along with slow conversion of illite/mica to both chlorite and vermiculite. Randomly interstratified illite–smectite and occasional kaolinite and metahalloysite are not authigenic, the former is considered to be inherited from Ordovician shales, the latter from previous interglacial paleosols eroded by incoming ice during Wisconsinan-age ice advances (Mahaney and Ermuth, 1975; Mahaney et al., 2014). However, despite the high ratio of carbonates:silicates (approximately 80:20%) in fluvial sediment, chloritization of illite and slow genesis of vermiculite in middle to Early Holocene profiles (Mahaney and Sanmugadas, 1986) are documented in the basin. Clay mineral genesis is accompanied by degradation of plagioclase and mica, very minor decrease in quartz, and in horizons with lower pH, near complete removal of calcite. Stratigraphically similar soils in several catchments of southcentral Ontario contain randomly interstratified illite–smectite inherited from Ordovician shales but lack ‘free’ smectite in any quantity. Wetting depth in middle to Early Holocene fluvial soils in the region, estimated from root depth and thickness of B horizons (Mahaney and Sanmugadas, 1986), allows the throughput of soil moisture sufficient to remove smectite. Thus, the presence of moderate to large amounts of smectite in a soil stratigraphically similar to other Middle and Early Holocene soils in the basin with pedons lacking expandable clay minerals, invokes a means of explaining its genesis. Of all the soil forming factors (Birkeland, 1999)—climate, biota, topography, lithology and time—only biota and soil climate are the two variables that could

190

W.C. Mahaney / Geomorphology 228 (2015) 189–199

standard (R46), that is explored here. A regional study like this is important since it adds to the bioclimatic literature first introduced by Jenny (1941) and this may well be the first time in the literature that clay genesis has been shown to be related to fluctuating soil moisture regulating certain aspects of soil chemistry, in all likelihood caused by a hydrophilic fern, with possible applications elsewhere. 2. Regional geology and field area The Rouge River and Little Rouge Creek basins of South-Central Ontario (Fig. 1B) drain across till-encrusted Ordovician shales of the Whitby formation, a gray to black noncalcareous shale (Liberty, 1955, 1964). Quaternary deposits mask most of the shale with most information derived from well logs and a prominent outcrop near Twyn Rivers Drive on Little Rouge Creek (Fig. 1B), 3 km north of Lake Ontario. The mean annual temperature in the area is 7 °C, with extremes of − 34 °C and 40 °C. Mean annual precipitation is 860 mm and mean annual actual evapotranspiration is 530 mm (Brown et al., 1968; Phillips and McCulloch, 1972) yielding a mean annual water surplus of 330 mm. Hence, the climate is humid with sufficient soil water available for leaching. Available soil moisture in the basin is unknown and can only be estimated from solution of carbonate in Ah/B horizon complexes, followed by precipitation of carbonate in C horizons, as an estimate that normal leaching is capable of throughput of moisture mostly to at least root depth (ca. 50 cm). In the R43 profile discussed here this depth is greatly reduced to, at most, a few cm— a response to the presence of a hydrophilic fern capable of reducing soil water to slow leaching allowing the genesis of Ca-smectite. More recent data obtained from the Ministry of the Environment (1984) averaged from broad regions along the north shore of Lake Ontario provide mean climatic data as follows: mean annual precipitation (MAP)— 900 mm, mean evaportranspiration (AE) —550 mm, runoff— 250–300 mm, and snowfall —100–150 mm. These values are approximate and while similar to previously recorded climatic data, the latter increase in precipitation and evapotranspiration presumably reflects global warming. The mean water surplus calculated at 330 mm is probably declining somewhat with recent warming but sufficient to provide soils at or beyond field capacity most of the year. The sandy to silt loam textures of the R43 and R46 profiles normally yield about ~5% water content. 3. Materials and methods

Fig. 1. A, location map of the field sites in southern Ontario based on GIS data (after Mahaney et al., 2014); b, Topography of the landslide (R43) and 8 m terrace section (R46) in the Rouge River Catchment; C, Maidenhair fern (Adiantum pedantum).

intervene to offset the throughput of meteoric water into the soil system. The only biotic element present on the landslide mass is the maidenhair fern (Adiantum pedantum) (Fig. 1C), an uncommon plant in the drainage, and a species preferring wet/shaded, coarse sandy habitat with a known positive effect on evapotranspiration (Jones, 1987; Paris and Windham, 1988; Cody and Britain, 1989). It is the relationship between the presence of this fern and the clay mineral composition of the resident soil (R43), compared with an equivalent pedological

The R43 profile was excavated to expose fresh material. All samples were collected from a profile face cleaned back 10–15 cm. Soil descriptions follow guidelines set out by the Canada Soil Survey Committee (1998) and Birkeland (1999). Correlations with the US soil taxonomic system are made using the guidelines of the Soil Survey Staff (NSSC, 1995) for comparative purposes, for example, Ah = A, Bm = Bw. The Cu horizon relates to fresh, unconsolidated and unweathered (cf. ‘u’ material; Hodgson 1976). Therefore, soil description and stratigraphy presented here are similar to what was discussed in Mahaney and Sanmugadas (1986). Soil color assessments are based on Oyama and Takehara (1970) soil chips. At least 500 g samples were collected at the sites to allow for particle size, clay mineral, and chemical analyses. These were used for laboratory work including particle size analysis following procedures outlined by Day (1965), and chemical extractions outlined below. Samples were wet sieved to separate sands from clay and silt. The clay fraction was analyzed by hydrometer following a method described by Mahaney (1990). Soluble salts were measured by electrical conductivity (Bower and Wilcox, 1965) and pH by electrode in a 1:5 solution of distilled water. Calcium carbonate was determined by acid dissolution of carbonate and measurement of evolved CO2 (Nelson, 1982). Organic carbon was measured following procedures of Walkley and Black (1934) using the Degtjareff method and total nitrogen by the Kjeldahl method (Bremner, 1965).

W.C. Mahaney / Geomorphology 228 (2015) 189–199

The mineralogy of the clay fraction (b 2 μm) was analyzed by XRD using a Toshiba ADG-301H instrument following methods described by Whittig (1965). Selected sands were subsampled and analyzed under the light microscope where after a smaller group of samples was subjected to SEM/EDS analysis using a JEOL-840-JSM with an Energy Dispersive Spectrometer (EDS). The SEM/EDS analysis follows methods outlined by Vortisch et al. (1987) and Mahaney (2002). Photomicrographic imagery was obtained at accelerating voltages of 10–20 keV; X-ray microanalysis was acquired at an accelerating voltage of 10–15 keV so that concentrations of Fe could be assessed. All imagery is in the secondary mode. The maidenhair fern (A. pedantum), found only at the R43 site described here, has a range across much of northeastern North America, mostly confined to moist localities. Aside from a preference for moist soil, the species may colonize saturated soils following mass wasting events (landslides) (Cody and Britain, 1989), a subject warranting further investigation. Palynological investigations in southern Ontario rarely produce fossil spores of A. pedantum (Personal communication, J.A. McAndrews, 7.21.2014). The plant usually grows in moist, shady woods often around rocks with no particular soil preference as reported in a U.S. Government website (http://www.nps.gov/plants/pubs/chesapeake/plant/1303. htm). Sarah Finkelstein (personal communication, 7.23.2014) reported that Adiantum spores are occasionally observed in pollen records but the distribution of species of Adiantum in southern Ontario is unknown. If A. pedantum is restricted to moist, occasionally wet sites with coarse grained texture, its occurrence at the time of the mass wasting event or soon thereafter is not unexpected. Since it seems that workers with the plant have not carried investigations over into the soil sphere, it behooves future researchers to investigate soils associated with the species, especially the clay mineral composition of relevant lower horizons and parent material with which it is associated. Recent interest among ecologists (van Dam & Heil, 2011) to explore more fully AG–BG (above ground–below ground) plant/soil relationships to better understand root → soil processes/interactions, seems appropriate with the plant/soil interactions reported here. The chemical and mineralogical analysis of soils related to A. pedantum, as hypothesized here, could be further explored to understand biotic-produced changes in soil hydrology linked to clay mineral genesis, a form of little explored biomineralization. 4. Results 4.1. Stratigraphy The inverted stratigraphy of the Rouge Valley is typical of southcentral Ontario where the floodplain is floored with silty sand covering three-fifths to four-fifths of the valley bottom and older terraces stretch up along the valley sides to the tableland, a vertical distance of ~100 m. The floodplain soils are Fluvisols, typical of young pedons in the early stages of genesis (Brewer and Walker, 1969; Alexander, 1974; Ahmad et al., 1977; Mahaney, 1990), ranging from Ck/Cuk to Ah/Ck/Cuk profiles with increasing distance from the stream channels. Overbank deposits in the floodplain are common, giving rise to pedostratigraphic columns with punctuated Ahbk/Cubk horizons forming over periods of a century or more (Mahaney and Sanmugadas, 1986). At approximately 3 m above stream level, discontinuous terraces reveal more advanced pedogenesis with Ah/Ck/Cuk profiles containing sharp horizon boundaries, greater root frequency and stronger colors indicative of more advanced oxide release in thin A/C horizons (Entisols NSSC, 1995; Regosols, CSSC, 1998) and increased precipitation of carbonate in C horizons. Redox potentials are higher here and waterlogged profiles are less common than on the floodplain. At higher elevations above stream level the ~ 8 m terrace, which houses the Twyn Rivers formation of Mahaney and Sanmugadas (1986), reveals still better developed profiles (Figs. 2 and 3) with enhanced color differentiation, greater depth and the presence of Bk

191

Fig. 2. The R43 profile. Horizon designations mostly follow the Canada Soil Survey Committee (1998); the Cu designation follows Hodgson (1976) and Birkeland (1999).

or Bt horizons indicative of Brunisols (Inceptisols) or Luvisols (Alfisols), respectively. Less common, as in R43 described here, the B horizon contains enough carbonate to qualify as Bk that is probably the result of a reduced soil hydrological regime and less leaching caused by the resident fern population. On the basis of weathering characteristics, depth of profile, and horizon properties in the 8 m terrace (site R12) of Rouge River (Mahaney and Sanmugadas, 1986), Twyn River-age deposits are considered to be of Middle Holocene age. While not definitive, relative age data indicate that the terrace sands were deposited sometime in the early-to-mid postglacial period. On the basis of similar elevation and soil profile characteristics, site R46 in the 8 m terrace is considered coeval to or close in age with the landslide on a terrace located 6 m above water level and within the middle group of terraces. The landslide mass, sourced from an adjoining 18-m high terrace (Rouge formation of Mahaney and Sanmugadas, 1986), provided the parent material for the R43 profile. 4.2. Soil profiles Although the two profiles—R43 and R46—key out differently, the former a typical Brunisol (cf. Inceptisol) and the latter a Luvisol (cf. Alfisol), both have formed under similar macro-climate in topographically high and well-drained situations with uniform lithologies and similar age, the only variable being a different biota with drier soil climate in R43. The R43 profile weathered under a beech (Fagus grandifolia) and maple (Acer saccharum) forest with a representative understory of grass, sedge and fern (A. pedantum) whereas R46 has a biological history connected with hemlock (Tsuga Canadensis) and pine (Pinus strobus), and without a fern component in the vegetation. The R43 profile (Fig. 2) formed in a typical clast-supported, masswasted deposit considered to be of Middle Holocene age with a source off the high terrace at site R15 (Mahaney and Sanmugadas, 1986). Because the sites are nearly contemporaneous with the eastern hemlock blight (Foster et al., 2006), it is possible that both the mass wasting and alluvial incision events were caused by excess surface runoff when the forest cover was disturbed with the demise of this species.

192

W.C. Mahaney / Geomorphology 228 (2015) 189–199

Fig. 3. Stratigraphy at Site R46. The soil cover on the 8-m terrace is formed in a 5-m thick clast-supported alluvium that becomes matrix-supported near the surface. The alluvium grades across a sharp, truncated contact into a 1.25-m thick bed of sand and silty sand of the Scarborough formation (~90 ka) that overlies shale of the Whitby formation (Ordovician).

The R46 section (Fig. 3) consists of 2 m of Ordovician shale (Whitby formation) overlain, across an unconformity, with 1.25-m-thick beds of alternating clay, silty-sand and sand which belong to the Scarborough formation of Early Wisconsinan age. Each bed in this glaciolacustrine member highlights rapid fluxes of deep water to near-shore/beach to deep water judging from particle size variations from bottom to top of the deposit, most probably occurring during the early stage of Lake Scarborough sedimentation. A second unconformity separates these Early Wisconsinan glaciolacustrine beds from approximately 6 m of clast-supported mid-Holocene alluvium, the topmost 0.6 m of which houses a mid-Holocene soil, correlated to the Rouge chronosequence of Mahaney and Sanmugadas (1986). Both R43 and R46 exhibit similar colors down-profile, the upper epipedons (Ah horizons) ranging from brownish black (10YR 2/2) and brownish gray (10YR 4/2) in R43, to brownish black (10YR 2/2) in R46, which indicate that the steady state of soil organic matter has been reached in both pedons. The lower epipedons (B horizons) yield variable shades of brown (10YR 4/4, 4/6) in both profiles indicating similar degrees of weathering, colors indicating principally hydroxide and oxide release. The Ck horizons in both profiles vary slightly with a dull yellowish brown (10YR 5/3) in R43 and 10YR 5/4 in R46, the enriched chroma in the latter suggesting slight elevation in hydroxides/oxides. The parent materials in both profiles are yellowish brown (10YR 5/3) in R43 and dull yellow (2.5Y 6/3) in R46, the 10YR color in R43

suggestive of reworked oxides from source sediment. There is no evidence of mottling/gleying in the profiles but none is expected since the textures are free draining (higher in R46) and bedrock is at considerable depth (7 m in R46 and ~5 m in R43). The buried Ckb horizon in R43 with a dark brown (10YR 3/4) color has been truncated by the slide mass and clearly represents a soil remnant, perhaps from a Brunisol (Inceptisol) or Regosol (Entisol) that survived emplacement of the landslide deposit. Clearly the buried soil in the 6-m terrace formed over an unknown time window long enough to produce a C horizon. Following emplacement of the landslide debris sheet, a somewhat more fully developed Inceptisol formed in the landslide deposit from the Middle Holocene to the present. Admittedly, the two soils –R43 and R46– are offset in age by an unknown time but the presence of a B horizon in the surface pedon places both within the ~ 5–8 kyr Middle Holocene window. The brown coatings on sands in the Ckb horizon contain some carbon, and because the sediment effervesces, it is likely that carbonate was translocated during pedogenesis prior to truncation during emplacement of the landslide deposit. 4.3. Particle size Grain size curves for the two profiles—R43 and R46—depict contrastingly different particle size distributions reflecting the different geomorphic/pedologic processes that produced the soils. The R43 profile

W.C. Mahaney / Geomorphology 228 (2015) 189–199

(Fig. 4) possesses a wide window of size distributions across all horizons that possibly indicate the wide range of particle size materials in the source sediments and various degrees of liquidity in the mass-wasted material. These data yield a marginally 5% higher percentage of clay at the 9 phi line (see Fig. 4) in the Bk horizon suggestive of clay movement down profile but the lack of clay films on ped faces argues against an argillic horizon. Assuming a random size distribution of materials in the landslide mass at the start of pedogenesis, lower sand and silt and somewhat greater clay percentages in the Bk horizon, compared with all other horizons sampled, suggest that minor weathering similar to R46 has occurred over the total time of morphogenesis. The R46 profile (Fig. 5), on the other hand, reveals a tighter window of particle sizes horizon-to-horizon but with slightly parabolic characteristics suggestive of water transport modified by weathering. Close knit particle size distributions like this further suggest little variation in water velocity at the close of lag gravel deposition, with differences in grain size distributions primarily due to pedogenesis. Whereas in contrast with R43, the clear increase of clay in the Bt horizon and the presence of thin clay/organic films on peds, there is sufficient clay translocation to support an argillic horizon designation. If the silt fraction may be taken as an aeolian indicator, R46 contains 13% higher silt in the surface compared with horizons down-profile, whereas R43 has 20% less silt in the Ah horizon compared with the lower horizon group. While particle size by itself is not definitive, the silt cap in R46 suggests, but does not prove, locally derived aeolian input. Because R46 is subject to slow infusion of slopewash along the terrace surface, the increase in silt may not depict airfall-influx of sediment including clay, the presence of which has been considered crucial to the presence of smectite in similar soils in adjacent New York State (Arnold and Cline, 1961). Given the different aspects of the two sites (Fig. 1), however, with R43 east-facing and R46 west facing it is possible with a prevailing NW wind system that the west-facing site might likely have been a sink of air-influxed silt at times of high wind and open surfaces in the basin

193

to provide a higher percentage of b 63 μm sediment for inclusion in the R46 pedon. If the silt is aeolian in origin the source is likely within the catchment as INAA analysis of other profiles in the Rouge catchment has shown no variation in chemical composition to suggest longdistance sources (Mahaney and Hancock, 1993a,b). 4.4. Clay mineralogy The mineralogy of the b2 μm fraction (Table 1; Fig. 6A–C) was analyzed to determine the clay mineral composition, weathering pathways from hydrous mica to other 2:1 clay minerals (including smectite), and distribution of common primary minerals resident in both soil systems. The distributions of metahalloysite, kaolinite, calcite and plagioclase exhibit detrital signatures judging from their near random concentrations. Illite in R43 exists only in trace quantity in the Cu horizon (Fig. 7A), remains in small amounts through much of the profile (Fig. 7B), increasing in the upper epipedon. The moderate concentration of smectite in the R43-Bk (Fig. 7C) suggests that the dampened, fern-affected soil water hydrology has allowed its genesis, and with reduced leaching, the maintenance of this clay in the weathering sequence. The trace to small quantities of vermiculite and chlorite present in R43 are probably derived from weathering of smectite, although some chlorite might possibly have been translocated from hydrothermal rocks in the nearby Canadian Shield or from previous weathering episodes. The R46 profile exhibits a somewhat similar ratio of detrital to pedogenic minerals. Declining kaolinite upward in the profile is probably detrital and given the limited weathering time does not likely represent a weathering sequence. Metahalloysite and illite–smectite appear detrital as well. Illite is uniformly distributed in small quantity upward in the profile where it transitions to trace quantity in the epipedon. Smectite was not detected whereas vermiculite and chlorite are likely weathering products derived from illite. Weathering of plagioclase in the Bt horizon is possibly a product of hydrolysis accelerated by

Fig. 4. Particle size curves for horizons in the R43 profile. Particle size grades follow the Wentworth System for sand/silt (63 μm) and the USDA for silt/clay (b2 μm). Note the parent material (Cuk horizon) of the landslide mass contains less than 0.5% of fine silt plus clay (b4 μm) fraction. The Bk and the Ck horizons where most of the smectite resides in R43 contain ~5% and b1% clay size material, respectively.

194

W.C. Mahaney / Geomorphology 228 (2015) 189–199

Fig. 5. Particle size distributions in the R46 profile. Sediment grade sizes are as outlined above.

adsorbed water in the heavy clay texture present there. Calcite in the clay fraction appears to produce a trend closely related to the chemical composition of the profile. 4.5. Soil chemistry Selected chemical properties (Table 2) were chosen to shed light on biotic influences between the two sites. Total organic matter, somewhat higher in the surface epipedon of R43, decreases with depth in a similar fashion to R46. Organic carbon follows apace at about 60% the concentration of organic matter in each horizon. Percentages are somewhat higher in the Ah1 horizon of R43, decreasing with depth, and in tandem with similar concentrations in R46. The higher organic carbon in R43's surface epipedon may be explained by the deciduous canopy supplying greater leaf litter compared with R46. Total nitrogen follows a similar

Table 1 Mineralogya of the b2 μm fraction of soils at sites R43 and R46.

4.6. SEM/EDS

Site

Horizon

Depth

K

MH

I

S

I-S

V

Chl

Q

P

O

Calc

R43

Ah1 Ah2 Bk Ck Cuk Ckb Ah Bt Ck1 Ck2 Cuk

0–8 8–17 17–32 32–60 60–115 115+ 0–15 15–35 35–50 50–62 62+

x tr tr tr – tr tr tr x x x

– – – tr – – – – – ? tr

xx x x x tr x tr x x x x

tr tr xx x – tr – – – – –

x tr x x – tr x xx x x x

x x tr tr – x x xx x x x

x x tr tr – tr x tr tr – –

xxx xx xx x xx xx xx xx x xx xx

tr x x x x x x tr x x x

– – – – – – – tr – – –

– tr tr – tr tr – tr – xx x

R46

pattern, somewhat higher in the R43-Ah1 horizon, then decreasing in the Bk horizon, a pattern somewhat different than in R46 where nitrogen increases in the Bt horizon, possibly connected with greater movement of clay and organic carbon from the Ah to the Bt horizon. The soil reaction varies between the two profiles with pH registering alkaline down-profile in R43. In R46, pH varies from slightly acidic in the surface becoming greatly alkaline with depth. Total salts measured by electrical conductivity are low, which in part, parallels concentrations of CaCO3 which only vary slightly with depth in R43. Conductivity measures only total salts which in this case probably represent nitrates, chlorine and sulfate not present in the parent materials. Data for percent CaCO3 in R46 is not available but percent carbonate measured by effervescence shows that the solum (A + B horizons) is free of carbonate, registered strength increasing from the Bt/Ck1 contact downward, presumably a reflection of the prevailing coniferous vegetation and lack of a fern population.

a Minerals are: kaolinite (K), metahalloysite (MH), illite (I), smectite (S), illite-smectite (I-S), vermiculite (V), chlorite (Chl), quartz (Q), plagioclase (P), orthoclase (O) and calcite (Calc). Semi-quantitative mineral amounts are calculated as follows: – = nil, tr = trace, x = small amount, xx = medium and xxx = abundant.

Selected sands in R43 were subjected to analysis by SEM/EDS to determine the effect of mass wasting on quartz and other minerals and to search for weathering microfeatures and coatings that might relate to the XRD results. Many grains suffered severe abrasion and fracture patterns, presumably from transport and impact as evidenced on imagery in Fig. 7. Starting with the Bk horizon (Fig. 7A–C) and working down profile it is evident that most grains studied are angular with varying degrees of etching and abrasion. About 15% of all grains studied in the pedon and parent material are cracked, many severely with some showing only remnant fracture faces, the latter possibly the result of bedrock release (Mahaney, 2002). Cracked minerals (Fig. 7A), on the other hand, are without doubt products of mass wasting (Mahaney, 2002), and despite the short drop for the mass, one-sixth of all grains analyzed carry this microtexture. Abraded grains with pockets of clay minerals and

W.C. Mahaney / Geomorphology 228 (2015) 189–199

195

Fig. 6. A, XRD results for the R43-Cuk horizon; B, XRD results for the Ck horizon; C, XRD results for the R43-Bk horizon.

alteration products are common also in both the Bk and Ck horizons (Fig. 7B, C), the clay component carrying 2:1 Si/Al ratios estimated from EDS spectra with small concentrations of Ca and very minor Fe suggesting mixed concentrations of Ca-smectite and nontronite (Fe-bearing smectite), respectively. Severely etched grains, mostly plagioclases, make up b 10% of the imaged population and some grains in the Ck horizon carry pockets packed with clay (Fig. 7D). Despite the data showing that many grain surfaces are abraded, even polished during transit, some clay resides in depressions possibly forming post-depositionally in the coarse part of the matrix material. Plagioclase and quartz dominate within the parent material (Cuk horizon; Fig. 7E, F) along with minor calcite and mica, principally biotite, all nearly bereft of adhering particles and with clean uncoated surfaces. Clean grains lacking coats in the parent material contrast with significant clay coated grains in the soil, which indicate that clay genesis occurred under understory vegetation composed of A. pedantum.

4.7. Bioclimatic model Comparison of R43 with R46 is made to test biotic differences between sites where both profiles have formed under similar macroclimate, in similar topographic settings and lithology and with similar age relationships. While the macro-biota is different – maple, beech forest at R43 (Fig. 8) and hemlock, pine forest at R46 – the maidenhair fern is present only at R43. Given the fern has an inordinate evapotranspiration capacity (Page, 2002) there is no instrumentation in place to measure the actual soil water loss. The mean soil water is probably a large but unknown percentage of the annual water surplus (soil moisture plus surface runoff) and much of this is probably taken up by the fern given that Ca-smectite is found throughout the profile from trace concentrations in the Ah horizon to moderate quantities at depth (Table 1). The impact of the fern on the throughput of moisture in R43 must be sufficient to reduce the soil moisture common in the forest soil of R46, which in the latter must be at or near field capacity much of the time. In R43, soil moisture must be reduced to a level close to

the permanent wilting point, or at least sufficient to produce an elevated Si/Al ratio allowing smectite genesis. 5. Discussion Ferns can be notorious for water loss out of soils and may be more important than geomorphic/pedogenic variables in calculations of water loss. For example, ferns (Jamesonia sp.?) are prolific on the high Holocene-age terraces of the Eastern Mérida Andes and are known to control downward movement of soil water (Mahaney et al., 2007) by generating increased actual evapotranspiration. In sunny, arid and windy areas of the high Andean terraces, ferns may uptake excessive soil moisture out of roots extending to the base of thick Ah horizons, a process that might explain the general lack of B horizons in Late Glacial pedons (~12 ka; Mahaney et al., 2001a, 2001b). Despite high precipitation in excess of 1.0 m/yr, sufficient porosity to allow soil water movement, and evidence of leaching within the A horizon complex, B horizons are found only in soils with sandy textures and hence high soil moisture movement. The lack of B horizons in soils that otherwise have the right lithology, biotic composition, topographic settings and time to form them can be only explained by the presence of a species with a high water requirement. The B horizon enigma has been explored previously in both the Eastern and Western Mérida Andes by Mahaney and Kalm (1996) and Mahaney et al. (2000, 2001a, 2001b); however, the vegetation complex that might be involved was never tested until Mahaney et al. (2007) focused on the ferns, a subject that clearly needs further investigation. Farther afield in humid environments where smectite is not expected, its occurrence might be explained by the presence of hydrophilic plants like A. pedantum, a bio-regulator that reduces the throughput of soil moisture with reduced leaching offsetting the Si:Al ratio leading to the genesis of one or more species of smectite. The primary and clay mineral analysis of the R43 profile indicates that it has mainly a detrital mineral distribution, especially focusing on the 1:1 (Si:Al = 1:1) clays, which are likely reworked from previous interglacial weathering events or even pre-glacial weathering environments. This interpretation is supported by the illite and illite–smectite

196

W.C. Mahaney / Geomorphology 228 (2015) 189–199

Fig. 7. SEM imagery of the R43 profile in the lower epipedon (B horizon), subsoil (C horizon) and parent material (Cu). A, quartz inclusion in feldspar from the coarse sand fraction in the Bk horizon with long central crack (right of center) and secondary cracks (arrows) in upper right partly filled with clay; B, severely abraded and fracture ridden quartz of the fine grade fraction in the Bk horizon with thin clay accumulations on flat fracture faces and within narrow grooves (lower right); C, quartz with heavy abraded features and clay packed in pockets (center and upper left); D, well weathered, etched oligoclase plagioclase grain (center) aligned with quartz carrying an Mg coating (left) in the Ck horizon; E, very angular and well fractured oligoclase grain (center) from the fine sand fraction of the Cuk horizon adjacent to a subangular, well abraded quartz; F, fresh, and well fractured, angular quartz with deep grooves and conchoidal fractures that appear radially oriented next to an impact center.

down-profile distributions as well as uniform quartz and plagioclase concentrations showing near unvarying amounts. The lack of chlorite in the parent material of R43 and its appearance in small to trace amounts in the profile could amount to recrystallization, mostly occurring within the Ah horizon complex. The presence of vermiculite in the Ah horizons and its known affinity with coniferous species (Birkeland, 1999) suggests sufficient soil water in the surface horizons favoring hydrolysis and weathering of illite or conversion of illite–smectite to vermiculite. The calcite distributions indicate leaching below the Ah group and possible precipitation in the Ck horizon. The absence of calcite in the clay fraction of the Cu horizon shows that it did not reside in the fine fraction of the rubble parent material at time of emplacement but is present in trace quantities higher in the profile, either as a weathered/precipitated product or the result of air-infall activity. The R43 profile discussed herein, and its clay mineral composition, is unique in the Rouge Catchment and may be unique within many drainage basins in South-Central, Ontario. If landslide debris remains somewhat unstable after emplacement, it may be that the fern population

can withstand instability better than other species, similar perhaps to a species such as aspen (Populus tremuloides, Ives, 1941), which may explain its ability to colonize and grow on mass wasted rubble. Of the 75-odd sites studied and analyzed in the Rouge Catchment over the last 35 years (Mahaney and Ermuth, 1975; Mahaney and Sanmugadas, 1986; Mahaney and Terasmae, 1988; Mahaney and Hancock, 1993a, 1993b, Mahaney et al., 2014), no documented occurrence of A. pedantum is known except at R43. Furthermore, smectite as a 2:1 (Si:Al = 2:1) clay mineral has not been identified in other soils across the evolutionary soil sequence from the floodplain through the terrace soils to the tableland, the throughput of soil water apparently sufficient to inhibit its genesis by weathering of illite. The pedological origin of smectite, and other 2:1 and 2:1:1clay minerals in R43, balanced against a parent material with only a small concentration of illite, means that the soil is chemically active beneath a stand of maple-beech with an understory of ferns, sedges and grasses. While the forest canopy over both sites may have changed over the last several millennia, the understory vegetation, including the fern at

W.C. Mahaney / Geomorphology 228 (2015) 189–199

197

Fig. 8. Bioclimatic model restricting soil wetting depth based on present regional climatic/geomorphic/pedologic parameters of temperature, precipitation (Prec.), evaportranspiration (ET) and surface runoff. The landslide sourced sediment from the Halton Till, Thorncliffe formation and Scarborough formation, all of Wisconsinan age. The pedostratigraphy of the R43 soil shows that it penetrates the slide mass across a disconformity of probable century length into the middle terrace comprising a remnant of the post-Twyn Rivers soil.

R43, have most likely been in place since the landslide event. Given the high concentration of smectite in the Bk and Ck horizons of R43, clay genesis is likely long-lived starting near or closely following emplacement of the landslide. Correlation of the R43 profile with R46 in the Middle Holocene terrace system suggests that the landslide had to occur sometime between 5 and 8 ka. The close correlation between R43 and R46 is mirrored by the detrital distribution of 1:1 clays and with minor exceptions of trace amounts of illite–smectite, vermiculite and chlorite in R46, the pattern up-profile is similar to R43 with the exception of nil smectite in the former soil. Quartz and plagioclase have similar distributions in both profiles, partly detrital and partly due to little time for weathering in a humid, middle latitude climate locality. Slight differences of mainly

Table 2 Selected soil chemical parameters for the R43 and R46 profiles, Rouge River Catchment, South-Central, Ontario, Canada. Site

Horizon Depth (cm)

Organic carbon (%)

Organic matter (%)

N (%) 0.25 0.19 0.09 0.04 0.02 –

R43 Ah1 Ah2 Bk Ck Cuk Ckb

0–8 8–17 17–32 32–60 60–115 115 +

3.20 1.91 1.95 0.35 0.13 –

5.5 3.3 3.4 0.6 0.2 –

R46 Ah Bt Ck1 Ck2 Cuk

0–15 15–35 35–50 50–60 60 +

2.19 2.50 2.18 3.84 2.45

3.53 3.75 3.83 3.65 3.12

– = Insufficient sample. a R46 CaCO3 estimated (scale 0–4) with 10% HCl. b Electrical conductivity (μS/cm).

CaCO3 (%)

14.38 13.83 15.10 12.64 14.08 – –a 0.11 0 0.42 0 0.04 3 0.03 4 b 0.01 4

pH (1:5)

ECb

7.2 7.2 7.3 7.8 8.4 –

.06 .02 .03 .03 .06 –

6.2 6.6 7.4 8.0 8.4

.03 .02 .05 .05 .04

trace concentrations of calcite in the clay fraction of the two profiles may result from the landslide source material which emanates from an acidic soil cover under pine forest on the higher elevation an Early Holocene terrace. The age of the Ckb horizon in R43 is something of an enigma, its particle size possibly influenced by emplacement of the slide mass that buried it. The particle size data in Fig. 4 indicates a silt loam texture with 18–60% sand, 35–77% silt and 1–5% clay. If the landslide mass was saturated at time of emplacement it is possible, with truncation and destruction of the pre-existing Ah/B horizon complex, that organic compounds may have been mobilized and translocated into the underlying horizon, thus producing the brown (10YR 4/4) color resident in the Ckb horizon. It is also possible that the Ckb horizon lost some of its clay content judging by the presence of slopewash sediment adjacent to the landslide deposit. Moreover, the age of the landslide/R43 pedon is within the Middle Holocene (6–10 m terrace elevation range), its correlation with the slightly higher 8-m terrace at R46 and other documented soils in the Rouge chronosequence (Site R12 in Mahaney and Sanmugadas, 1986) indicating a similar stage of development. The most important question to come out of this research concerns the age of the middle group of fluvial terraces in the Rouge and other catchments in southern Ontario, as identified here and by others (Karrow, 1967; White, 1975; Mahaney and Sanmugadas, 1986; Mahaney and Hancock, 1993a, 1993b). White (1975) briefly mentions terraces 80 km NW of the Rouge catchment and Karrow (1967) describes the floodplain and terraces reaching to ~ 45 ft above stream level in the Rouge catchment where the highest Holocene terraces above stream level are actually somewhat higher (~18 m, ~60 ft). Segments of the 14–18 m terraces are found throughout the Rouge and Little Rouge catchments (Fig. 1B) and this erosional level is fully ~20 m (65 ft) below the tableland surface, on average, with valley deepening occurring closer to Lake Ontario. The highest terraces are younger than the last stage of Lake Iroquois (~11 ka) and older than the middle

198

W.C. Mahaney / Geomorphology 228 (2015) 189–199

terrace group of deposits, as documented herein, indicating that a considerable volume of glacial sediment was eroded during the two-step episodic erosion cycle that produced the various levels. The earliest erosional step rapidly produced the high terrace system as the Rouge system formed following deglaciation. Soils formed in these older deposits fall into similar soil orders as defined with the younger group but with marginally thicker profiles, finer texture and slightly higher content of extractable Fe and Al. The second erosional step came later and produced a suite of terraces between the 6 and 10 m levels, tentatively correlated with one prominent landslide (R43), of presumed Middle Holocene age. These deposits contain Brunisols (Inceptisols) and Luvisols (Alfisols) equivalent to the pedons in R43 and R46, respectively. A view of both Rouge catchments—Rouge River and Little Rouge Creek—(Fig. 1B) shows on average that valley widths take up ~4/5th of the valley volume indicating a long erosional episode during the Holocene, requiring perhaps 5 kyr to remove a large volume of the Early to Middle Holocene fluvial record leaving terrace remnants behind on valley slopes. Continual near-decade flooding of the floodplain has led over time to significant erosion of sediment leaving a soil transect comprising C/Cu, C/D profiles proximal to the streams becoming Ah/C/Ahb/Cox pedosequences on the distal reaches of the floodplain, The only uncalibrated 14C dates on these thin/shallow-depth Ahb horizons are b 300 yr BP (Mahaney and Terasmae, 1988) with dating complicated by the presence of carbonate and the hard water effect (Mahaney and Sanmugadas, 1986). With stream meandering, these Fluvisols are eroded and reformed following decade after decade of episodic overbank erosion and sedimentation inhibiting the development of B horizons. Dating these terraces by radiocarbon has proved impossible owing to the lack of fossil organic material suitable to establish age control. 6. Conclusions Changes in soil hydrological regimes are known in other areas, principally in the Mérida Range of the northwestern Venezuelan Andes, where ferns of the genus Jamesonia are known to affect evapotranspiration rates to the extent where soil water throughput in the epipedons (Ah horizons) is retarded. This presumably affects weathering at depth and prohibits the evolution of B horizons which otherwise would be expected to form in this tropical alpine environment with more than 1 m of precipitation per yr. In the Andean case, the lack of a frozen substrate at any time of the year means that the fern communities can maintain a high rate of evaportranspiration even during the dry season (Jan–March). It appears that the southern Ontario example of a distorted soil hydrological regime in R43 bears a certain similarity to the Andean example cited with the added expectation that soil epipedons are frozen during winter months. Other than a change in clay mineralogy distributions, no other profile manifestations seem to be affected by the presence of a hydrophilic species of fern in the Rouge Catchment but a thorough study of ferns and their distributions in catchments of southern Ontario has not been carried out. The presence of smectite in the R43 profile of the Rouge River and its genesis related to the biotic factor in soil formation may have wide ranging applications in other environments where hydrophilic species of plants control the genesis of Si-rich clay minerals and thus control other aspects of soil chemistry and mineralogy. Acknowledgments The work was carried out with funding from Quaternary Surveys, Toronto (Grant-2014(16)), and minor research grants from York University (MRG-1994_61). Critical discussion with students in my Sedimentary Methods course (2012–2013) helped to shape the content of this paper. I am indebted to J.A. McAndrews (ROM, Toronto) and Sarah Finkelstein (University of Toronto) for information on fern habitat and distribution in southern Ontario. Critical reviews by Dick Arnold

(Cornell University), Andrew Plater and three anonymous reviewers are greatly appreciated. Identification of Adiantum pedantum in the Rouge catchment was made by Mike Boyer (retired, Biology Dept., York University). Preparation of Figs. 1A and 1B by Andrew Stewart (Strata Consulting, Toronto) and Pedro Costa (University of Lisbon) is greatly appreciated.

References Ahmad, M., Ryan, J., Paeth, R.C., 1977. Soil development as a function of time in the Punjab River plains of Pakistan. Soil Sci. 41, 1162–1166. Alexander, E.B., 1974. Extractable iron in relation to soil age on terraces along the Truckee River, Nevada. Soil Sci. Soc. Am. Proc. 38, 121–124. Arnold, R.W., Cline, M.G., 1961. Origin of a surficial deposit in soils of Eastern Fulton County. N. Y. Soil Sci. Soc. Am. Proc. 25 (3), 240–242. Birkeland, P.W., 1999. Soils and Gemorphology. Oxford University Press, Oxford, U.K. (430 pp.). Bower, C.A., Wilcox, L.V., 1965. Soluble salts. In: Black, C.A. (Ed.), Methods of Soil Analysis, part 2, American Society Agronomy, Madison, Wi, pp. 933–951. Bremner, J.R., 1965. Total nitrogen. In: Black, C.A. (Ed.), Methods of Soil Analysis. American Society of Agronomy, Madison, Wi, pp. 149–176. Brewer, R., Walker, P.H., 1969. Weathering and soil development on a sequence of river terraces. Aust. J. Soil Res. 29, 293–305. Brown, D.M., et al., 1968. The climate of Southern Ontario. Climatological Studies NO. 5. Dept of Transport, Meteorological Branch, Ottawa. Canada Soil Survey Committee (CSSC), 1998. The Canadian System of Soil Classification. NRC Research Press, Ottawa, Canada (Publ. 1646, 187 pp.). Cody, W.J., Britain, D.M., 1989. Ferns and allies of Canada. Research Branch, Agriculture Canada Pub 1829/E, Ottawa, Ontario (429 pp.). Day, P.E., 1965. Particle fractionation and particle size analysis. In: Black, C.A. (Ed.), Methods of Soil Analysis. American Society of Agronomy, Madison, Wi, pp. 545–567. Foster, D.R., Oswald, W.W., Faison, E.K., Doughty, E.E., Hansen, B.C.S., 2006. A climatic driver for abrupt mid-Holocene vegetation dynamics and the hemlock decline in New England. Ecology 87 (12), 2959–2966. Hodgson, J.M., 1976. Soil survey field handbook, Soil Survey Tech. Monograph No. 5. Rothamsted Experimental Station, Harpenden, Herts, UK (99 pp.). Ives, R.L., 1941. Forest replacement rates in the Colorado headwaters area. Bull. Torrey Bot. Club 68, 407–408. Jackson, L.J., Ellis, C., Morgan, A.V., McAndrews, J.H., 2000. Glacial lake levels and eastern Great Lakes Palaeo-Indians. Geoarchaeology 15, 415–440. Jenny, H., 1941. Factors of Soil Formation. McGraw-Hill, NY 281 pp. Jones, David L., 1987. Encyclopaedia of Ferns. Timber Press, Portland, Oregon. Karrow, P.F., 1967. Pleistocene geology of the Scarborough area. Ontario Geological Survey Rept. 46 (108 pp.). Liberty, B.A., 1955. Studies of the Ordovician system in Central Ontario. Proc. Geol. Assoc. Can. 7 (1), 139–147. Liberty, B.A., 1964. Upper Ordovician stratigraphy of the Toronto area. Guidebook, Geology of Central Ontario. American Association of Petroleum Geologists, pp. 43–53. Mahaney, W.C., 1990. Ice on the Equator. Wm Caxton Ltd, Ellison Bay, Wisc. Mahaney, W.C., 2002. Atlas of sand grain surface textures and applications. Oxford University Press, Oxford, UK (237 pp.). Mahaney, W.C., Ermuth, F.E., 1975. The effects of agriculture and urbanization on the natural environment. Geographical Monographs, No. 7 (152 pp.). Mahaney, W.C., Hancock, R.G.V., 1993a. Late Quaternary stratigraphy and geochemistry of the R47 section, Rouge River Basin, south-central Ontario, Canada: correlation with Scarborough Bluffs. J. Quat. Sci. 8 (2), 167–178. Mahaney, W.C., Hancock, R.G.V., 1993b. Glacial geology and geomorphology of the Rouge River basin, Southern Ontario, Field Trip Guidebook. 3rd Annual International Geomorphological Conference. McMaster University, Hamilton, Ontario, Canada (41 pp.). Mahaney, W.C., Kalm, V., 1996. Field guide for the international conference on quaternary glaciation and paleoclimate. In: Mahaney, W.C., Kalm, V. (Eds.), June 21–July 1, 1996, Quaternary Surveys, Toronto (79 pp.). Mahaney, W.C., Sanmugadas, K., 1986. Soil development as a function of time in the Rouge River Basin, south-central Ontario. Géog. Phys. Quatern. 40, 207–216. Mahaney, W.C., Terasmae, J., 1988. Notes on radiocarbon-dated Holocene soils in the Rouge River basin, south-central Ontario. Acta Geol. Hung. 31 (1–2), 153–163. Mahaney, W.C., Milner, M.W., Sanmugadas, K., Kalm, V., Bezada, M., Hancock, R.G.V., 2000. Late Quaternary deglaciation and Neoglaciation of the Humboldt Massif, northern Venezuelan Andes. Z. Geomorphol. 122, 209–226. Mahaney, W.C., Kalm, V., Bezada, M., Hütt, G., Milner, M.W., 2001a. Stratotype for the Mérida Glaciation at Pueblo Llano in the northern Venezuelan Andes. J. S. Am. Earth Sci. 13, 761–774. Mahaney, W.C., Milner, M.W., Russell, S., Kalm, V., Bezada, M., Hancock, R.G.V., Beukens, R., 2001b. Peleopedology of Middle Wisconsin/Weichselian paleosols in the Merida Andes, Venezuela. Geoderma 104, 215–237. Mahaney, W.C., Dirszowsky, R.W., Milner, M.W., Harmsen, R., Finkelstein, S.A., Kalm, V., Bezada, M., Hancock, R.G.V., 2007. J. S. Am. Earth Sci. 23, 46–60. Mahaney, W.C., Hancock, R.G.V., Milan, A., Pulleyblank, C., Costa, P.J.M., Milner, M.W., 2014. Reconstruction of Wisconsinan-age ice dynamics and compositions of Southern Ontario glacial diamictons, glaciofluvial/lacustrine and deltaic sediment. Geomorphology 206, 421–439.

W.C. Mahaney / Geomorphology 228 (2015) 189–199 Ministry of Environment, 1984. The Hydrogeology of Southern Ontario, 2nd ed. Government Printer, Toronto. National Soil Survey Center (NSSC), 1995. Soil survey laboratory information manual. Soil Survey Investigations Report no. 45, version 1.00, U.S.D.A. (305 pp.). Nelson, R.E., 1982. Carbonate and gypsum. In: Black, C.A. (Ed.), Methods of Soil Analysis Part 2. American Society Agronomy, Madison, Wi, pp. 181–198. Oyama, M., Takehara, H., 1970. Standard Soil Color Charts. Japan Research Council for Agriculture Forestry and Fisheries. Page, C.N., 2002. Ecological strategies in fern evolution: a neopteridological overview. Rev. Palaeobot. Palynol. 119, 1–33. Paris, Cathy A., Windham, Michael D., 1988. A biosystematic investigation of the Adiantum pedantum complex in Eastern North America. Syst. Bot. 13, 240–255. Phillips, D.W., McCulloch, J.A.W., 1972. The climate of the Great Lakes Basin. Climatological Studies No. 20. Environment Canada.

199

Van Dam, N.M., Heil, M., 2011. Multitrophic interactions below and above ground: en route to the next level. J. Ecol. 99, 77–88. Vortisch, W., Mahaney, W.C., Fecher, K., 1987. Lithology and weathering in a paleosol sequence on Mt. Kenya, East Africa. Geol. Palaeontol. 21, 245–255. Walkley, A., Black, I.A., 1934. An examination of the Degtjareff method for determining soil organic matter, and a proposed modification of the chromic acid titration method. Soil Sci. 37, 29–38. White, O.L., 1975. Quaternary geology of the Bolton area, southern Ontario. Ontario Geological Survey Report. 117 (119 pp.). Whittig, L.D., 1965. X-ray diffraction techniques for numerical identification and mineralogical composition. In: Black, C.A. (Ed.), Methods of Soil Analysis American Society of Agronomy, Madison, Wisc, pp. 671–696.