Functional Ecology

Title: Multiple abiotic and biotic drivers of long-term wood decomposition within and among species in the
Accepted Article
semiarid inland dunes: a dual role for stem diameter

Enkhmaa Erdenebilega, b
, Congwen Wanga, b, Xuehua Yea, Qingguo Cuia, Juan Dua, Zhenying Huanga, *,

Guofang Liua, *, Johannes H. C. Cornelissenc

Short Title: Drivers of wood decomposition rates

aState Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of

Sciences, Beijing 100093, PR China

bUniversity of Chinese Academy of Sciences, Beijing 100049, China
cSystems Ecology, Department of Ecological Science, VU University Amsterdam, De Boelelaan 1085, 1081 HV

Amsterdam, The Netherlands

enkhmaa_05@yahoo.com,wangcw@ibcas.ac.cn, yexuehua@ibcas.ac.cn, cinkgo@ibcas.ac.cn, du@ibcas.ac.cn

zhenying@ibcas.ac.cn, liugf@ibcas.ac.cn, j.h.c.cornelissen@vu.nl

*Correspondence author: Zhenying Huang, Guofang Liu; Address: State Key Laboratory of Vegetation and

Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China. Phone:

86-10-62836634; Fax: 86-10-62836634; Email: zhenying@ibcas.ac.cn for ZH, liugf@ibcas.ac.cn for GL.

Number of words in the summary: 384

Number of words in the manuscript: 9322

(Introduction: 1364; methods: 1738; results: 661; discussion: 1961; acknowledgements: 103)

Number of references: 66

Number of figures: 4; Number of tables: 2

ACKNOWLEDGMENTS

We thank Jianlin Zhang, Hongmei Mao and Zhaoren Wang for their assistance in the setup of the litter

incubation experiment. We thank the National Field Research Station for Ordos Grassland Ecosystems in Inner

Mongolia of China and the Ordos Sandland Ecological Research Station, CAS for providing the research site

and relevant climate data. This work has been supported by grants of the National Key Research and

Development Program of China (2016YFC0500501, 2018YFE0182800) and the National Science Foundation of

This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process, which may lead to
differences between this version and the Version of Record. Please cite this article as doi:
10.1111/1365-2435.13559
This article is protected by copyright. All rights reserved
 China (NSFC, 31470712) to GL, the CAS President's International Fellowship Initiative (PIFI, 2018VCA0014)
Accepted Article
for JHCC and the CAS-TWAS President's Fellowship for International PhD Students for EE. We also declare

that all coauthors have no conflict of interest.

AUTHORS’ CONTRIBUTIONS

G.L., X. Y. and Z.H. conceived the study; E.E., G.L., Q.C. and J.D. collected the field data; E.E., G.L. and

J.H.C.C. analysed the data and wrote the first draft of the paper; all authors contributed to revisions.

DATA ACCESSIBILITY

Data are available from the Dryad Digital Repository https://doi.org/10.5061/dryad.3tx95x6c2.

This article is protected by copyright. All rights reserved

Accepted Article
DR XUEHUA YE (Orcid ID : 0000-0002-5625-9877)
PROFESSOR ZHENYING HUANG (Orcid ID : 0000-0001-7746-7539)
MR GUOFANG LIU (Orcid ID : 0000-0001-7746-7539)

Article type : Research Article

Editor : Laura Yahdjian
Section : Ecosystems Ecology

Type of article: Full paper

Title: Multiple abiotic and biotic drivers of long-term wood decomposition within and among species in
semiarid inland dunes: a dual role for stem diameter

Abstract
1. Litter decomposition in sunny, semiarid and arid ecosystems is controlled by both biotic factors including
litter traits and abiotic factors including UV light, but for wood decomposition it still remains uncertain which
of these environmental factors are the predominant controls among different woody species. In these dry
ecosystems it is likely that the stem diameter and spatial position of the dead wood are of particular importance
especially where wood can be buried vs. exposed due to substrate displacement by wind. Here we focus on the
fact that stem diameter can affect decomposition rates both via the relative surface exposure to sunlight or soil
and via higher resource quality of narrower stems to decomposers.
2. In a field manipulation experiment, we investigated the relative importance of litter position (sand burial vs.
surface vs. suspended above the surface), UV radiation (block vs pass), and stem diameter class (<2, 2-4, 4-8,
8-13, and 13-20 mm) on the mass loss of woody litters of four shrub species in an inland dune ecosystem in
northern China.
3. We found that after 34 months of in situ incubation, the mass loss of buried woody litters was three times
faster than those of suspended and surface woody litters (53.5 ± 2.7 %, 17.0 ± 1.0 % and 14.4± 1.2 %,
respectively). In surface and suspended positions, litter decomposition rates were almost equally low and most
mass loss was during the first two years, when bark was still attached and UV radiation had no significant
effect on woody litter mass loss. These findings suggest that sand burial is the main environmental driver of
wood decomposition via its control on microbial activity. Moreover, wood N and diameter class were the
predominant factors driving woody litter decomposition. A key finding was that wider stems had slower litter

This article is protected by copyright. All rights reserved

Accepted Article
decomposition rates both directly (presumably via greater relative surface exposure), but also indirectly via
their higher wood dry matter content or lower wood N; these effects were modulated by litter position.
4. Our findings highlight a dual role of stem diameter on wood decomposition, i.e. via relative surface
exposure and via wood traits. The accuracy and confidence of global carbon cycling models would be
improved by incorporating the different effects of stem diameter on woody litter decomposition and
belowground wood decomposition processes in drylands.

KEYWORDS
dryland, functional traits, litter position, wood litter decomposition, sand burial, shrub encroachment, stem
diameter

This article is protected by copyright. All rights reserved

Accepted Article
1. INTRODUCTION
Litter decomposition is a key process in the carbon (C) and nutrient cycles in terrestrial ecosystems (Swift,
Heal, & Anderson, 1979; Berg & McClaugherty, 2008). Generally, woody litter decomposition is mainly
controlled by substrate quality, climate, and decomposing organisms (Swift, Heal, & Anderson, 1979; Harmon
et al., 1986; Cornwell et al., 2009). Wood decomposition within mesic temperate or tropical ecosystems is
strongly modulated by biotic factors such as soil fauna and microbes due to their high abundances and activity
at suitable environmental moisture regime (Torres & González, 2005; González, Gould, Hudak, &
Hollingsworth, 2008; Bradford et al., 2014; Liu et al., 2015a; Ulyshen, 2016). Microbial wood decomposition
is mainly controlled by fungi that are highly specialized in resource requirements (Rayner & Boddy, 1988),
although Wu et al. (2018, 2019) found that the decomposition rate of woody debris in subtropical forests was
accelerated by an abiotic factor, i.e. UV light. By contrast, lower vegetation canopy cover, i.e. high irradiance,
and lower annual precipitation could constrain decomposer abundances and activity. Therefore, there is
accumulating evidence that litter decomposition in semiarid and arid ecosystems is controlled by both biotic
and abiotic factors (Austin & Vivanco, 2006; King, Brandt, & Adair, 2012; Liu et al., 2018). The abiotic
conditions are a consequence of the dynamics of these ecosystems, where litter can either be strongly exposed
to sunlight or get buried by wind-displaced soil material. This has been recognized in recent research on litter
decomposition in sunny drylands, which has highlighted the importance of several abiotic mechanisms related
to light exposure and burial, including photodegradation, photo priming effect and litter-soil mixing (Austin &
Vivanco, 2006; Lin & King, 2014; Hewins & Throop, 2016; Liu et al. 2018). Austin & Ballaré (2010)
suggested that lignin inhibits litter decomposition in mesic ecosystems while it enhances litter decomposition
in arid ecosystems due to photochemical processes. Also, photodegradation of plant litter can reduce the
structural and chemical bottleneck imposed by lignin in secondary cell walls therby increasing microbially
driven litter decomposition (Austin, Méndez, & Ballaré, 2016). These are mechanisms that may explain why
litter decomposition in drylands is faster than expected from biotic mechanisms alone (Throop & Archer,
2009). Indeed, several studies reported that the rate of decomposition in sunny drylands is accelerated by
stronger solar exposure including UV radiation (Austin & Vivanco, 2006; King, Brandt, & Adair, 2012) and
also by burial under displaced soil material (Austin, Araujo, & Leva, 2009; Liu et al., 2015b, Liu et al. 2018)
as compared to decomposition of surface litter.
Changes in plant species have strong effects on biogeochemical processes via plant-soil feedback (Hobbie,
2015) and this must certainly be true for changes in the relative abundance of herbaceous vs. woody plants,
which possibly mediate changes in incubation microclimate (Gottschall et al., 2019). Shrub encroachment, a
common phenomenon in arid and semiarid grasslands, affects soil C and N accumulation and turnover
(Hibbard, Archer, Schimel, & Valentine, 2001). Moreover, there exist many vegetation types in drylands in
which woody plants are common and likely important to biogeochemical cycling, such as desert, inland dunes,
semidesert, woodland, shrubland and savanna. It is therefore surprising that research on woody debris

This article is protected by copyright. All rights reserved

Accepted Article
decomposition in arid and semiarid ecosystems is scarce when compared with non-woody plant material such
as leaves or herbaceous stems. The sparse evidence that has accumulated suggests that, compared to mesic
ecosystems, woody debris decomposes slowly in water-limited ecosystems such as in dry woodland, savanna,
shrubland, semi-desert or desert, even if and where the decomposition rates are increased substantially due to
termite activity (Ebert & Zedler, 1984; Milton & Dean, 1996; Andersen, Stricker, & Nelson, 2016). For
example, Milton and Dean (1996) found that dead shrubs lying on the soil in an arid shrubland had woody
litter mass half-lives of 9 to 18 years for different species. Moreover, Lyons and McCarth (2010) found there
was no difference in Juniperus ashei wood decomposition in open and shaded habitats after 29 months of
incubation. These observations seem a contradiction with the findings for herbaceous litter, which showed
faster than expected decomposition due to sun exposure or burial, as explained above. Here we argue that the
effect of solar radiation on wood decomposition should be weak or significantly decreased due to the lower
surface area to volume ratio of wood, i.e. its relatively low exposure to solar radiation compared to that of
herbaceous litter in arid and semiarid regions. Moreover, bark could provide a powerful protective layer
against photodegradation of the wood beneath. As for leaf litter, there is very scarce knowledge about the
effects of burial on wood decomposition, but one desert study showed that soil movement mediated burial
provides an optimal microclimate with higher moisture and relatively stable temperature for microbial wood
degradation (Moorhead & Reynolds, 1993). In contrast, Moroni et al. (2015) suggested that wood buried by
soil, litter, or ground vegetation in wide-ranging forest ecosystems was efficiently preserved, i.e. slowly
decomposed. However, this review included relatively few dryland sites and our understanding of
decomposition processes of buried wood in drylands is still poor.
While the above litter positions and related decomposition processes have focused on herbaceous vs.
woody litter in general terms, there is now much literature showing that species differ greatly in their traits as
well as in the “afterlife effects” these traits have on decomposition. This is the case both for leaf litter
(Kazakou, Vile, Shipley, Gallet, & Garnier, 2006; Cornwell et al., 2008; Bakker, Carreño-Rocabado, &
Poorter, 2011) and for woody litter (Cornwell et al., 2009; Freschet, Weedon, Aerts, van Hal, & Cornelissen,
2012; Pietsch et al., 2014; Zuo et al., 2018; Hu et al., 2018). For instance, wood nitrogen was the best predictor
of the decomposition rates of woody debris across angiosperm clades based on a global meta-analysis
(Weedon et al., 2009). Also, thicker stems tend to be decomposed more slowly than thinner ones and this
difference has been attributed to the lower surface to volume ratio of the former. Indeed, stem diameter could
explain 41% of the variance of wood decomposition rates of 15 Neotropical tree species (van Geffen, Poorter,
Sass-Klaassen, van Logtestijn, & Cornelissen, 2010). Recently, Hu et al. (2018) found that wood nitrogen and
diameter could explain approximately half of the global variation in wood decomposition rates. Moreover,
wood density or wood dry matter content controlled the rate of wood decomposition in diverse forests
(Freschet, Weedon, Aerts, van Hal, & Cornelissen, 2012; Pietsch et al., 2014; Liu et al., 2015a; Zuo et al.,
2018). However, these studies have not considered the possibility that stem diameter and stem tissue quality

This article is protected by copyright. All rights reserved

Accepted Article
traits (e.g. dry matter content) may be confounded. That is to say, stem diameter might affect decomposition
rates not only via relative exposure, but also because younger, narrower stems might have better resource
quality to decomposers. Additionally, as the outermost covering of wood material, bark plays an important role
in stem protection against insect or pathogen damage (Rosell, 2016). Moreover, bark may have a moisture-
retention function in woody debris and it could increase microbially driven wood decomposition (Dossa,
Paudel, Cao, Schaefer, & Harrison, 2016; Ulyshen, Müller, & Seibold, 2016). Recently, Dossa et al. (2018)
found that bark can enhance coarser WD decomposition but slows twig decomposition in some species. Bark
thickness may affect both microbial invasion and microenvironment (Cornwell et al., 2009). Rosell (2016)
showed that bark thickness is strongly positively correlated with stem sizes. Thus, it is likely that bark
thickness affects woody litter decomposition, directly or via it relation with diameter, but empirical data are
lacking. The same is true for the effect of bark traits on wood decomposition in general (but see Zuo et al.,
2018).
It still remains uncertain to which extent litter position vs. initial litter traits of different species, or their
interactions, predominantly control the decomposition of woody litter in arid and semiarid environments.
Therefore, to fill this research gap, we designed an experimental study to examine the effects of litter position
(sand burial, suspended above the surface, and in a surface litter layer) on the decomposition of woody litters
from four shrub species with different diameters (φ: <2, 2-4, 4-8, 8-13, 13-20 mm) in an inland dune
ecosystem, where the effects of litter position via UV radiation on aboveground wood decomposition were
explicitly examined through experimental UV blocking. Thus, the question is how UV radiation and litter
positions together influence the decomposition of woody litters across different diameters in a semiarid dune
ecosystem? To answer this question, we hypothesized that: (1) buried woody litters would have higher
decomposition rates than surface and suspended woody litters due to stable and favorable moisture regime and
well developed microbial community; (2) UV radiation would increase the decomposition rate of woody litters;
(3) bark thickness would be negatively correlated with the rate of wood decomposition; (4) woody litters of
smaller diameter, with higher relative surface exposure to external microbes and sunlight, and possibly higher
resource quality, should be decomposed faster than wider ones.

2. MATERIALS AND METHODS

2.1 Study site
The field work was conducted at the Ordos Sandland Ecological Research Station (OSERS) in the Mu Us
inland sand dune area, Institute of Botany, Chinese Academy of Sciences (Inner Mongolia, China, 39°29′
37.6″N, 110°11′29.4″E, 1290 m a.s.l.). Site details can be found in Erdenebileg et al. (2018). Briefly, in this
site mean annual temperature is 6.2 °C and mean annual precipitation is 369 mm, 80% of which falls during
the growing season from April to August. Soil texture in this area is aeolian sand. Soil fertility is low with low
N concentration. Since strong wind occurs in winter and early spring, both wind erosion and sand burial are
common processes in this area. The woody vegetation in this region comprises perennial subshrubs (dominated

This article is protected by copyright. All rights reserved

Accepted Article
by Artemisia ordosica Krasch and Hedysarum laeve Maxim.). In order to combat land desertification, two
shrubs, i.e. Salix psammophila C. Wang et Ch. Y. Yang and Caragana korshinskii Kom., have been planted by
local governments for over 20 years in some areas in order to stabilize the dunes, but large areas with strong
sand dune dynamics remain in this region, so burial of dead wood of all four shrubs mentioned is a widespread
phenomenon.

2.2 Litter collection and functional trait measurements

The four shrub species used in this study are A. ordosica, C. korshinskii, H. laeve and S. psammophila. For
each species, recently senesced finer woody litters were collected from 30-50 individual plants from the
beginning of October through the middle of November 2015. For each species we collected stem, branches and
twigs from each of five diameter classes (φ <2, 2-4, 4-8, 8-13, 13-20 mm), where the total length collected per
class was inversely related to diameter in order to obtain broadly equal total mass across the diameter classes
(Figure S1). For measurements of initial undecomposed or decomposed woody litter density that may correlate
to woody litter decomposition rates, initial and decomposed woody litter subsamples (collected at subsequent
harvests; see below) were immersed in water in plastic containers for 4 days to be fully saturated and to ensure
homogeneous filling of air spaces (Freschet, Weedon, Aerts, van Hal, & Cornelissen, 2012; also see wood
water absorption curve based on the largest wood size, Figure S2). Initial and decomposed litter volumes were
measured using the Archimedes’ principle of water displacement (details in Williamson & Wiemann, 2010).
The saturated samples were then gently blotted dry with filter paper and weighed to obtain saturated mass,
subsequently oven-dried at 65 °C for 72 h to obtain dry mass. Wood dry matter content (Wood DMC, g g-1)
was calculated as dry mass divided by its saturated mass and initial or decomposed wood density (g cm-3) was
calculated as litter dry mass divided by saturated litter volume. Because bark could affect wood decomposition
by providing a protective barrier around the wood (Cornwell et al., 2009; Dossa et al., 2018), bark thickness
(mm) and bark dry matter content (Bark DMC, g g-1) were also determined. Bark thickness was measured
using digital calipers. Bark was separated from the initial wood subsamples and then was immersed into water
to be saturated for one day. The saturated bark samples were gently blotted dry with filter paper and weighed to
obtain saturated bark mass, subsequently oven-dried at 65 °C for 24 h to obtain its dry mass.

2.3 Experimental design

In order to assess the influences of UV radiation and litter position (i.e. suspended above the surface to
represent the position of the litter still attached to the shrub; on the soil surface; buried below the sand) on
woody litter decomposition, we used five subplots (two UV treatments × two aboveground positions and one
sand burial) to examine responses of woody litter decomposition to litter position or UV radiation. We adopted
a UV screen treatment design based on the method reported by Brandt, King, Hobbie, Milchunas, &
Sinsabaugh (2010). The detailed description can be seen in Erdenebileg et al. (2018). Briefly, we

This article is protected by copyright. All rights reserved

Accepted Article
experimentally manipulated UV radiation received by the aboveground litter by utilizing two pairs of steel
frames (l × w × h: 440 × 240 × 50 cm) carrying plastics with removable louver sheets that either block or pass
UV radiation. Considering the distinct optical properties, two types of plastic materials were used: UV-
transparent acrylic (UV pass, which transmits approximately 90% of UV-A and UV-B radiation, Zhongshan
Good Life Sun Sheet Co., Ltd., China) and UV-absorbing polycarbonate (UV block, which blocks
approximately 95% of UV radiation, Zhongshan Good Life Sun Sheet Co., Ltd., Guangzhou, China). Screens
were custom-designed and constructed in a louvered design to allow for penetration of rainfall to the litter
layer, to allow free air movement and avoid heating. A logger (iButton) was laid in each incubation site to
automatically record temperature and relative humidity at intervals of two hours (DS1923, Maxim, USA). We
acknowledge that the readers on relative humidity obtained from loggers could not represent gravimetric
moisture content, but they could still be a valid proxy of gravimetric moisture content based a microcosm-
based experiment conducted by Wang, Throop, & Gill (2015).
We used the litterbag approach to capture the mass loss of woody litter of the four species for five
diameter classes at different harvests, i.e. 6, 12, 18, 24 and 34 months (from December 2015 through
September 2018) by sealing more than 5 g of pre-weighed air-dried woody litter into each litterbag. This
amount depends on the actual situation because woody litters are quite distinct from leaf litters. That is to say,
we cut shrub wood into fixed 10 cm long segments before sealing litters into bags. Three subsamples for each
species were used to correct air-dried litter mass via initial air-dried and oven-dried mass (at 65°C for 48 h).
The size of the litterbags was 15 cm × 20 cm with a mesh size of 2-mm for the top side exposed to the sun, and
a 0.5-mm mesh for the bottom side in order to prevent loss of small woody litter fragments (Baker & Allison,
2015). We acknowledge that the commonly used litter bag method might cause small change in
microenvironment, which could affect wood decomposition to a small extent. The number of replicates was 5.
Therefore, the number of litterbags used was 2500 (4 species × 5 diameter classes × 5 treatments × 5 harvest
times × 5 replicates). The top 10-cm layer of soil had first been cleared of plant material and removed before
the litterbags were laid out, after which the soil layer was mixed and put back on, (see Figure S3a,b). The 500
litterbags were placed in each of the 5 subplots either in suspended position supported by a fishing net at 10 cm
distance below the UV screen (see Figure S3c), on the soil surface (see Figure S3d) or buried in the sandy soil
at 10 cm depth. We retrieved litterbags after 5 periods of incubation (see above), cleaned the samples in the
laboratory by brushing, oven-dried them at 65°C for 72 h to obtain dry masses.

2.4 Data analysis

All statistical tests except for the structural equation modeling (SEM) were carried out in R (v3.3.1, R Core
Team 2016). Bark thickness, N concentration and C: N ratio of initial wood litter and k values for litter
decomposition were Log10-transformed to satisfy the assumptions of normal distribution. A three-way
repeated-measures ANOVA with incubation period as a repeated factor was used to examine the effects of
species, diameter class, treatment (sand burial and combinations of UV radiation and litter position) and all

This article is protected by copyright. All rights reserved

Accepted Article
their interactions on the woody litter mass loss. In order to disentangle the influence of UV radiation on wood
decomposition, a four-way repeated-measures ANOVA was used to examine the effects of species, diameter
class, UV radiation, litter position (suspended and soil surface), and all their interactions on wood mass loss;
the sand burial treatment was excluded from this analysis. For each of the 5 harvests, Tukey HSD tests with
Bonferroni correction were used to examine the differences in wood mass loss between the three positions, the
differences between UV treatments for each of surface and suspended positions and the differences between
surface and suspended positions for each of the UV treatments. Tukey HSD tests with Bonferroni correction
were also used to examine the differences in bark and wood traits among diameter classes for each species and
among species for each diameter class. We fitted six litter decomposition trajectories reported by Cornwell &
Weedon (2014) via the litterfitter package (https://github.com/cornwell-lab-unsw/litterfitter) on our wood litter
decomposition data, most of which obeyed a negative exponential model of fractional mass loss through time.
Thus, we carried out nonlinear regressions using the ‘nls’ function to calculate the k values (yr-1) based on
Olson’s (1963) negative exponential model (Yt = Y0 e-kt, where t is incubation time and Y0 and Yt represent
mass remaining at incubation times 0 and t, respectively) across all incubation times and replicates for each
combination of diameter class, species and treatment. Pearson correlation analysis was used to test the
relationships between the respective bark and wood traits and between these traits and k values of woody litter
decomposition. We used multi-model inference to determine how changes in rates of wood decomposition
were related to influential bark and wood traits. We quantified the effects of all possible combinations of
predictors on k values using the Dredge function of MUMIn package (Bartoń, 2009). Prior to running full
models, we standardized the predictor variables using the standardize function in the arm package, which is
essential for interpreting parameter estimates (Grueber et al., 2011). We excluded models with highly
correlated predictor variables (|r|>0.5) to reduce any collinearity problem among explanatory variables, so that
the candidate models only contained one of a pair of highly correlated variables (Dormann et al., 2013). The
best candidate models were ordered based on AICc and we used the model averaging procedure on top order
models with ΔAICc ≤ 4 criteria (Anderson & Burnham, 2002). In the multi-model inference, we transformed
the five diameter classes into a continuous predictor variable using the simple mean of each range (also see
Abbott & Crossley, 1982). There were strong correlations on the k values for wood decomposition between UV
block and UV pass on surface or suspended location (r = 0.93, 96, both P <0.001, respectively). Therefore, the
k values were averaged across surface or suspended location in SEM. Using the SEM, we tested how wood
decomposition rates among three positions were affected by diameter directly and indirectly, i.e. via changing
wood N, DMC and bark thickness, which we considered to be plausible based on possible causal relationships.
The SEM was fitted using IBM SPSS Amos 21 (Amos Development Corporation, Chicago, IL, USA). The
significance level was set at P <0.05 except for the χ2 test of model fit in the SEM at P > 0.05.

This article is protected by copyright. All rights reserved

Accepted Article
3. RESULTS
3.1 Initial wood litter traits
The morphological and chemical traits of initial wood litter differed among the four species and among five
diameter classes (Figure 1). Overall, there was remarkable covariation of several wood and bark traits with
diameter class. The initial bark thickness and wood DMC increased with increasing diameter class especially
for A. ordosica; initial wood DMC was significantly positively correlated to bark thickness or wood density
(Figure 1a, c, d; Table S1). A. ordosica had lower bark DMC than the other species for most of the diameter
classes (Figure 1b). Initial wood N decreased with increasing diameter class for most species (Figure 1e). The
woody litter of C. korshinskii and H. laeve had higher N concentration of initial litter than the two other
species for each of the five diameter classes (Figure 1e, f). Initial wood C: N ratio was a proxy for initial wood
N concentration due to the stable C concentration (r =1.00, P < 0.001; Table S1).

3.2 Rates of wood litter decomposition

The three-way repeated-measures ANOVA showed that position/UV treatment (F(3, 400) = 2503, P < 0.001),
diameter class (F(4, 400) = 548, P < 0.001), species (F(3, 400) = 87.6, P < 0.001), and all their interactions had
significant effects on woody litter mass loss (Table 1). However, when only aboveground positions (surface
and suspended) were considered in the four-way repeated-measures ANOVA, UV radiation did not have a
main effect on woody litter mass loss and the interaction between UV and position was marginally significant
(Table S2). UV radiation significantly affected mass loss of surface woody litters, but it did not change mass
loss of suspended woody litters (Table S3). As we expected, after 12, 24 and 34 months of incubation, the
mass losses of buried woody litters were higher than those in the aboveground treatments; they ranged 12.7–
47.3 %, 25.7–61.5 % and 31.9–73.9 % with averages of 25.7 ± 2.1 %, 43.0 ± 2.6 % and 53.5 ± 2.7 %,
respectively (Figure 2). By contrast, after 12, 24 and 34 months of incubation, the mass losses under surface
and suspended conditions were 2–15 %, 6-23 %, 7–27 % and 5-13 %, 11-23 %, 12–27 %, respectively (Figure
2; Table S4). During the whole incubation, litter mass loss of suspended woody litter was significantly higher
than that on the surface (17.0 ± 1.0 % vs. 14.4 ± 1.2 %, respectively; Table S2). Overall, the mass loss of
buried woody litters was three times faster than those of suspended and surface woody litters (Figure 2).

3.3 Correlations between bark and wood traits and decomposition rates
The k value for woody litter decomposition was higher towards smaller diameter classes, particularly in the
sand burial treatment (Figure 3). Pearson’s correlation analyses indicated that k values were strongly
negatively correlated with initial bark thickness, wood dry matter content (wood DMC), wood C: N ratio and
diameter class in all treatments; the k values were significantly negatively correlated with initial wood density
on the surface and in the sand burial treatment but not in the suspended treatment (all P < 0.05) (Table S5).
The k values were significantly positively correlated with initial N concentration in all treatments (all P < 0.05;
Table S5). The bark DMC did not exhibit significant correlation with decomposition rate for any of the

This article is protected by copyright. All rights reserved

Accepted Article
treatments (Table S5). The structural equation modelling showed that diameter, initial wood DMC and N had
direct effects on k values, among which diameter had the stronger influence on decomposition rates of buried
wood (Figure 4). Diameter also had a strong indirect effect on k values by affecting initial wood DMC directly
or indirectly via wood density in the surface treatment. There was a marginally significant positive trend of
diameter on k values via wood N in all the treatments (P = 0.069; Figure 4). The full model analyses showed
higher importance of initial N concentration (or C: N ratio) and diameter class in explaining variation in the k
values for most treatments; for the surface litter under UV blocking, the higher importance was owed to initial
wood DMC and N concentration (or C: N ratio) (Table 2). Generally, the density of decomposed wood litter
was negatively correlated with wood mass loss, in which the stronger relationships were observed in sand
burial than in the other two positions (Figure S4).

4. DISCUSSION
Here, we assessed the interactive effects of woody litter diameter, position (sand burial vs. surface vs.
suspended) and UV exposure on decomposition rate in a semiarid inland dune ecosystem across four shrub
species over almost three years. Among the various abiotic and biotic drivers addressed here we found that
woody litter position (suspended, surface, buried), stem diameter and wood traits, and the interactions between
these factors, all had substantial effects on decomposition rates, while differences in exposure to UV radiation
did not affect decomposition rates significantly. A key finding was that stem diameter not only had direct
effects on decomposition rate, presumably via surface area to volume ratio, but also indirect effects via
differences in wood traits, especially wood dry matter content and wood N. Below we will discuss these
findings in more detail in the context of previous literature.

4.1 Environmental rivers of woody litter decomposition

In terrestrial ecosystems, differences in litter decomposition during different periods of incubation are
associated to the seasonal changes in solar radiation, temperature, and moisture (Lin, Scarlett, & King, 2015;
Wang, Liu, Wang, & Chen, 2015). We found that relatively fast rates of wood decomposition occurred in the
subsequent summers (Figure 2; Table S4), when environments conditions benefitted wood decomposition
(Figure S5). As hypothesized (1), buried woody litter decomposition was faster in the entire decomposition
period due to relatively stable and higher moisture as compared with those aboveground (Figure 2; Table S4).
As expected, we found that the mass losses of buried woody litters during 34 months of incubation were three
times faster than those of suspended and surface woody litters (Figure 2). This supports the reported pattern
that the difference between the rate of buried and surface wood decomposition increased with decreasing
precipitation (Smyth et al., 2016). The accelerating effect of sand burial on wood decomposition is even
stronger compared with previous studies in similar dry habitats, in which litter burial less than doubled the rate
of litter decomposition compared to that on the soil surface (Santos, Elkins, Steinberger, & Whitford, 1984;
Vivanco & Austin, 2006; Austin, Araujo, & Leva, 2009; Liu et al., 2015b). Also, this finding contrasts with the

This article is protected by copyright. All rights reserved

Accepted Article
fact that buried wood had slower decomposition rate than that on the soil surface in diverse forest ecosytems
(Romero, Smith, & Fourqurean, 2005; Fasth, Harmon, Sexton, & White, 2011). The likely reason is that the
sand burial exposes dead wood to an abundant microbial community and to relatively beneficial humidity to
microbially driven decomposition. The previous work on soil enzymatic activity in the Mu Us inland dunes
showed larger values for polyphenoloxidase, peroxidase and polyphenol oxidase at 0-5 cm of soil depth than
those at 5-10 cm of soil depth in the S. psammophila community (45.00 ± 3.07 vs. 24.24 ± 0.68 ug g-1 h-1,
730.85 ± 24.95 vs. 354.42 ± 69.35 ug g-1 h-1, 318.14 ± 35.2 vs. 153.01 ± 11.3 ug g-1 h-1, respectively; Wang,
2018). Indeed, the fungal/bacterial biomass ratios based on the phospholipid fatty acid (PLFA) technique
nearby the incubation site were 0.378 ± 0.015, 0.381 ± 0.018, 0.398 ± 0.019 and 0.382 ± 0.018 at 0-5, 5-10,
10-15 and 15-20 cm of soil depth, respectively (unpublished data), which indicates that this dry region hosts
relatively high fungal abundance (Chen et al., 2014). According to High Throughput DNA Sequencing
methodology, the dominant fungal species in the top soil (0-10 cm) of Artemisia ordosia vegetation on Ordos
plateau were Ascomycota (64.6%), Mortierellomycota (12.3%) and Basidiomycota (8.4%), accounting for
85.3% of total number of OTUs (unpublished data). Since especially Ascomycota are predominantly
saprophytic, these complementary observations imply that relatively faster buried wood decompositon could
be owed to fungal activity. There was no visible evidence of termites in our study (observations inside and
outside litterbags by the authors), contrasting to strong termite effects on wood decomposition in hot drylands
(Buxton, 1981). In forests, slow decomposition of dead wood burial has been associated with exposure to low
temperatures in peat or after being overgrown by mosses. These differences emphasize the different driving
mechanisms of litter decomposition in different biomes.
In contrast with the fast rate of buried wood decomposition, aboveground decomposition rates were very
low in the soil surface and suspended position with exposure to solar radiation; most mass loss in these
treatments occurred during the first two years (Table S4). This relatively slow decomposition is consistent with
previous studies (Buxton, 1981; Abbott & Crossley, 1982; Milton & Dean, 1996) and can explain why almost
20 % of organic carbon is owed to standing and surface litters in inland dune ecosystems (Li, 2006). We
acknowledge that the litter bags could intercept a small portion of solar radiation, so in our experiment we may
have slightly underestimated the effect of photodegradation on wood mass loss. Longer-term studies are
needed to investigate the trajectory of aboveground wood decomposition more comprehensively (Prescott
2005; Cornelissen et al., 2012). Many studies have shown that woody plant encroachment increases fluxes of
soil C and N (McCulley, Archer, Boutton, Hons, & Zuberer, 2004; Hibbard, Archer, Schimel, & Valentine,
2001). This present study highlights that relatively fast decomposition of buried woody litter may be an
important contributor to this phenomenon.
Within the aboveground positions, woody litter decomposition rates were somewhat faster in suspended
litter than in surface litter (Figure 2; Table S4), consistent with previous studies (Lin & King, 2014;
Erdenebileg et al., 2018). In dryland, decomposition of suspended litters is likely strongly determined by
abiotic factors, such as photodegradation and wind abrasion, but also by a favorable combination of

This article is protected by copyright. All rights reserved

Accepted Article
temperature or moisture at night promoting microbial decomposition (Wang et al., 2017). However this
conflicts with the finding that termite-mediated decomposition rate of ground-placed wood debris was over 4
times greater than that in suspended condition in tropical rainforest (Law, Eggleton, Griffiths, Ashton, & Parr,
2019). These findings imply that contrasting influential factors determined the carbon turnover between
ecosystems varying in water limitation and soil fanua abundance. As revealed by our UV blocking treatments,
UV radiation alone had no significant effect on woody litter decomposition (Table S2), which is inconsistent
with our hypothesis (2); this hypothesis had been informed by the findings of Erdenebileg et al. (2018), who
found that UV radiation had a significant effect on leaf litter decomposition. The discrepancy between leaf
litter and woody litter in response to UV light may be due to the contrasting architecture between leaf and
wood tissues. UV radiation did not affect suspended wood decomposition but it decreased surface wood
decomposition, suggesting that UV radiation possibly decreases the activity of colonizing microbes (Wang,
Liu, Wang, & Chen, 2015). This study reinforces how different plant tissues decay in very different ways and
how these differences also depend on their vertical position.
Soil-litter mixing and photodegradation are mechanisms that may explain why higher than expected non-
woody litter decomposition occurred in drylands (Austin & Vivanco, 2006; Hewins & Throop, 2016; Liu et al.
2018). In addition, Liu et al. (2018) found that relative exposure of the leaf surface to sunlight per unit leaf
mass (i.e., specific leaf area) could be a predominant factor in explaining variation in leaf litter decomposition.
As noted above, together these suggest that sand burial is a main environmental factor in driving woody litter
decomposition in shrub-dominated drylands.

4.2 Controls of initial litter quality on woody litter decomposition

Litter decomposition rates depend on both incubation environment and the afterlife legacy of functional traits
of different species through litter substrate quality (Swift, Heal, & Anderson, 1979; Harmon et al., 1986;
Cornwell et al., 2009). We found that the rates of woody litter decomposition increased with increasing initial
wood N concentration or decreasing initial C: N ratio in all treatments; initial wood N concentration and C: N
ratio were highly correlated due to their log-transformation and to the stable C concentration. Our observations
confirm previous reports that litter with higher nutritional quality (e.g. higher N concentration and lower C: N
ratio) can decompose relatively fast (Weedon et al., 2009; van Geffen, Poorter, Sass-Klaassen, van Logtestijn,
& Cornelissen, 2010; Liu et al., 2015a). Wood dry matter content (wood DMC), as a key trait representing
internal litter structure, is known to be a negative predictor of wood decomposition rate (Freschet et al., 2012;
Liu et al., 2015b) and our findings add to this body of evidence. However, a novel finding of our work,
revealed by structural equation modeling, was that variation in wood DMC was one of the two different
mechanisms by which stem diameter strongly controlled wood decomposition, with thinner stems having
lower initial DMC or higher N and faster decomposition. The other mechanism in our study, consistent with
our hypothesis (4) based on previous literature, was that the lower wood surface area to volume ratio for larger
diameter stems likely lowered the rate of colonization by microbial decomposer communities (Cornwell et al.,

This article is protected by copyright. All rights reserved

Accepted Article
2009), thereby reducing their decompositon rate (van Geffen, Poorter, Sass-Klaassen, van Logtestijn, &
Cornelissen, 2010; Hu et al., 2018). We acknowledge that fungal colonization can be facilitated through cuts
and wounds. In wider stems, the possibility of colonizing the stem through the cut ends may be relatively
higher than in thinner stems. Thus, the effects of stem diameter on wood decomposition would be expected to
be stronger if the cut surfaces would be sealed in order to prevent fungal colonization. However, this is
inconsistent with the finding that half-buried stems decomposed relatively faster than buried stems in
temperate deciduous forest with 1030 mm of annual precipitation (Oberle et al., 2018), emphasizing that burial
effects on woody litter decomposition is dependent on the precipitation and associated soil mositure of
different ecosystems. The relationships between wood density and the rate of decomposition are contrasting
between different studies. For instance, Liu et al. (2015a) showed a negative relationship in a rainforest
ecosystem. By contrast, wood density could not explain variation in rates of decomposition at local or global
scales in some other studies (Weedon et al., 2009; van Geffen, Poorter, Sass-Klaassen, van Logtestijn, &
Cornelissen, 2010). Probably the negative relationship only features in those cases where wood density is
strongly correlated with DMC, as the volume basis at which mass is expressed in wood density can at best only
be indirectly involved in dry matter decomposition. The model averaging procedure and structural equation
modeling demonstrated that initial wood N (or wood C: N ratio) and diameter class were predominant and
mostly independent controls on wood decomposition rates. This result at a local scale is consistent with the
finding at gobal scale reported by Hu et al. (2018). These findings together confirm the importance of initial
wood N (or wood C: N ratio) and diameter as key wood traits in predicting wood decomposition rates.
Different from previous literature, however, we revealed a strong indirect effect of stem diameter on wood
decomposition via wood DMC and N, an effect that strongly depended on litter position. Indeed, the dual role
of stem diameter on decompositon of woody litter was particularly strong under buried condition and
incorporating this role may help to improve the accuracy and confidence of modeling global carbon cycling in
drylands.
The bark is of considerable importance as an environment filter for community assembly through habitat
access and provision during the early stage of wood decomposition (Zuo et al., 2016). Previous studies showed
mixed effects of bark thickness on the abundance of decomposers, for instance negative effects on bark beetles
in spruce (Wainhouse, Cross, & Howell, 1990) and positive effects, probably via bark fissuredness on wide-
ranging macro-invertebrate populations across 11 temperate tree species (Zuo et al., 2016). In our study, with
very small stem diameters, invertebrates did not play any significant role in decomposition. In SEM, bark
thickness did not regulate woody litter decomposition although it was signficantly negatively correlated with
rates of wood decomposition due to its tight relationship with diameter (Table S5; Figure 4). By contrast, any
direct effect of bark on wood decomposition aboveground was not apparent. Together our findings, which
were inconsistent with our hypothesis (3), suggest that the effects of bark on wood decomposition rates are
distinct between mesic and dry ecosystems due to contrasting moisture regimes. Additionally, the recalcitrant

This article is protected by copyright. All rights reserved

Accepted Article
outer bark might shelter the softer inner bark and wood below from mainly abiotic UV degradation in
drylands. Future studies on the effects of bark on wood organic turnover should therefore focus on how such
effects vary in importance and mechanisms across diverse terrestrial ecosystems with contrasting precipitation
and irradiance regimes. For instance, bark removal experiments are needed to disentangle the effects of bark
protection on wood decomposition on the soil surface or suspended above it.

5. Conclusion
Over 34 months of incubation, the decomposition rates of buried woody litters were three times faster than
those of aboveground woody litters of the same four shrub species in an inland dune ecosystem. UV had no
significant influence on the rates of wood decomposition, suggesting that the importance of UV radiation for
wood decomposition is negligible compared to that for leaf litters. Among several litter traits previously linked
to litter quality, initial wood N concentration (or C: N ratio) and diameter class were predominant factors
driving wood decomposition. A key finding (see hypothesis 4) was that stem diameter played a dual role in
determining wood decomposition: directly via surface area to volume ratio and indirectly via its link with
wood dry matter content and wood N. This dual role of diameter on wood decomposition and, thereby, as a
promotor of belowground organic fluxes, may be of particular importance in grasslands subject to shrub
encroachment and other shrub-dominated semiarid and arid ecosystems. The accuracy and confidence of
global carbon cycling model would be improved by incorporating the different effects of stem diameter on
woody litter decomposition and belowground wood decomposition processes in drylands.

REFERENCES
Abbott, D.T. & Crossley Jr, D.A. (1982) Woody litter decomposition following clear-cutting. Ecology, 63, 35–
42.
Andersen, D.C., Stricker, C.A. & Nelson, S.M. (2016) Wood decay in desert riverine environments. Forest
Ecology & Management, 365, 83–95.
Anderson, D.R. & Burnham, K.P. (2002) Avoiding pitfalls when using information-theoretic methods. The
Journal of Wildlife Management, 66, 912–918.
Austin, A.T. & Vivanco, L. (2006) Plant litter decomposition in a semi-arid ecosystem controlled by
photodegradation. Nature, 442, 555–558.
Austin, A.T., Araujo, P.I. & Leva, P.E. (2009) Interaction of position, litter type, and water pulses on the
decomposition of grasses from the semiarid Patagonian steppe. Ecology, 90, 2642–2647.
Austin, A.T., Ballaré, C.L. (2010) Dual role of lignin in plant litter decomposition in terrestrial ecosystems.
Proceedings of the National Academy of Sciences of the United States of America, 107, 4618–4622.
Austin, A.T., Méndez M S, Ballaré C L. Photodegradation alleviates the lignin bottleneck for carbon turnover
in terrestrial ecosystems. Proceedings of the National Academy of Sciences of the United States of
America, 113, 4392–4397.

This article is protected by copyright. All rights reserved

Accepted Article
Baker, N.R. & Allison, S.D. (2015) Ultraviolet photodegradation facilitates microbial litter decomposition in a
Mediterranean climate. Ecology, 96, 1194–2003.
Bakker, M.A., Carreño-Rocabado, G. & Poorter, L. (2011) Leaf economics traits predict litter decomposition
of tropical plants and differ among land use types. Functional Ecology, 25, 473–483.
Bartoń, K. (2009) MuMIn: multi-model inference, R package version 0.12. 0. http://r-forge. r-project.
org/projects/mumin/.
Berg, B. & McClaugherty, C. (2008) Plant litter: decomposition, humus formation, carbon sequestration.
Berlin: Springer.
Bradford, M.A., Warren II, R.J., Baldrian, P., Crowther, T.W., Maynard, D.S., Oldfield, E.E. & King, J.R.
(2014) Climate fails to predict wood decomposition at regional scales. Nature Climate Change, 4, 625–
630.
Brandt, L.A., King, J.Y., Hobbie, S.E., Milchunas, D.G. & Sinsabaugh, R.L. (2010) The role of
photodegradation in surface litter decomposition across a grassland ecosystem precipitation
gradient. Ecosystems, 13, 765–781.
Buxton, R.D. (1981) Termites and the turnover of dead wood in an arid tropical environment. Oecologia, 51,
379–384.
Chen, D.M., Mi, J., Chu, P.F., Cheng, J.H., Zhang, L.X., Pan, Q.M., Xie, Y.C. & Bai, Y.F. (2015) Patterns and
drivers of soil microbial communities along a precipitation gradient on the Mongolian Plateau.
Landscape Ecology, 30, 1669–1682.
Cornelissen, J.H.C., Sass-Klaassen, U., Poorter, L., van Geffen, K., van Logtestijn, R.S., van Hal, J.,
Goudzwaard, L., Sterck, F.J., Klaassen, R.K.W.M., Freschet, G.T., van der Wal, A., Eshuis, H., Zuo, J.,
de Boer, W., Lamers, T., Weemstra, M., Cretin, V., Martin, R., den Ouden, J., Berg, M.P., Aerts, R.,
Mohren, G.M.J. & Hefting, M.M. (2012). Controls on coarse wood decay in temperate tree species:
birth of the LOGLIFE experiment. Ambio, 41, 231–245.
Cornwell, W.K., Cornelissen, J.H.C., Amatangelo, K., Dorrepaal, E., Eviner, V.T., Godoy, O., Hobbie, S.E.,
Hoorens, B., Kurokawa, H., Pérez-Harguindeguy, N., Quested, H.M., Santiago, L.S., Wardle, D.A.,
Wright, I.J., Aerts, R., Allison, S.D., van Bodegom, P., Brovkin, V., Chatain, A., Callaghan, T.V., Díaz,
S., Garnier, E., Gurvich, D.E., Kazakou, E., Klein, J.A., Read, J., Reich, P.B., Soudzilovskaia, N.A.,
Vaieretti, M.V. & Westoby, M. (2008). Plant species traits are the predominant control on litter
decomposition rates within biomes worldwide. Ecology Letters, 11, 1065–1071.
Cornwell, W.K., Cornelissen, J.H.C., Allison, S.D., Bauhus, J., Eggleton, P., Preston, C.M., Scarff, F.,
Weedon, J.T., Wirth, C. & Zanne, A.E. (2009) Plant traits and wood fates across the globe: rotted,
burned, or consumed? Global Change Biology, 15, 2431–2449.
Cornwell, W.K., Weedon, J.T. 2014. Decomposition trajectories of diverse litter types: a model selection
analysis. Methods in Ecology and Evolution, 5, 173–182.

This article is protected by copyright. All rights reserved

Accepted Article
Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G. & Münkemüller, T. (2013)
Collinearity: a review of methods to deal with it and a simulation study evaluating their performance.
Ecography, 36, 27–46.
Dossa, G.G., Paudel, E., Cao, K., Schaefer, D. & Harrison, R.D. (2016) Factors controlling bark
decomposition and its role in wood decomposition in five tropical tree species. Scientific Reports, 6,
34153.
Dossa, G.G.O., Schaefer, D., Zhang, J.L., Tao, J.P., Cao, K.F., Corlett, R.T., Cunningham, A.B., Xu, J.C.,
Cornelissen, J.H.C. & Harrison, R.D. (2018) The cover uncovered: Bark control over wood
decomposition. Journal of Ecology, 106, 2147–2160.
Ebert, T.A. & Zedler, P.H. (1984) Decomposition of ocotillo (Fouquieria splendens) wood in the Colorado
Desert of California. American Midland Naturalist, 111, 143–147.
Erdenebileg, E., Ye, X.H., Wang, C.W., Huang, Z.Y., Liu, G.F. & Cornelissen, J.H.C. (2018) Positive and
negative effects of UV irradiance explain the interaction of litter position and UV exposure on litter
decomposition and nutrient dynamics in a semi-arid dune ecosystem. Soil Biology & Biochemistry, 124,
245–254.
Fasth, B.G., Harmon, M.E., Sexton, J. & White, P. (2011). Decomposition of fine woody debris in a deciduous
forest in North Carolina. The Journal of the Torrey Botanical Society, 138, 192–207.
Freschet, G.T., Weedon, J.T., Aerts, R., van Hal, J.R. & Cornelissen, J.H.C. (2012) Interspecific differences in
wood decay rates: insights from a new short-term method to study long-term wood decomposition.
Journal of Ecology, 100, 161–170.
González, G., Gould, W.A., Hudak, A.T. & Hollingsworth, T.N. (2008) Decay of aspen (Populus tremuloides
Michx.) wood in moist and dry boreal, temperate, and tropical forest fragments. Ambio, 37, 588–597.
Gottschall, F., Davids, S., Newiger-Dous, T.E., Auge, H., Cesarz, S. & Eisenhauer, N. (2019) Tree species
identity determines wood decomposition via microclimatic effects. Ecology and Evolution, 9, 12113–
12127.
Grueber, C.E., Nakagawa, S., Laws, R.J. & Jamieson, I.G. (2011) Multimodel inference in ecology and
evolution: challenges and solutions. Journal of Evolutionary Biology, 24, 699–711.
Harmon, M.E., Franklin, J.F., Swanson, F.J., Sollins, P., Gregory, S.V., Lattin, J.D., Anderson, N.H., Cline,
S.P., Aumen, N.G., Sedell, J.R., Lienkaemper, G.W., Cromack, K.Jr. & Cummins, K.W. (1986)
Ecology of coarse woody debris in temperate ecosystems. Advances in Ecological Research, 15, 133–
302.
Hewins, D.B. & Throop, H.L. (2016) Leaf litter decomposition is rapidly enhanced by the co-occurrence of
monsoon rainfall and soil-litter mixing across a gradient of coppice dune development in the
Chihuahuan Desert. Journal of Arid Environments, 129, 111–118.
Hibbard, K.A., Archer, S., Schimel, D.S. & Valentine, D.W. (2001) Biogeochemical changes accompanying
woody plant encroachment in a subtropical savanna. Ecology, 82, 1999–2011.

This article is protected by copyright. All rights reserved

Accepted Article
Hobbie, S. E. (2015). Plant species effects on nutrient cycling: revisiting litter feedbacks. Trends in Ecology &
Evolution, 30, 357–363.
Hu, Z.H., Michaletz, S.T., Johnson, D.J., McDowell, N.G., Huang, Z.Q., Zhou, X.H., & Xu, C.G. (2018).
Traits drive global wood decomposition rates more than climate. Global Change Biology, 24, 5259–
5269.
Kazakou, E., Vile, D., Shipley, B., Gallet, C., & Garnier, E. (2006). Co‐variations in litter decomposition,
leaf traits and plant growth in species from a Mediterranean old‐field succession. Functional Ecology,
20, 21–30.
King, J., Brandt, L. & Adair, E. (2012) Shedding light on plant litter decomposition: Advances, implications
and new directions in understanding the role of photodegradation. Biogeochemistry, 111, 57–81.
Law, S., Eggleton, P., Griffiths, H., Ashton, L. & Parr C. (2019) Suspended dead wood decomposes slowly in
the tropics, with microbial decay greater than termite decay. Ecosystems, 22, 1176–1188.
Li, C.P. (2006) Carbon dynamics of Artemisia ordosica Krasch. shrublands under different habitat conditions
in Mu Us Sandland, Inner Mongolia [In Chinese with English abstract]. Ph.D thesis. Institute of Botany,
Chinese Academy of Sciences.
Lin, Y. & King, J.Y. (2014) Effects of UV exposure and litter position on decomposition in a California
grassland. Ecosystems, 17, 158–168.
Lin, Y., Scarlett, R.D. & King, J.Y. (2015) Effects of UV photodegradation on subsequent microbial
decomposition of Bromus diandrus litter. Plant and Soil, 395, 263–271.
Liu, G.F., Cornwell, W.K., Cao, K., Hu, Y., Van Logtestijn, R.S.P., Yang, S., Xie, X., Zhang, Y., Ye, D., Pan,
X., Ye, X., Huang, Z., Dong, M. & Cornelissen, J.H.C. (2015a) Termites amplify the effects of wood
traits on decomposition rates among multiple bamboos and dicot woody species. Journal of Ecology,
103, 1214–1223.
Liu, G.F., Cornwell, W.K., Pan, X., Ye, D., Liu, F., Huang, Z., Dong, M. & Cornelissen, J.H.C. (2015b)
Decomposition of 51 semidesert species from wide-ranging phylogeny is faster in standing and sand-
buried than in surface leaf litters: implications for carbon and nutrient dynamics. Plant and Soil, 396,
175–187.
Liu, G.F., Wang, L., Jiang, L., Pan, X., Huang, Z.Y., Dong, M. & Cornelissen, J.H.C. (2018) Specific leaf area
predicts dryland litter decomposition via two mechanisms. Journal of Ecology, 106, 218–229.
Lyons, K.G. & McCarthy, W.A. (2010) Early decomposition of Ashe juniper (Juniperus ashei) wood in open
and shaded habitat. Rangeland Ecology & Management, 63, 359–365.
McCulley, R.L., Archer, S.R., Boutton, T.W., Hons, F.M. & Zuberer, D.A. (2004) Soil respiration and nutrient
cycling in wooded communities developing in grassland. Ecology, 85, 2804–2817.
Milton, S.J. & Dean, W.R.J. (1996) Rates of wood and dung disintegration in arid South African rangelands.
African Journal of Range & Forage Science, 13, 89–93.

This article is protected by copyright. All rights reserved

Accepted Article
Moroni, M.T., Morris, D.M., Shaw, C., Stokland, J.N., Harmon, M.E., Fenton, N.J. & Hagemann, U. (2015)
Buried wood: a common yet poorly documented form of deadwood. Ecosystems, 18, 605–628.
Moorhead, D.L. & Reynolds, J.F. (1993) Changing carbon chemistry of buried creosote bush litter during
decomposition in the Northern Chihuahuan Desert. American Midland Naturalist, 130, 83–89.
Oberle, B., Covey, K.R., Dunham, K.M., Hernandez, E.J., Walton, M.L., Young, D.F., Zanne, A.E. (2018)
Dissecting the effects of diameter on wood decay emphasizes the importance of cross-stem conductivity
in Fraxinus americana. Ecosystems, 21, 85–97.
Olson, J.S. (1963) Energy storage and the balance of producers and decomposers in ecological systems.
Ecology, 44, 322–331.
Pietsch, K.A., Ogle, K., Cornelissen, J.H.C., Cornwell, W.K., Bönisch, G., Craine, J.M., Jackson, B.G., Kattge,
J., Pelter, D.A., Penuelas, J., Reich, P.B., Wardle, D.A., Weedon, J.T., Wright, I.J., Zanne, A.E., Wirth,
C. (2014) Global relationship of wood and leaf litter decomposability: the role of functional traits within
and across plant organs. Global Ecology & Biogeography, 23, 1046–1057.
Prescott, C.E. (2005) Do rates of litter decomposition tell us anything we really need to know? Forest Ecology
& Management, 220, 66–74.
R Core Team. (2016) R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. URL https://www.R-project.org/.
Rayner, A.D.M., & Boddy, L. (1988) Fungal decomposition of wood. Its biology and ecology. John Wiley &
Sons Ltd.
Romero, L.M., Smith, T.J. & Fourqurean, J. W. (2005) Changes in mass and nutrient content of wood during
decomposition in a south Florida mangrove forest. Journal of Ecology, 93, 618-631.
Rosell, J.A. (2016) Bark thickness across the angiosperms: more than just fire. New Phytologist, 211, 90–102.
Santos, P.F., Elkins, N.Z., Steinberger, Y. & Whitford, W.G. (1984) A comparison of surface and buried
Larrea tridentata leaf litter decomposition in North American hot deserts. Ecology, 65, 278–284.
Smyth, C.E., Titus, B., Trofymow, J.A., Moore, T.R., Preston, C.M., Prescott, C.E. & the CIDET Working
Group (2016) Patterns of carbon, nitrogen and phosphorus dynamics in decomposing wood blocks in
Canadian forests. Plant and Soil, 409, 459–477.
Swift, M.J., Heal, O.W. & Anderson, J.M. (1979) Decomposition in Terrestrial Ecosystems. Oxford:
Blackwell.
Throop, H.L. & Archer, S.R. (2009) Resolving the dryland decomposition conundrum: some new perspectives
on potential drivers. In Progress in botany, Springer, Berlin, Heidelberg, 171–194.
Torres, J.A. & González, G. (2005) Wood Decomposition of Cyrilla racemiflora (Cyrillaceae) in Puerto Rican
dry and wet forests: A 13‐year case study. Biotropica, 37, 452–456.
Ulyshen, M.D., Müller, J. & Seibold, S. (2016) Bark coverage and insects influence wood decomposition:
Direct and indirect effects. Applied Soil Ecology, 105, 25–30.
Ulyshen, M.D. (2016) Wood decomposition as influenced by invertebrates. Biological Reviews, 91, 70–85.

This article is protected by copyright. All rights reserved

Accepted Article
Van Geffen, K.G., Poorter, L., Sass-Klaassen, U., Van Logtestijn, R.S. & Cornelissen, J.H.C. (2010) The trait
contribution to wood decomposition rates of 15 Neotropical tree species. Ecology, 91, 3686–3697.
Vivanco, L. & Austin, A.T. (2006) Intrinsic effects of species on leaf litter and root decomposition: a
comparison of temperate grasses from North and South America. Oecologia, 150, 97–107.
Wainhouse, D., Cross, D.J. & Howell, R.S. (1990) The role of lignin as a defense against the spruce bark
beetle Dendroctonus micans: effect on larvae and adults. Oecologia, 85, 257–265.
Wang, L.X., Throop, H.L. & Gill, T. (2015) A novel method to continuously monitor litter moisture – A
microcosm-based experiment. Journal of Arid Environments, 115, 10–13.
Wang, J., Liu, L.L., Wang, X. & Chen, Y. (2015) The interaction between abiotic photodegradation and
microbial decomposition under ultraviolet radiation. Global Change Biology, 21, 2095–2104.
Wang, J., Liu, L.L., Wang, X., Yang, S., Zhang, B.B., Li, P., Qiao, C.L., Deng, M.F. & Liu, W.X. (2017).
High night‐time humidity and dissolved organic carbon content support rapid decomposition of
standing litter in a semi‐arid landscape. Functional Ecology, 31, 1659–1668.
Wang, J.W. (2018) Analysis of soil enzyme activity of typical plant community in the eastern margin of Mu
Us Sandland [In Chinese with English abstract]. Master thesis. Shanxi University.
Weedon, J.T., Cornwell, W.K., Cornelissen, J.H.C., Zanne, A.E., Wirth, C. & Coomes, D.A. (2009) A global
meta‐analysis of wood decomposition rates: a role for trait variation among tree species?. Ecology
Letters, 12, 45–56.
Williamson, G.B. & Wiemann, M.C. (2010) Measuring wood specific gravity… correctly. American Journal
of Botany, 97, 519–524.
Wu, C.S., Zhang, Z.J., Wang, H.K., Li, C., Mo, Q.F. & Liu, Y.Q. (2018) Photodegradation accelerates coarse
woody debris decomposition in subtropical Chinese forests. Forest Ecology & Management, 409, 225–
232.
Wu, C.S., Wang, H.K., Mo, Q.F., Zhang, Z.J., Huang, G.X., Kong, F.Q., Liu, Y.Q. & Wang, G.G. (2019)
Effects of elevated UV-B radiation and N deposition on the decomposition of coarse woody debris.
Science of the Total Environment, 663, 170–176.
Zuo, J., Berg, M.P., Klein, R., Nusselder, J., Neurink, G., Decker, O., Hefting, M.M., Klassen, S.U.,
Logtestijn, R.S.P.V., Leo, G.W., Hal, J.V., Sterck, F.J., Poorter, L. & Cornelissen, J.H.C. (2016) Faunal
community consequence of interspecific bark trait dissimilarity in early-stage decomposing logs.
Functional Ecology, 30, 1957–1966.
Zuo, J., Hefting, M.M., Berg, M.P., van Logtestijn, R.S., van Hal, J., Goudzwaard, L., Liu, J.C., Sass-
Klaassen, U., & Cornelissen, J.H.C. (2018). Is there a tree economics spectrum of decomposability? Soil
Biology & Biochemistry, 119, 135–142.

This article is protected by copyright. All rights reserved

Accepted Article
Table 1 Results from the three-way repeated measures ANOVA of predictive variables on wood mass losses in
the semiarid dune ecosystem.
Variation Dfn, Dfd F P value
Between-subjects effects Species (S) 3, 400 87.63 < 0.001
Diameter class (DC) 4, 400 548.51 < 0.001
Litter treatment (LT) 4, 400 2503.19 < 0.001
S × DC 12, 400 13.32 < 0.001
S × LT 12, 400 4.26 < 0.001
DC × LT 16, 400 33.32 < 0.001
S × DC × LT 48, 400 2.80 < 0.001
Within-subjects effects Time (T) 4, 1587 1324.03 < 0.001
T×S 12, 1587 5.19 < 0.001
T × DC 16, 1587 21.36 < 0.001
T × LT 16, 1587 100.99 < 0.001
T × S × DC 48, 1587 0.87 0.717
T × S × LT 48, 1587 1.09 0.319
T × DC × LT 64, 1587 2.38 < 0.001
T × S × DC × LT 192, 1587 1.18 0.056
The predictive variables include species, diameter class, litter treatment (i.e. environment in terms of position
and UV exposure; see Table 2), and all their interactions. Dfn and Dfd denote numerator and denominator
degrees of freedom, respectively. The P values in bold denote significant terms.

This article is protected by copyright. All rights reserved

Accepted Article
Table 2 Effects of initial traits on wood decomposition rates via model averaging in the semiarid dune
ecosystem.
Treatment Source Estimate Std. Error z value P value RVI
Suspended & UV block Intercept -1.181 0.017 66.11 <0.001 -
Diameter -0.170 0.035 4.48 <0.001 1.00
Initial wood C: N ratio -0.110 0.108 1.01 0.311 0.52
Initial wood N 0.100 0.107 0.93 0.355 0.48
Suspended & UV pass Intercept -1.172 0.018 59.64 <0.001 -
Diameter -0.159 0.081 1.95 0.052 0.83
Initial wood C: N ratio -0.097 0.095 1.01 0.314 0.54
Initial wood N 0.084 0.095 0.88 0.379 0.46
Initial wood DMC -0.030 0.068 0.44 0.659 0.17
Surface & UV block Intercept -1.288 0.029 41.45 <0.001 -
Initial wood DMC -0.366 0.059 5.74 <0.001 1.00
Initial wood N 0.119 0.122 0.97 0.334 0.52
Initial wood C: N ratio -0.109 0.121 0.90 0.371 0.48
Surface & UV pass Intercept -1.317 0.027 44.65 <0.001 -
Diameter -0.441 0.059 6.95 <0.001 1.00
Initial wood C: N ratio -0.062 0.078 0.77 0.440 0.45
Initial wood N 0.058 0.077 0.74 0.459 0.43
Sand burial Intercept -0.564 0.013 40.43 <0.001 -
Diameter -0.281 0.028 9.47 <0.001 1.00
Initial wood C: N ratio -0.041 0.043 0.92 0.356 0.53
Initial wood N 0.036 0.043 0.84 0.403 0.47
Model-averaged coefficients and importance values for explanatory variables predicting wood decomposition
rates (k values) are shown for models based on all treatments in terms of litter position and UV exposure. RVI
denotes relative variable importance. Diameter classes were transformed into a continuous predictive variable.
The P values in bold denote significant terms.

This article is protected by copyright. All rights reserved

Accepted Article
Figure legends
FIGURE 1 Differences in initial bark and wood traits among species and diameters in the semiarid dune
ecosystem. Bark DMC: bark dry matter content, Wood DMC: saturated wood dry matter content. The bars
represent means ± SE (n=5 for bark traits or n=3 for other traits). The same lowercase denotes no significant
difference in bark or wood traits among species within each of the diameter classes, while the same uppercase
denotes no significant difference among diameter classes within each species at P < 0.05.
FIGURE 2 Effects of treatment, species and diameter on wood mass losses over 34 months in the semiarid
dune ecosystem. The bars represents mean ± SE (n = 5). The same Greek symbol denotes no significant
difference among three positions including sand burial, soil surface and suspended locations at P < 0.05. The
same lowercase denotes no significant difference between UV treatments for soil surface and suspended
position respectively and the same uppercase denotes no significant difference between the surface and
suspended positions for each of the UV treatments at P < 0.05.
FIGURE 3 Effects of treatment and diameter on k values for wood decomposition in the semiarid dune
ecosystem. The bars represent means ± SE (n = 4 species). The same lowercase denotes no significant
difference in k values among species within each of the diameter classes and the same uppercase denotes no
significant difference in k values among diameter classes within each of the litter treatments at P < 0.05.

FIGURE 4 The structural equation model (SEM) depicting pathways by which wood traits may influence k
values in the semiarid dune ecosystem. The model parameters in the SEM were χ2df=3 = 1.013, P = 0.798, GFI
= 0.987, RMSEA < 0.001. The red and blue arrows represent positive and negative relationships, respectively.
The widths of the arrows are proportional to the strengths of the path coefficients. The double headed arrows
between ksuspended and ksurface indicate there were no causal relationships of wood decomposition rates between
these two positions; just a correlation because both were driven by the same traits. The value in parentheses
denotes the P value.

This article is protected by copyright. All rights reserved

Initial bark thickness (mm)
(a) Species (S): P<0.001 Diameter class (DC): P<0.001 S x DC: P=0.003

1.5

Aa
Artemisia ordosica
Caragana korshinskii

Aa

b
Hedysarum laeve

1.0

Aa
Salix psammophil

b
Aa
b
Aa

b
a

Aa
AB

b
b

AB
0.5

AB
Ba
a

a
BC
Ba

a
BC
Ba

AB
b

Ba

a
ab
Ba

C
C
b
C
0.0
(b)

Initial bark DMC (g g−1)

0.8
Species (S): P<0.001 Diameter class (DC): P=0.001 S x DC: P<0.001

Aa
0.6

Aa
Aa

Aa

Aa
Aa
Aa

Aa
Aa

Aa
b

Aa
Aa

Aa

Aa
Aa
Aa
0.4

Bb
Bb

Bb

Bb
0.2
0.0

(c)
Initial wood DMC (g g−1)

Species (S): P=0.047 Diameter class (DC): P<0.001 S x DC: P=0.091

0.8

Aa
Aa
Aa

Aa

Aa

Aa
a

Aa
Aa

AB
0.6

a
a

Aa

AB

Aa
a
AB
Aa
Aa

AB
Aa
b

b
Aa

Ba
0.4

Bb
0.2
0.0

(d)
Aa
1.5

Species (S): P<0.001 Diameter class (DC): P<0.001 S x DC: P=0.09

Initial wood N (%)

Aa
Aa
1.0

Aa
Aa
Ab

Aa

b
b

Aa
Aa
Ac

Ab
b
0.5

AB

b
Ad

AB

c
Bb

Bc
Ac
Ab
Ac

Ac
0.0

0−2 2−4 4−8 8−13 13−20

Diameter classes (mm)

 Fraction of wood mass loss

0.0
0.2
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.8
1.0

Aa Aa Aa Aa Aa

0.4 α β
α β
α β
α β
0.6 α β

Aa Aa Aa Aa Aa

12
Aa Aa Aa Aa Ba
γ

β
β
β
β
Aa Aa Aa Aa Aa

Sand burial

α
α
α

Aa Aa Aa Aa Aa

γ
β
α β
α β
β

Surface & UV pass

Surface & UV block
Ba Aa Aa Aa Aa

24
Aa Aa Aa Aa Aa

Suspended & UV pass

Suspended & UV block

β
β
β
β
β

Aa Aa Aa Aa Aa

α
α
α
α
α
Artemisia ordosica

Aa Aa Aa Aa Aa

β
β
β
β
β

Aa Aa Aa Aa Aa

34
Aa Aa Aa Aa Aa

β
β
β
β
β

Aa Aa Aa Aa Aa
α
α
α

Aa Aa Aa Aa Aa

α β
α β
β
β
β

Aa Aa Aa Aa Aa

12
Aa Aa Aa Aa Aa

β
β
β
β
β

Aa Aa Aa Aa Aa

α
α
α
α
α

Aa Aa Aa Aa Aa

β
β
β
β
β

Aa Aa Aa Aa Aa

24
Aa Aa Aa Aa Aa

β
β
β
β
β

Aa Aa Aa Aa Aa

α
α
α
α
α
Caragana korshinskii

Aa Aa Aa Aa Aa

β
β
β
β
β

Aa Aa Aa Aa Aa

34
Aa Aa Aa Aa Aa

β
β
β
β
β

Aa Aa Aa Aa Aa
α

Aa Aa Ba Aa Aa
α γ

α β
α β
α β
β

Aa Aa Aa Aa Aa

12
Aa β Aa Aa Aa Aa
β
β
β
β

Aa Aa Aa Aa Aa
α
α
α
α

Aa Aa Aa Aa Aa
β
β
β
β

Aa Aa Aa Aa Aa
24
Incubation time (months)
Aa Aa Aa Aa Aa
β
β
β
β

α β αβ

Aa Aa Aa Aa Aa
Hedysarum laeve

α
α
α
α
α

Aa Aa Aa Aa Aa
β
β
β
β
β

Aa Aa Aa Aa Aa
34

Aa Aa Aa Aa Aa
β
β
β
β
β

Aa Aa Aa Aa Aa
α
α

Aa Aa Aa Aa Aa
α β
α β
α β
β
β

Aa Aa Aa Aa Aa
12

Aa Aa Aa Aa Aa
β
β
β
β
β

Aa Aa Aa Aa Aa
α
α
α
α
α

Aa Aa Aa Aa Aa
β
β
β
β
β

Aa Aa Aa Aa Aa
24

Aa Aa Aa Aa Aa
β
β
β
β
β

Aa Aa Aa Aa Aa
α
α
α
α
α
Salix psammophila

Aa Aa Aa Aa Aa
β
β
β
β
β

Aa Aa Aa Aa Aa
34

Aa Aa Aa Aa Aa
β
β
β
β
β

Aa Aa Aa Aa Aa

13−20 mm 8−13 mm 4−8 mm 2−4 mm 0−2 mm

 0.6
Treatment (T): P<0.001
Diameter class (DC): P<0.001
T x DC: P=0.184

Aa
Sand burial

0.5
Surface & UV block
Surface & UV pass
Suspended & UV block
0.4 Suspended & UV pass

a
AB
k values (year−1)

a
BC
0.3

a
BC

a
C
0.2

Ab
Ab
Ab

Ab

AB b
b

AB b
AB

b
Bb
0.1

AB
Bb

Bbb
Bb

Bb
AB
Bb

Bb
Bb
Bb
Bb

Bb
0.0

0−2 2−4 4−8 8−13 13−20

Diameter classes (mm)
 R2=0.24 Wood N 0.29 (<
0.001) R2=0.90
1) ksand burial
.00
(<0

1)
6
0.3 R2=0.76

.00
<0
-0
.5
R2=0.84

3(
1 Bark thickness
(0

0.6
.0

-0.
69
)

87
ksurface

(<0

0.001)
.00
-0.57 (

1)
0.008)

0.87 (<
0.84 (0.005)

Diameter
-0.
40

.014)
-0.52 (0
(0.

1)
07

.00
1)

<0
1(
)
ksuspended 0 31

0.6
6 (0.
0.2
R2=0.85

) Wood density
(< 0.001
0.72
R2=0.37
R2=0.82 Wood DMC