Comments

Ecology, 91(6), 2010, pp. 18671869

2010 by the Ecological Society of America

Functional leaf traits and biodiversity

effects on litter decomposition
in a stream: comment
ANTOINE LECERF1

AND

JOHN KOMINOSKI2

In a recent paper, Schindler and Gessner (2009) tested

how functional diversity of leaf litter explains litter
diversity effects on decomposition. Their study builds on
a large body of experimental evidence that decomposition rates of litter species mixtures are not simply
predicted by the component species decomposing in
isolation (Gartner and Cardon 2004, Hattenschwiler
and Gasser 2005, Kominoski et al. 2007, Lecerf et al.
2007). Such litter diversity effects are thought to be
explained by the presence of heterogeneous litter (in
terms of leaf degradability) in mixtures, causing
deviation (positive or negative) in decomposition of
mixtures as compared to the average single-species
decomposition of the component litter species (Gartner
and Cardon 2004, Hattenschwiler et al. 2005). Functional diversity represents a valuable framework for
testing this hypothesis. Schindler and Gessner (2009) is
the rst published paper that addresses how functional
litter diversity inuences in-stream litter decomposition.
Their manipulative experiment, with homogeneous
and heterogeneous mixtures of litter enclosed in coarseand ne-mesh bags, tested for decomposition rates of
individual litter species from monospecic and mixedspecies assemblages. Based on species-specic decay
rates expected from published and unpublished data,
they split up a set of nine litter species spanning a broad
range of decay rates into three functional leaf groups
(slow-, medium-, or fast-decomposing species). Homogeneous mixtures consisted of three litter species taken
from a single functional group, and heterogeneous
mixtures consisted of one species from each of three
functional groups. The planned comparison of homogeneous versus heterogeneous mixtures was used to test
for the effect of functional litter diversity on litter
Manuscript received 8 June 2009; revised 4 September 2009;
accepted 10 September 2009. Corresponding Editor: P. M.
Groffman.
1 Universite de Toulouse, UPS, INP, CNRS, EcoLab
(Laboratoire dE cologie Fonctionnelle), 29 Rue Jeanne
Marvig, F-31055 Toulouse, France. E-mail: lecerf@cict.fr
2 Department of Forest Sciences, University of British
Columbia, Vancouver, British Columbia V6T 1Z4 Canada.

decomposition. As mass loss differences between homogeneous and heterogeneous litter mixtures were found to
be nearly signicant, Schindler and Gessner (2009)
argued that evidence was weak in support of the idea
that functional litter dissimilarity matters. Rather, we
believe that the conclusions of Schindler and Gessner
(2009) are conservative and were likely explained by
their choice of experimental design, data analysis, and
interpretation. We offer several suggestions that would
have altered the results and interpretations of Schindler
and Gessner (2009) and that we anticipate will assist
ecologists in adequately testing for effects of functional
diversity.
Recent years have seen major methodological advances in the assessment of functional diversity of
species assemblages (Petchey and Gaston 2006, Villegier
et al. 2008). One recurrent criticism of functional group
richness (i.e., the most popular index of functional
diversity) is the inherent subjectivity of selecting a
number of functional groups and optimal cut-off limits
(Petchey and Gaston 2006). In addition, there is a
nontrivial risk of misclassifying species when functional
groups are built from a priori information. The
functional group classication of Schindler and Gessner
(2009) resulted in overlapping litter functional groups
and large within-group variations. Decay constants
estimated for individual litter species incubated in
single-species, coarse-mesh bags during the experiment
(cf. Schindler and Gessner 2009: Table 2) showed that
two species (alder, maple) were apparently misclassied,
and two other species (hazel, sycamore) decomposing at
very similar rates were classied as functionally distinct.
Similarly, litter decomposition from ne-mesh bags had
substantial variability within litter functional group
categories, which led to striking overlap (cf. Schindler
and Gessner 2009: Table 2). Thus, we suggest that
functional group richness based on a priori classication
of litter species may be ineffective at identifying realized
functional diversity of litter mixtures and, hence,
ineffective at addressing its effects on litter decomposition.
To assess the genuine extent of functional distinctiveness between homogeneous and heterogeneous mixtures
in Schindler and Gessner (2009), we quantied functional litter diversity for all litter mixtures on the basis of
exponential decay constant determined in coarse-mesh
bags from Table 2 in Schindler and Gessner (2009). We
calculated functional richness (range of species trait
values divided by a maximum range), functional
regularity (an index derived from the Bulla O index;
Mouillot et al. 2005), and functional divergence
(variance of species trait values) following the multifaceted framework for functional diversity assessment

1867

1868

COMMENTS

proposed by Mason et al. (2005) and Villegier et al.

(2008). All calculations were based on ln-transformed
trait values to achieve a normal distribution. Overall, the
three functional diversity indices failed to univocally
distinguish between homogeneous and heterogeneous
mixtures (Mann-Whitney, U  15, P . 0.08; two sample
t test, jt28j , 2.0, P . 0.05). Functional richness of 9 of
the 27 heterogeneous mixtures (0.070.43) fell within the
range of values for the homogeneous mixtures (0.10
0.20). Functional regularity of 17 of the 27 heterogeneous treatments (0.500.97) fell within the range of
values for the homogeneous mixtures (0.760.97).
Functional divergence of 8 of the 27 heterogeneous
treatments (0.070.81) fell within the range of values for
the homogeneous treatments (0.130.39). Additionally,
functional richness and functional divergence in a few
heterogeneous mixtures (notably, alderhazelsycamore
and alderwillowsycamore) were much lower than any
value for homogeneous mixtures. These indices demonstrate that the planned comparison between homogeneous and heterogeneous mixtures was spurious and,
therefore, that the experimental design in Schindler and
Gessner (2009) lacks power to detect effects of
functional litter diversity on decomposition.
We believe there are additional reasons why signicant effects of functional litter diversity on decomposition were not detected by Schindler and Gessner (2009).
The authors interpretation of nonsignicance for the
interaction between litter mixtures and litter functional
group category (P 0.06) and litter species (P 0.25)
was conservative in spite of statistical issues. First,
ANOVA results testing difference between homogeneous and heterogeneous mixtures (cf. Schindler and
Gessner 2009: Table 4) were likely inuenced by the
unbalance between the number of homogeneous (n 3)
and heterogeneous (n 27) litter mixtures (Shaw and
Mitchell-Olds 1993). A more balanced design would
have made the test more robust against violation of the
assumption of homogeneity of variance (Quinn and
Keough 2002). Standard deviations estimated from the
error bars (SE) in Fig. 5 provide evidence of heteroscedasticity. Second, a common issue among biodiversityecosystem functioning studies, including Schindler
and Gessner (2009), is the condition of independence in
ANOVA for mixed-species treatments as a result of
interactive effects among component species (Gartner
and Cardon 2004, Hattenschwiler and Gasser 2005,
Kominoski et al. 2007, Lecerf et al. 2007). This problem
could have been resolved by either using statistics for
correlated data (e.g., linear mixed-effects models, but see
Ives and Zhu 2006) or by simply conducting individual
ANOVAs for each litter species (e.g., Hattenschwiler
and Gasser 2005). Both an unbalanced design and
nonindependent observations are likely to lead to
incorrect Type I error rates, warranting extreme caution
when interpreting nearly signicant P values.

Ecology, Vol. 91, No. 6

The conclusion of Schindler and Gessner (2009) that

the effects of functional litter diversity on decomposition
are minimal is partly based on their interpretation of the
percent sums of squares in ANOVA (Spehn et al. 2005,
Lecerf et al. 2007). Their results basically indicated that
functional litter diversity (cumulative sums of squares:
3%) was a much less important factor than litter species
identity (cumulative sums of squares: 40%) in their
analysis. However, as a measure of relative effect size,
percent sums of squares does not identify how much the
decomposition rate of individual litter species was
modied in heterogeneous mixtures. In fact, in spite of
the subtle differences in functional litter diversity
between homogeneous and heterogeneous groups, Fig.
5 suggests that litter functional diversity had an effect on
decomposition and was dependent on litter species
identity and functional group category. For example,
there was a synergistic effect of litter functional diversity
on elm (the fastest-decomposing species) and an
antagonistic effect of litter functional diversity on beech
(the slowest-decomposing species).
In conclusion, our primary objective is not to speculate
on whether or not functional litter diversity is important
in explaining decomposition. Rather, we discuss how a
priori species classications, statistical analysis choice,
and conservative interpretations might lead to underestimating ecologically relevant effects; in this case, the
effects of litter functional diversity on in-stream decomposition. Future studies should take advantage of recent
methodological advances in the assessment of functional
diversity of species assemblages (Petchey and Gaston
2006, Villegier et al. 2008). We also encourage ecologists
interested in testing for functional diversity to consider
the statistical issues and solutions we present.
Literature cited
Gartner, T. B., and Z. G. Cardon. 2004. Decomposition
dynamics in mixed-species leaf litter. Oikos 104:230246.
Hattenschwiler, S., and P. Gasser. 2005. Soil animals alter plant
litter diversity effects on decomposition. Proceedings of the
National Academy of Sciences (USA) 102:15191524.
Hattenschwiler, S., A. V. Tiunov, and S. Scheu. 2005.
Biodiversity and litter decomposition in terrestrial ecosystems. Annual Review of Ecology, Evolution and Systematics
36:191421.
Ives, A. R., and J. Zhu. 2006. Statistics for correlated data:
phylogenies, space, and time. Ecological Applications 16:20
32.
Kominoski, J. S., C. M. Pringle, B. A. Ball, M. A. Bradford,
D. C. Coleman, D. B. Hall, and M. D. Hunter. 2007.
Nonadditive effects of leaf-litter species diversity on breakdown dynamics in a detritus-based stream. Ecology 88:1167
1176.
Lecerf, A., G. Risnoveanu, C. Popescu, M. O. Gessner, and E.
Chauvet. 2007. Decomposition of diverse litter mixtures in
streams. Ecology 88:219227.
Mason, N. W. H., D. Mouillot, W. G. Lee, and J. B. Wilson.
2005. Functional richness, functional evenness and functional
divergence: the primary components of functional diversity.
Oikos 111:112118.

June 2010

COMMENTS

Mouillot, D., N. W. H. Mason, O. Dumay, and J. B. Wilson.

2005. Functional regularity: a neglected aspect of functional
diversity. Oecologia 142:353359.
Petchey, O. L., and K. J. Gaston. 2006. Functional diversity:
back to basics and looking forward. Ecology Letters 9:741
758.
Quinn, G. P., and M. J. Keough. 2002. Experimental design
and data analysis for biologists. Cambridge University Press,
Cambridge, UK.
Schindler, M., and M. O. Gessner. 2009. Functional leaf traits
and biodiversity effects on litter decomposition in a stream.
Ecology 90:16411649.
Shaw, R. G., and T. Mitchell-Olds. 1993. ANOVA for
unbalanced data: an overview. Ecology 74:16381645.
Spehn, E. M., et al. 2005. Ecosystem effects of biodiversity
manipulations in European grasslands. Ecological Monographs 75:3763.
Villegier, S., N. W. H. Mason, and D. Mouillot. 2008. New
multidimensional functional diversity indices for a multifaceted framework in functional ecology. Ecology 89:2290
2301.

_______________________________

1869

June 2010

COMMENTS

_______________________________

Ecology, 91(6), 2010, pp. 18691871

2010 by the Ecological Society of America

Functional leaf traits and biodiversity

effects on litter decomposition in a
stream: reply
MARK O. GESSNER1
Experiments testing for effects of litter mixing on
decomposition have been an important component of
efforts to elucidate the signicance of biodiversity for
ecosystem functioning. Outcomes of these experiments
with mixed litter have been frustratingly mixed as well,
however, with some suggesting positive effects on
decomposition rate (i.e., acceleration), some negative
effects, and some no effect at all (Gartner and Cardon
2004, Hattenschwiler et al. 2005, Gessner et al. 2010).
How may these inconsistencies be reconciled? One idea
is that diversity effects can arise only when functional
traits of litter included in mixtures are sufciently
dissimilar to elicit litter-specic responses of microbial
decomposers or detritivores (Hoorens et al. 2003, Smith
and Bradford 2003, Epps et al. 2007, Schindler and
Gessner 2009). Under this assumption, heterogeneous
mixtures composed of dissimilar litter types are expected
Manuscript received 5 November 2009; accepted 3 December
2009. Corresponding Editor: P. M. Groffman.
1 Department of Aquatic Ecology, Eawag: Swiss Federal
Institute of Aquatic Science and Technology, 8600 Dubendorf, Switzerland; and Institute of Integrative Biology (IBZ),
ETH Zurich, Switzerland. E-mail: gessner@eawag.ch

1869

to show strong diversity effects on decomposition. In

contrast, decomposition of mixed litter composed of
species that share functional traits (homogenous mixtures) is predicted not to deviate signicantly from the
average decomposition rate of component species. In
light of the conicting results from previous experiments
(Gartner and Cardon 2004), we set out to test this
hypothesis with litter from broad-leaved forest trees
decomposing in a stream, and we concluded that our
experiment provides some, but not overwhelming,
support for the dissimilarity hypothesis (Schindler and
Gessner 2009).
The commentary by Lecerf and Kominoski (2010) on
our test of functional diversity as a predictor of
decomposition in litter mixtures (Schindler and Gessner
2009) outlines ideal circumstances for conducting and
analyzing such experiments, it expresses concern that
our experiment and interpretations do not perfectly
reect these circumstances, and it suggests alternative
ways to approach the question. I appreciate the
comments by Lecerf and Kominoski (2010) and
welcome the opportunity to respond. Below I will
address each of the main concerns, reconcile or qualify
apparent differences in appraisal, and clarify or justify
our choices. In their concluding paragraph, Lecerf and
Kominoski (2010) assert that their primary objective is
not to speculate on whether or not functional litter
diversity is important in explaining decomposition.
Overall, however, their commentary implies that clear
diversity effects would have emerged in heterogeneous
litter mixtures had we chosen a different experimental
design, data analysis, and interpretation (Lecerf and
Kominoski 2010). I will argue that this assertion is
insufciently substantiated and that our previous
conclusions are well balanced and robust.
The most critical remark made by Lecerf and
Kominoski (2010) relates to overlap in dissimilarity
among litter types between our homogenous and
heterogeneous litter mixtures. This observation is
correct. Consequently, we explicitly discussed our results
in light of this important point (Schindler and Gessner
2009). Lecerf and Kominoski (2010) enhance our
discussion in that they make the argument quantitative
by calculating functional diversity indices and comparing them statistically between our homogenous and
heterogeneous litter mixtures. A limitation of this
quantitative analysis is, however, that it ignores the
specic pattern in our data set. Specically, the overlap
in dissimilarity we noted was primarily due to one fastdecomposing (alder) and one slow-decomposing (sycamore) litter species, which decomposed much more
slowly or more rapidly, respectively, than suggested by
previous results from the same (e.g., Hieber and Gessner
2002) or a similar (Gessner and Chauvet 1994) stream.
Even so, alder litter behaved exactly as predicted for a
fast-decomposing species under the dissimilarity hy-

1870

COMMENTS

pothesis in that mass loss was faster in heterogeneous

than in homogenous mixtures (Schindler and Gessner
2009: Fig. 5a). Ash, in contrast, showed identical
decomposition rates in homogenous and heterogeneous
mixtures. This is counter to expectations under the
dissimilarity hypothesis because ash was unequivocally
classied as a rapidly decomposing species. Thus, the
species that caused overlap in functional dissimilarity
between homogenous and heterogeneous litter mixtures
(alder and sycamore) were not the same species
responsible for our failure to detect unequivocal
diversity effects on decomposition in the heterogeneous
litter mixtures (e.g., ash). Therefore, although formally
correct and generally useful, the functional diversity
analysis by Lecerf and Kominoski (2010) falls short of
elucidating the causes behind the outcome of our
experiment. Clearly, overlap in dissimilarity between
homogenous and heterogeneous litter mixtures was not
the key driver accounting for the only marginally
signicant difference in decomposition rate observed
by Schindler and Gessner (2009).
An important point to realize is that our experiment
was not designed to maximize contrast among species in
heterogeneous litter mixtures, and we avoided choosing
litter from the regional species pool comprising different
life forms. Instead the experiment was conceived to
reect the natural range of major litter types during
autumn leaf fall in temperate forest ecosystems, including forest streams. As a consequence, our choice of litter
was restricted to the relatively narrow range covered by
leaves from broad-leaved deciduous trees. We also
included species anticipated to decompose at intermediate rates, which reduces overall dissimilarity in heterogeneous litter mixtures. Greater contrasts between litter
types can easily be created, for instance by mixing
grasses and forbs (Hoorens et al. 2003), or by mixing
litter of nitrogen-xing species that have low concentrations of recalcitrant compounds with conifer needles
containing high concentrations of lignin and other
polyphenolics such as tannins (Fyles and Fyles 1993).
If the dissimilarity hypothesis holds, then diversity
effects are more likely detected in experiments involving
only two such extreme end members than in the
experiment reported in Schindler and Gessner (2009).
Thus, it is certainly possible that clearer diversity effects
arise in some heterogeneous litter mixtures than
observed by Schindler and Gessner (2009), but the
results of our experiment suggest that the underlying
mechanisms may not be powerful enough to play out
very strongly in litter mixtures from trees of temperate
mixed deciduous forests.
Lecerf and Kominoski (2010) further argue that
selection of discrete functional groups is inherently
subjective (see also Petchey et al. 2009). They propose
using instead continuous metrics that capture functional
diversity, or dissimilarity, of species mixtures to increase

Ecology, Vol. 91, No. 6

chances of detecting functional diversity effects on

ecosystem processes. This is a valid point. The suggestion boils down to substituting a continuous for a
categorical variable and using regression analysis instead
of analysis of variance. Although there is no doubt that
such a regression approach could be powerful to test the
dissimilarity hypothesis, it also poses its own inherent
challenges. How to measure and express functional
diversity on a continuous scale is an active area of
research that is hotly debated at present. Deep
disagreement exists about the metrics to use (e.g.,
Petchey and Gaston 2009, Poos et al. 2009) and, more
important, effects of functional diversity on ecosystem
processes hinge strongly on the traits selected to
calculate functional diversity metrics (Petchy et al.
2009). Both realities introduce an element of subjectivity. It is in this light that we opted for the simplest
possible, traditional approach and assigned leaf species
to one of three well-established decomposition classes.
This classication has been widely used since the seminal
work on litter decomposition in streams by Petersen and
Cummins (1974). Since the number of functional groups
is xed to three in this classication, our choice was not
subjective, as Lecerf and Kominoski (2010) imply.
The nal main point raised by Lecerf and Kominoski
(2010) relates to statistical issues, particularly the fact
that unbalanced experimental designs such as the one we
used (Schindler and Gessner 2009) are less powerful
than balanced designs. Unfortunately, the unbalanced
design reects the nature of litter mixing experiments,
and of many biodiversityecosystem functioning experiments in general. This, together with other factors that
reduce power of ANOVAs, could have inuenced our
ability to show litter heterogeneity to be a key driver of
litter diversity effects on decomposition, as noted in
Schindler and Gessner (2009). However, the argument
of statistical power should not be overrated given that
our design involved a total of 150 experimental units for
the comparison of homogenous and heterogeneous litter
mixtures, far more than in most ecological eld
experiments. In response to another statistical issue
raised by Lecerf and Kominoski (2010), it is also
important to clarify that our main conclusion (i.e.,
rather weak evidence for diversity effects on decomposition in heterogeneous litter mixtures) was primarily
supported by the marginally signicant (P 0.06)
interaction effect we obtained in an ANOVA (see
Schindler and Gessner 2009: Table 4), reecting the
pattern shown in Fig. 5 of that paper. In contrast to the
assertion by Lecerf and Kominoski (2010), we did not
primarily base our argument on the percent of sums of
squares accounted for by different factors in the
ANOVA, although it is worth noting that the term of
most interest, viz. the interaction between litter mixture
(i.e., homogenous vs. heterogeneous mixture) and decay
category (i.e., functional group), accounted for ,2% of

June 2010

COMMENTS

the total variation in litter mass loss. Lecerf and

Kominoski (2010) conclude their paragraph on statistical issues by stating that extreme caution is warranted
when interpreting nearly signicant P values. I
concur. This is, therefore, precisely what we did.
The experiment presented in Schindler and Gessner
(2009) represents one of the rst attempts to explicitly
test the dissimilarity hypothesis. It provides some hints in
support of the hypothesis, but overall we found
insufcient evidence to draw rm conclusions about
the role of functional litter diversity in determining
decomposition rates of litter mixtures. While I remain
convinced that the conceptual basis and statistical
approach of our study is sound and, especially, that
the conclusions drawn are fully appropriate, I concede to
Lecerf and Kominoski (2010) that many alternative ways
can be envisaged to approach the problem of whether
litter heterogeneity in mixtures determines the occurrence and strength of diversity effects on decomposition.
One alternative is to measure trait dissimilarity of species
mixtures based on a range of traits (e.g., lignin, nitrogen,
and phosphorus concentrations) and use regression
analysis to relate the average dissimilarity values of the
mixtures to decomposition rate, as discussed above. In
addition to such alternative experimental designs,
specic tests may be conceived to elucidate possible
biological mechanisms that could cause diversity effects
on decomposition. So far, mechanisms underlying
diversity effects on ecosystem processes have been largely
explored by detecting patterns in data sets by computational means (Hector et al. 2009). This has yielded
important insights. But greatest progress in the future
will likely be made when the biological processes that
lead to diversity effects are unraveled. Several such
mechanisms have been identied (Hattenschwiler et al.
2005, Gessner et al. 2010) and await experimental testing.
Acknowledgments
I am grateful to Jessica Hines and Matty Berg for
constructive comments and suggestions on the manuscript on
short notice.
Literature cited
Epps, K. Y., N. B. Comerford, J. B. Reeves III, W. P. Cropper,
Jr., and Q. R. Araujo. 2007. Chemical diversity highlighting a
species richness and ecosystem function disconnect. Oikos
116:18311840.

1871

Fyles, J. W., and I. H. Fyles. 1993. Interaction of Douglas-r

with red alder and salal foliage litter during decomposition.
Canadian Journal of Forest Research 23:358361.
Gartner, T. B., and Z. G. Cardon. 2004. Decomposition
dynamics in mixed-species leaf litter. Oikos 104:230246.
Gessner, M. O., and E. Chauvet. 1994. Importance of stream
microfungi in controlling breakdown rates of leaf litter.
Ecology 75:18071817.
Gessner, M. O., C. M. Swan, C. K. Dang, B. G. McKie, R. D.
Bardgett, D. H. Wall, and S. Hattenschwiler. 2010. Diversity
meets decomposition. Trends in Ecology and Evolution, in
press.
Hattenschwiler, S., A. V. Tiunov, and S. Scheu. 2005.
Biodiversity and litter decomposition in terrestrial ecosystems. Annual Review of Ecology, Evolution and Systematics
36:191218.
Hector, A., T. Bell, J. Connolly, J. Finn, J. Fox, L. Kirwan, M.
Loreau, J. McLaren, B. Schmid, and A. Weigelt. 2009. The
analysis of biodiversity experiments: from pattern toward
mechanism. Page 105120 in S. Naeem, D. E. Bunker, A.
Hector, M. Loreau, and C. Perrings, editors. Biodiversity,
ecosystem functioning, and human wellbeing: an ecological
and economic perspective. Oxford University Press, Oxford,
UK.
Hieber, M., and M. O. Gessner. 2002. Contribution of stream
detrivores, fungi, and bacteria to leaf breakdown based on
biomass estimates. Ecology 83:10261038.
Hoorens, B., R. Aerts, and M. Stroetenga. 2003. Does initial
litter chemistry explain litter mixture effects on decomposition? Oecologia 137:578586.
Lecerf, A., and J. Kominoski. 2010. Functional leaf traits and
biodiversity effects on litter decomposition in a stream:
comment. Ecology 91:18671869.
Petchey, O. L., and K. J. Gaston. 2009. Dendrograms and
measures of functional diversity: a second installment. Oikos
118:11181120.
Petchey, O. L., E. J. OGorman, and D. F. B. Flynn. 2009. A
functional guide to functional diversity. Page 4959 in S.
Naeem, D. E. Bunker, A. Hector, M. Loreau, and C.
Perrings, editors. Biodiversity, ecosystem functioning, and
human wellbeing: an ecological and economic perspective.
Oxford University Press, Oxford, UK.
Petersen, R. C., and K. W. Cummins. 1974. Leaf processing in a
woodland stream. Freshwater Biology 4:343368.
Poos, M. S., S. C. Walker, and D. A. Jackson. 2009.
Functional-diversity indices can be driven by methodological
choices and species richness. Ecology 90:341347.
Schindler, M. H., and M. O. Gessner. 2009. Functional leaf
traits and biodiversity effects on litter decomposition in a
stream. Ecology 90:16411649.
Smith, V. C., and M. A. Bradford. 2003. Do non-additive
effects on decomposition in litter-mix experiments result from
differences in resource quality between litters? Oikos 102:
235242.