During oil-sands mining all vegetation, soil, overburden,
and oil sand is removed, leaving pits several kilometers wide
and up to 100 m deep. Th ese pits are reclaimed through a
variety of treatments using subsoil or a mixed peat-mineral soil
cap. Using nonmetric multidimensional scaling and cluster
analysis of measurements of ecosystem function, reclamation
treatments of several age classes were compared with a range
of natural forest ecotypes to discover which treatments had
created ecosystems similar to natural forest ecotypes and at
what age this occurred. Ecosystem function was estimated
from bioavailable nutrients, plant community composition,
litter decomposition rate, and development of a surface organic
layer. On the reclamation treatments, availability of nitrate,
calcium, magnesium, and sulfur were generally higher than in
the natural forest ecotypes, while ammonium, P, K, and Mn
were generally lower. Reclamation treatments tended to have
more bare ground, grasses, and forbs but less moss, lichen,
shrubs, trees, or woody debris than natural forests. Rates of
litter decomposition were lower on all reclamation treatments.
Development of an organic layer appeared to be facilitated
by the presence of shrubs. With repeated applications of
fertilizers, measured variables for the peat-mineral amendments
fell within the range of natural variability at about 20 yr. An
intermediate subsoil layer reduced the need for fertilizer and
conditions resembling natural forests were reached about 15 yr
after a single fertilizer application. Treatments over tailings sand
receiving only one application of fertilizer appeared to be on a
diff erent trajectory to a novel ecosystem.
Recreating a Functioning Forest Soil in Reclaimed Oil Sands in Northern Alberta:
An Approach for Measuring Success in Ecological Restoration
S. M. Rowland, C. E. Prescott,* and S. J. Grayston University of British Columbia
S. A. Quideau University of Alberta
G. E. Bradfi eld University of British Columbia
River regions of northern Alberta, of which 311 billion barrels in
the Athabasca region are believed to be recoverable with current
surface mining technology (Government of Alberta, 2007:
www.albertacanada.com/industries/924.html). During mining,
all vegetation and soil cover is removed and about 15 to 50 m
of overburden is stripped away to reveal the oil sand, which is
then itself stripped away and transported to the extraction plant.
Th is creates open pits several kilometers wide and up to 100 m
deep into which unwanted mine residues may be tipped. Th e oil
companies operating in the region are obliged to reclaim their
operational land as set out within the Land Surface Conservation
and Reclamation Act 1973 and the Environmental Protection
and Enhancement Act 1992 (Government of Alberta, 1999).
Provincial guidelines for reclamation require the oil companies to
establish commercial forest within the natural range of ecotypes
found in the Central Mixed Wood Sub-Region of the Boreal
Forest and, secondarily, wildlife habitat capabilities similar to
predisturbance conditions (Government of Alberta, 1999). Th e
reclamation of the Athabasca oil sands mines is a challenging soil-
forming and landscaping exercise on a massive scale.
Th e reclamation treatments used by the oil companies use the
mine residues in combination with overburden, stock-piled soil
and peat, to create soil-like materials capable of supporting stable
plant and microbial communities. Over the last 30 yr seven main
prescriptions have been used; these include peat-mineral mix or
subsoil or both over tailings sands, overburden, Clearwater shale
or lean oil sands (Table S1).
During removal, the shale overburden becomes fragmented and
increases in volume, allowing water to percolate through it, leading
to weathering, slumping, and erosion (Stolte et al. 2000). For these
reasons all saline overburden is capped with 80 cm of nonsaline
overburden or till. Th e properties of nonsaline overburden vary;
those salvaged near the surface (up to 1 m below the soil surface)
here are an estimated 1.6 trillion barrels of oil within the oil
sands located in the Peace River, Cold Lake, and Athabasca
Abbreviations: FH, fermented/humifi ed surface organic matter; MRRP, multi-response
permutation procedures; NMS, nonmetric multidimensional scaling; P-M, peat-
mineral; PRS, plant root simulator.
S.M. Rowland, C.E. Prescott, and S. J. Grayston, Dep. of Forest Sciences, Univ. of British
Columbia, 3041-2424 Main Mall, Vancouver, BC V6T 1Z4, Canada; S.A. Quideau, Dep.
of Renewable Resources, Univ. of Alberta, 442 Earth Science Bldg., Edmonton, AB
T6G 2H1, Canada; G.E. Bradfi eld, Dep. of Botany, Univ. of British Columbia, 3529-6270
University Blvd., Vancouver, BC V6T 1Z4, Canada.
Copyright © 2009 by the American Society of Agronomy, Crop Science
Society of America, and Soil Science Society of America. All rights
reserved. No part of this periodical may be reproduced or transmitted
in any form or by any means, electronic or mechanical, including pho-
tocopying, recording, or any information storage and retrieval system,
without permission in writing from the publisher.
Published in J. Environ. Qual. 38:1580–1590 (2009).
Received 10 July 2008.
*Corresponding author (firstname.lastname@example.org).
© ASA, CSSA, SSSA
677 S. Segoe Rd., Madison, WI 53711 USA
TECHNICAL REPORTS: ECOSYSTEM RESTORATION
Rowland et al.: Recreating a Functioning Forest Soil in Reclaimed Oil Sands in Northern Alberta 1581
have variable clay content and pH between 4.9 and 6.6. Th e
overburden 1 to 2 m below the soil surface has higher pH (8)
and higher base-cation content (Danielson et al., 1983). Before
being mined, the sand is compressed due mainly to the eff ect of
historic glaciation, but after mining and processing the sand oc-
cupies a greater volume than before. Th us even after the removal
of the oil component, the mining residues occupy a greater space
than previously, thus reclaimed sites other than wetlands tend to
be raised areas rather than depressions.
Th e organic component of reconstructed soils is predomi-
nantly peat, which is available in large amounts in the mining
footprint. Peat is usually added as a 20 to 50 cm capping to
the reconstructed soils to increase their organic C and nutrient
content, and improve their water-holding capacity. Th e peat-
mineral soil is coarsely mixed during the salvage operation dur-
ing which peat is over-stripped to a maximum depth of 3 m and
so includes 25 to 50% by volume of mineral materials (Fung
and Macyk, 2000). Where lean oil sands (unprocessed oil sands
with <6% oil by mass) and saline overburden, rather than tail-
ings sand, is being reclaimed, an intermediate capping layer of
up to 80 cm of nonsaline subsoil material is used to bury the
material and protect it from water and plant root ingress (Oil
Sands Vegetation Reclamation Committee, 1998; Stolte et al.,
2000; C. Qualizza, personal communication, 2005). Th ere is as
yet no standard prescription or “recipe” for reclamation—the oil
companies adopt diff erent methods of reclamation with diff er-
ent materials, diff erent cap depths and, experimentally, replace
peat with salvaged forest fl oor from upland areas (McMillan et
al., 2007; C. Welham personal communication, 2005).
During the fi rst growing season after capping, barley (Hordeum
vulgare L.) is sown to provide quick vegetation cover and erosion
control. Barley is a poor competitor and is readily invaded by local
fl ora within the fi rst few years (Hardy BBT Ltd, 1990). Natural
plant recolonization from adjacent reclaimed sites has been suc-
cessful for grasses and forbs, but shrub and tree establishment has
not been signifi cant and tree planting is deemed necessary (Hardy
BBT Ltd, 1990; W. Tedder, personal communication, 2005).
Nursery-grown white spruce [Picea glauca (Moench) Voss] and as-
pen (Populus tremuloides Michx. Lamb) are usually planted on sites
being reclaimed for forestry, usually at the next available planting
season. Some companies practice fertilization at tree planting and
at regular intervals postplanting; others do not fertilize.
Th e overall aim of this research is to determine (i) if the
reclamation practices are creating ecosystems which appear to
be on a predictable path to become functioning forest ecosys-
tems similar (i.e., within the natural range of variability) to
those naturally present in the region (Fig. 1), (ii) which type(s)
of amendments are most successful, and (iii) by what age this
occurs. An estimate of the natural range of variability for key
response variables in reference (or “target”) natural forest eco-
types were compared with those of a number of oil sands rec-
lamation treatments across a range of ages. Th e response vari-
ables measured included two of the three major attributes of
reclamation identifi ed by Ruiz-Jaen and Aide (2005): vegeta-
tion structure and ecological processes. Vegetation was assessed
through ground cover estimates, and ecological processes were
assessed by measuring nutrient bioavailability, litter decompo-
sition rates, and development of a surface organic layer.
Materials and Methods
Th e study area is within the Wood Buff alo region of north-
ern Alberta, Canada. Th e study area covers an area of about 135
km north-south and 50 km east-west, centered approximately
35 km north of Fort McMurray (57°00´ N, 111°28´ W; 369 m
above sea level). Daily average temperatures range from –18.8°C
in January to +16.8°C in July, and only 2 mo of the year are
deemed frost-free. Average annual precipitation is 455 mm of
which 342 mm falls as rain (Environment Canada Climate Nor-
mals 1971–2000; www.climate-weatheroffi ce.ec.gc.ca).
Th e oil sands are the delta deposits of an ancient tropical
sea, subsequently compressed by extensive glacier formations
and invaded by oil migrating from elsewhere. Th e local Clear-
water shale formation north of Fort McMurray lies over the
oil-rich sand deposits and is saline (Mossop, 1980; Marshall,
1982); these consolidated deposits have strongly sodic or saline
pore fl uids where electrical conductivity exceeds 6 dS/m (Stolte
et al., 2000; Purdy et al., 2005) and contain highly dispersive
sodium-clays that are stable only while confi ned in situ. Th e
shales are physically stable while at an angle of repose not ex-
ceeding 7° or 12%, resulting in a gently undulating landscape.
Glacial till from the Pleistocene Epoch-era forms a clay-rich
mineral soil layer over this except in some places where fl uvial
sand persists (Stolte et al., 2000), supporting a mosaic of com-
munities from bogs and rich or poor fens to dry pine forests.
Natural soils within the oil sands are classifi ed broadly as
Brunisols, Luvisols, Gleysols, or Organic, this following a
moisture gradient from xeric to subhydric and from aerobic
to increasingly anaerobic and reducing conditions (Turchenek
and Lindsay, 1982; Oil Sands Vegetation Reclamation Com-
mittee, 1998). Th is gradient in soil development is caused
by impeded water fl ow due to increasing clay content in the
mineral subsoils, from glacio-fl uvial sands to glacio-lacustrine
muds to fi ne-clay glacial till. Gray Luvisols have developed on
medium to fi ne-textured lacustrine deposits and till, and Dys-
tric Brunisols have developed on coarser parent material such
as glaciofl uvial outwash and eolian sands. Organic soils devel-
oped on poorly drained lowlands overlaying glacial deposits
(Natural Regions Committee, 2006) provide the bulk of the
raw peat that is salvaged for use in reclamation.
Th e dominant tree species in the Central Mixed Wood Sub-
Region of the Boreal Forest are aspen, poplar (Populus balsamifera
L.), white spruce, black spruce [P. mariana (Mill.) Britton,Sterns
and Poggenb.], jack pine (Pinus banksiana Lamb), balsam fi r
[Abies balsamea (L.) P. Mill], and tamarack [Larix laricina (Du
Roi) Koch] (Rowe, 1972). Th e forest has been classifi ed into sev-
eral ecotypes according to landscape, drainage, and soils, with
diff erent plant communities developing with diff erent propor-
tions of tree species including jack pine, aspen, and white spruce
(Beckingham and Archibald, 1996). Ecotypes with the potential
end-use of commercial forestry are denoted by the predominant
1582 Journal of Environmental Quality • Volume 38 • July–August 2009
indicator species- blueberry (b) (Vaccinium spp.), Labrador tea-
mesic (c), low bush cranberry (d) (V. macrocarpon Ait), dogwood
(e) (Cornus spp.), and horsetail (f) (Equisetum spp.) (Oil Sands
Vegetation Reclamation Committee, 1998).
Twenty-nine reclaimed sites were selected, representing a
range of age (3–34 yr) and treatments practised by the three
major oil companies operating in the Athabasca region around
Fort McMurray, Syncrude Canada Ltd, Albian Sands Energy
Inc., and Suncor Energy Inc. (Table 1). All but one of these
are long-term monitoring plots established between 2000 and
2006; the remaining plot (WA5, a recolonized peat waste area
in the Suncor facility) was established for this study.
Eighteen natural sites were selected from the long-term
monitoring plots; these ranged from xeric/nutrient-poor sites
(ecotype a1) to mesic/nutrient medium-to-rich sites (ecotype
d1) (Table S2). Th ree replicates of each of six natural ecotypes
were drawn from ecotypes a1, b1, b3, d1, d2, and d3. Eco-
types c, e, and f were not included because ecotype c is not
found locally, and ecotypes e and f represent very wet forests
that are not likely to be created by the reclamation treatments.
One plot representing a b3 ecotype was destroyed mid-study
by the expansion of a mine. All of the reclaimed and several
of the natural forest sites are within the mine lease areas of
Syncrude, Suncor, and Albian Sands; the remainder are natural
forest sites located on public land outside of the leased mining
areas. Each long-term monitoring plot measures 10 by 40 m;
all measurements and samples were taken around the plot pe-
rimeter (within 2 m of the plot) to preserve the integrity of the
environment within the plot.
Unless stated otherwise, “soil” is the term used to refer to the
upper mineral component on a natural site or the original cap-
ping layer, whether a peat-mineral (P-M) mixture or subsoil,
of a reclamation treatment. Soil is, therefore, the mineral-rich
component found closest to the surface, capable of supporting
vegetation growth and developing an organic layer over its sur-
face. By this terminology the capping layer is analogous with
the upper mineral layer in a natural soil.
Soils were sampled soon after spring thaw in June 2005, with
the exception of Albian and FF plots which were sampled in June
2006. Sampling was semi-systematic; a composite sample was
made from 10 soil cores (7-cm diam.) taken from four locations
down each long side, and one at each short side of the plot. As
each core was extracted it was laid out on a plastic bag and the
litter from the top of the core was hand-picked into a plastic bag.
Th e depth of the surface organic layer (hereafter referred to as FH)
was measured before it was put into a separate plastic bag. Th e F
and H layers were harvested together because they could not be
distinguished in the fi eld with confi dence. On younger reclaimed
sites and most subsoil-capped treatments, the organic layer oc-
curred patchily as a thin black fi lm without an obvious F layer. On
some reclaimed sites close examination was necessary to determine
where the thin FH ended and the original P-M mixture began,
the FH generally being darker than the P-M. Any live vegetation
was removed at this stage. Finally a handful of soil from the upper
10 cm was taken. Samples were composited × 10, well-mixed by
hand, and stored in resealable plastic bags on ice until the end of
the day. Litter and FH were then weighed wet. All bulk samples
were frozen (–18°C) at the Syncrude facility until transported to
the laboratory for subsequent drying and analysis.
Moisture Content, pH, and Carbon/Nitrogen Ratio
On receipt at the lab, the frozen samples of soil, FH, and litter
were defrosted overnight and a weighed subsample was placed in
a drying oven (105°C) for 72 h and reweighed to obtain gravi-
metric moisture content and total dry mass. A second subsample
was dried at about 50°C for several days before being analyzed
for total C and N by automated dry combustion using a Carlo-
Erba elemental analyzer with the remainder analyzed for pH in
water [pH(w)]. All samples of the FH and soil were sieved (<4
mm) to remove stones and roots before analysis.
Development of the Surface Organic Layer
Th e mass of the FH layer in each plot was estimated from
the original wet FH sample mass adjusted by accounting for
gravimetric moisture to obtain dry mass per m2 (the original
sample being a composite of 10 7-cm cores, total area harvest-
ed 0.038 m2). Th e dry mass (i.e., g/0.038 m2) and the mean
(n = 10) FH depth were transformed to natural log ln(x+1), to
achieve a normal distribution. Th e (transformed) FH dry mass
and depth, pH, total N, total C, and moisture content were
regressed against site age to fi nd a model for FH development.
Th e age by which FH development would fall within the natu-
ral range of variability was estimated from the model.
Mass loss of aspen leaf litter was measured to compare litter de-
composition rates across the range of reclamation treatments and
natural ecotypes. In September 2005, fresh-fallen or recently se-
nesced litter samples of naturally-growing aspen were collected from
a forest within the study area. Th e litter was stored in paper bags and
allowed to dry indoors for at least 48 h before use. A subsample was
stored dry for later analysis of total C and N concentrations.
Fig. 1. A model for ecosystem restoration; arrows indicate one possible
restoration trajectory, or predictable path, to a successful conclusion.
Rowland et al.: Recreating a Functioning Forest Soil in Reclaimed Oil Sands in Northern Alberta 1583
Litter bags were made from 10 by 10 cm squares of nylon mesh
fabric (mesh size 0.8 by 0.125 mm) sewn on three sides with poly-
ester thread. Litter samples of approximately 0.50 g were placed
inside a bag with a numbered metal tag and the open end was
sealed using a hot-glue gun and general-purpose glue sticks.
In September 2005 seven bags per site were placed on top of
the soil on a single replicate of each reclamation treatment and
natural forest ecotype (except plots Albian and FF which were
not added to the study until 2006). All bags were retrieved af-
ter 365 ( ± 4) d, air-dried, and the contents were removed and
weighed. Only three bags were usable from treatment H as the
bags had become buried by tailings sands which had concreted
to the decomposing aspen litter. Th e sand in these bags was
carefully removed with a fi ne sable brush. Th ere was no visible
contamination in any of the other bags.
In situ Bioavailable Nutrients
Plant root simulator (PRS) probes (Western Ag Innovations
Inc., Saskatoon, SK, Canada) were used to measure bioavailabil-
ity of nutrients within the soil. Th e probes consist of cation- and
anion-exchange resin membranes encased in a plastic holding
device, which are inserted into soil to measure nutrient supply
in situ with minimal disturbance (Qian and Schoenau, 2002).
Th e PRS probes were inserted in four random locations on each
plot, with pairs of probes (one cation, one anion) at each loca-
tion, a total of eight probes per plot. Th e probes were fi rst in-
stalled in June 2006 and retrieved after 44 ( ± 6) d, whereupon
fresh PRS probes were installed and incubated for a further 51
( ± 6) d. Probes were washed with deionized water, bulked ac-
cording to treatment and sent to Western Ag Innovations for
analysis. Anions and cations from the probes were eluted for 1
h using 0.5 N HCl/2 mol L–1 KCl. Th e eluate was analyzed for
levels of ammonium (NH4+) and nitrate (NO3–) using automat-
ed colorimetry. Inductively-coupled plasma (ICP) spectropho-
tometry/atomic absorption spectrometry (AAS)/fl ame emission
spectrometry (FES) was used to measure levels of P, K, S, Ca,
Mg, Al, Fe, Mn, Cu, Zn, B, and Pb in the 0.5 N HCl/2 mol L–1
KCl eluate. Th e totals for the two burial periods were combined
and converted to micromoles per 10 cm2 per total burial period.
Th e PRS probe nutrient capture is relative among sites, it does
not equate to total soil nutrient fl ux over the term of incubation
and there is no linear relationship of capture rate vs. time.
During July and early August 2006 groundcover was surveyed
on all plots except plot 64 which had been destroyed. Ten 1 by
1 m quadrats were laid systematically around each plot starting
at one corner and at 10-m intervals along each long side and at
the 5-m midpoint of the ends. Within each quadrat a visual es-
timate of all layers of groundcover <2 m above ground level was
scored as percent cover (one 10 by 10 cm square = 1%), classifi ed
as: pine, spruce, broadleaf, woody shrubs, forbs, grasses, mosses,
lichens, woody debris (>5 cm long or >1 cm diam.) and bare
ground. Aerial cover above 2 m was estimated visually and the
percent cover was added to the score for that vegetation type.
Th e experimental design was based on the availability of
reclamation treatments; there are numerous confounding
factors including uneven numbers of replicates (sometimes
only one per treatment), age since reclamation, age at time of
planting, planting species, proximity to other reclaimed sites
or disturbed areas (related to natural colonization potential),
fertilization treatment, peat type, and prior stock-piling (if re-
corded), altitude, and aspect. Th is created problems with using
classical statistical methods to identify treatment eff ects. Where
possible, data were analyzed using SAS 9.1 (SAS Institute Inc.,
Cary, NC) to identify models and treatment eff ects using clas-
Table 1. Reclamation treatments (L. Leskiw, personal communication, 2005; C. Qualizza, personal communication, 2005, 2006; W. Tedder, personal
WA5 (29) 36(11) 37(24)
1 × 250–350kg/ha†
1 × 300 kg/ha†
4 × 250kg/ha/yr†
(N: P: K)
(10: 30: 15)
† Lanoue (2003).
1584 Journal of Environmental Quality • Volume 38 • July–August 2009
sical statistical methods. PC-Ord 5.0 (MJM Software Design,
Gleneden Beach, OR) had specifi c relevance to biological com-
munity data (McCune and Grace, 2002) and was used to apply
nonparametric analyses where appropriate.
Nonmetric Multidimensional Scaling
An ordination technique known as nonmetric multidimen-
sional scaling (NMS), was chosen as it is able to manipulate a
large range of variables, normality and linear relationships are not
a prerequisite for analysis (McCune and Grace, 2002), and it has
been used to evaluate community-environment relationships in
ecosystems (Laughlin and Abella, 2007). It produces an image of
the plots in an ordination space that provides a visual aid to inter-
preting similarity among entities, those more similar being clus-
tered closer together. Groundcover data from quadrat surveys were
transformed from raw percent cover to 2/pi arcsine square-root(x)
as this has the eff ect of equalizing the class limits for very high and
very low proportion values while maintaining a broad spread of
classes for intermediate values (McCune and Grace, 2002).
Th e NMS ordinations were undertaken using PC Ord v.5.0
(McCune and Grace 2002, and references therein). Two data ma-
trices were used, one containing groundcover data and the second
containing chemical data for the soils (pH(w), C/N ratio, moisture
content, bioavailable nutrients) as well as categorical information
(treatment or age class, the latter by 5-yr increments, except natural
ecotypes which were arbitrarily assigned to a top age class). Mass loss
by decomposition was not used in this analysis since only one rep-
licate per treatment had received the bags, so creating an imbalance
in the spread of data. An NMS ordination was performed on the
groundcover data, and the chemical variables in the second matrix
were subsequently correlated and overlaid on the NMS axes. Th is
suited our objective of assessing the restoration treatments primarily
on the degree of vegetation recovery followed by an independent
evaluation of relationships with potential causal factors.
Th e distance measure used was Sorensen (Bray-Curtis). Th e
“auto-pilot” and “medium thoroughness” options were selected.
Th is undertook a series of repeated computations (iterations) to
adjust the position of entities (plots) in ordination space. It began
with a four-dimensional solution and reduced to one dimension
for each of 50 runs with real data starting with random coordi-
nates, then 50 runs with shuffl ed data (that is, shuffl ed within
columns–each column being one variable). Th is shuffl ing forms
the basis for a Monte Carlo test of signifi cance of each dimen-
sionality (McCune and Grace, 2002). Th e best solution was the
number (n) of dimensions beyond which additional dimensions
do little to reduce stress. Stress is the departure from monoto-
nicity in the plot of distance in the original n-dimensional space
vs. distance in the ordination space, thus the closer data points
lie to a monotonic line, the lower is the stress (McCune and
Grace, 2002). A fi nal stress of <20 was deemed satisfactory for
ecological community data (McCune and Grace, 2002). A plot
of stress vs. iteration (and the maximum number of iterations
was 200) showed if and when the stress declined to a constant,
and at that point was deemed to be stable, or if it fl uctuated
(in which case, more iterations required), with a fi nal instabil-
ity value automatically calculated from the standard deviation
in stress over the preceding x iterations (McCune and Grace,
2002); an instability criterion of 0.00001 was set as a “target” by
the auto-pilot mode. Th e auto-pilot then ran a fi nal, best solu-
tion (low stress, low instability value) using starting coordinates
taken from an automatically-saved fi le for a previous run that
had produced the better solution in n dimensions.
After the fi rst NMS ordination, stress levels were too high
(>35, but stable), so pine and spruce were additively combined
to create a new variable of “conifer” that was transformed as
before. Th e NMS ordination was repeated with this new vari-
able in the groundcover matrix.
Th e resulting three-dimensional image was rotated on screen
until a subjective decision was made to select a “best view” show-
ing a relatively clear separation between groups. Th e categories
within the viewing options were then changed from “treatment”
to “age class” and the new image was viewed. Correlation coef-
fi cients (r) expressing linear relationships of each variable with
each axis were revealed by selecting the Pearson Correlation op-
tion that, in PC-Ord 5.0, becomes available by viewing a two-
dimensional, rather than three-dimensional, image; the square
(r2) expresses the proportion of variation on that axis that is “ex-
plained” by the variable in question (McCune and Grace, 2002).
Multi-Response Permutation Procedures
Multi-response permutation procedures (MRPP) were un-
dertaken in PC Ord 5.0. Th e MRPP is a nonparametric analy-
sis of within-group and between-group diff erences, to test for
signifi cant diff erences among two or more groups. Th e results
provide a measure of heterogeneity or homogeneity while giv-
ing a P value to distinguish between pairs of groups where n > 1
(McCune and Grace, 2002). Th e MRPPs were undertaken fi rst
by treatment vs. natural ecotypes (grouped together) and, next,
by treatment (where n > 1) vs. each ecotype from a1 to d3.
Th e distance measure chosen was Sorensen (Bray Curtis). A
test statistic, the chance-corrected within-group agreement was
examined—this indicates whether entities within groups are
identical to one another (value = 1.0) or as diff erent as would
be expected by chance (value = 0) or more diff erent than would
be expected by chance (value < 0), with typical community
ecology values being <0.30 (McCune and Grace, 2002).
To further examine relationships among the reclamation
treatments and the natural ecotypes a cluster analysis using
PC Ord v.5.0 was performed on the ground-cover (biological)
data. In this method, cluster groups of plots were identifi ed
from a dendrogram and inspected for which reclaimed sites
were clustered with natural ecotypes, and which were not. As
in the NMS, the environmental (chemical) data were used in a
subsequent step to help interpret the clustering results.
Moisture Content, pH and Carbon/Nitrogen Ratios
Soil moisture in the reclamation prescriptions increased in order:
F < H < Albian < B < I < WA5 < M < A < E < FF (Fig. S1). Th is
Rowland et al.: Recreating a Functioning Forest Soil in Reclaimed Oil Sands in Northern Alberta 1585
rank order will be followed in each of the graphs. Generally, natural
ecotypes followed the expected trend of increasing soil moisture as
the class moved from a1 to d3. In six of the reclaimed treatments (I,
WA5, M, A, E, and FF), soil moisture was higher than the natural
range; in the other treatments (F, H, Albian, and B), soil moisture
was within the natural range of variability for the b3 to d3 ecotype
mineral soils. Litter moisture was generally similar to FH moisture
in all reclamation treatments, and FH moisture in the reclamation
treatments was within the range of natural variability.
Soils in reclamation treatments were 1 to 2 pH units higher
than natural ecotypes, except treatment FF which retained
near-natural pH values (Fig. S2). On reclaimed sites (except
FF) the pH(w) of litter ranged from 6.3 to 6.9, FH 6.3 to 7.2,
and soil 6.1 to 7.5; in the natural ecotypes the pH(w) of litter
ranged from 5 to 6.3, FH 4.9 to 5.8, and soil 4.4 to 5.5. Th e
pH(w) of litter in natural ecotypes increased from pine (a1)
to aspen and forbs with spruce (b1, b3, d1) and fell again as
spruce and Sphagnum dominated (d3).
Th e C/N ratios of FH on reclamation treatments were be-
low the natural range of variability (Fig. S3). An abandoned
peat dump WA5 had the lowest C/N ratio in litter, soil, and
the second lowest in FH. Th e C/N ratios in reclamation treat-
ments generally followed the order: litter > soil > FH, which
diff ers from the pattern in natural ecotypes where the C/N ra-
tios decreased with depth (litter > FH > soil). Th e C/N ratio of
natural soils decreased as site moisture status increased and the
C/N ratios for all three d-ecotypes was low (<20).
Mass loss of aspen litter on the reclamation treatments ex-
cept A, I, and WA5 was below the range of natural variability
(Fig. 2). Treatment B (plot 3) had the lowest mass loss (11.5%).
Using SAS 9.1 a model was sought to explain decomposition
rate of aspen litter across all treatments in Year 1 using variables
of moisture, pH(w), and C/N ratios of both FH and soil. Th e
only variable of signifi cance (P = 0.0021) was soil C/N ratio
(Fig. S4). Th e model (r2 = 0.5608) is:
%mass loss = 48 – 0.80(C/NSOIL)
Th ere appeared to be two clusters. Group (i) describes soils of lower
(≤20) C/N ratio that are found on ecotypes d1, d2, and d3, and
treatment WA5. Group (ii) comprises other reclamation treatments,
and ecotypes a1, b1, and b3, with higher soil C/N ratios.
Development of the Surface Organic Layer
Th e dry mass of FH on all reclaimed treatments fell be-
low the range of natural variability (Fig. 3). Across the natural
ecotypes the FH dry mass sample range was 168 to 1479 g
(equivalent to 4.4–38.4 kg/m2) for a1 to d3 sites. Th e FH dry
mass on natural sites varied signifi cantly and negatively with
pH(w) (P < 0.05, r2 = 0.3881).
Th e FH development on reclaimed sites was best predicted
with a model using site age (transformed to log10) and FH dry
mass (transformed to natural log ln) (P < 0.0001, r2 = 0.8282).
ln(dryFH+1) = –0.25 + 3.46 × log10(age)
pH(w) and total carbon content of FH were also signifi cant
(P < 0.05) but each variable contributed just 1.5% to the model
(r2 = 0.8579). Th e FH moisture, FH depth, and FH total N
were not signifi cant. Applying the model as a best guess (since
17% of the variability remains unexplained), the minimum
time by which FH will be within the lower range of natural
variability (i.e., 168 g) is 36 yr.
In situ Bioavailable Nutrients
Reclaimed sites generally had high NO3
and K. Sodium and Mn were much reduced, but Ca, Mg, Fe, S,
B, and Al were all generally elevated on reclamation treatments
(Table S3). Copper was generally below detectable limits across
all ecotypes and treatments, but trace amounts were detectable
on some reclamation treatments. Zinc was generally present at
very low amounts across all plots but at slightly elevated levels
on reclamation treatments. Th e levels of nutrients on most rec-
lamation treatments was outside (above or below) the natural
range of variability. Albian had extremely low P and K but very
high Fe and S. Albian and WA5, both unfertilized, had higher
ecotypes. Repeatedly fertilized sites H and I, compared with the
once-only fertilized treatments (A, B, E, F, M, FF), generally had
slightly higher mean values or an extended range of values (as
indicated by the standard deviation) of most nutrients.
Th ere was a general trend for reclaimed sites to have more
bare ground, grasses, and forbs but less cover of moss, lichen,
shrubs, broadleaves, spruce, pine, or woody debris than natu-
ral sites. Th e only treatments that had any lichen cover were H
and I. Th ere was considerable variability within and among eco-
types and treatments for all 10 groundcover classes measured.
Th ere was also variability with age irrespective of treatment, with
younger and older reclaimed sites having fewer groundcover
classes than middle-age ones, and with an apparent “tailing off ”
of cumulative groundcover after a peak at the 21 to 25 yr age
class (Fig. 4). Albian and WA5 were the least similar to the oth-
er treatments, with Albian having almost nil plant cover while
WA5 was densely populated by shrubs and broadleaves; note
that WA5 was the sole representative of the 26 to 30 yr age class,
accounting for the high broadleaf and shrub values for that age
class. Among the natural ecotypes, d1 appeared to be the most
diff erent, having the lowest mean values for moss and lichen but
the highest mean values for forbs, shrubs, and broadleaves.
Inclusion of groundcover data into the model for FH de-
velopment on reclamation treatments indicated that shrub
cover was a contributing factor (P < 0.0001) contributing 10%
to the model, whereas age and pH contributed about 50 and
20%, respectively. Woody debris and pine were also signifi cant
(P < 0.01) but each contributed just 1.5% to the model. Th e
full model (r2 = 0.8469) was:
ln(dryFH + 1) = 0.15 + 0.03(age) + 0.40(pH(w)) +
0.04(shrubs) + 0.03(pine) + 0.09(woody debris)
− but low NH4
– than the fertilized reclamation treatments and the natural
1586 Journal of Environmental Quality • Volume 38 • July–August 2009
Nonmetric Multidimensional Scaling
Nonmetric multidimensional scaling produced a solution
in three dimensions with a fi nal stress of 11 and a fi nal insta-
bility of 0.00001 after 81 iterations. Most of the variability
was expressed in axis 3 (r2 = 0.416) which was correlated with
mosses, lichen, grass, forbs, conifer, woody debris, pH(w), B,
Ca, and Na (Table 2). Th e Monte Carlo test indicated that the
proportion of 50 randomized runs with stress less than or equal
to the observed stress (the probability that a similar fi nal stress
would be achieved by chance) was 0.0196, or 2%.
Th e ecotypes clustered together with some reclaimed sites from
treatments E, H, and I (Fig. 5). Other reclamation treatments
were more distant, so less similar to the ecotypes. In general, re-
claimed sites by 21 to 25 yr, irrespective of treatment, were more
similar to a natural forest and less similar to younger reclaimed
site (Fig. 6). Th ere were exceptions—the oldest plots for treat-
ments A and B remained distant from the natural ecotypes after
24 and 19 yr, respectively. Th e eldest sites (age-class 31–35 yr)
also appeared less similar to natural ecotypes and diff erent from
the other reclaimed plots, projecting forward in the fi gure.
Multiresponse Permutation Procedures
Th e test of chance-corrected within-group agreement, de-
scribing within-group homogeneity compared to random expec-
tation (McCune and Grace, 2002) was 0.17. Treatments E and
I as groups were indistinguishable from one another (P = 0.71)
and not signifi cantly diff erent from the natural ecotypes group
(P > 0.05). Treatment I, as a group, was not signifi cantly diff er-
ent from ecotypes b1and b3; treatment E, as a group, was not
signifi cantly diff erent from ecotypes b1, d1 and d2 (P > 0.05).
Treatment H (a tailings sand treatment) as a group diff ered from
the natural ecotypes group (P = 0.0004), although several plots
exhibited similar results to treatment I. Th is treatment was also
diff erent from all other treatments (P < 0.05) except E and I.
Treatments A and B (both tailings sand treatments) as groups
were indistinguishable from each other (P = 0.58) and diff er-
ent from the natural ecotypes (P < 0.00001). Pairs of treatment
groups that were signifi cantly diff erent from each other were A
vs. H, B vs. H, A vs. I, and B vs. I (P < 0.05).
Age-class data were used to identify temporal changes in relation-
ships. Th e cluster dendrogram was pruned at the 50% level, reveal-
ing four distinct clusters, or groups (Table 3). Treatments E, H and I
clustered more closely with natural ecotypes than other treatments.
Th e reclamation treatments diff ered from natural forests
and from each other in the physical and chemical characteris-
tics measured. Broadly, the presence of tailings sands at depth
led to reduced moisture content, while a P-M mix at the sur-
face was associated with higher moisture unless the P-M mix
was directly over tailings sands as in treatment H. Subsoil caps
(treatments B, F) were generally dry.
Fig. 4. Cumulative mean ground cover classes by ecotype or age class.
Fig. 3. Mass of surface organic matter per square meter at the start of
the growing season at (left) natural and (right) reclaimed sites by
treatment. Error bars represent one standard deviation.
Fig. 2. Mean mass loss of aspen leaf litter during 1 yr of decomposition,
(left) at natural ecotypes and (right) reclamation treatments, ranked
in order of increasing soil moisture. Error bars indicate one standard
deviation. The mean value for treatment H is the mean of three litter
bags; all other treatment means are for seven litter bags.
Rowland et al.: Recreating a Functioning Forest Soil in Reclaimed Oil Sands in Northern Alberta 1587
Th e high pH of the P-M mixtures at most reclamation treat-
ments when compared to the natural soils can be attributed
to the inherently high pH of salvaged mineral materials of the
Athabasca oil sand region (Fung and Macyk, 2000), and mix-
ing this alkaline material with surface organic material during
overstripping. Windblown deposits from exposed tailings may
also have contributed. Treatment FF was the only treatment with
a pH similar to the natural ecotypes. Th is P-M mix had an un-
usually high C content (191 g OC kg–1) and was comprised of
undecomposed, naturally acidic fi bric peat (Hemstock, 2008).
Th e reclamation treatments diff ered from natural ecotypes in
their nutrient status. In all treatments, fertilized or unfertilized, N
was present at levels at or above the natural range. Addition of N
in fertilizers should be revisited in light of these observations of
abundant nitrate on unfertilized sites. Th ose treatments with lev-
els of both K and P close to the natural ranges were: (i) those with
a shallow P-M cap that received fertilizer for several years (treat-
ments H and I), (ii) those with a shallow P-M cap over nonsaline,
clay-rich subsoil and overburden, fertilized once (treatment E),
and (iii) those with a deep mineral-soil cap laid over tailings sands
and fertilized once (treatment B). Th is includes both young and
more mature plots, indicating that it is a treatment eff ect rather
than an age eff ect. Other P-M treatments were characterized by
very low K and/or P levels, which may indicate that these nutri-
ents, rather than N, are defi cient at these sites.
Development of a surface organic layer varied among the
treatments although the diff erences between treatments were
not large. Th e model suggested that it would take at least 36 yr
to achieve an FH layer similar to that found in an a1 ecotype,
this being at the lowest end of the natural range of variability.
In the fi eld, only plot 30 (treatment H) had a dry mass FH
close to the lower end of the natural range; this plot was 34 yr
old. Interestingly, Frouz et al., (2001) found that the develop-
ment of forest fl oor (L, F, and H layers) increased for the fi rst
30 yr after reclamation of two open-cast coal mines in eastern
Europe, after which they appeared to stabilize.
Th e model indicated that shrubs were important in the devel-
opment of FH, so planting shrubs is recommended to accelerate
forest-fl oor development. Castro et al. (2004) also noted that
shrub cover aided reforestation in a harsh environment by miti-
gating water stress through reduced solar radiation to the fl oor,
and also helped prevent seedling mortality from browsing. It
has also been shown that the leeward sides of shrubs benefi t soil
with wind and thereby creating localized changes in microcli-
mate and plant litter deposition (Mummey et al., 2002). Shrub
cover may play a similar role in reclaimed oil sands sites with an
additional benefi t of facilitating FH development.
Litter decomposition appeared to be depressed on the recla-
mation treatments compared with that on natural ecotypes de-
spite moisture levels generally above the natural range of vari-
ability. Decomposition rate was especially low on treatment B,
a subsoil capping mixture laid over tailings sands. In treatment
F, where subsoil is laid as a cap over overburden, although it
was drier than treatment B, decomposition was faster but still
below than that of most P-M treatments. Th is suggests that if
tailings sand, which has a higher hydrocarbon content than
+ pools and soil organic matter, presumably by interfering
Table 2. Pearson linear correlations (r) of variables vs. ordination axes
indicating direction of relationship, positive or negative.
Fig. 6. Nonmetric multidimensional scaling plot in three dimensions,
by age class or grouped natural ecotypes.
Fig. 5. Nonmetric multidimensional scaling plot in three dimensions,
by treatment or grouped natural ecotypes. This three-
dimensional image should be viewed like a hollow cube or a box,
looking down into it from above with its top surface (the widest
part of the image) in the same plane as this page. Each plot is
represented by a single dot, some are partly hidden. The plots are
suspended in space with those closer to the top appearing larger
than those found farther along axis 1 (into the page). The pink
overlay contains those plots within the natural cluster.
1588 Journal of Environmental Quality • Volume 38 • July–August 2009
overburden, is reclaimed with a subsoil cap, the tailings sand
may reduce that subsoil’s inherent microbial activity. Oil ad-
dition to soil results in high C/N ratios which is unfavorable
to microbial activity, though this can be overcome by nutrient
addition (Choi et al.2002). Th is hypothesis is being tested in a
parallel study of soil microbial diversity and function at these
same sites (P. Dimitriu, personal communication 2007). Tail-
ings sand, in contrast to overburden and peat, has been shown
to be devoid of ectomycorrhizae, but mycorrhizae can be re-
stored using peat additions (Bois et al. 2005).
Plant diversity on reclamation treatments appeared to re-
main fairly stable before declining with advancing canopy
closure at about age 31 to 35, when understorey plants disap-
peared. Th is is analogous to the decline in ruderal species noted
by Frouz et al., (2008) as trees came to dominate postmining
sites in the Czech Republic, about 30 yr into succession. Th ere
was some evidence that grasses form a higher proportion of
plant cover on our reclaimed sites than on the natural sites;
this has also been reported on highly disturbed reclaimed sites
where the plant species originally sown were still evident after
14 (Norman et al., 2006), 25 (Rayfi eld et al., 2005), and 45 yr
(Hodačová and Prach, 2003).
Some of the reclaimed sites appear to be on a path to be-
coming functioning ecosystems similar to those naturally pres-
ent in the boreal region; others do not. Th e best-performing
P-M treatments are E, H, and I. Th ere was a line of separation
at the 16 to 20 yr age range where reclaimed sites E, H, and
I shifted away from their reclaimed site status and toward the
natural range. Th ese treatments also showed high diversity with
8 to 10 diff erent groundcover types. Treatment E appeared to
become similar to natural ecotypes within 15 to 20 yr, and
Treatments H and I within 20 to 25 yr.
Th e worst-performing treatments, A and B (both covering
tailings sand), appear to remain in the early reclamation phase
for at least 25 yr. Th ese treatments appear to be on a trajectory
to a novel ecosystem that may or may not achieve the objectives
of restoring commercial forestry and wildlife habitat potential
within an appropriate timescale. It may be possible to divert
these treatments to a desired trajectory by adding more fertil-
izer in the early years as per treatments H and I.
Since there was only one replicate in each of treatments F, M,
FF, WA5, and Albian, we are less confi dent in drawing conclu-
sions as to their candidacy but it looks very likely that Albian,
an outlier separate from all other treatments or natural ecotypes,
will take decades to shift out of its reclamation state. Treatment
WA5 was clustering with the natural ecotypes, but it is not fea-
sible on a large scale since it is a deep peat treatment and the
amount of available peat is limited. As for treatments M (age 6)
and FF (age 4), which diff er in the type of peat used (mesic vs.
fi bric), there is little evidence yet that either of these young sites
is moving toward a natural condition. Th ey should continue to
be monitored to determine relative advantages of the diff erent
peat types, and fertilized to overcome low availability of K and
(on treatment M) Mn. Treatment F (age 14) may be of particular
interest when peat resources decline since it has no surface peat
amendment and, with the exceptions of K and Mn, it is broadly
similar to the range of variability in plant nutrient chemistry of
natural ecotypes. Shrubs could be planted on this treatment to
enhance its diversity and boost its restoration trajectory.
Th e approach used—comparing two ecological attributes
(plant community and ecological processes) on young reclaimed
sites with those on a range of older, natural forest ecotypes— ap-
pears to have been successful in allowing us to determine which
treatments are creating functional ecosystems similar to the target
ecotypes. Th e comparison would have benefi tted from inclusion
of natural sites that had been disturbed (burned, blown down,
or logged) within the last 20 yr. Studying natural ecosystem re-
covery would allow us to see whether ecosystem development on
reclaimed sites mirrors that on natural sites of the same age since
disturbance. Th is comparison was originally planned, but was
not possible with the available resources. Additional information
on microbial diversity, organic matter quality, and hydrological
function is being collected at the sites used in this study, which
will improve our ability to measure and predict the success of
reclamation treatments in the oil sands.
Nonmetric multidimensional scaling proved to be a useful
technique for analyzing data from this unbalanced experimental
design—it rapidly produces results that are visually informative
and can be quickly interpreted. Reclamation treatments were
still distinguishable after 20 to 25 yr and treatment diff erences
created consistent, observable diff erences in response variables.
Results of the cluster analysis confi rmed those of NMS.
It was interesting that some of the older reclaimed sites which
looked like forests on casual observation, and might have been
assumed to have been successfully restored, were in fact not on
a trajectory to becoming similar to one of the target ecotypes.
Table 3. Groups of plots as defi ned by cluster analysis, by ecotype and treatment by age class.
Plots 1 2 7 10 16 19 20 21 23 24 26 27 28
29 30 32 34 49 50 57 62 63
a1 b1 b3 d2 d3
4 8 25 43 61 WA5 Albian3 12 14 17 36 37 38 39 40 42
46 75 86 87 88 89 FF
Age class and treatment
AlbianA, A, A, I, FF
B, B, B, E, M
A, A, B, F, H
I, I, I
E, I, WA5
Rowland et al.: Recreating a Functioning Forest Soil in Reclaimed Oil Sands in Northern Alberta 1589
Instead, they appear to be developing into a new nontarget eco-
type. Does this mean that restoration of these sites has been un-
successful? It may be that expectation of restoring an ecosystem
precisely like that which was perturbed is unrealistic. Perhaps
when dealing with a profoundly-disturbed landscape on this
scale, recreating an ecological landscape (i.e., a mosaic of dif-
ferent communities comprising forest, forest edge, grassland) of
functioning ecosystems is a more realistic and suffi cient goal.
Some of the reclamation treatments applied to oil sands sites
appear to develop into ecosystems with similar plant community
composition and soil nutrient availability to those which occur
naturally in this area. Other reclamation treatments did not closely
resemble the natural forest ecotypes and may become novel ecosys-
tems. It appears that treatments that use: (i) a repeatedly-fertilized
peat-mineral mixture, or (ii) a peat-mineral mixture fertilized once
and with underlying subsoil over clean overburden, are develop-
ing into a functioning forest soil capable of supporting ecosystem
processes which mimic those of the natural boreal forest.
Implications for Practice
Sites reclaimed with a peat-mineral cap require fertilizer with
P, K, and trace elements including Mn to provide an early boost
to ecosystem development. Annual fertilizer treatments for the
fi rst 5 yr after planting may be suffi cient to boost a peat-mineral-
reclaimed site to a status similar to a natural forest by age 20
to 25, whether over tailings sands or not. Repeated addition of
fertilizer will assist plant growth and litter production, and hence
development of a surface FH layer. Lesser applications of fertil-
izer may be appropriate on nontailings sand sites capped with
subsoil, with or without a peat-mineral layer at the surface, and
with shrubs included in the planting mixture on those sites with-
out a peat-mineral surface layer. Th e present reserves of relatively
clean and unaltered peat and glacial till resources are unlikely to
provide suffi cient capping material for future reclamation on the
scale currently proposed, so alternative prescriptions requiring
smaller amounts and/or alternative inputs are needed. Th e prac-
tices of adding N in fertilizer should be revisited in light of the
high levels of soil N measured in the reclamation treatments.
Supplemental Information Available
Table S1 describes the materials used for reclamation on oil
sands sites. Table S2 describes the ecotypes used in this study.
Figure S1 shows the moisture content of soils at each of the
sites. Figure S2 shows the pH of soils at each of the sites. Figure
S3 shows the V/N ratios of soils at each of the sites. Figure S4
shows the relationship between soil C/N ratio and mass loss
of litter. Table S3 shows the bioavailability of nutrients in soils
at each of the sites. Th is material is available free of charge at
Th is study was funded by NSERC/CSNRG, with fi nancial,
administrative, technical, and fi eld support provided by Syncrude
Canada Ltd, Albian Sands Energy Inc. and Suncor Energy Inc.
Many corporate personnel assisted with mine orientation, site
access and safety, and on-site supervision. Special thanks are paid
to Clara Qualizza (Syncrude), Kristina Nordstrom (Albian), and
Wayne Tedder (Suncor) for advice and assistance throughout the
project, and Jennifer Lloyd (University of Alberta) for technical
support in the fi eld and laboratory.
Beckingham, J.D., and J.H. Archibald. 1996. Field guide to ecosites of
northern Alberta. Canadian Forest Serv., Edmonton.
Bois, G., Y. Piche, M.Y.P. Fung, and D.P. Khasa. 2005. Mycorrhizal inoculum
potentials of pure reclamation materials and revegetated tailing sands
from the Canadian oil sand industry. Mycorrhiza 15:149–158.
Castro, J., R. Zamora, J.A. Hódar, J.M. Gómez, and L. Gómez-Aparicio. 2004.
Benefi ts of using shrubs as nurse plants for reforestation in Mediterranean
mountains: A 4-year study. Restor. Ecol. 12:352–358.
Choi, S.-C., K.K. Kwon, J.H. Sohn, and S.-J. Kim. 2002. Evaluation of
fertilizer additions to stimulate oil biodegradation in sand seashore
mescocosms. J. Microbiol. Biotechnol. 12:431–436.
Danielson, R.M., S. Visser, and D. Parkinson. 1983. Plant growth in four
overburden types used in the reclamation of extracted oil sands. Can. J.
Soil Sci. 63:353–361.
Frouz, J., B. Keplin, V. Pizl, K. Tajovsky, J. Stary, A. Lukesova, A. Novakova,
V. Balik, L. Hanel, J. Materna, C. Duker, J. Chalupsky, J. Rusek, and
T. Heinkele. 2001. Soil biota and upper soil layer development in two
contrasting post-mining chronosequences. Ecol. Eng. 17:275–284.
Frouz, J., K. Parch, V. Pizl, L. Hanel, J. Stary, K. Tajovsky, J. Materna, V.
Balik, J. Kalcik, and K. Rehounkova. 2008. Interactions between soil
development, vegetation and soil fauna during spontaneous succession
in post mining sites. Eur. J. Soil Biol. 44:109–121.
Fung, M.Y.P., and T.M. Macyk. 2000. Reclamation of oil sands mining areas. p.
755–774. In R.I. Barmhisel et al. (ed.) Reclamation of drastically disturbed
lands. Agron. Monogr. 41. ASA, CSSA, and SSSA, Madison, WI.
Government of Alberta. 1999. Conservation and reclamation information
letter guidelines for reclamation to forest vegetation in the Athabasca oil
sands region. C and R/IL/99–1.Gov. of Alberta, Edmonton, AB.
Hardy, BBT., Ltd. 1990. Natural plant invasion into reclaimed oil sands
mine sites. Rep. RRTAC 90–3. Alberta Land Conserv. and Reclamation
Council, Edmonton, AB.
Hemstock, S. 2008. Plant productivity, soil microorganisms, and soil nitrogen
cycling in peat amendments used for oil sands reclamation. M.Sc. thesis.
Univ. of Alberta.
Hodačová, D., and K. Prach. 2003. Spoil heaps from brown coal mining: Technical
reclamation versus spontaneous revegetation. Restor. Ecol. 11:385–391.
Lanoue, A. 2003. Phosphorus content and accumulation of carbon and
nitrogen in boreal forest soils. M.Sc. thesis, University of Alberta.
Laughlin, D.C., and S.R. Abella. 2007. Abiotic and biotic factors explain
independent gradients of plant community composition in ponderosa
pine forests. Ecol. Modell. 205:231–240.
Marshall, I. 1982. Mining, land use and the environment (I); a Canadian
overview. Lands Directorate, Environ. Canada, Ottawa.
McCune, B., and J.B. Grace. 2002. Analysis of Ecological Communities. MJM
Software, Gleneden Beach, OR.
McMillan, R., S.A. Quideau, M. MacKenzie, and O. Biryukova. 2007.
Nitrogen mineralization and microbial activity in oil sands reclaimed
boreal forest soils. J. Environ. Qual. 36:1470–1478.
Mossop, G.D. 1980. Geology of the Athabasca oil sands. Science (Washington,
Mummey, D.L., P.D. Stahl, and J.S. Buyer. 2002. Soil microbiological
properties 20 years after surface mine reclamation: Spatial analysis of
reclaimed and undisturbed sites. Soil Biol. Biochem. 34:1717–1725.
Natural Regions Committee. 2006. Natural regions and subregions of Alberta.
Compiled by D.J. Downing and W.W. Pettapiece. Publ. T/852. Gov.of
Alberta, Edmonton, AB.
Norman, M.A., J.M. Koch, C.D. Grant, T.K. Morald, and S.C. Ward. 2006.
Vegetation succession after bauxite mining in western Australia. Restor.
Oil Sands Vegetation Reclamation Committee. 1998. Guidelines for
1590 Journal of Environmental Quality • Volume 38 • July–August 2009 Download full-text
reclamation to forest vegetation in the Athabasca oil sands region.
Provincial Gov. of Alberta, Edmonton, AB.
Purdy, B.G., S.E. Macdonald, and V.J. Lieff ers. 2005. Naturally saline boreal
communities as models for reclamation of saline oil sand tailings. Restor.
Qian, P., and J.J. Schoenau. 2002. Practical applications of ion exchange resins in
agriculture and environmental soil research. Can. J. Soil Sci. 82:9–21.
Rayfi eld, B., M. Anand, and S. Laurence. 2005. Assessing simple versus
complex restoration strategies for industrially disturbed forests. Restor.
Rowe, J.S. 1972. Forest regions of Canada. Publ. 1300.Dep. of the Environ./
Canadian Forest Serv., Information Canada, Ottawa.
Ruiz-Jaen, M.C., and T.M. Aide. 2005. Restoration success: How is it being
measured? Restor. Ecol. 13:569–577.
Stolte, W.J., S.L. Barbour, and C.D. Boese. 2000. Reclamation of saline-sodic waste
dumps associated with the oilsands industry. p. 385–395. In A. Etmanski
(organizer) Global Land Reclamation/Remediation 2000 and Beyond. Proc.
of the Canadian Land Reclamation Assoc. 25th Annual Meet., Edmonton,
AB, Canada. 17–20 Sept. 2000. Canadian Land Reclamation Assoc., Int.
Affi liation of Land Reclamationists, Edmonton, AB.
Turchenek, L.W., and J.D. Lindsay. 1982. Soils inventory of the Alberta oil
sands environmental research program study area. Rep. 122. Alberta Res.
Council, Edmonton, AB.