ARTICLE IN PRESS
Journal of
Arid
Environments
Journal of Arid Environments 61 (2005) 13–25
www.elsevier.com/locate/jnlabr/yjare
Variation in nitrogen economy of two Stipa
species in the semiarid region of northern China
Z. Yuana, L. Lia,Ã, X. Hana, S. Wana,b, W. Zhangc
aLaboratory of Quantitative Vegetation Ecology, Institute of Botany, The Chinese Academy of Sciences,
Xiangshan, Beijing 100093, China
bEnvironmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6422, USA
cSchool of Biological Sciences, The Flinders University of South Australia, G.P.O. Box 2100, Adelaide,
SA 5001, Australia
Received 16 March 2004; received in revised form 11 May 2004; accepted 11 August 2004
Abstract
How effectively plants utilize nitrogen (N), the most limiting nutrient in natural ecosystems,
will largely determine the success of plants in intra- and inter-specific competition. Theory
suggests that there should be a trade-off between the two components of nitrogen-use
efficiency (NUE, total net primary production per unit N absorbed), i.e., nitrogen productivity
(A) and mean residence time (MRT) of nitrogen in plant tissues, and that this trade-off
depends on the N availability of the habitats. A field experiment was conducted in 20
grasslands in the semiarid region of northern China to examine A, MRT and NUE in two
Stipa species (S. grandis and S. krylovii) in habitats with different soil N availability. Our
results showed substantial differences in the N economy between S. grandis in dry and N-poor
habitats and S. krylovii in wetter and N-rich habitats. S. grandis had a significantly higher A
but relatively lower MRT than the less productive species, S. krylovii. NUE of S. grandis was
higher than that of S. krylovii due to the fact that the increment of A was higher than the
decrement of MRT. Within each species, no parameter correlated with the concentration of
total soil nitrogen. There was a negative relationshipbetween A and MRT within species in
relation to soil water and nitrogen. Our results suggest that the trade-off between A and MRT
ÃCorresponding author. The Chinese Academy of Sciences, Ecology Center, Institute of Botany, No. 20
Nanxincun, Xiangshan, Haidian District, Beijing 100093, China. Tel.: +86-1062-591-431.
E-mail addresses: zyyuan@ns.ibcas.ac.cn (Z. Yuan), linghao@ns.ibcas.ac.cn (L. Li).
0140-1963/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jaridenv.2004.08.002

---

ARTICLE IN PRESS
14
Z. Yuan et al. / Journal of Arid Environments 61 (2005) 13–25
depends on the range of soil N availability of the habitats and can apply to grassland
vegetations with a narrow range of soil N availability.
r 2004 Elsevier Ltd. All rights reserved.
Keywords: Habitat preference; Mean residence time; Nitrogen-use efficiency; Nitrogen productivity; Plant
strategies
1. Introduction
It has well been documented that soil nitrogen (N) availability often limits
plant growth in terrestrial ecosystems (Chapin, 1980; Vitousek and Howarth, 1991).
The strategy and effectiveness of N use are critical for the success of plants in
intra- and interspecies competition in natural ecosystems, particularly in habitats
with low N availability (Small, 1972). N use efficiency (NUE) has been simply
defined as the biomass production per unit absorbed N (Hirose, 1975; Vitousek,
1982), which is an important parameter to understand adaptation of plants to
different N availabilities (Tateno and Chapin, 1997; Aerts and Chapin, 2000).
However, Berendse and Aerts (1987) suggested that, due to evolutionary trade-offs,
plant characteristics that tend to enhance growth rate and dry matter production are
negatively correlated with traits that helpto reduce nitrogen losses. Therefore they
further redefined NUE as the product of the nitrogen productivity (A, growth rate
per unit N in the plant; Ingestad, 1979) and the mean residence time of N in the
plant (MRT) to make the concept more biologically relevant. The separation of
NUE into A and MRT facilitates a functional interpretation of NUE in terms of
different N economies (Garnier and Aronson, 1998; Eckstein et al., 1999; Aerts and
Chapin, 2000).
Plants may improve NUE through increasing either A or MRT or both. However,
previous studies suggested that a trade-off exists between A and MRT: species in N-
rich habitats tended to have higher A and shorter MRT whereas species in N-poor
habitats have lower A and longer MRT (Aerts, 1990; Eckstein and Karlsson, 1997;
Garnier and Aronson, 1998). For example, two evergreen shrubs, Erica tetralix and
Calluna vulgaris, occurring in N-poor habitats, had a longer MRT and a lower A
than a grass species, Molinia caerulea, in heathlands (Aerts, 1990). Both components
(A and MRT) changed proportionally among these species, resulting in a similar
NUE in both life-forms. The results suggest a trade-off between A and MRT. The
proposed trade-offs were found in studies focusing on interspecific differences (Aerts,
1990; Eckstein and Karlsson, 1997; Yasumura et al., 2002) but not in other studies
dealing with congeneric or within-species comparisons (Aerts and de Caluwe, 1994b;
Weih et al., 1998; Hikosaka and Hirose, 2001; Nakamura et al., 2002). However,
these studies dealing with variations in A and MRT among plants with the same life
form were carried out under controlled conditions. We still do not know if the trade-
off between A and MRT exists among plants growing in their natural habitats and, if
it does, whether it is related to habitat fertility. Furthermore, all studies on NUE
sensu Berendse and Aerts (1987) so far have exclusively addressed the species from

---

ARTICLE IN PRESS
Z. Yuan et al. / Journal of Arid Environments 61 (2005) 13–25
15
lowland mires or subarctic tundra, but to our knowledge, the species from the
semiarid region have not been studied. To test the existence of the hypothesized
trade-off between A and MRT, we examined the variation in N economy within and
between two broadly distributed Stipa species in the semiarid region of northern
China.
2. Materials and methods
2.1. Sites and species
The study was carried out in the semiarid region of northern China. We chose 20
relatively well-preserved grasslands throughout Duolun County (1151560–1161510E,
411490–421270N), which is located on the southern edge of the Hunshandake
Sandland in the center of Inner Mongolia Autonomous Region, China (Fig. 1). This
area belongs to a typical agro-pastoral ecotone with typical semiarid monsoon
climate of moderate temperature zone. Mean annual precipitation is around
385.5 mm and mean annual temperature is 1.6 1C, with mean monthly temperature
Fig. 1. Location of the sites.

---

ARTICLE IN PRESS
16
Z. Yuan et al. / Journal of Arid Environments 61 (2005) 13–25
ranging from –18.3 1C in January to 18.5 1C in July. Frequently, there is hard wind in
winter and spring. Snow accumulated on the floor remains until mid-may.
Within the 20 grasslands, we set upa total of 108 samp
ling sites in stable
communities dominated by monocultures of each of these two species: 56 for S.
grandis and 52 for S. krylovii. All the sites are old permanent grasslands with a wide
range of soils regarding their nitrogen content and their water supply. The
managements, very different from each other (grazing in autumn for some), have
been applied for many years. Species composition and vegetation structure of these
sites are representative of Stipa grasslands. The two Stipa species we selected are
dominant species in Inner Mongolia Grassland of northern China: S. grandis is tall
bunchgrass and can grow to almost 1 m at the peak of the growing season (late
August). S. Krylovii is smaller than S. grandis. They are both perennial C3 grasses
and the above-ground parts of these species were wholly dead until early spring from
autumn. In the semiarid region of northern China, the two Stipa species often form
nearly pure stands.
2.2. Soil and vegetation sampling
The soils of each of these sites were sampled (5 cm diameter) at 0–20 cm depth and
analysed for water moisture and total nitrogen (Kjeldahl). Percent soil moisture was
determined by mass loss upon drying at 105 1C for 24 h. The units used to present the
soil nitrogen are mg gÀ1 dry soil. We harvested above-ground plant parts in August
and November 2003 at each site. In August, when the vegetation was at its maximum
height, we mowed a 1 Â 1 m2 quadrat in the above-ground vegetation per site and
removed non-Stipa species and dead material from the samples. After drying for 48 h
at 70 1C in an oven, we measured the total dry mass of each quadrat sample and that
of 30 Stipa shoots selected randomly from the quadrat. These data were used to
calculate above-ground Stipa biomass per unit area and per shoot. In November,
when the above-ground parts of the Stipa species were dead, we randomly sampled
30 dead shoots and measured their dry weights, as above. All shoot samples were
milled, and their total N concentration was measured with an NC analyzer (KDY-
9820, Tongrun Ltd., China). Nitrogen contents of green shoots were calculated from
N concentrations and dry weights in July; those of dead shoots were calculated from
the equivalent November data.
2.3. Calculations
Calculations of A, MRT and NUE were based on the assumption that the Stipa
stand was at a steady state: matter inflow equaled matter outflow in the stand on an
annual basis. We considered that our study sites met this criterion because they were
mature climax Stipa communities. We calculated A, MRT and NUE as follows:
A ¼ NPP=NPOOL;
(1)
MRT ¼ NPOOL=NLOSS;
(2)

---

ARTICLE IN PRESS
Z. Yuan et al. / Journal of Arid Environments 61 (2005) 13–25
17
NUE ¼ A Â MRT ¼ NPP=NLOSS;
(3)
where NPP, NPOOL and NLOSS are net primary productivity, mean N content in a
growing season, and N loss, respectively.
We used the above-ground biomass of dead shoots in November for NPP, because
the populations would be stable and the amounts of carbon and N translocated from
roots to shoots in spring would have equilibrated with those reallocated from shoots
to roots in autumn. NPOOL of above-ground plants was calculated as the mean value
of the highest N content and the lowest N content. The lowest N content in above-
ground plants (green parts) of the perennial Stipa species was zero, because the
above-ground parts of those species were wholly dead until early spring from
autumn. In consequence, the NPOOL was calculated by mean of zero (in November)
and the highest N content (in August), i.e. half of the N content per green shoot in
August. Thus NUE is the reciprocal of N concentration in dead shoots and is equal
to the product of A and MRT.
When the above-ground parts of those grass species were wholly dead, part of
tissue N was resorbed into storage organs. The fraction of the N pool annually
resorbed prior to dead was then calculated as N resorption efficiency (NRE)
(Yasumura et al., 2002; Escudero and Mediavilla, 2003):
NRE ¼ ðmax NPOOL À N content in NovemberÞ= max NPOOL;
(4)
where max NPOOL denotes N pool at the seasonal maximum (in August).
Additionally, the N concentration of senesced parts (in November) was directly
used as an indicator of N resorption proficiency (NRP) sensu Killingbeck (1996).
NRP can be viewed as a measure of complete N resorption in terms of proximity to a
theoretical lower limit on N concentrations in senesced plant parts.
2.4. Statistical analyses
Statistical tests were performed using SPSS version 10.0 (SPSS Inc., Chicago, IL,
USA). NRE data were arcsin-transformed before analyses. All data sets had a
variance that was not significantly different from the normal distribution (p40.05).
Difference between two variables was tested by student’s t-test. The Spearman
correlation was employed to assess the relationship among the various parameters
related to soil nitrogen and soil water moisture.
3. Results
The above-ground biomass per unit area in August was not different between the
two Stipa species (F=1.704, p=0.195). However, the fast-growing species, S. grandis,
had higher above-ground biomass per shoot in August than the slow-growing
species, S. krylovii (po0.001, Table 1). S. grandis was restricted to the dry sites and
infertile habitats. In contrast, S. krylovii occurred mostly in wetter or fertile habitats
(Fig. 2). There were significant differences in soil water content (po0.01) and in soil

---

ARTICLE IN PRESS
18
Z. Yuan et al. / Journal of Arid Environments 61 (2005) 13–25
Table 1
Above-ground biomass, N concentration and N content of green shoots in August and dead shoots in
November, N use traits, averaged across sites (see Fig. 1)
Parameter
Condition
Stipa species
S. grandis
S. krylovii
Above-ground biomass
Green (g m–2)
113.9478.79a
98.8276.74a
Green (g shoot–1)
1.6870.08a
0.4570.02b
Dead (g shoot–1)
1.7570.05a
0.4470.02b
N concentration
Green (mg g–1)
5.0570.10a
6.5970.13b
Dead (mg g–1)
2.4770.06a
3.4970.09b
N content
Green (mg m–2)
569.51744.79a
647.30744.70b
Green (mg shoot–1)
8.5970.48a
2.9970.16b
Dead (mg shoot–1)
4.4370.16a
1.4670.06b
N use traits
A
452.74723.08a
276.09714.25b
MRT
1.0970.04a
1.2370.05b
NUE
417.84710.50a
306.2978.93b
Data are means7SE. Different letters between means indicate statistical difference at 5% level (Students’
t-test).
4
4
) -1
3
3
2
2
1
1
Aboveground biomass (g shoot
0
0
0
3
6
9
12
0
0.5
1
1.5
2
2.5
soil moisture (%)
soil nitrogen (mg g-1)
Fig. 2. Variation in above-ground biomass per shoot in August in relation to soil moisture and total
soil nitrogen ( S. grandis, J S. krylovii). Each point represents a sampling site in the twenty grasslands
(Fig. 1).
nitrogen (po0.001) between the habitats dominated by the two Stipa species. N
concentration differed significantly between species (po0.01). In both species, green
shoots had significantly higher N concentration than the dead shoots (po0.001)
(Table 1).

---

ARTICLE IN PRESS
Z. Yuan et al. / Journal of Arid Environments 61 (2005) 13–25
19
The variations in NUE with soil water moisture and soil nitrogen were similar to
those in A (Fig. 3). S. grandis had a significantly higher A (po0.001), but lower MRT
(po0.05) than S. krylovii. NUE ranged from 260.65 to 666.77 g gÀ1 in S. grandis and
from 167.89 to 431.86 g gÀ1 in S. krylovii (Fig. 3). Results of ANOVA showed that
NUE differed significantly between species (po0.01). The lower NUE in S. krylovii
was mainly due to a lower A (Table 1).
There was a significant difference both in NRE and N resorption proficiency
(NRP) between the two species. NRE was significantly lower in S. grandis than in S.
krylovii (Fig. 4). High resorption proficiency implies low concentrations because a
low N concentration in senesced parts is an evidence of high proficiency, and vice
versa (Killingbeck and Whitford, 2001). NRP was higher in S. grandis than in S.
krylovii—this contrasted against NRE (Fig. 4).
NUE of S. grandis and A of S. krylovii correlated negatively with soil water
moisture across different sites (Table 2). However, there were no significant
correlations between any N use trait and soil nitrogen in either species. MRT of the
two Stipa species did not correlate significantly with either soil water moisture or soil
nitrogen. There were no significant correlations of resorption efficiency (NRE) with
either soil water moisture or soil nitrogen. NRP of S. grandis significantly correlated
with soil water moisture, but not with soil nitrogen. NRP of S. krylovii was not
significantly correlated with soil water moisture or soil nitrogen (Table 2).
The species with higher NPP lost more N than species with lower NPP. The same
pattern was found within species across different sites (Fig. 5a). There was a negative
relationshipbetween A and MRT within S. grandis (r=–0.810, po0.01), S. krylovii
(r=–0.736, po0.01), and among all sites (r=–0.725, po0.01) (Fig. 5b).
4. Discussion
We examined the nitrogen use efficiency (NUE) in two perennial grasses, S.
grandis and S. krylovii, by distinguishing two components: nitrogen productivity (A)
and mean residence time (MRT) according to the NUE index of Berendse and Aerts
(1987). Our results showed that there were substantial differences in NUE and
habitat preference between the two Stipa species in the semiarid region of northern
China, which was consistent with previous assumptions (Chapin, 1980; Vitousek,
1982). However, relationshipbetween A and MRT among habitats was inconsistent
with the predictions of Eckstein and Karlsson (2001).
The N-poor sites were dominated by productive species (S. grandis) with large A
values (Table 1, Fig. 3). The possible advantages of large A in potentially productive
species in infertile habitats have not previously been considered. A higher A is
associated with rapid growth, a relatively large investment of N in photosynthetic
tissue, an efficient photosynthetic N use in leaves, and a relatively small proportion
of carbon used for respiration (Garnier and Aronson, 1998; Lambers et al., 1998).
The lower A of S. krylovii probably is caused by morphological differences in
response to differences in nutrient conditions and self-shading effects in the canopy
(Aerts and de Caluwe, 1994a), resulting in lower photosynthetic activity (Hirose and

---

ARTICLE IN PRESS
20
Z. Yuan et al. / Journal of Arid Environments 61 (2005) 13–25
1000
1000
800
800
)-1 600
600
yr -1
(g g
400
400
A
200
200
0
0
0
3
6
9
12
0
0.5
1
1.5
2
2.5
soil moisture (%)
soil nitrogen (mg g-1)
4
4
2
2
MRT (yr)
0
0
0
3
6
9
12
0
0.5
1
1.5
2
2.5
soil moisture (%)
soil nitrogen (mg g-1)
800
800
600
600
)-1
400
400
NUE (g g
200
200
0
0
0
3
6
9
12
0
0.5
1
1.5
2
2.5
soil moisture (%)
soil nitrogen (mg g-1)
Fig. 3. Variation in nitrogen productivity (A), mean residence time of nitrogen (MRT) and nitrogen use
efficiency (NUE) in relation to soil moisture and total soil nitrogen ( S. grandis, J S. krylovii).

---

ARTICLE IN PRESS
Z. Yuan et al. / Journal of Arid Environments 61 (2005) 13–25
21
80
6
NRE
NRP
60
b
4
a
b
)-1
E (%) 40
R
a
N
NRP (mg g
2
20
0
0
S. grandis
S. krylovii
Fig. 4. NRE and NRP of S. grandis and S. krylovii. Means with SE. Differences between pairs referred to
that between species and were tested using the Students’ t-test after arcsin transformation. Different letters
indicate a statistical difference at p=0.05.
Table 2
Correlation coefficients between each N use trait (A, MRT and NUE), N resorption (NRE and NRP) and
soil moisture or soil nitrogen
Species
N use traits
Soil moisture
Soil nitrogen
S. grandis (n=56)
A
À0.020
À0.011
MRT
À0.147
À0.140
NUE
À0.317*
À0.256
NRE
À0.111
À0.103
NRP
0.305*
0.175
S. krylovii (n=52)
A
À0.326*
À0.035
MRT
0.225
0.077
NUE
À0.247
À0.033
NRE
0.146
0.090
NRP
À0.040
À0.057
Two Stipa species
A
À0.200
À0.144
MRT
0.073
0.095
NUE
À0.396**
À0.266
NRE
0.073
0.095
NRP
0.132
0.092
*po0.01, **po0.001.
Werger, 1987). Our results were not consistent with the conclusions of Nakamura et
al. (2002), who found no positive association between A of Carex and habitat
fertility. In contrast to Carex species of wetland, the Stipa species we used grew in

---

ARTICLE IN PRESS
22
Z. Yuan et al. / Journal of Arid Environments 61 (2005) 13–25
4
1000
800
)-1
)-1 600
yr

s
hoot 2
g
-1
(
P
400
P
(g g
N
A
200
0
0
0
2
4
6
8
10
0
1
2
3
4
(a)
N loss (mg shoot-1)
(b)
MRT (yr)
Fig. 5. (a) Relationship between net primary productivity (NPP) and nitrogen loss. (b) Relationship
between nitrogen productivity (A) and mean residence time of nitrogen (MRT).  S. grandis, J S. krylovii.
dry and N-poor habitats. Soil water availability plays a more important role than
soil N availability in regulating the ecological performance of the Stipa species in the
semiarid area. Thus the relationshipbetween A and habitat fertility in wetland may
not be applied to grassland vegetation in the semiarid region. Our results suggest
that large A is not necessarily a competitive advantage for plants in N-rich habitats
in the semiarid area. According to Berendse and Aerts (1987), however, plants that
are ecologically successful in N-rich habitats are more likely to have large A values.
Leaf life spans and resorption efficiency have been identified to be the two factors
in affecting MRT (Garnier and Aronson, 1998; Eckstein et al., 1999). The leaf life
spans in the two Stipa species are similar because they are same generic species. Thus
the different MRT between the two species could be attributable to the difference in
N resorption efficiency. NRE was significantly lower in S. grandis than S. krylovii
(Fig. 4), leading to the shorter MRT in S. grandis.
NUE of the two Stipa species observed in our study (Table 1) was relatively high
compared with other species in previous studies (Eckstein and Karlsson, 1997;
Va´zquez de Aldana and Berendse, 1997; Meuleman et al., 2002; Nakamura et al.,
2002). A high NUE is generally considered to be an adaptation to habitats with low
soil N availability (Chapin, 1980). Therefore, the high NUE in the two Stipa species
is associated with the infertile soils at the study sites.
Our results showed that there was a trade-off between A and MRT in the two
Stipa species (Fig. 5b), which was consistent with the results of Eckstein and
Karlsson (1997), who found a negative relationshipbetween A and MRT in 14
species in northern Sweden. Nevertheless, conclusions from other studies are
inconsistent (e.g. Aerts and de Caluwe, 1994b; Va´zquez de Aldana and Berendse,
1997; Weih et al., 1998; Nakamura et al., 2002). The inconsistency could be largely
contributed to the differences in defining NUE, the indices to measure its
parameters, and vegetation units are considered (single species vs. communities)

---

Document Outline

• Variation in nitrogen economy of two Stipa species in the semiarid region of northern China
  • Introduction
  • Materials and methods
    • Sites and species
    • Soil and vegetation sampling
    • Calculations
    • Statistical analyses
  • Results
  • Discussion
  • Acknowledgements
  • References

---