Carbon dynamics in the deciduous broadleaf tree Erman’s birch (Betula ermanii) at the subalpine treeline on Changbai Mountain, Northeast China

Abstract

Premise of the study:
The growth limitation hypothesis (GLH) and carbon limitation hypothesis (CLH) are two dominant explanations for treeline formation. The GLH proposes that low temperature drives the treeline through constraining C sinks more than C sources, and it predicts that non-structural carbohydrate (NSC) levels are static or increase with elevation. Although the GLH has received strong support globally for evergreen treelines, there is still no consensus for deciduous treelines, which experience great asynchrony between supply and demand throughout the year.

Methods:
We investigated growth and the growing-season C dynamics in a common deciduous species, Erman's birch (Betula ermanii), along an elevational gradient from the closed forest to the treeline on Changbai Mountain, Northeast China. Samples were collected from developing organs (leaves and twigs) and main storage organs (stems and roots) for NSC analysis.

Key results:
Tree growth decreased with increasing elevation, and NSC concentrations differed significantly among elevations, organs, and sampling times. In particular, NSC levels varied slightly during the growing season in leaves, peaked in the middle of the growing season in twigs and stems, and increased continuously throughout the growing season in roots. NSCs also tended to increase or vary slightly in developing organs but decreased significantly in mature organs with increasing elevation.

Conclusions:
The decrease in NSCs with elevation in main storage organs indicates support for the CLH, while the increasing or static trends in new developing organs indicate support for the GLH. Our results suggest that the growth limitation theory may be less applicable to deciduous species' growth than to that of evergreen species.

Full text

American Journal of Botany 0(0): 1–8, 2018; http://www.wileyonlinelibrary.com/journal/AJB © 2018 Botanical Society of America • 1
e alpine treeline is one of the most striking terrestrial ecolog-
ical boundaries and is thought to be caused by low temperatures
(~6.7°C) that restrict growth during the growing season (Körner,
2003a; Körner and Paulsen, 2004). e existence of such a uniform
isocline implies that a single common functional process underlies
its formation worldwide. Studies to date have largely focused on
the eects of low temperatures on the carbon (C) source–sink bal-
ance (Tranquillini, 1979; Körner, 2003b; Hoch and Körner, 2012),
leading to the proposal of two hypotheses to explain the functional
mechanisms involved. e C limitation hypothesis (CLH) proposes
that the decrease in temperature with increasing elevation reduc-
es C gain (i.e., photosynthesis), resulting in treeline formation at
higher elevations where C gains are insucient to compensate for
the requirements of C sinks (e.g., growth, respiration, and C loss-
es; Wardle, 1993). Alternatively, the growth limitation hypothesis
(GLH) states that low temperature constrains meristematic activity
(i.e., organ formation) and drives treeline formation (Körner, 1998).
erefore, an understanding of how C dynamics in trees change
with increasing elevation is essential for evaluating the mechanisms
involved in treeline formation.
As the primary products of photosynthesis, non- structural
carbohydrates (NSCs, mainly including starch and soluble sug-
ars) account for most of the stored C in a plant and thus reect the
source–sink balance (Chapin etal., 1990; Körner, 2003a; Martínez-
Vilalta etal., 2016). Consequently, changes in NSC concentrations
in trees with increasing elevation have generally been accepted as a
Carbon dynamics in the deciduous broadleaf tree Erman’s
birch (
Betula ermanii
) at the subalpine treeline on Changbai
Mountain, Northeast China
Qing-Wei Wang1,2,4, Lin Qi1, Wangming Zhou1, Cheng-Gang Liu3, Dapao Yu1,4, and Limin Dai1
RESEARCH ARTICLE
Manuscript received 25 August 2017; revision accepted 14
December 2017.
1 Key Laboratory of Forest Ecology and Management,Institute
of Applied Ecology,Chinese Academy of Sciences, Shenyang,
110164, China
2 Forestry and Forest Products Research Institute, 1 Matsunosato,
Tsukuba, Ibaraki 305-8687, Japan
3 Key Laboratory of Tropical Plant Resources and Sustainable
Use,Xishuangbanna Tropical Botanical Garden,Chinese Academy
of Sciences, Mengla, 666303, China
4 Authors for correspondence: (e-mail: wangqw08@gmail.com;
yudp2003@iae.ac.cn)
PREMISE OF THE STUDY: The growth limitation hypothesis (GLH) and carbon limitation
hypothesis (CLH) are two dominant explanations for treeline formation. The GLH proposes
that low temperature drives the treeline through constraining C sinks more than C sourc-
es, and it predicts that non- structural carbohydrate (NSC) levels are static or increase with
elevation. Although the GLH has received strong support globally for evergreen treelines,
there is still no consensus for deciduous treelines, which experience great asynchrony
between supply and demand throughout the year.
METHODS: We investigated growth and the growing- season C dynamics in a common decid-
uous species, Erman’s birch (Betula ermanii), along an elevational gradient from the closed
forest to the treeline on Changbai Mountain, Northeast China. Samples were collected from
developing organs (leaves and twigs) and main storage organs (stems and roots) for NSC
analysis.
KEY RESULTS: Tree growth decreased with increasing elevation, and NSC concentrations
diered signicantly among elevations, organs, and sampling times. In particular, NSC levels
varied slightly during the growing season in leaves, peaked in the middle of the growing sea-
son in twigs and stems, and increased continuously throughout the growing season in roots.
NSCs also tended to increase or vary slightly in developing organs but decreased signicantly
in mature organs with increasing elevation.
CONCLUSIONS: The decrease in NSCs with elevation in main storage organs indicates support
for the CLH, while the increasing or static trends in new developing organs indicate support
for the GLH. Our results suggest that the growth limitation theory may be less applicable to
deciduous species’ growth than to that of evergreen species.
KEY WORDS Betulaceae; carbon balance; growth limitation; leaf habit; non-structural carbo-
hydrates; organ dependence; seasonal variation; treeline formation mechanisms
Citation: Wang, Q.-W., L. Qi, W. Zhou, C.-G. Liu, D. Yu, and
L. Dai. 2018. Carbon dynamics in the deciduous broadleaf tree
Erman’s birch (Betula ermanii) at the subalpine treeline on Changbai
Mountain, Northeast China. American Journal of Botany 0(0): 1–8.
doi: 10.1002/ajb2.1006

2 • American Journal of Botany
proxy for testing the GLH and CLH (Shi etal., 2008; Fajardo etal.,
2012; Hoch and Körner, 2012), whereby no decrease in NSC concen-
trations in any organ indicates support for the GLH and a decrease
in NSC concentrations indicates support for the CLH. To date, many
studies have found support for the GLH in various treelines, most of
which were composed of evergreen species (e.g., Hoch and Körner,
2003, 2005, 2012; Shi etal., 2008; Dawes etal., 2015). However, evi-
dence for deciduous species is mixed; for instance, Larix potaninii
was controlled by the growth limitation at the eastern Himalayas
treeline (Shi etal., 2008), whereas L. decidua at the Swiss treeline
had great aboveground growth responses to elevated CO2, support-
ing the CLH (Dawes etal., 2013). Even within a given species (e.g.,
Nothofagus pumilio), treelines appear to be supportive of the GLH
(Fajardo etal., 2013; Piper etal., 2016) and the CLH (Fajardo and
Piper, 2014, 2017). It is questionable whether the prevalent physi-
ological mechanism (growth limitation) most accepted to explain
treeline formation applies to deciduous treeline species.
C storage of deciduous species may have higher seasonal dy-
namics in relation to elevation than evergreen species as a result
of leaf habit (Martínez- Vilalta etal., 2016). e dominant explana-
tion is that deciduous trees, seasonally shedding leaves, generally
experience great asynchrony between supply, which occurs only in
the growing season, and demand, which occurs throughout the year
(Chapin etal., 1990). Accordingly, NSC concentrations would de-
crease to a minimum during the early growing season, increase as
assimilation occurs, and nally reach a peak toward the late grow-
ing season (Schadel et al., 2009). Moreover, deciduous broadleaf
species with a lower leaf mass per area (LMA) are more susceptible
to frequent mechanical damage and tissue loss, making C demand
high and variable (Sveinbjörnsson etal., 1992). By contrast, ever-
green species can produce a stable C supply throughout the year,
resulting in only a slight seasonal uctuation in C storage (Dickson,
1989; Kozlowski, 1992; Hoch, 2015). us, NSC accumulation driv-
en by growth limitation in deciduous species should clearly take
place early in the growing season, when growth is more intense.
Surprisingly, previous studies that have examined the GLH have
measured NSCs at the end of the growing season, which may fail to
reect “growth limitations” when growth is not actually occurring
(Shi etal., 2008; Yu etal., 2014). For example, Hoch and Körner
(2012) investigated 13 treeline sites worldwide, composed of trees
with dierent leaf habits (10 evergreen and four deciduous species),
and found that NSCs increased with elevation in all species except
the deciduous N. pumilio (in Chile), in which the NSC concentra-
tions decreased in leaves but increased in branches at the end of the
growing season. is also suggests that a single snapshot of changes
in NSC concentrations with increasing elevation cannot be used to
determine whether growth or C limitation is occurring. However,
the seasonal dynamics of NSCs in deciduous treeline species are
still not well understood.
C storage of deciduous species in relation to elevation may
also vary dierently among organs throughout the growing sea-
son (Hoch etal., 2002; Piper etal., 2016). Newly developing organs
(e.g., leaves and twigs) oen have greater seasonal oscillations in C
storage than major storage organs (e.g., stems and roots; Martínez-
Vilalta etal., 2016), due to higher meristematic activity (e.g., cell
division and elongation). Although recent studies have found that
developing organs could quickly become C- autonomous in the
absence of any underlying stress (Keel and Schädel, 2010; El Zein
etal., 2011; Landhäusser, 2011), it is unknown whether this occurs
at treelines. Consequently, it has been suggested that studies using
NSC concentrations to assess C or growth limitations should focus
primarily on developing organs (Piper etal., 2016). However, many
previous studies have investigated NSC concentrations in major
storage organs (Hoch and Körner, 2003; Fajardo etal., 2011, 2012;
Lenz etal., 2014), and it has also been shown that low temperatures
and other stressors may have a greater eect on NSC in major stor-
age organs than in developing organs, due to the need to translocate
C over longer distances (Pratt and Jacobsen, 2017). For instance,
Piper etal. (2016) found that NSCs increased signicantly in leaves
with increasing elevation but tended to decrease in branches of
treeline species growing in a Mediterranean climate (Chile). ere-
fore, consideration of the organs sampled may also be crucial for
evaluating growth and C limitations in treeline species.
In the present study, we aimed to evaluate C dynamics along an
elevation gradient in a common deciduous broadleaf species, Er-
man’s birch (Betula ermanii), which is the predominant species at
treeline on Changbai Mountain, the highest mountain in Northeast
China. Although Erman’s birch is the dominant treeline species in
the subalpine zone in East Asia (Gansert etal., 1999), it has not been
included in previous global synthesis studies on treeline formation
(Harsch etal., 2009; Hoch and Körner, 2012; Martnez- Vilalta etal.,
2016) because of the lack of C storage data (Yu et al., 2014). We
focused on two treeline theories in relation to C balance, the GLH
and CLH, by investigating tree growth and changes in NSC con-
centrations with increasing elevation in developing organs (leaves
and twigs) and mature organs (stems and roots) of Erman’s birch
throughout the growing season. In support of the GLH, we expect-
ed that C storage in all organs would increase with elevation, be-
cause low temperature constrains C sinks more than C sources at
higher elevations; alternatively, in support of the CLH, we expected
that NSC accumulation would decrease with increasing elevation,
knowing that deciduous species are more prone to tissue losses due
to physical damage and, consequently, increased risks of C limita-
tion. Furthermore, we expected that NSC concentrations would
vary in all organs throughout the growing season because of the
great asynchrony between C supply and demand even when the
growing season is short, and that the magnitude of such change
would be greater in developing organs than in mature organs be-
cause of their higher metabolic activity.
MATERIALS AND METHODS
Study site
e study was conducted on the north slope of the Changbai Moun-
tain Natural Reserve (41°3′–42°28′N, 127°9′–128°55′E), Jilin Prov-
ince, Northeast China. ere are four dened vegetation zones along
the elevational gradient in this region, including Korean pine (Pinus
koraiensis) and broadleaved mixed forest (740–1100 m a.s.l.), conifer
forest (1100–1800 m a.s.l.), Erman’s birch forest (1800–2000 m a.s.l.),
and alpine tundra (above 2000 m a.s.l.). For Erman’s birch, the cen-
tral distribution area occurs at 1900 m a.s.l.; the edge of the closed
forest (hereaer “timberline”) occurs at 1950–2000 m a.s.l.; and the
treeline, where tree heights are >3 m, occurs at 2018 m a.s.l. e
climate at the treeline is characterized by severe cold, high humidity,
and strong winds (Yu etal., 2014), with a mean annual temperature
of −2.3°C to −3.8°C; a frost- free period of about 65–70 d; annual
precipitation ranging from 1000 to 1100 mm, most of which occurs
from June to September; and annual wind speeds ranging from 6 to

2018, Volume 0 • Wang et al.—Carbon storage in a deciduous treeline species • 3
10 m s−1, with gales sometimes lasting >200 d. e growing season
at the treeline generally starts at the end of May or in early June and
ends when the rst severe frost occurs in late September.
Sampling
We established a 30 × 20 m sample strip at three dierent elevations
encompassing the closed forest (1908 m a.s.l.), the timberline (1976
m a.s.l.), and the treeline (2018 m a.s.l.) for Erman’s birch. We con-
ducted sampling on three sunny days: 20 June, 3 August, and 8 Sep-
tember, 2010, which represented the early, middle, and late growing
season, respectively. At each elevation, we selected ve similarly
aged trees (~7.0 cm diameter at breast height) with heights >3 m.
Each tree was separated from the others by ≥10 m. On each tree, we
sampled (1) mature leaves, (2) young branches (1–2 cm diameter;
hereaer “twigs”), (3) stem xylem (segments 1.5 cm long, measured
from the outermost stem section toward the pith at breast height),
and (4) roots (0.5–1.0 cm diameter). Each organ had 10 replicates
per elevation site and sampling time. To avoid light eects on NSC
concentrations, we sampled leaves from nonshaded leading branch-
es on the upslope side (Li etal., 2001). We determined twig age by
counting the number of twig nodes, from which we then measured
mean annual shoot growth. We removed the bark and phloem from
the twig, stem, and root samples with a knife because bark accounts
for little of a tree’s NSC stock (Chantuma etal., 2009) and phloem
mainly functions in C transport (Hartmann and Trumbore, 2016).
We took two cores from opposite sides of the stem on each tree
using a 0.5 mm stem corer (Suunto, Vantaa, Finland). Because it
has previously been shown that starch and sugar concentrations
rapidly decline from the outermost tree ring toward the pith, where
concentrations are very low or even undetectable (Yu etal., 2014),
we collected the outermost (youngest) 1.5 cm segments as the stem
sapwood. In the case of root wood samples that were <1 cm diam-
eter, we considered the whole xylem to be active (sapwood; Hoch
etal., 2002).
We collected samples around noon to minimize the eects of
diurnal uctuations and stored them in a cool box in the eld. We
then heated them in a microwave oven (40 s at 600 W) within 6 h
of sampling to denature the enzymes. In the laboratory, we dried
half the organ samples to a constant mass at 70°C (~48 h), ground
them to a ne powder, and then stored them over silica gel at 4°C.
We used the remaining samples to determine the leaf mass per area
(LMA, g m−2) and woody organ densities because NSC concentra-
tions expressed in terms of mass are aected by wood density. We
measured leaf area with a leaf area meter (CI- 203; CID Bio- Science,
Camas, Washington, USA) and calculated LMA as leaf dry mass
divided by leaf area. We then dried the remaining woody organ
samples to a constant mass at 70°C (~72 h) to determine their dry
weight. Based on Archimedes’ principle, the volume of the woody
samples was measured by submerging the woody organs in water
(22°C) in a glass beaker with a scale. e dierence caused by sub-
mersion of the sample could be converted to volume, since water
density equals the unit at the standard temperature and pressure.
Finally, we calculated wood density as the ratio of dry mass of the
woody organs to their volume.
Chemical analysis
We dened NSCs as the sum of total soluble sugars and starch. e
concentrations of total soluble sugars and starch were determined
using the anthrone method, as described by Li etal. (2008). Briey,
we placed the powdered material in a 10 mL centrifuge tube with
5 mL of 80% ethanol. en we incubated the mixture at 80°C for
30 min and centrifuged it at 5000 × g for 5 min. Aer repeating
this process twice, we spectrophotometrically measured the com-
bined supernatants for soluble sugar (within 30 min) at 620 nm
using anthrone reagent and calculated the concentrations of solu-
ble carbohydrate from standard regression equations using glucose
as a standard. Starch was released by boiling the residue in 2 mL
distilled water for 15 min and then adding 2 mL of 9.2 M HClO4
solution at room temperature and leaving it for 15 min to hydrolyze
the starch. We then added an additional 4 mL of distilled water to
the tube and centrifuged the mixture at 5000 × g for 10 min, follow-
ing which we extracted the pellet again with 2 mL of 4.6 M HClO4
solution. We then spectrophotometrically analyzed the combined
supernatants for starch at 620 nm using anthrone reagent with glu-
cose as a standard. We calculated starch concentration by multi-
plying glucose concentrations by a conversion factor of 0.9 (Osaki
etal., 1991). e concentrations of soluble sugar and starch were
expressed on a dry- matter basis (% dm).
Temperature records
We monitored canopy air and soil temperatures at the three ele-
vations using microclimatic loggers (−30 to +50 °C; HOBO H8
Pro temperature loggers; Onset, Bourne, Massachusetts, USA).
One logger was placed 2 m above the ground, avoiding direct sun-
light; the other was placed at a soil depth of 10 cm under full can-
opy shade, following a previously published protocol (Körner and
Paulsen, 2004). Both loggers recorded the temperature at 30 min
intervals from 1 January to 31 December 2010. e beginning of the
growing season was dened as the date on which the daily mean soil
temperature at a depth of 10 cm rst exceeded 3.2°C, and the end of
the growing season was dened as the date on which the daily mean
soil temperature rst dropped below 3.2°C (following Körner and
Paulsen, 2004).
Statistical analysis
We tested all data (NSC, soluble sugar, and starch concentrations) for
normality with the Kolmogorov- Smirnov test and log- transformed
them to meet the assumption of normality where required. We then
used two- way repeated- measures analysis of variance (ANOVA) to
analyze the eects of elevation and sampling date on the concentra-
tions of NSCs and the soluble sugar and starch components for each
organ type. We also used one- way ANOVA to analyze variations in
annual shoot length, LMA, and wood density across all elevations.
Signicant dierences in variables among elevations were evaluated
using Tukey’s multiple range test. We performed all statistical tests
using SAS version 8.1 (SAS Institute, Cary, North Carolina, USA)
and considered results signicant at the 5% level.
RESULTS
Growth
e growing season was 12 d shorter at the treeline than in the
closed forest, and tree diameter and height also tended to be lower
at the treeline (Table1). e annual increase in twig length over the

4 • American Journal of Botany
previous 3 yr decreased signicantly, by an average of 29.4%, with
increasing elevation (F2, 27 = 43.45, P < 0.0001; Fig.1) but was not af-
fected by age (F2, 27 = 1.41, P = 0.2502). In terms of functional traits,
LMA increased signicantly with increasing elevation (F2, 42 = 30.56,
P < 0.0001), whereas wood density was similar across all elevations
and woody organ types (Table2).
Variation in NSC concentrations during the growing season
e concentrations of NSCs, soluble sugars, and starch in each organ
were signicantly aected by the sampling date (Table3). ere was
also a signicant interaction between the sampling date and elevation,
indicating that the seasonal variation in NSC concentration within
each organ varied among birch trees growing at dierent elevations.
NSC concentrations in leaves varied only slightly during the grow-
ing season, due to the seasonal increase in soluble sugars being oset
by the decrease in starch (Fig.2). By contrast, the concentrations of
NSCs and starch in twigs and stem wood showed signicant variation
during the growing season, peaking in the middle of the season (Au-
gust) and then decreasing toward the end of the season (September),
with this pattern being more pronounced at 1976 m a.s.l. In root wood,
there was a continuous increase in the NSC and starch concentrations,
though the extent of this decreased with increasing elevation. e con-
tribution of soluble sugars to the NSC concentration was >50% at all
elevations and increased with increasing elevation (Appendix S1; see
Supplemental Data with this article).
Variations in NSC concentrations with increasing elevation
ere was signicant variation in the concentrations of NSCs, sol-
uble sugars, and starch in each organ with increasing elevation,
though there were some exceptions (NSCs and starch in twigs, and
sugars in twigs and roots; Table 3). NSC concentrations in leaves
increased signicantly in June and September and increased slight-
ly in August with increasing elevation (Fig.2). By contrast, these
trends were not observed in twigs, where the lowest values were re-
corded at 1976 m a.s.l. in September. For mature organs (stem wood
and root wood), NSC concentrations decreased signicantly with
increasing elevation. Furthermore, variations in the concentrations
of sugars in stems and starch in roots tended to be consistent with
those observed for NSC concentrations in each of these organs.
DISCUSSION
Carbon dynamics have been studied intensely in evergreen treeline
species, with ndings to date supporting the GLH. However, the
ndings about mechanisms of treeline formation in deciduous
FIGURE1. Variation in twig growth with increasing elevation: mean an-
nual growth (± SE, n = 10) of twigs of Erman’s birch (Betula ermanii) trees
growing at dierent elevations on Changbai Mountain, Northeast China
(1908 m a.s.l. = closed forest; 1976 m a.s.l. = timberline; 2018 m a.s.l. =
treeline). Growth was compared across elevations using Tukey’s multiple
range test. Dierent letters within an age group indicate signicant dif-
ferences among elevations (P < 0.05).
TABLE2. Variation in the leaf mass per area (LMA, g m−2) and wood density
(g cm−3) of twigs, stem sapwood, and root wood in adult Erman’s birch (Betula
ermanii) trees growing at three elevations on Changbai Mountain, Northeast
China (1908 m a.s.l. = closed forest; 1976 m a.s.l. = timberline; 2018 m a.s.l. =
treeline).
Elevation
(m a.s.l.)
Plant organ
LMA Twig Stem sapwood Root wood
1908 46.90 (1.10)C0.56 (0.02) 0.67 (0.02) 0.47 (0.02)
1976 56.13 (1.90)B0.59 (0.01) 0.62 (0.01) 0.53 (0.02)
2018 63.83 (1.50)A0.60 (0.02) 0.65 (0.01) 0.47 (0.02)
F2, 42 30.56 2.92 1.02 3.57
P <0.0001 0.09 0.15 0.06
Notes: Values are means, with SE in parentheses (n = 15 elevation−1). Measurements were
compared across elevations using Tukey’s multiple range test (F and P values are shown).
Different superscript letters indicate statistically significant differences among elevations
(P < 0.05).
TABLE1. Characteristics of the study sites on Changbai Mountain, Northeast
China, in terms of elevation, temperature, and the growth of Erman’s birch (Betula
ermanii).
Variable Closed forest Timberline Treeline
Elevation (m a.s.l.) 1908 1976 2018
Mean air/soil temperature (°C)
Annual −3.0/4.0 −3.0/3.1 −3.4/−1.3
Growing season 11.8/9.9 11/8.7 10.8/8.2
January −17.5/0.4 −17.3/−0.2 −17.6/−14.2
August 12.9/12.8 13/11.6 12.6/11.0
Growth
L ength of growing
season (days) 141.0 136.0 129.0
M ean diameter at
breast height
(cm) 8.1 (1.3) 7.2 (1.8) 6.0 (1.6)
Mean height (m) 6.4 (1.6) 6.5 (2.1) 4.5 (0.9)
Notes: Mean temperatures within the tree crown (air) and at a soil depth of 10 cm were
monitored at 30 min intervals with temperature loggers (HOBO H8 Pro) from 1 January
to 31 December, 2010. Values before and after the slash symbol represent mean air
temperature and mean soil temperature, respectively. Growth variables are shown as
means, with SE in parentheses. The beginning of the growing season was defined as the
date on which the daily mean soil temperature at a depth of 10 cm first exceeded 3.2°C,
and the end of the growing season was defined as the date on which the daily mean soil
temperature first fell below 3.2°C (Körner and Paulsen, 2004).

2018, Volume 0 • Wang et al.—Carbon storage in a deciduous treeline species • 5
species are contradictory (i.e., Shi etal., 2008; Hoch etal., 2012;
Fajardo etal., 2013, 2017; Piper etal., 2016). In the present study,
we found that shoot growth in Erman’s birch decreased signi-
cantly with increasing elevation, while the C dynamics of this de-
ciduous species depended on elevation, organ type, and sampling
time. NSC concentrations increased with increasing elevation in
leaves, showed no variation in twigs, and decreased in stems and
roots with increasing elevation. Such divergence among organs is
not consistent with the results of previous studies on evergreen
species, in which NSC concentrations have been found to increase
with elevation in all organs (Shi etal., 2008; Hoch and Körner,
2012).
Growing- season dynamics in NSC concentrations among
organ types
As we originally hypothesized, NSC concentrations in organs dra-
matically varied throughout the growing season, irrespective of el-
evation (Fig.1). However, unexpectedly, we found that the highest
amplitude of NSC oscillations during this period occurred not in
leaves but rather in twigs and stems. ese ndings do not support
the idea that developing organs are C- autonomous (Landhäuss-
er, 2011) but may reect signicant amounts of mobilized C due
to the large biomass of aboveground woody organs in this spe-
cies. Furthermore, this pattern may not be solely explained by the
source–sink framework, which accounts for imbalances between
supply (photosynthesis) and demand (growth and respiration)
and predicts that NSC dynamics throughout the season should be
larger in leaves and roots (Martínez- Vilalta etal., 2016). is pat-
tern may also be explained by organ functions and their roles in
whole- plant C dynamics. Leaves act as the main source of carbo-
hydrates and have high metabolic rates, high concentrations of in-
termediary metabolites, and large amounts of living cells requiring
turgor maintenance (Sala etal., 2012); this is particularly true for
deciduous broadleaf species, which need to produce and reserve
enough C to meet the entire year’s demand in a short time (Fajardo
etal., 2013). In roots, NSC concentrations increased continuously
rather than decreasing in the early season to support early growth.
us, it is possible that (1) twig growth is supported mainly by
current assimilates rather than by root C storage or that (2) roots
play the main long- term storage role with intermediate osmotic
and metabolic demands (Martínez- Vilalta etal., 2016), as reected
by intermediate concentrations of NSCs and the soluble sugar and
starch fractions.
Twigs and stems are responsible for C translocation between the
sites of C assimilation (leaves) and C storage (roots) (Hoch etal.,
2002). For instance, C storage was lower in the early growing season
due to the high levels of early growth, increased in the middle of
the growing season when C uptake was highest, and then decreased
in the late growing season as C was transported to the roots (Ap-
pendix S1). Alternatively, in the context of structural and functional
relationships, it is possible that seasonal dynamics in stems are the
result of selection for storage traits, which gives rise to trade- os in
transport and biomechanics traits (e.g., cavitation resistance; Pratt
and Jacobsen, 2017).
Our results are inconsistent with those of previous studies on
evergreen species (Hoch etal., 2002; Zhu etal., 2012; Dang etal.,
2015). For instance, Hoch etal. (2002) found that NSC concentra-
tions in the leaves and stems of Pinus cembra generally decreased
throughout the growing season. However, our results are in line
with Piper etal.’s (2016) ndings for the temperate deciduous broad-
leaf species N. pumilio, suggesting that the source–sink framework
cannot completely explain the growing- season dynamics of NSCs
in deciduous species, though this conclusion requires further inves-
tigation at dierent sites.
Elevational trends in NSC concentrations among organ types
Changes in NSC concentrations with increasing elevation depend
on organ types. ere was no change in the NSC concentrations in
leaves and twigs with increasing elevation, whereas stems and roots
showed a decreasing trend (Fig.2). e change in NSC concentra-
tion in leaves would have been larger had it been calculated on a
volume basis, because the leaf matter area increases with increasing
elevation; however, this would have had a minimal eect on woody
organs, because wood densities were similar among elevations
(Table 2). According to the growth limitation theory, the trends
observed in leaves and twigs indicate that under low temperature,
the production of carbohydrates via photosynthesis exceeds the
demand for growth at high elevations in Erman’s birch. Also, tree
growth, including the diameter and twig length, decreased with
elevation (Table1 and Fig.1). us, the results for leaves and twigs
support the GLH at the treeline on Changbai Mountain. is is con-
sistent with the ndings of previous studies on deciduous species,
TABLE3. Two- way repeated- measures analysis of variance to test the eects of month (June, August, and September) and elevation (1908, 1976, and 2018 m a.s.l.)
on the concentrations of non- structural carbohydrates (NSCs) and soluble sugar and starch components in dierent organs of Erman’s birch (Betula ermanii) trees on
Changbai Mountain, Northeast China. Signicant dierences (P < 0.05) are in bold.
Organ Factors df
NSCs Sugars Starch
FPFPFP
Leaf Sampling date (D) 2 6.02 0.0076 318 <0.0001 229.36 <0.0001
Elevation (E) 2 128.51 <0.0001 234.73 <0.0001 235.55 <0.0001
D × E 4 11.52 <0.0001 137.53 <0.0001 69.66 <0.0001
Twig D 2 38.46 <0.0001 33.03 <0.0001 19.05 <0.0001
E 2 0.85 0.4503 1.46 0.2714 1.76 0.2138
D × E 4 4.9 0.0049 21.77 <0.0001 3.98 0.0129
Stem wood D 2 52.63 <0.0001 8.28 0.0018 79.31 <0.0001
E 2 29.1 <0.0001 21.74 0.0001 5.22 0.0234
D × E 4 2.56 0.0642 11.11 <0.0001 13.39 <0.0001
Root wood D 2 116.33 <0.0001 67.8 <0.0001 64.67 <0.0001
E 2 83.76 <0.0001 0.41 0.6697 78.74 <0.0001
D × E 4 7.51 0.0005 5.13 0.0039 12.06 <0.0001

6 • American Journal of Botany
such as B. platyphylla and L. potaninii (Shi etal., 2008), N. pumilio
and L. decidua (Fajardo etal., 2013), and B. ermanii (Yu etal., 2014).
On the other hand, the decrease of NSCs in stems and roots with
elevation indicates that C sources were more restricted by low tem-
peratures than by C sinks, supporting the CLH. is is opposite of
the conclusion from developing organs, perhaps because of the foli-
ar habit and organ functions of deciduous species (Piper and Fajar-
do, 2014). It has been shown that deciduous species generally have
higher requirements for C storage than evergreens to allow for the
replacement of leaf loss when facing stress (Hoch etal., 2003; El Zein
etal., 2011; Givnish etal., 2014). Particularly
at higher elevations where the abiotic stress-
ors are more intense (Table 1), C demand
would also become stronger—evidence that
soluble sugars contributed to greater pro-
portions of NSC in woody organs (especially
roots) at the upper elevational limit (Appen-
dix S1). is high demand for C may poten-
tially lead to C starvation in storage organs
(i.e., roots; Landhäusser and Lieers, 2012;
Piper and Fajardo, 2014). Furthermore, ma-
ture organs are important C storage sites that
account for most NSCs in adult trees, since
their biomass proportion (storage volume) is
much higher than that invested in leaves or
twigs (Fajardo etal., 2013). is means that
although NSCs of developing organs were
higher than those of mature storage organs
and increased with elevation, they might not
be sucient to make up for decrease in C
storage of stems and roots with elevation.
Empirical support for both the GLH and
the CLH has also been found elsewhere (Shi
etal., 2008; Fajardo etal., 2013; Yu etal., 2014;
Piper etal., 2016), but evidence from decidu-
ous treelines is still generally scarce and pro-
vides contradictory support for either hypoth-
esis. For instance, a free- air CO2 enrichment
experiment at a Swiss treeline showed that
L. decidua had high aboveground growth re-
sponses to elevated CO2, providing support for
the CLH (Dawes etal., 2013). However, Fajar-
do and Piper (2014, 2017) found inconclusive
support for the relative importance of either C
limitation or growth limitation (Fajardo etal.,
2013; Piper etal., 2016) in N. pumilio. ese
seemingly contradictory results may be part-
ly ascribed to the sampling- time dependency
and organ dependency of C storage in relation
to elevation. More importantly, however, these
results imply that using NSC to distinguish
between C and growth limitations at treelines
in deciduous species is not as straightforward
as it is in evergreen species (Fajardo and Piper,
2017). Further investigations are necessary to
make the denitive explanation for deciduous
species clear.
It is notable that changes in NSCs with el-
evation were not consistent during the grow-
ing season, particularly in leaves and twigs
(Table 3 and Fig. 2). If temperature is the only driver of treeline
formation, seasonality should slightly aect elevational variation in
NSCs. us, our results suggest that local factors appear to modu-
late C dynamics, as observed in other treelines (e.g., drought eects
on Mediterranean deciduous treelines; Piper etal., 2016). Indeed,
on Changbai Mountain, soils are characterized by a high content
of volcanic oat stone, a thin depth, and a low water- holding ca-
pacity, making them unfavorable for tree growth. Despite relatively
high precipitation in this region, the radial growth of birch trees was
positively aected by temperature and precipitation (Yu etal., 2007),
FIGURE2. Variation in carbon storage with sampling time and elevation: mean concentrations
(± SE, n = 5) of non- structural carbohydrates (NSCs; circles, June), soluble sugars (squares, Au-
gust), and starch (triangles, September) during the growing season in dierent organs of Erman’s
birch (Betula ermanii) trees growing at dierent elevations on Changbai Mountain, Northeast
China (1908 m a.s.l. = closed forest; 1976 m a.s.l. = timberline; 2018 m a.s.l. = treeline). Statistical
dierences in mean concentrations among elevations were tested using Tukey’s multiple range
test within each sampling time and organ type, indicated by dierent letters (P < 0.05). The initial
statistics for these data are presented in Table3.
0
5
10
15
20
June
August
September
0
2
4
6
8
0
2
4
6
0
3
6
9
12
NSC concentrations (%, dw)
NSCSugars Starch
Twig Stem wood Root wood
ba
c
Leaf
Elevation
ab a
b
ba
a
bab
a
b
ab
a
ab b
a
a
b
a
a
b
a
a
b
a
b
a
c
ba
ab
ab
a
ab
b
ca
b
bb
abb
a
ba
ab
ba
a
a
b
a
a
b
a
b
c
a
ba
b
a
b
c
a
b
b
ba
b
bb
a
bb
a
1908 197620181908 19762018 19081976 2018

2018, Volume 0 • Wang et al.—Carbon storage in a deciduous treeline species • 7
and shoot increment was not associated with its C status but was sig-
nicantly related to high δ13C values (Yu etal., 2014). Furthermore,
the mean soil temperature (8.2°C; Table1) at the treeline was higher
than that in global climate studies (6.7°C; Körner and Paulsen, 2004),
indicating that the Erman’s birch treeline may be located below the
predicted elevation. ese facts suggest that water stress may aect
the treeline structure together with low temperature. However, phys-
iological mechanisms associated with water stress remain unclear.
CONCLUSIONS
Our results show that C dynamics in the deciduous broadleaf species
Erman’s birch were signicantly aected by elevation, organ type,
and sampling time. Although NSC concentrations varied through-
out the growing season, the magnitude of such variation was greater
in twigs and stems than in leaves and roots. C dynamics in relation
to elevation appear to be supportive of the two explanations: de-
veloping organs support the GLH, in that there was no decrease in
NSC concentrations of leaves and twigs with elevation; while main
storage organs support the CLH, in that there was a decrease in C
storage with increasing elevation. ese ndings suggest that using
NSC concentration as a proxy for distinguishing the GLH and the
CLH is not a straightforward way to explain treeline formation in
deciduous species as it is for evergreen species. e growth limita-
tion theory for treeline formation must be revisited. e importance
of C allocation and its function needs to be emphasized in future
studies.
ACKNOWLEDGEMENTS
We thank Ms. J. Tian, J. Jia, H. Ding, and J. Liu for providing tech-
nical assistance in the eld; Drs. K. Hikosaka, H. Günter, and A.
Fajardo for providing constructive comments on the manuscript;
and Ms. K. Martinez for improving the English language of the
manuscript. is project was funded by the National Natural Sci-
ence Foundation of China (nos. 41571197, 41701052); Special
Research Project of the Institute of Applied Ecology, Chinese Acad-
emy of Sciences (no. Y5YZX151YD); and Key Laboratory of Forest
Ecology and Management, Institute of Applied Ecology, Chinese
Academy of Sciences (no. LFEM2016- 05). We thank two anony-
mous reviewers for valuable comments that improved the manu-
script.
SUPPORTING INFORMATION
Additional Supporting Information may be found online in the
supporting information tab for this article.
LITERATURE CITED
Chantuma, P., A. Lacointe, P. Kasemsap, S. anisawanyangkura, E. Gohet, A.
Clément, A. Guilliot, et al. 2009. Carbohydrate storage in wood and bark
of rubber trees submitted to dierent level of C demand induced by latex
tapping. Tree Physiology 29: 1021–1031.
Chapin, F. S., E. D. Schulze, and H. A. Mooney. 1990. e ecology and economics
of storage in plants. Annual Review of Ecology and Systematics 21: 423–447.
Dang, H. S., K. R. Zhang, Q. F. Zhang, and Y. M. Xu. 2015. Temporal variations of
mobile carbohydrates in Abies fargesii at the upper tree limits. Plant Biology
17: 106–113.
Dawes, M. A., F. Hagedorn, I. T. Handa, K. Streit, A. Ekblad, C. Rixen, C. Körner,
et al. 2013. An alpine treeline in a carbon dioxide- rich world: Synthesis of a
nine- year free- air carbon dioxide enrichment study. Oecologia 171: 623–637.
Dawes, M. A., C. D. Philipson, P. Fonti, P. Bebi, S. Hättenschwiler, F. Hagedorn,
and C. Rixen. 2015. Soil warming and CO2 enrichment induce biomass shis
in alpine tree line vegetation. Global Change Biology 21: 2005–2012.
Dickson, R. 1989. Carbon and nitrogen allocation in trees. Annales Des Sciences
Forestières 46: 631–647.
El Zein, R., P. Maillard, N. Bréda, J. Marchand, P. Montpied, and D. Gérant. 2011.
Seasonal changes of C and N non- structural compounds in the stem sap-
wood of adult sessile oak and beech trees. Tree Physiology 31: 843–854.
Fajardo, A., and F. I. Piper. 2014. An experimental approach to explain the south-
ern Andes elevational treeline. American Journal of Botany 101: 788–795.
Fajardo, A., and F. I. Piper. 2017. An assessment of carbon and nutrient limita-
tions in the formation of the southern Andes tree line. Journal of Ecology
105: 517–527.
Fajardo, A., F. I. Piper, and L. A. Cavieres. 2011. Distinguishing local from global
climate inuences in the variation of carbon status with altitude in a tree line
species. Global Ecology and Biogeography 20: 307–318.
Fajardo, A., F. I. Piper, and G. Hoch. 2013. Similar variation in carbon storage be-
tween deciduous and evergreen treeline species across elevational gradients.
Annals of Botany 112: 623–631.
Fajardo, A., F. I. Piper, L. Pfund, C. Körner, and G. Hoch. 2012. Variation of
mobile carbon reserves in trees at the alpine treeline ecotone is under envi-
ronmental control. New Phytologist 195: 794–802.
Gansert, D., K. Backes, and Y. Kakubari. 1999. Altitudinal and seasonal variation
of frost resistance of Fagus crenata and Betula ermanii along the Pacic slope
of Mt. Fuji. Japan. Journal of Ecology 87: 382–390.
Givnish, T. J., S. C. Wong, H. Stuart-Williams, M. Holloway-Phillips, and G. D.
Farquhar. 2014. Determinants of maximum tree height in Eucalyptus species
along a rainfall gradient in Victoria, Australia. Ecology 95: 2991–3007.
Harsch, M. A., P. E. Hulme, M. S. McGlone, and R. P. Duncan. 2009. Are treelines
advancing? A global meta- analysis of treeline response to climate warming.
Ecology Letters 12: 1040–1049.
Hartmann, H., and S. Trumbore. 2016. Understanding the roles of nonstructural
carbohydrates in forest trees—from what we can measure to what we want
to know. New Phytologist 211: 386–403.
Hoch, G. 2015. Carbon reserves as indicators for carbon limitation in trees.
In: U. Lüttge and W. Beyschlag [eds.], Progress in botany. vol 76, 321–346.
Springer, Cham, Switzerland.
Hoch, G., and C. Körner. 2003. e carbon charging of pines at the climatic
treeline: A global comparison. Oecologia 135: 10–21.
Hoch, G., and C. Körner. 2005. Growth, demography and carbon relations of
Polylepis trees at the world’s highest treeline. Functional Ecology 19: 941–
951.
Hoch, G., and C. Körner. 2012. Global patterns of mobile carbon stores in trees
at the high- elevation tree line. Global Ecology and Biogeography 21: 861–871.
Hoch, G., M. Popp, and C. Körner. 2002. Altitudinal increase of mobile carbon
pools in Pinus cembra suggests sink limitation of growth at the Swiss treeline.
Oikos 98: 361–374.
Keel, S. G., and C. Schädel. 2010. Expanding leaves of mature deciduous forest
trees rapidly become autotrophic. Tree Physiology 30: 1253–1259.
Körner, C. 1998. A re- assessment of high elevation treeline positions and their
explanation. Oecologia 115: 445–459.
Körner, C. 2003a: Alpine plant life: Functional plant ecology of high mountain
ecosystems, 2nd ed. Springer, Berlin, Germany.
Körner, C. 2003b. Carbon limitation in trees. Journal of Ecology 91: 4–17.
Körner, C., and J. Paulsen. 2004. A world- wide study of high altitude treeline
temperatures. Journal of Biogeography 31: 713–732.
Kozlowski, T. T. 1992. Carbohydrate sources and sinks in woody- plants. Botan-
ical Review 58: 107–222.
Landhäusser, S. M. 2011. Aspen shoots are carbon autonomous during bud
break. Trees 25: 531–536.

8 • American Journal of Botany
Landhäusser, S. M., and V. J. Lieers. 2012. Defoliation increases risk of carbon
starvation in root systems of mature aspen. Trees- Structure and Function 26:
653–661.
Lenz, A., Y. Vitasse, G. Hoch, and C. Körner. 2014. Growth and carbon relations
of temperate deciduous tree species at their upper elevation range limit. Jour-
nal of Ecology 102: 1537–1548.
Li, M.-H., G. Hoch, and C. Körner. 2001. Spatial variability of mobile carbohy-
drates within Pinus cembra trees at the alpine treeline. Phyton- Annales Rei
Botanicae 41: 203–213.
Li, M.-H., W.-F. Xiao, S.-G. Wang, G.-W. Cheng, P. Cherubini, X.-H. Cai, X.-L.
Liu, et al. 2008. Mobile carbohydrates in Himalayan treeline trees I. Evidence
for carbon gain limitation but not for growth limitation. Tree Physiology 28:
1287–1296.
Martínez-Vilalta, J., A. Sala, D. Asensio, L. A. Galiano, G. N. Hoch, S. Palacio, F.
I. Piper, et al. 2016. Dynamics of non- structural carbohydrates in terrestrial
plants: A global synthesis. Ecological Monographs 86: 495–516.
Osaki, M., T. Shinano, and T. Tadano. 1991. Redistribution of carbon and nitro-
gen compounds from the shoot to the harvesting organs during maturation
in eld crops. Soil Science and Plant Nutrition 37: 117–128.
Piper, F. I., and A. Fajardo. 2014. Foliar habit, tolerance to defoliation and their
link to carbon and nitrogen storage. Journal of Ecology 102: 1101–1111.
Piper, F. I., B. Vinegla, J. C. Linares, J. J. Camarero, L. A. Cavieres, and A. Fajar-
do. 2016. Mediterranean and temperate treelines are controlled by dierent
environmental drivers. Journal of Ecology 104: 691–702.
Pratt, R. B., and A. L. Jacobsen. 2017. Conicting demands on angiosperm
xylem: Tradeos among storage, transport and biomechanics. Plant, Cell &
Environment 40: 897–913.
Sala, A., D. R. Woodru, and F. C. Meinzer. 2012. Carbon dynamics in trees:
Feast or famine? Tree Physiology 32: 764–775.
Schädel, C., A. Blöchl, A. Richter, and G. Hoch. 2009. Short- term dynamics of
nonstructural carbohydrates and hemicelluloses in young branches of tem-
perate forest trees during bud break. Tree Physiology 29: 901–911.
Shi, P., C. Körner, and G. Hoch. 2008. A test of the growth- limitation theory for
alpine tree line formation in evergreen and deciduous taxa of the eastern
Himalayas. Functional Ecology 22: 213–220.
Sveinbjornsson, B., O. Nordell, and H. Kauhanen. 1992. Nutrient relations of
mountain birch growth at and below the elevational tree- line in Swedish
Lapland. Functional Ecology 6: 213–220.
Tranquillini, W. 1979: Physiological ecology of the alpine timberline: Tree exis-
tence at high altitude with special reference to the European Alps. Springer,
Berlin, Germany.
Wardle, P. 1993. Causes of alpine timberline: A review of the hypotheses. In: J.
Alden, J. L. Mastrantonio, and S. Odum [eds.], Forest development in cold
climates, 89–103. Plenum Press, New York, New York, USA.
Yu, D. P., G. G. Wang, L. M. Dai, and Q. L. Wang. 2007. Dendroclimatic analysis
of Betula ermanii forests at their upper limit of distribution in Changbai
Mountain, Northeast China. Forest Ecology and Management 240: 105–113.
Yu, D. P., Q. L. Wang, J. Liu, W. Zhou, L. Qi, X. Wang, L. Zhou, et al. 2014. For-
mation mechanisms of the alpine Erman’s birch (Betula ermanii) treeline on
Changbai Mountain in Northeast China. Trees—Structure and Function 28:
935–947.
Zhu, W. Z., M. Cao, S. G. Wang, W. F. Xiao, and M. H. Li. 2012. Seasonal dynam-
ics of mobile carbon supply in Quercus aquifolioides at the upper elevational
limit. PLOS One 7: e34213.