Disentangling direct and indirect effects of water table drawdown on above‐ and belowground plant litter decomposition: consequences for accumulation of organic matter in boreal peatlands
ABSTRACT Pristine peatlands are carbon (C)-accumulating wetland ecosystems sustained by a high water table (WT) and consequent anoxia that slows down decomposition. Persistent WT drawdown as a response to climate and/or land-use change affects decomposition either directly through environmental factors such as increased oxygenation, or indirectly through changes in plant community composition. This study attempts to disentangle the direct and indirect effects of WT drawdown by measuring the relative importance of environmental parameters (WT depth, temperature, soil chemistry) and litter type and/or litter chemical quality on the 2-year decomposition rates of above- and belowground litter (altogether 39 litter types). Consequences for organic matter accumulation were estimated based on the annual litter production. The study sites were chosen to form a three-stage chronosequence from pristine (undrained) to short-term (years) and long-term (decades) WT drawdown conditions at three nutrient regimes. The direct effects of WT drawdown were overruled by the indirect effects through changes in litter type composition and production. Short-term responses to WT drawdown were small. In long-term, dramatically increased litter inputs resulted in large accumulation of organic matter in spite of increased decomposition rates. Furthermore, the quality of the accumulated matter greatly changed from that accumulated in pristine conditions. Our results show that the shift in vegetation composition as a response to climate and/or land-use change is the main factor affecting peatland ecosystem C cycle, and thus dynamic vegetation is a necessity in any model applied for estimating responses of C fluxes to changing environment. We provide possible grouping of litter types into plant functional types that the models could utilize. Furthermore, our results clearly show a drop in soil summer temperature as a response to WT drawdown when an initially open peatland converts into a forest ecosystem, which has not yet been considered in the existing models.
- SourceAvailable from: ncbi.nlm.nih.gov[show abstract] [hide abstract]
ABSTRACT: The garlic stalk is a byproduct of garlic production and normally abandoned or burned, both of which cause environmental pollution. It is therefore appropriate to determine the conditions of efficient decomposition, and equally appropriate to determine the impact of this decomposition on soil properties. In this study, the soil properties, enzyme activities and nutrient dynamics associated with the decomposition of garlic stalk at different temperatures, concentrations and durations were investigated. Stalk decomposition significantly increased the values of soil pH and electrical conductivity. In addition, total nitrogen and organic carbon concentration were significantly increased by decomposing stalks at 40°C, with a 5∶100 ratio and for 10 or 60 days. The highest activities of sucrase, urease and alkaline phosphatase in soil were detected when stalk decomposition was performed at the lowest temperature (10°C), highest concentration (5∶100), and shortest duration (10 or 20 days). The evidence presented here suggests that garlic stalk decomposition improves the quality of soil by altering the value of soil pH and electrical conductivity and by changing nutrient dynamics and soil enzyme activity, compared to the soil decomposition without garlic stalks.PLoS ONE 01/2012; 7(11):e50868. · 3.73 Impact Factor
- Journal of Geophysical Research 11/2012; 118:1-13. · 3.17 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: a b s t r a c t Peatland restoration has been implemented on sites exploited for horticultural peat for over a decade in Eastern Canada. However, little is known about nutrient dynamics and microbial processes in this region. Belowground nitrogen (N) and phosphorus (P) transformations and carbon utilisation by microorgan-isms were examined in a harvested peatland 10 years after restoration measures were implemented to assess whether restoration is returning the peatland to a state that falls within the natural range of variation found in a neighbouring bog. N mineralisation rates were almost 10-fold higher in the surface (0e10 cm) compared to the subsurface (10e20 cm) layers for all sites and were highly variable within sites. P pools were small (<0.02 mg g À1 dry peat) and mineralisation rates of P were low in all sections. In the surface layer, the net mineralisation and ammonification rates appeared to be highest in unrestored sites but lowest in restored sites. In contrast, the capacity of microorganisms in using different carbon (C) sources, also described as microbial functional diversity, was highest at restored sites but lowest at unrestored sites. The preferable C sources varied between sites and were significantly correlated with aboveground vegetation composition. Our study suggests that microbial activity and nutrient trans-formations differ between natural and unrestored harvested peatlands. Our results indicate that the presence of vegetation regrowth in the unrestored area of a peatland alters belowground processes by stimulating microbial activity and increasing the uptake of nutrients, leading to smaller pools of inor-ganic N available in the peat. When restoration has been carried out, microbial activity is even higher than in natural conditions, possibly leading to high immobilization of N, and net mineralisation rates are very low. This research indicates that while belowground processes have shifted from unrestored con-ditions following restoration, they do not appear to be fully re-established to a degree similar to natural conditions 10 years post-restoration.Soil Biology and Biochemistry 05/2013; 64:37-47. · 3.65 Impact Factor
Disentangling direct and indirect effects of water table drawdown on
above- and belowground plant litter decomposition:
Consequences for accumulation of organic matter in boreal peatlands.
Petra Straková1,2, Timo Penttilä2, Jukka Laine3 and Raija Laiho1
1Peatland Ecology Group, Department of Forest Sciences, University of Helsinki, Finland
2Finnish Forest Research Institute, Vantaa Research Unit, Finland
3Finnish Forest Research Institute, Parkano Research Unit, Finland
Climate change, decomposition, litter quality, peatlands, plant litter, organic matter
accumulation, temperature, water table drawdown.
Pristine peatlands are carbon (C) accumulating wetland ecosystems sustained by a high water
table (WT) and consequent anoxia that slows down decomposition. Persistent WT drawdown
as a response to climate and/or land-use change affects decomposition either directly through
environmental factors such as increased oxygenation, or indirectly through changes in plant
community composition. This study attempts to disentangle the direct and indirect effects of
WT drawdown by measuring the relative importance of environmental parameters (WT
depth, temperature, soil chemistry) and litter type and/or litter chemical quality on the 2-year
decomposition rates of above- and belowground litter (altogether 39 litter types).
Consequences for organic matter accumulation were estimated based on the annual litter
production. The study sites were chosen to form a three stage chronosequence from pristine
(undrained) to short-term (years) and long-term (decades) WT drawdown conditions at three
The direct effects of WT drawdown were overruled by the indirect effects via changes in
litter type composition and production. Short-term responses to WT drawdown were small. In
long-term, dramatically increased litter inputs resulted in large accumulation of organic
matter in spite of increased decomposition rates. Further, the quality of the accumulated
matter greatly changed from that accumulated in pristine conditions.
Our results show that the shift in vegetation composition as a response to climate and/or
land-use change is the main factor affecting peatland ecosystem C cycle and thus dynamic
vegetation is a necessity in any models applied for estimating responses of C fluxes to
changing environment. We provide possible grouping of litter types into plant functional
types that the models could utilize. Further, our results clearly show a drop in soil summer
temperature as a response to WT drawdown when an initially open peatland converts into a
forest ecosystem, which has not yet been considered in the existing models.
Published in Global Change Biology, vol. 18, pages 322–335, 2012.
Decomposition is one of the key processes in element cycling in most ecosystems. In
peatlands, there has been a long-term imbalance between litter production and decomposition
caused by high water tables (WT) and consequent anoxia. This has resulted in peatlands
being a significant sink of carbon (C) from the atmosphere (e.g., Gorham 1991; Schulze &
Lowering of the WT, because of climatic or land-use changes, promotes several changes
in peatland environmental conditions that have direct effects on decomposition (Laiho et al.
2006). Increased soil aeration has a positive effect, while increased peat compaction,
lowering of soil pH and drop in temperature may, in turn, have a negative effect. Yet, there is
increasing evidence that the indirect effects on decomposition via changes in the structure of
plant communities may have much more impact on ecosystem C cycling than any direct
effects of environmental changes (Straková et al. 2010).
Lowering of WT induces changes in the plant community structure (Weltzin et al. 2000;
2003; Robroek et al. 2007; Breeuwer et al. 2009), that can eventually lead to a complete
replacement with species adapted to the new conditions (Laine et al. 1995). Such changes
tend to be more pronounced in sites with more nutrients, and intensify over time since the
WT drawdown (Laine et al. 1995). Following that, quantity and quality of the above- and
belowground litter produced after the WT drawdown, as well as the location (depth
distribution) of the belowground litter, greatly differ from that produced in pristine conditions
(Laiho et al. 2003; Straková et al. 2010, Murphy et al. 2009 ). Litter quality (relative
proportions of soluble and recalcitrant compounds and nutrients) is a key factor in C cycle,
for it determines the quality of the substrate as a source of energy and nutrients for
decomposer microorganisms. Thus, different litter materials may have vastly differing rates
of decomposition (e.g., Hobbie 1996; Thormann et al. 2001; Cornwell et al. 2008). Changes
in vegetation composition following a persistent WT drawdown may therefore result in
overall shifts in litter quality and decomposability in peatlands (Dorrepaal et al. 2005;
Straková et al. 2010). In addition to the “fate” of the existing peat deposit, the C balance of
fresh litter inputs affects the C sink/source function of peatlands following persistent water
table drawdown (Laiho 2006). The extent of the role of the changed litter inputs and their
decomposition rates on the C balance of different peatland types have not been explored yet,
in spite of their significance.
The aim of this study was to evaluate the effect of WT drawdown (at different time scales:
years and decades after the persistent WT drawdown) on plant litter decomposition. We
determined decomposition rates of the litter types typical of boreal peatland sites with
varying nutrient and WT regimes, and estimated the relative importance of environmental
parameters and litter type and/or litter chemical quality on the decomposition dynamics.
Further, we described the short-term accumulation rate of organic matter at the different WT
regimes. To estimate C sink/source behaviour of drained peatlands, we compared our results
with literature data on heterotrophic soil respiration.
We postulated that WT drawdown has dual effects on plant litter decomposition:
(1) direct, made by improved environmental conditions for aerobic decomposers, that will be
reflected as an increase in litter decomposition rate at plant species level, and
(2) indirect, through changes in plant community structure, that will be reflected in the
decomposition rate at the community level and/or in the short-term accumulation rate of
We presumed that the direct effects will dominate in the short-term (no big changes in
vegetation yet), while the indirect effects will dominate in the long-term (dramatic changes in
vegetation composition, additional effects of the vegetation changes on soil properties).
Further, we expected the effects be more pronounced in nutrient-rich (fen) than nutrient-poor
(bog) sites, in line with changes in vegetation composition (e.g., Laine et al. 1995), tree stand
development (Minkkinen et al. 2001), litter quality and inputs (Straková et al. 2010),
microbial community composition and activity (Peltoniemi et al. 2010; Straková et al. 2011).
We assume N and P be more readily available at the minerotrophic fen sites (nutrient-rich)
compared to the bog (nutrient-poor), enhancing microbial activities and thus instigating
greater mass loss (McClaugherty et al. 1985; Taylor et al. 1989; Váv?ová et al. 2009).
This article focuses on the first two years of the decomposition process, what is, for most
litter types, a decadal movement from litter to soil organic matter. In a peatland ecosystem
the first two years are very much still the initial stage of litter decomposition, and may not
reflect the long-term C accumulation accurately. We will validate the further behaviour of the
accumulated organic matter at the different WT regimes after obtaining longer-term data
from the continuation of this study.
2. Materials and methods
2.1. Study sites
The research was carried out at Lakkasuo, a raised bog complex in Central Finland (61°48'N,
24°19'E, c. 150 m.a.s.l.). Annual precipitation in this area is 710 mm, of which about one-
third falls as snow. The average annual temperature sum (threshold value 5°C) is 1160 degree
days and average temperatures for January and July are ?8.9 and 15.3°C, respectively
(Finnish Meteorological Institute, Juupajoki weather station 1961-1990).
We had three study sites with differing nutrient regimes: bog (ombrotrophic, i.e. fed solely
by precipitation; nutrient-poor), oligotrophic fen (minerotrophic, i.e. additionally fed by
groundwater inputs; moderate nutrient regime) and mesotrophic fen (minerotrophic; nutrient-
rich). Each of those consisted of a pristine control plot, a plot with short-term, c. 4 years,
water table drawdown (STD), and a plot with long-term, c. 40 years, water table drawdown
(LTD) (Laine et al. 2004). Together, these plots formed a gradient from a wet pristine
peatland to a drying system and finally to a peatland forest ecosystem (Laiho et al. 2003.
Within each site, all plots supported the same plant community and had similar soil
composition and structure before the WT drawdown. The pristine and LTD plots covered
about 900 m2, and the STD plots about 500 m2.
The water tables in the manipulated plots were lowered by ditching. LTD had been
achieved with practical-scale drainage for forestry in 1961 (i.e., to improve tree growth), and
STD with new ditches for this experimental purpose in 2001. Short-term WT drawdown had
led to the average WT being 10 cm (bog) to 20 cm (fen) deeper than in the corresponding
pristine plots, which is close to the estimate given by Roulet et al. (1992) for the short-term
impact of climate change on WT in northern peatlands. In the LTD plots, the average WT
was 15 (bog) to 40 (fen) cm deeper than in the pristine plots. We assume that the initial post-
drainage drop in WT was close to that observed in our STD plots, and that the further
lowering was caused by increased evapotranspiration by the growing tree stands (Sarkkola et
al. 2010). The difference between fen and bog also largely derives from the higher tree stand
evapotranspiration in fens where the tree stands develop faster (Minkkinen et al. 2001). As
the WT depth is clearly different (usually lower) next to a drainage ditch than is the average
of the drained area (Grieve et al. 1995; Schlotzhauer & Price 1999), no measurements were
made next to the ditches (minimum distance > 1 m for STD and 10 m for LTD).
Short-term, four-year WT drawdown had a rather small effect on vegetation composition:
Sphagnum moss and sedges had suffered while shrubs had flourished together with pine
(Pinus sylvestris) and birch (Betula pubescens) (Straková et al. 2010; Jukka Laine, Eeva-
Stiina Tuittila and Harri Vasander, unpublished data). In long-term, 40 years, the changes in
vegetation composition were dramatic: WT drawdown had resulted in conversion of an open
peatland dominated by Sphagnum and graminoids into a forest ecosystem dominated by pine
and birch. The vegetation change was associated, besides a decrease in WT, with a drop in
pH and increase in nutrient concentration of surface peat (Straková et al. 2010).
2.2. The litter material
We collected altogether 39 litter types (plant species and part/organ) that included foliar litter,
roots and woody parts of vascular plants, and mosses (Appendix 1), and reflected the
dominant species at the different nutrient and WT regimes, as well as different plant groups
with distinctive chemical composition (Straková et al. 2010). Litter of Betula nana,
Eriophorum vaginatum and P. sylvestris (altogether 6 types) was present at all plots
(“common litter”), and could be used to evaluate the direct effect of WT drawdown on litter
decomposition, including a possible change in litter quality at the litter type level. The other
litter types included were typical of certain nutrient and WT regimes (“specific litter”)
(Appendix 1; Anttila 2008), and thus reflected the indirect effects. Vascular plant litter was
collected by harvesting senescent leaves, needles or dead branches from living plants. As
young green stem parts of Sphagnum moss were shown to decay at faster rate than litter
(recently dead stem parts, Limpens & Berendse 2003), we collected moss litter by cutting a
3-5 cm thick layer beyond the living moss with scissors (thus, excluding both the upper green
and the lower, already decomposing, layers). The collected litter was further sorted and any
green or visually decomposing parts were removed. Harvesting took place in September and
October 2004 or 2005 during the highest natural litter fall at our sites (Anttila 2008).
For the belowground litter of Carex lasiocarpa, C. rostrata, E. vaginatum and B. nana,
whole living plants were collected at the study sites and cultivated in containers filled with
expanded clay and water from May to October 2005. To simulate the natural variation in
nutrient regimes between the different peatland sites, fertilizer was added to the containers
monthly at two concentration levels. Higher concentration was used for plants from the
nutrient-rich mesotrophic fen, lower concentration for the plants from the nutrient-poor bog
and oligotrophic fen. Roots that were used to represent belowground litter were harvested
from the plants at the end of the cultivation period. Pine roots were harvested from 3 years
old Scots pines, cultivated at a tree-nursery, and divided into two size classes: 0-2 mm and 2-
Each litter type was air-dried at the room temperature (20 °C) to constant mass (about 92-
94% dry mass) and gently mixed. Sub-samples were withdrawn to determine initial litter
quality and dry mass content (Straková et al. 2010). Detailed chemical characterization of the
different litter types is presented by Straková et al. (2010). In short, non-graminoid foliar
litters had a high concentration of nutrients and extractives. Graminoids and mosses were rich
in holocellulose-comprising sugars and lignin-like compounds (Klason lignin, CuO oxidation
phenolic products). Sphagnum species of sections Acutifolia and Palustria (mostly hummock
species) displayed a higher concentration of cellulose and lignin-like compounds and lower
concentration of hemicellulose than the species of section Cuspidata (mostly lawn-level and
hollow species). Woody litters were marked by a high concentration of Klason lignin.
2.3. Litter decomposition
Litter decomposition rates were determined in the natural environment. This means that each
specific litter type was decomposing at the plot where it had been produced and collected
(except for pine roots that originated from a tree-nursery), in conditions where that kind of
litter would fall and naturally be decomposing. We used the litterbag method, which, in spite
of some known sources of inaccuracy (Domisch et al. 2000; Taylor 1998, Kurz-Besson et al.
2005) is the most useful and widely used method for determining mass loss rates of different
materials in situ. To minimize the negative effect of air-drying on litter decomposition
(Taylor 1998), litterbags were remoistened with surface water of the given site before
installation. We assume that this helped the microbial communities typical of the site to re-
colonize the litter. Also, the mesh size of the nylon bags used was 1 x 1 mm to prevent
physical losses of the material but to allow small mesofauna typical of the sites (Silvan et al.
2000) to enter the bags. There are generally not bigger decomposers at our sites and thus the
mesh size 1 x 1 mm did not prevent their effects on litter fragmentation, which was the
concern of Cotrufo et al. (2010). For each plot, mostly 4-5 replicates were prepared per each
litter type for annual recovery; the number of replicates was lower for some litter types or
plots due to limitations in litter amounts (Appendix 1).
For the litterbag incubation, locations were selected following the given plant species
abundance, i.e. each litterbag was installed in the conditions where the given litter type would
naturally decompose. Litterbags containing moss litter were installed under the living parts of
moss shoots of the given species, where fresh moss litter is naturally formed and begins to
decompose. The other aboveground litters were placed horizontally on the litter layer surface
where the litters naturally fall, always in touch with some fallen litter of the given litter type.
As the decomposition process may be affected by interactions (either positive or negative)
between different litter types when decomposing as mixtures (Gartner & Cardon 2004),
contact with other typically associated litters was also ensured. Belowground litterbags were
installed vertically in the peat profile at the given depth (0-10 to 20-30 cm below the soil
surface, see Appendix 1) near living plants of the species in question. Installation took place
in October-November 2004. To capture for possible between-year variation in litter
decomposition rates, another set of litterbags was installed in October 2005 for a 1-year
period. Some litterbags with belowground litter were installed in October 2005 and 2006 (see
Appendix 1). Incubation periods presented in here are 1 and 2 years: the first recovery cohort
was collected 12 months after installation and the second one 24 months. The samples are a
subset from an ongoing long-term study.
After each recovery, litterbags were transported to a laboratory where the content was
cleaned by removing all additional (ingrowth) materials, weighted to determine the remaining
mass and gently homogenized before sub-sampling. Dry mass content was determined by
drying two sub-samples at 105 °C overnight. Decomposition rates were expressed as dry
mass loss after each incubation period (Appendix 1).
2.4. Environmental parameters
This section describes measurements of environmental parameters that were tested in this
study as potential predictors of variation in litter decomposition rates.
WT depth was continuously recorded at all pristine and LTD plots using Ott (Kempten,
Germany) WT recorders. Position of the decomposing litter relative to the WT was estimated
based on those continuous measurements, and monthly measurements at the exact locations
where litterbags were installed.
Temperature was monitored in 2-4 hour intervals using temperature loggers (i-Button
DS1921G, MaximIntegrated Products) at the same locations where the studied litter was
decomposing. The loggers were thus installed in soil at 10 and 20 cm depth for the
belowground litters, in moss patches about 5 cm from the surface for the moss litters, and in
the surface litter layer for the other litter types in which case they were protected from direct
sunlight by a thin layer of litter. Daily mean temperatures and the cumulative temperature
sums over 0 ºC threshold for the specific incubation periods were calculated from these local
measurements. If local data were not available for some dates or locations, mean daily soil
temperature values were estimated from the values of adjacent measurements or the closest
locations of the same plot and litter type. Daily variation in temperature was calculated as
standard deviation of the temperature records within a day.
Soil samples were collected at all plots using a box sampler from the 0-30 cm surface peat
layer. The peat-cores were cut into 10-cm layers, and element concentrations were analyzed
as for the litter materials (Straková et al. 2010).
To capture purely environmental effects on decomposition we used pine cellulose as a
standard material. Unlike common litter the quality of which may change as a response to
WT drawdown, cellulose had identical chemical composition at all nutrient and WT regimes.
Mesh bags with cellulose strips were installed at the same locations as the litterbags and the
mass loss rates were measured for the same periods as that for litters. After each recovery, the
cellulose strips were gently cleaned, dried at 105 °C overnight and weighted to determine the
dry mass loss.
2.5. Data analyses
2.5.1. Direct effects of WT drawdown and site nutrient regime on litter decomposition
The direct effects (i.e., induced mainly by the environmental changes and reflected at plant
species level) of WT drawdown and site nutrient regime on litter decomposition were
estimated using 1) decomposition rates of the common litter (litter types common to all WT
and nutrient regimes), 2) decomposition rates of cellulose as a standard material, and 3) the
relative effect of environmental factors in the models of litter decomposition (described in
Factorial ANOVA followed by Tukey´s post-hoc comparison was carried out using
Statistica for Windows version 6.1 (StatSoft, 2003). Mass loss of the common litter was used
as a single continuous response variable, and site nutrient and WT regime, litter type,
incubation layer and incubation period (1 or 2 years) were used as categorical predictors
(factors). Separate tests were performed for aboveground and belowground litter.
Correspondingly, to estimate the direct effects of site nutrient and WT regime and their
interactions on cellulose decomposition, mass loss of cellulose was used as a single
continuous variable, and site nutrient and WT regime, incubation layer and incubation period
(1 or 2 years) as factors.
2.5.2. Indirect effects of WT drawdown on litter decomposition
To estimate the indirect effects of WT drawdown we calculated litter decomposition rates at
the community level that are mediated by changes in plant community structure.
Decomposition rates of different litter types were weighted by their inputs presented by
Straková et al. 2010. Further, we calculated the relative effect of litter type (PFT) in the
models of litter decomposition (described in detail later).
2.5.3. Models of litter decomposition
Models were constructed to identify factors controlling the variation in the litter mass loss.
Because of the hierarchical data structure, a mixed (multilevel) model approach was used
(Goldstein 1995). We identified three hierarchical levels, or levels of clustering, in the data:
(1) site, (2) incubation location, (3) recovery cohort (year 1 and year 2; this made the model
follow a repeated measures design). The models thus had the following form:
yijk??ijk??1x1ijk??2x2ijk+ ??nxnijk+ vk+ujk+?ijk (1)
where yijk is the cumulative mass loss for the incubation period i within incubation location j
in site k. The fixed part consists of intercept ?, and site, weather and litter quality
characteristics x1ijk–xnijk with parameters ?1??n. In the random part, vk is the variance derived
from site k, ujk is the variance associated with different incubation locations j, and ?ijk
accounts for the within-incubation location variation between different incubation periods (1
or 2 years; recovery cohort).
The estimation was done using MLwiN software (Rasbash et al. 2004), which estimates
the fixed and random parameters simultaneously. We applied the restricted iterative
generalized least square (RIGLS) method. The significance of the variables was evaluated
based on their parameter standard error (parameter value should be at least twice its SE).
The parameter SE was also used to group different litter types into plant functional types
(PFTs, Box 1996). Litter types whose mass loss rates did not significantly differ (based on
SE) were grouped into a PFT so that the different PFTs significantly differed from each other.
The value of ?2×log-likelihood was used to compare models of increasing complexity.
The factors included in the final models were selected based on the amount of explained total
variation in litter mass loss and the value of ?2×log-likelihood. The goodness of fit was
further evaluated based on residuals. Different models were constructed for the belowground
litters and for the aboveground litters that included mosses.
Simple decomposition rate coefficients were estimated for the different litter types using
the exponential decay function (Olson 1963) as described in Váv?ová et al. (2009). The
goodness of fit was evaluated based on residuals. The fit of our litter materials was generally
poor and thus the percentage of dry mass loss after each incubation period will be used to
express the decomposition dynamics.
2.5.4. Temperature patterns in the decomposing litter
To estimate the effects of site nutrient and WT regime, their interactions, litter type and
incubation layer on variation in temperature patterns in the decomposing litter, variation
partitioning was performed by redundancy analysis (RDA) using Canoco for Windows
version 4.5 (ter Braak & Šmilauer 2002). Measured values of daily mean temperature or daily
variation in temperature in the decomposing litter were used as response variables (a separate
variable for each day). A group of binary variables describing either site nutrient or WT
regime, their interaction term, litter type or incubation layer was kept at a time as explanatory
variables, while the other groups were used as covariables (see Table 1).
2.6. Simulation of organic matter accumulation
To calculate short-term accumulation of organic matter (amounts of different litter types
remaining at the sites after 2 years of decomposition), the measured litter mass loss rates were
applied to the litter inputs presented by Straková et al. (2010).
3.1. Variation in environmental conditions for decomposition
3.1.1. Water table
In pristine conditions, fens had higher WT with smaller variation during the vegetation
season compared to the bog (Fig. 1). Ditching had greater impact on WT in the initially
wetter fens compared to the bog. At fens, the ditching had resulted in average WT being 20
(years) to 40 (decades) cm deeper than in the pristine plots. At the bog, the WT was only 10
(years) to 15 (decades) cm deeper following the ditching compared to the pristine plot (Fig.
3.1.2. Litter temperature
WT drawdown influenced temperature patterns in the decomposing litter. Daily mean
temperature, cumulative temperature sums as well as variation in temperature within a day
(estimated for mid May - mid October) decreased following the long-term WT drawdown
(Table 2; see also Fig. 2). The decrease in daily mean temperature following the long-term
WT drawdown was greatest in the spring and summer months (April - July): by 1-2 ºC for the
surface litter layer and by 3-4 ºC for the soil layers. Only in the autumn months (September-
October) the daily mean temperature of the decomposing litter at the pristine plots was
somewhat lower compared to that at the long-term drained plots, the difference was generally
less than 1 ºC.
There was a strong effect of litter type and/or incubation layer on temperature patterns in
the litter (Table 1). Daily mean temperature during summer and daily variation in temperature
generally decreased in the direction: surface > moss > soil 10 cm > soil 20 cm (Table 2; see
also Fig. 2). Opposite pattern was found for daily mean temperature during winter (not
3.1.3. Cellulose decomposition
Decomposition of cellulose (standard material) was used to capture for purely environmental
effects on the decomposition process. The best environmental conditions for decomposition
were generally in moss patches and soil 0-10 cm layer; worst on the surface litter layer of the
pristine plots, in hollows of the pristine bog, and in the deepest (20-30 cm) soil layer
WT drawdown had a positive effect on cellulose decomposition (p < 0.001), and the effect
was more pronounced in the long-term (decades) WT drawdown conditions. The effect of
WT regime on cellulose decomposition was greatest in the surface and 0-10 cm soil layer,
and decreased with increasing soil depth. In the surface layer, site nutrient regime had a
positive effect on decomposition (p < 0.01) that was highest at the mesotrophic fen and
lowest at the bog.
3.2. Litter decomposition
Decomposition rates varied considerably between the different litter types (Appendix 1-3).
For the aboveground litters, mass loss rates decreased in the direction: 1) foliar litter (broad-
leaved arboreal plants > minerotrophic graminoids > needle-leaved arboreal plants >
ombrotrophic graminoids), 2) moss (feather moss > lawn species of Sphagnum > hummock
species of Sphagnum), 3) woody litters. For the belowground litters, mass loss rates
decreased in the direction: 1) minerotrophic graminoids, 2) fine roots (< 2 mm) of trees and
shrubs, 3) ombrotrophic graminoids, 4) thicker roots (2-10 mm) of trees.
There was no between-year variation in 1-year decomposition rates.
3.2.2. Direct effects of WT drawdown
WT drawdown had a direct positive effect on litter decomposition (p < 0.001): the mass loss
of common litter increased following the short-term WT drawdown, and increased further
following the long-term WT drawdown. The effect of WT drawdown on belowground litter
decomposition decreased with increasing soil depth (p < 0.001).
The effect of WT drawdown on aboveground litter had different patterns at sites with
different nutrient regimes (p < 0.01). At the nutrient-rich mesotrophic fen the effect appeared
already after the short-term WT drawdown, while at the nutrient-poor bog and the
oligotrophic fen it appeared only after the long-term WT drawdown. Such differences were
not observed for the belowground litter.
Of the site environmental parameters, soil N concentration and WT drawdown (STD,
LTD) for aboveground litters (positive correlations) and installation depth for belowground
litters (negative correlation; effect decreases with increasing soil depth) proved to be the best
predictors of litter decomposition rates (Table 3). However, these parameters accounted for
only less than 2% of the total variation in litter decomposition rates.
3.2.3. Indirect effects of WT drawdown
Litter decomposition rates at the community level (decomposition rates of different litter
types weighted by their input) increased following WT drawdown (Fig. 3). At the fen sites
the increase appeared already after the short-term WT drawdown, being most dramatic at the
nutrient-rich mesotrophic fen, there the rates were even higher than those in the long-term
drained conditions. At the nutrient-poor bog site the increase appeared only after the long-
term WT drawdown.
Litter type accounted for about 65% of the total variation in litter decomposition rate,
which was far more than what was the effect of site environment (less than 2%). Litter type
as such (39 types) captured for all the variation in initial litter quality in relation to variation
in litter decomposition rates. For aboveground litters, our grouping of litter types into PFTs
based on their decomposition rates (Table 4, 5 types; see also Fig. 4) required further
inclusion of the concentration of extractable compounds and holocellulose to lignin ratio in
the litter to the model (Table 3). These litter quality parameters showed positive correlation
with the decomposition rates. When litter types or PFTs were not included in the model, the
litter quality parameters that were most related to mass loss rates included concentration of
total extractives and N (positive correlation with mass loss), Klason lignin and p-
hydoxyphenols (lignin-like compounds, negative correlation with mass loss). These
parameters accounted for about 40% of the total variation in aboveground litter mass loss (not
3.2.4. Effects of site nutrient regime
Site nutrient regime had an effect on litter decomposition at the surface litter layer and at the
0-10 cm soil depth (p < 0.001). Within the fens, the decomposition increased with increasing
nutrient availability, being higher at the nutrient-rich mesotrophic fen. However, at the
nutrient-poor bog the decomposition was as high as at the mesotrophic fen (surface litter
layer) or even higher (0-10 cm soil depth).
In pristine conditions and after the short-term WT drawdown, decomposition rates at the
community level were higher at the mesotrophic fen compared to the nutrient-poorer bog and
the oligotrophic fen. After the long-term WT drawdown, the rates were very similar at all
3.3. Accumulation of organic matter
Following changes in litter inputs (Fig. 3 in Straková et al. 2010), the amount of accumulated
organic matter increased dramatically after the long-term WT drawdown and its composition
greatly changed (Fig. 5). The clearest effect was an increase in remains of tree litters (leaves,
needles, branches, cones) following the long-term WT drawdown. The accumulated organic
matter consisted mainly of Sphagnum and graminoids at the pristine plots and after the short-
term WT drawdown. Those materials were reduced after long-term WT drawdown and other
moss species, mostly P. schreberi increased in amounts.
There are three main factors that influence decomposition dynamics in situ: (1) quality of
litter as the substrate for decomposing organisms, (2) the type and abundance of the
decomposers, and (3) the environmental conditions (e.g., temperature, moisture, oxygen and
nutrient availability, pH) under which the decomposers live and assimilate the litter (Belyea
1996; Laiho 2006). In this study we focused on the effects of changing peatland hydrology on
litter decomposition at plant species (i.e. affected mainly by the environmental changes;
direct effects) and community levels (i.e. additionally affected by the successional changes in
vegetation community; indirect effects) at sites with different nutrient regimes. As peatlands
contain a major proportion of the terrestrial C pool, predictions of their C cycle under a
changing climate and/or land-use are of great importance.
4.1. Direct effects of WT drawdown overruled by the indirect effects
As hypothesized, WT drawdown had direct positive effects on litter decomposition rates, but
in the long-term (decades) they were overruled by the indirect effects via changes in plant
community composition and production. This resulted in large accumulation of organic
matter at the long-term drained plots, in spite of increased decomposition rates of the litter.
Our results show that litter type or PFT as such may predict, to a large extent, the variation
in litter decomposability. This finding is supported by earlier studies (e.g., Hobbie 1996;
Thormann et al. 2001; Dorrepaal et al. 2005; Bragazza et al. 2007; Cornwell et al. 2008). We
propose that the variation in litter quality between the different materials (Straková et al.
2010) and consequent differences in activity and composition of microbial communities
(Thormann et al. 2004; Peltoniemi 2010; Straková et al. 2011) are largely responsible for the
variation in decomposition rates; the litter type/litter quality effect is, in the active surface
layers, stronger than the effect of the environment. For example, the specific chemical quality
of Sphagnum moss suppresses its decomposition (e.g., Bragazza et al. 2007), in spite of
favourable environmental conditions for decomposition provided by moss patches as shown
by our results on cellulose decomposition.
Though no single chemical parameter could predict the variability in decomposition
dynamics associated with such different materials as included in this study (see also Bragazza
et al. 2007), reasonable predictions were obtained with a few chemical parameters:
concentrations of total extractives and N (positive correlation with mass loss), Klason lignin
and p-hydoxyphenols (lignin-like compounds, negative correlation with mass loss). This
gives the general characteristic of a substrate that the decomposers prefer to utilize: rich in
nutrients, with high proportion of easily degradable compounds relative to the recalcitrant
We found total extractives (i.e. sum of nonpolar- (dichloromethane-) and polar- (acetone-,
ethanol- water-) extractives) being somewhat better related to litter mass loss than water
extractives used in earlier decomposition studies (Gholz et al. 2000; Preston & Trofymow
2000; Trofymow et al. 2002; Váv?ová et al. 2009) or in decomposition models predicting
litter mass loss (e.g., Moorhead et al. 1999; Liski et al. 2005). This finding may be influenced
by the number of different materials used in this study, which was higher than in any earlier
It is noteworthy that besides the increased soil aeration (direct effect of WT drawdown),
environmental conditions for decomposition are in long-term further changed through the
changes in vegetation (indirect effect of WT drawdown). The litter layer might, together with
increasing tree canopy, serve as protection against UV-B radiation that has negative effects
on litter decomposers (Gehrke et al. 1995; Duguay & Klironomos 2000), as well as insulation
keeping favourable moisture conditions. Such effects were not explicitly measured in this
study and thus cannot be separated from the direct effects of WL drawdown.
This study shows a drop in temperature associated with the site forestation, possibly due to
tree canopy shading and evaporative cooling effect of trees. We did not find any negative
effect of such decrease in temperature on litter decomposition rates (cf. Dorrepaal et al.
2009). The temperature effect was possibly overruled by other effects, i.e. litter type and soil
characteristics, as those also changed following WT drawdown. Also, fresh litter represents
the youngest, most easily decomposable organic matter that has low temperature sensitivity
(Karhu et al. 2010).
4.2. Effects of site nutrient regime and soil depth
As expected, litter decomposition rates at the community level correlated positively with site
nutrient regime. Contrary to that, decomposition of the common litter at the nutrient-poor bog
was not slower compared to that at the fens, even though environmental conditions for
decomposition were indeed worse at the bog, as shown by our results on cellulose
decomposition. One possible explanation of this finding is adaptation and specialization of
decomposers to plant species characteristic of a given community, a “home field advantage”
(Hunt et al. 1988; Gholz et al. 2000; Bragazza et al. 2007). Litter types included here in
common litter represent plant species more typical of bogs: B. nana, E. vaginatum and P.
sylvestris. Such litters then decompose more slowly in communities that have lower
abundance of comparable plant and associated microbial decomposers, independent of
favourable environmental conditions. This mechanism may in fact be also included in the
increase of decomposition rates of common litter following WT drawdown as the species are
also more typical of the drained plots (indirect effect of WT drawdown).
Decomposition rates of belowground litter decreased with depth. It is noteworthy that it
was the distance from the soil surface that determined decomposition rates, rather than the
distance from the WT. Soil compaction with associated change in the soil pore size
distribution towards higher proportion of small pores that may protect litter against microbial
attack (Breland & Hansen 1996), as well as decrease in litter summer temperature with soil
depth (see Table 2) may slow down the decomposition process in deeper soil layers,
independently of the WT position.
4.3. Accumulation of organic matter
In long-term, dramatically increased litter inputs (Straková et al. 2010) resulted in large
accumulation of organic matter in spite of increased decomposition rates. This emphasizes
the significance of litter production: if the inputs are high, organic matter accumulates at a
site despite the high litter decomposition rates.
The C balance of a drained peatland depends on 1) the rate of decomposition of the ‘‘old
C’’ (peat accumulated before the WT drawdown), and 2) the rates of inputs and
decomposition of the ‘‘new C’’ (biomass produced by the changed vegetation after the WT
drawdown) under the new environmental conditions (Laiho 2006). If the accumulation
(inputs – decomposition losses) of the new organic matter exceeds the decomposition losses
from the old peat, the peatland will remain a sink of C. If not, then the peatland will become a
source of C to the atmosphere. The estimated annual C loss via heterotrophic soil respiration
ranges from 145-670 g m-2 in boreal forestry-drained peatlands (Ojanen et al. 2010). The
annual C inputs via litter production estimated in this study ranged from 190 to 200 g m-2 for
the forestry-drained (LTD) plots, and from such inputs 110-130 g m-2 remained after 2 years.
Considering that relatively large part of the measured litter losses (10-30%) may still be
retained in the soil (Domish et al. 2000), it seems that a drained peat soil may, under certain
conditions defined mainly by the site nutrient regime (Ojanen et al. 2010), still act as a sink
of atmospheric C. When the whole ecosystem C balance is estimated, tree biomass
accumulation must be included in the calculations. In our study sites the tree stand retained
1300-5300 g m-2 C in the forestry-drained plots compared to 40-80 g m-2 C in the pristine
(undrained) plots (Anttila 2008).
Our assumptions are based on 2-year decomposition data only, however. Longer-term
decomposition results are needed to validate the further behaviour of the accumulated organic
matter at the different WT regimes. Also, the measured “old C” loss may not include the
possible priming effect of litter inputs on peat decomposition, as litter was removed from the
sites before each measurement (Ojanen et al. 2010). The possible role of this should also be
4.4. Time-scale of the effects
As presumed, the indirect effects of WT drawdown on litter decomposition, via changes in
plant community structure, dominated in the long-term (decades) relative to short-term
(years). The short-term changes reflect transient conditions where the direct effects of WT
drawdown are dominant: improved conditions for aerobic decomposition are linked with
unchanged or lowered amounts of organic matter inputs, most likely facilitating a net C loss
from the soil. In contrast, the long-term changes reflect a longer-lasting situation as the
ecosystem becomes adapted to the new conditions and the indirect effects of WT drawdown
get dominance: the increased litter inputs may at least partly compensate for the increased
rates of peat decomposition. So far, too few studies have considered the long-term aspect.
4.5. Implications for soil C modelling
Our results demonstrate that the shift in vegetation composition as a response to climate
and/or land-use change is the main factor affecting the peatland ecosystem C cycle (see also
Straková et al. 2010). Thus, dynamic vegetation is a necessity in any models applied for
estimating responses of C fluxes to changes in environmental conditions. It is noteworthy that
the time scale for vegetation changes caused by hydrological changes needs to extend to
We provide possible grouping of litter types into plant functional types based on their
decomposition rates (Table 4; 5 types for aboveground litter and 3 types for belowground
litter) that the models could directly utilize. As litter types within these groups may still
significantly vary in their quality, the PFT grouping suggested by Straková et al. (2010)
based on detailed chemical characterization of litter would probably provide even better
performance when the models allow for a narrower grouping.
Of the existing models, the dynamic global vegetation model LPJ-Why (Wania et al.
2009) includes 5 PFTs applicable for boreal peatlands: graminoids, Sphagnum, herbs, broad-
leaved and needle-leaved boreal plants, which compares rather well to our grouping.
Distinguishing the different sections of Sphagnum, and minerotrophic versus ombrotrophic
graminoids, would probably increase the accuracy of the model. The same holds for, e.g., the
ecosystem-level models by Zhang et al. (2002) and Bauer (2004). An updated version of the
Holocene Peat Model (HPM) uses 12 PFTs based on productivity and rooting characteristics
(Frolking et al. 2010). HPM now distinguishes within vascular PFT’s whether they are
minerotrophic or ombrotrophic, and within Sphagna their preference of microform, which
corresponds to our grouping. However, this model does not distinguish trees or belowground
litter. Since trees are an integral part of many present-day peatlands, and based on
palaeoecological studies characterized several currently wet treeless sites during a drier
climate period, they should be considered also in peatland models even if the general
perception of a peatland is wet, treeless, and Sphagnum-dominated. With trees, especially,
comes also the need to distinguish different litter types, since woody materials decompose at
rates different from the foliar litters of the same species.
The clear evidence of lowered soil temperature as a response to WT drawdown when an
initially open peatland converts into a forest ecosystem is another outcome of our study that
the models could directly utilize. In the existing models the possible lowering of soil
temperature as the long-term response of a peatland ecosystem to climatic warming has not
yet been considered.
Our materials generally showed a non-linear mass loss over the 2-year study period: most
of the mass was lost during the first year and further decomposition slowed down.
Decomposition of organic matter has often been described with the exponential decay, i.e.
negative exponential function, following Olson (1963), which implies that decomposition
approaches zero with time. This function may be suitable for the initial phase of
decomposition, but may not fit the later phases when decomposition gets slow while
considerable proportion of litter still remains (Latter et al. 1998). The fit of the negative
exponential function to our litter materials was generally poor, and especially so for
belowground litter. This suggests lower applicability of the negative exponential function in
peatlands where decomposition is generally suppressed by waterlogged conditions, compared
to mineral soil sites.
The study demonstrates that the direct effects of changing climate and/or land-use on
decomposition and C accumulation in peatlands are in long-term (decades) overruled by the
indirect effects via changes in vegetation community composition. Even though plant litter
decomposition rates increase following WT drawdown, the accumulation of new organic
matter also increases due to proportionally more increased litter inputs. The accumulation
may even exceed decomposition of the old peat at the nutrient-poor (bog) sites. Some boreal
peatlands may thus still act as a sink of atmospheric C under changing climate and/or land-
use, but the quality (chemical composition) of accumulated C will greatly differ from that
accumulated in pristine conditions. Longer-term decomposition results are needed to validate
the further behaviour of the accumulated organic matter at the different WT regimes.
This study was supported by the Academy of Finland and the Graduate School of Forest
Sciences. We thank Satu Repo, Alison Gillette, Heli Miettinen, and several other students,
trainees and laboratory technicians for their help with the field and laboratory work, and Juul
Limpens, Björn Berg and Tim Moore for their thorough and constructive comments on the
Anttila J (2008) Lyhyt- ja pitkäaikaisen kuivatuksen aiheuttamat muutokset soiden
maanpäällisen karikkeen tuotossa [Effects of short-term and long-term water-level
drawdown on the above-ground litter inputs into peatland ecosystems]. M.Sc. thesis,
University of Helsinki, Helsinki, 56 p.
Bauer IE (2004) Modelling effects of litter quality and environment on peat accumulation
over different time-scales. Journal of Ecology, 92, 661-674.
Belyea LR (1996) Separating the effects of litter quality and microenvironment on
decomposition rates in a patterned peatland. Oikos, 77, 529-539.
Box EO (1996) Plant Functional Types and Climate at the Global Scale. Journal of
Vegetation Science, 7, 309-320.
Bragazza L, Siffi C, Iacumin P, Gerdol R (2007) Mass loss and nutrient release during litter
decay in peatland: The role of microbial adaptability to litter chemistry. Soil Biology and
Biochemistry, 39, 257-267.
Breeuwer AJG, Robroek BJM, Limpens J, Heijmans MMPD, Schouten MGC, Berendse F
(2009) Decreased summer water table depth affects peatland vegetation. Basic and
Applied Ecology, 10, 330-339.
Breland TA, Hansen S (1996) Nitrogen mineralization and microbial biomass as affected by
soil compaction. Soil Biology and Biochemistry, 28, 655–663.
Cornwell WK, Cornelissen JH, Amatangelo K et al. (2008) Plant species traits are the
predominant control on litter decomposition rates within biomes worldwide. Ecology
Letters, 11, 1065-1071.
Cotrufo MF, Ngao J, Marzaioli F, Piermatteo D (2010) Inter-comparison of methods for
quantifying leaf litter decomposition rates. Plant and Soil, 334, 365-376.
Domisch T, Finér L, Laiho R, Karsisto M, Laine J (2000) Decomposition of Scots pine litter
and the fate of released carbon in pristine and drained pine mires. Soil Biology and
Biochemistry, 32, 1571–1580.
Dorrepaal E, Cornelissen JHC, Aerts R, Wallén B, van Logtestijn RSP (2005) Are growth
forms consistent predictors of leaf litter quality and decomposability across peatlands
along a latitudinal gradient? Journal of Ecology, 93, 817-828.
Dorrepaal E, Toet S, van Logtestijn RSP, Swart E, van de Weg MJ, Callaghan TV, Aerts R
(2009) Carbon respiration from subsurface peat accelerated by climate warming in the
subarctic. Nature, 460, 616–619.
Duguay KJ, Klironomos JN (2000) Direct and indirect effects of enhanced UV-B radiation on
the decomposing and competitive abilities of saprobic fungi. Applied Soil Ecology, 14,
Frolking S, Roulet NT, Tuittila E, Bubier JL, Quillet A, Talbot J, Richard PJH (2010) A new
model of Holocene peatland net primary production, decomposition, water balance, and
peat accumulation. Earth System Dynamics, 1, 1-21.
Gartner TB, Cardon ZG (2004) Decomposition dynamics in mixed-species leaf litter. Oikos,
Gehrke C, Johanson U, Callaghan TV, Chadwick D, Robinson CH (1995). The impact of
enhanced ultraviolet-B radiation on litter quality and decomposition processes in
Vaccinium leaves from the Subarctic. Oikos, 72, 213-222.
Gholz HL, Wedin DA, Smitherman SM, Harmon ME, Parton WJ (2000) Long-term
dynamics of pine and hardwood litter in contrasting environments: toward a global model
of decomposition. Global Change Biology, 6, 751-765.
Goldstein H (1995) Multilevel Statistical Models. Kendall's Library of Statistics 3. Hodder
Gorham E (1991) Northern peatlands: role in the carbon cycle and probable responses to
climatic warming. Ecological Applications, 1, 182-195.
Grieve IC, Gilvear DG, Bryant RG (1995) Hydrochemical and water source variations across
a floodplain mire, Insh Marshes, Scotland. Hydrological Processes, 9, 99-110.
Hobbie SE (1996) Temperature and plant species control over litter decomposition in
Alaskan tundra. Ecological Monographs, 66, 503-522.
Hunt HW, Ingham ER, Coleman DC, Elliot ET, Reid CPP (1988) Nitrogen limitation of
production and decomposition in prairie, mountain meadow and pine forest. Ecology, 69,
Karhu K, Fritze H, Hämäläinen K, Vanhala P, Jungner H, Oinonen M, Sonninen E, Tuomi
M, Spetz P, Kitunen V, Liski J (2010) Temperature sensitivity of soil carbon fractions in
boreal forest soil. Ecology, 91, 370-376.
Kurz-Besson C, Coûteaux MM, Thiéry JM, Berg B, Remacle J (2005) A comparison of
litterbag and direct observation methods of Scots pine needle decomposition measurement.
Soil Biology and Biochemistry, 37, 2315–2318.
Laiho R (2006) Decomposition in peatlands: Reconciling seemingly contrasting results on the
impacts of lowered water levels. Soil Biology and Biochemistry, 38, 2011-2024.
Laiho R, Vasander H, Penttilä T, Laine J (2003) Dynamics of plant-mediated organic matter
and nutrient cycling following water-level drawdown in boreal peatlands. Global
Biogeochemical Cycles, 17, 1053. doi:10.1029/2002GB002015
Laine J, Vasander H, Laiho R (1995) Long-term effects of water level drawdown on the
vegetation of drained pine mires in southern Finland. Journal of Applied Ecology, 32, 785-
Laine J, Komulainen VM, Laiho R et al. (2004) Lakkasuo – a guide to mire ecosystem.
University of Helsinki Department of Forest Ecology Publications, 31, 1-123.
Latter PM, Howson G, Howard DM, Scott WA (1998) Long-term study of litter
decomposition on a Pennine peat bog: which regression? Oecologia, 113, 94–103.
Limpens J, Berendse F (2003) How litter quality affects mass loss and N loss from
decomposing Sphagnum. Oikos, 103, 537-547.
Liski J, Palosuo T, Peltoniemi M, Sievänen R (2005) Carbon and decomposition model
Yasso for forest soils. Ecological Modelling, 189, 168-182.
McClaugherty CA, Pastor J, Aber JD, Melillo JM (1985) Forest litter decomposition in
relation to soil nitrogen dynamics and litter quality. Ecology, 66, 266-275.