Botanical Studies (2006) 47: 293-306.

*

Corresponding author: E-mail: linhuilong@lzu.edu.cn; Tel:

+86-931-13893330034; Fax: +86-931-8914086.

INTRODUCTION

The Qinghai-Tibet Plateau, located in southwest

China, has a unique, high-altitude natural environment

and is richly endowed with natural resources. Most of the

area is alpine steppe and meadow (1,627 million km

2

).

These ecosystems play an important role in agriculture,

water conservation, biodiversity, and ecological safety.

However, they are also extremely fragile on account of

steep inclines, sparse vegetation, severe physical air-

slaking conditions, strong solar radiation, loose soil

texture, low fertility, and eroded soil. In the past decades,

they have suffered many eco-environmental problems

caused by long-term overgrazing, irrational human

activities, and natural factors (climate change leading to

a decrease in rainfall). At present, 61.3% of this alpine

steppe and meadow has degenerated and is progressively

converting to barren land. This has dramatically reduced

agricultural production and resulted in poverty for the

human inhabitants. Ways to develop these ecosystems

commercially and environmentally are urgently needed.

Any solution will need to take into consideration both

the need to increase agricultural production and farmer

income and the requirement for long-term sustainability

of the farming system. Exploiting the plentiful availability

of plant germplasm may be a key factor in creating a

breakthrough in achieving these purposes (Ren and Lin,

2005).

The Qinghai-Tibet Plateau is home to 25 wild species

of the endemic genus Microula Benth. in the family Bor-

aginaceae. The other four species can only be found in

the alpine regions of Sikkim, Nepal, Bhutan, and Kashmir

(Wang et al., 1997a). Microula sikkimensis is one of the

25, a biennial herbage rich in

γ

-linoenic acid. As a steno-

topic species, it was mainly confined to the eastern rims of

the Qinghai-Tibetan Plateau (including Gansu, Qinghai,

and Sichuan Provinces) and was typically associated with

degraded alpine steppe and meadow, especially at the be-

ginning of the secondary succession (Wang et al., 2003b).

eCOlOgy

Seed yield predictions based on the habitat niche-

fitness of Microula sikkimensis, an endemic oil crop in

the Qinghai-Tibet Plateau

Huilong LIN*, Jizhou REN, and Qin WANG

College of Pastoral Agriculture Science and Technology and Gansu Grassland Ecological Research Institute, Lanzhou

University, Lanzhou 730000, Gansu, People’s Republic of China

(Received June 13, 2005; Accepted December 16, 2005)

ABSTRACT.

Microula sikkimensis is a biennial herb, found only in the eastern rims of the Qinghai-Tibetan

Plateau, and lends itself to multiple applications in medicine, food and fodder. Utilizing the seed production

of M. sikkimensis to develop a sustainable production ecosystem is a logical option for the degraded alpine

grassland in the Qinghai-Tibet Plateau. From 1994 to 2004, a field survey and transplanting trials were con -

ducted in eleven counties of three provinces for collection of habitat factor data. By multivariate statistical

analysis the habitat factors were condensed into seven key habitat factors which described an actual habitat

state. Introducing the niche theory into the research of M. sikkimensis, habitat niche-fitness (HNF) is de-

fined as the degree of similarity of an actual habitat state to the optimum habitat. A new model of HNF is

constructed to evaluate the adaptive extent of M. sikkimensis and the influence of habitat on seed yield with

the key habitat factors as dependent variables and factor weights as parameters. The results showed that the

values of HNF had a reasonable distribution and better reflected the varied differences under different habitat

conditions, and that the cultivation measures had the effect of increasing the value of HNF and seed yield, in-

creasing 14.26% and 99.61% at an average level, respectively. A seed yield prediction model was constructed

with HNF as a surrogate for composite environmental factors. The estimated seed yield agreed well with the

observed data, and the average of the absolute deviation percent was 5.46%, demonstrating the validity of the

model in predicting seed yield. The HNF model and seed yield prediction model evaluated the threshold value

of HNF, predicted the upper limit of seed yield for each study site and the limit seed yield, and have a wide

range of prospects for practical application in the similar regions of the Qinghai-Tibet Plateau.

Keywords: Habitat niche-fitness (HNF); Microula sikkimensis; Seed yield; The Qinghai-Tibet Plateau.

294

Botanical Studies, Vol. 47, 2006

It possesses multiple applications as medicine, food, and

fodder. Pharmaceutical experiments have indicated that

its seed oil can significantly decrease cholesterol and tri-

lycerides in blood serum, increase the ratio between total

cholesterol and high-density lipoprotein, prevent the ac-

cumulation of atheroma, preserve the structural integrity

of biological membranes and inner membranes of ves-

sels, and alleviate high blood fat (Li et al., 1999a, b). The

coarse stalks of M. sikkimensis contain abundant mineral

nutrient elements and crude protein (Wang et al., 2003b)

and can be processed into fodder. It has been estimated

that high profits could be made using this new feed re-

source to raise animals in winter and spring (Zhang and

Deng, 1999b). Therefore, exploiting this new promising

oil crop to develop a sustainable production ecosystem

is both a logical and natural option for the degraded al-

pine steppe and meadow ecosystems in the Qinghai-Tibet

Plateau. However, this plant is not yet utilized as a crop

because of its low seed output in nature; in fact, it is even

eliminated as a weed. Consequently, this species has faced

the threat of extinction for several decades. The initial

questions related to utilization and development of M. sik-

kimensis for farmers are: whether the location of interest

is a suitable habitat for cultivation of this species or not,

whether an acceptable seed yield can be harvested, and

what the potential seed yield will be when agricultural

cultivation technology is applied. Therefore, research on

habitat factors that determine seed yield and appropriate

habitats for cultivation of this species is required. Such

research could improve seed yield, encourage a potentially

new, ecologically sustainable industry, and predict where

the promising new crop will produce well.

A list of a species’ habitat requirements can be used

to predict the species’ presence or absence. From the

perspective of plant production, the factors that affect

plant growth fall into two categories: habitat factors—

of which temperature, water, and soil conditions are most

important—and agricultural cultivation technology. Under

the same cultivation conditions, it is obvious that habitat

factors play a key role in the existence and productivity

of a plant. These or associated habitat factors could be

used to identify suitable and unsuitable habitats for plant-

ing (Li and Lin, 1997; Lin and Li, 1998; Buggeman and

O’Nuallain, 2000; Liu et al., 2000). In general, the interac-

tions between habitat factors that affect the productivity

of plants are complex, multi-factorial, and often difficult

to describe mathematically, posing a challenge for seed

yield predictions based on traditional methods (Li and

Lin, 1997; Lin and Li, 1998). The fact that the seed yield

regressed directly with habitat factors under regression-

type models has lead them to be criticized for their empiri-

cal nature (Li and Ren, 1997; Scian and Bouza, 2005),

which created some difficulties in determining whether

M. sikkimensis could be introduced or not. It is difficult

to base an such an appraisal on direct judgments alone

(Jiang and Wang, 2004). Up to now, a number of scholars

have explored the physiological and ecological charac-

teristics of M. sikkimensis in such aspects as its distribu-

tion and population characteristics (Wang et al., 1997a;

1998a, b; 2003b), physiological traits (Niu, 1997; Wang

et al., 1997b, 1998c, d, e; Sun and Wang, 1998), chemical

composition in seed (An and Zheng, 1996; Fu et al., 1997,

1999; Zhen et al., 1997), medical value (Li et al., 1999a,

b), nutritional value (Zhang and Deng, 1999b), and breed-

ing of cultivars (Zhang and Deng, 1998, 1999a; Wang et

al., 2003a). However, quantitative research on distribution

regularities and habitat factors that determine seed yield

has not been done.

Understanding the ecological adaptability of the

species to different habitats is helpful for successful

conservation and is also useful for breeding for direct

utilization. Our work had three main goals: the first

was to extract the key habitat factors i.e., factors having

significant influence on the seed yield of M. sikkimensis

by Pearson Correlation and Partial Correlation Analyses.

The second goal was to discuss the meaning of habitat

niche-fitness (HNF) and construct its mathematical model

for this plant with the key habitat factors as dependent

variables and factor weights as parameters. The last was

to establish a prediction model for seed yield and to

validate the prediction model. It is hoped the HNF model

and seed yield prediction model will provide farmers

and practitioners with a viable tool for identifying likely

areas for cultivation of this species and applying the right

methods of fertilization, utilization, and development of

M. sikkimensis in similar regions of the Qinghai-Tibet

Plateau.

MATeRIAlS AND MeTHODS

From April 1994 to September 2004 field surveys

were carried out at eleven sites over a wide area of the

eastern rims of the Qinghai-Tibetan Plateau, using the

species’ responses to habitat factors as reliable criteria in

an analysis. In this study, transplanting trials adopting a

recommended cultivation technology resulted in habitat

modification. By transplanting trials at five sites selected

from the eleven field survey sites, the effects of habitat

modification were examined. Durations of field surveys

and transplanting trials (Month/Year) are shown in Table

1. Basic descriptive statistics of range, mean, and standard

error on the habitat factors are shown in Table 2.

Field surveys

The field survey sites were chosen in abandoned fields

of degenerated alpine steppe and meadow where M. sik-

kimensis had reached dominance in secondary succession.

The survey sites were situated in Hongyuan, Ruoergai,

Aba, and Maerkang Counties in Sichuan Province, Tian-

zhu, Hezuo, and Maqu Counties in Gansu Province, and

Menyuan, Haiyan, Gangcha, and Huangzhong Counties in

Qinghai Province (geographical coordinates from 100o08’

to 104o27’E and 31o50’ to 37o22’N). These sites were se-

lected because (1) they were the most representative sites

in degenerated alpine steppe and meadow; (2) they were

LIN et al. — The habitat niche-fitness of

Microula sikkimensis

295

within the major M. sikkimensis distribution areas; and

(3) their habitat characteristics varied greatly. Their me-

teorological and geographic features are shown in Table

1. Their weather typifies the continental plateau climate.

The growth season is very short, and total no-frost time is

70 to 190 days annually. Soil types are alpine steppe and

meadow soils. Consequently, it is difficult for trees to sur-

vive in this environment, which belongs to the alpine her-

bosa zone and is dominated by psychrotolerant vegetation.

The vegetative, flowering, and withering stages of M.

sikkimensis are late April to early May, late June to early

July, and mid to late September, respectively. In each field

survey site, five subplots of 1 m × 1 m were randomly se-

lected with a distance of 50 m between subplots. Soil and

vegetative attributes were observed once within subplots

in different growing season stages. The method of timing

and positioning, commonly accepted in agricultural sci-

ence, was used in observing the fields for field study of M.

sikkimensis growth and seed yield indexes. In mid Sep-

tember seeds of M. sikkimensis were harvested, dried, and

weighed to the nearest 0.01 g. Soil samples were taken

at the 0-30 cm layer in the subplots for different growing

stages. Soil analysis and extraction were carried out by

commercial laboratories adopting standard measurement

techniques (Honda, 1962; John, 1970; Black, 1979; Olsen

and Sommers, 1982; Gee and Bauder, 1986). Soil factors

measured were texture, soil moisture (SM), organic matter

(OM), soil acidity (pH), soil salinity (electrical conductiv-

ity, EC), total nitrogen (TN), alkaline hydrolytic nitrogen

(AHN), total phosphorus (TP), available phosphorus

(AVP), total potassium (TK), and available potassium

(AVK).

Transplanting trials

The transplanting trials were performed in locations

corresponding to the field survey sites in Hongyuan

County of Sichuan Province, Maqu and Tianzhu Counties

of Gansu Province, and Menyuan and Haiyan Counties

of Qinghai Province. Each transplanting plot at the dif-

ferent sites was 10 m × 10 m with 1 m buffer strips with

four replicates. Root tubers of M. sikkimensis supplied

by Gansu Grassland Ecological Research Institute via

quick regeneration multiplication tissue culture (Wang et

al., 2003a) were transplanted by hand in mid-April. Con-

sidering secure fertilizer criterion (Lu and Yang, 1997),

sheep manure at a rate of 7,500 kg/ha

were applied before

transplanting. Fertilizer nitrogen (urea, 46% N) at a rate

of 86 kg/ha and fertilizer phosphorus (super phosphate,

12% P

2

O

5

) at a rate of 54 kg/ha were split into three equal

amounts. The first amount was added during the land

preparation prior to transplanting; the second was added

30 days after transplanting, and the final amount at cyme

initiation. Because the soil of the study region is rich in

potassium, none was applied. Weeding was performed

Table 1. Meteorological and geographic features of the study sites.

Site

North latitude

(oN)

Altitude

(m)

Maximum

mean

temperature of

Jul. (°C)

Minimum

mean

temperature

of Jan. (°C)

Annual>0°C

accumulated

temperature

(°C)

Annual

mean

precipitation

(mm)

Annual

mean

temperature

(°C)

Duration of

field survey

(Month/

Year)

Duration of

transplanting

trial (Month/

Year)

Huangzhong 36o25’ 2667.5 12

-8

2065 527.6 2.8 4-9/2000

Maqu

34o24’ 2707.6 11.9

-15.4 1422.6 514.5 0.5 4-9/1997 4-9/2002

Hezuo

35o05’ 2915.7 12.6

-9.8 1729.3 558.1

0 4-9/1997

Tianzhu

37o18’ 3045.1 11.3

-11.4 1327.7 411.3 -0.2 4-9/1996 4-9/2001

Haiyan

36o56’ 3080

7.5

-18.1 1528.6 397.44 -0.3 4-9/1999 4-9/2004

Aba

32o54’ 3275 12.15

-5.85 1891.9 712

3.2 4-9/1995

Gangcha

37o20’ 3301.5 10.6

-13.6 1225.2 377.1 -0.6 4-9/2000

Ruoergai 33o20’ 3446

9.6

-9

1518.3 651.3 0.6 4-9/1994

Mengyuan 37o27’ 3471.6 10.7

-9.2 1404.4 615.5 1.4 4-9/1998 4-9/2003

Hongyuan 32o48’ 3504

10.1

-8.38 1432.2 728.4 1.1 4-9/1994 4-9/1997

Maerkang 31o50’ 3664.4 16.2

-0.4 3192.2 753.1 8.6 4-9/1995

Note: Meteorological data for each study site derived from located grassland research stations were calculated for a ten-year period

(1994-2004).

296

Botanical Studies, Vol. 47, 2006

thrice annually. The experimental field was not irrigated.

Other field management practices were identical to those

for other crops grown at this area. Cultivar density was 24

plants/m

2

in Hongyuan, 16.7 in Tianzhu, 23 in Maqu, 25

in Menyuan, and 20 in Haiyan, according to soil test rec-

ommendations.

The items measured in each plot of the transplanting

trials were the same as those in the field survey sites.

Data analysis methods

In order to identify habitat factors determining seed

yield, Pearson Correlation among various habitat factors

and Partial Correlation Coefficients between seed yield

and habitat factors were analyzed with SPSS (Statistics

Package for Social Science; SPSS, 1997). The principle

used was that the factors that played an important

role in the growth of M. sikkimensis would be kept as

target characters (key habitat factors) while those that

played the least important role would be discarded. The

Partial Correlation Analysis was used for measuring the

correlation between seed yield and each habitat factor

while eliminating the effects of all other habitat factors

in the dataset. Thus the distribution regularities could be

denoted by fewer key habitat factors.

The quantitative indexes of these key habitat factors

can be marked as x

1

, x

2

,…x

n

. The observation values of

each group in natural (field surveys) or transplanting trial

condition can be noted as X = (x

1

, x

2

,…x

n

). X stands for a

realized habitat state or a modified habitat state. Biologi -

cally, a crop will show certain adaptation to variables of

each key habitat factor, so the optimum value of key habi-

tat factor i can be marked as x

ai

(i = 1, 2, . . ., n). x

ai

can

be obtained from experimental observation (Li and Lin,

1997; Lin and Li, 1998). X

a

= (x

1a

, x

2a

,…x

na

) is a quantita-

Table 2. Descriptive Statistics of habitat variables.

Habitat variables

Abbr. Unit

Range

Mean value Standard error C.V. (%)*

North latitude

NL oN 31 o35’-37 o27’ 35o01’

2o17’

6.22

Altitude

AL m

2667.5-3664.4 3188.95

331.99

10.41

Maxium temperature of Jul.

MT

°C

17.5-26.2

11.33

2.17

19.15

Minimum temperature of Jan.

NT

°C

-38.1 – -20.4

-9.92

4.77

48.08

Annual mean temperature

AMT

°C

-0.6-8.6

1.55

2.64

179

Annual >0

°C

accumulated temperature AT

°C

1225.2-3192.2

1703.4

553.17

32.47

Annual average precipitation

AP mm

377-753.1

567.84

135.87

23.93

Soil moisture

SM

%

11.22-68.79

23.61

9.93

32.66

Soil clay grain content (<0.01 mm)

PS g/kg

107-159

129.5

22.5

17.37

Soil acidity

pH

4.8-8.5

6.8

2.34

34.4

Soil salinity (electrical conductivity)

EC dS/m

2-8

6

1.8

30

Soil total nitrogen

TN

%

0.16-0.70

0.27

0.13

48.15

Soil alkaline hydrolytic nitrogen

AHN mg/kg 21.20-50.19

30.72

10.96

35.68

Soil total phosphorus

TP %

0.045-0.155

0.08

0.03

37.5

Soil available phosphorus

AVP mg/kg

0.2-7.5

6.2

3.70

59.68

Soil organic matter

OM

%

6.05-22.5

12

7.75

64.58

Soil total potassium

TK

%

1.41-4.05

2.5

0.825

33

Soil available potassium

AV K mg/kg

133-315

198.5

62.75

31.61

Plant density

PD plants/m

2

6.67-51.20

20.19

12.31

60.97

Seed yield

Y kg/ha

50-710

227.94

153.51

67.35

Note: *Coefficient of variation: 0-15% (least variation), 15-35% (moderate), >35% (most varied).

LIN et al. — The habitat niche-fitness of

Microula sikkimensis

297

tive description of species attributes for the optimum habi-

tat requirements. The balance between requirement for

the optimum habitat and supply of a realized habitat in the

development of the species is an important characteristic

and can be measured by habitat niche-fitness (HNF). Tak-

ing M. sikkimensis as the study object and taking the key

habitat factors into consideration, we suggest that HNF for

M. sikkimensis be defined as the degree of similarity be-

tween supply of an actual habitat and requirement for the

optimum habitat, in which supply of the actual habitat and

requirement for the optimum habitat denote realistic habi-

tat conditions and species attributes, respectively. This is a

measurement of the "n-dimensional hypervolume" defined

by Hutchinson (1957). The mathematical model for HNF

can be expressed as follows:

)

,

,

(

K

X

X

f

F

a



.................................................. (1)

In this formula, the value of HNF F, which is in the

range of [0, 1], means the fitness degree of M. sikkimensis

in an actual habitat condition. The larger the value of F,

the higher the adaptive property for its habitat. Normally,

when the actual habitat state may change on the large-scale

eco-geographical regions, F value needs a wide distribu-

tion on the subset [0, 1]. f(X, X

a

, K) is the measurement of

the distance or the degree of similarity between two vec-

tors: X = (x

1

, x

2

,…x

n

) and X

a

= (x

1a

, x

2a

,…x

na

). However, in

these rain-fed farmland systems, the importance of vari-

ous key habitat factors to M. sikkimensis varies. Thus, we

have to consider the actual conditions of unequal weights

among the key habitat factors and extend Eq. (1), where

vector K = (k

1

, k

2

,…k

n

) is a set of weights of the extracted

key habitat factors and k

i

the weight coefficient of the key

habitat factor i. In this study, the weight coefficient of key

habitat factor is integrated by factor loadings derived from

the PRINCOMP procedure of SAS (SAS, 2000). The fac-

tor loading is the correlation coefficient between the prin-

cipal component and the key habitat factor, and its size

depicts the key habitat factor’s influence.

The data sets collected from the study sites were

used to construct a seed yield prediction model by

regression analyses of ln-transformed dependent (seed

yield per hectare) and independent (HNF) variables. Ln

transformations functioned by converting values to a

scale where the variance in the relationship was more

homogeneous for effective use of least-squares regression

(Steel and Torrie, 1980). The seed yield prediction model

with HNF as a surrogate for composite environmental

factors was validated based on the statistical and

biological requirements. Statistical validation was done

first through the coefficient of determination (R

2

), the

adjusted R

2

and the standard error of the estimate. For

biological verification, the observed seed yield per hectare

was collectively compared with the corresponding value

estimated by the seed yield prediction model with the

help of absolute deviation percent. The absolute deviation

percent was given by:

100

×

.

=

obs

obs

est

ADV

Y

Y

Y

e

............................. (2)

Where: e

ADV

= absolute deviation percent; Y

est

= estimated

seed yield per hectare by the seed yield prediction model;

Y

obs

= observed seed yield per hectare.

ReSUlTS

Key habit factors selection

The seed yield was significantly correlated with the

altitude (AL) and with the north latitude (NL). There was

also a negative correlation between AL and NL (Table 3).

In general, these survey regions with a high north latitude

also had a low altitude. Partial correlation coefficients be-

tween seed yield and NL and AL suggested that AL, which

was one of the indirect causes of the variability in seed

yield, could be chosen as a key habitat factor, and named

x

1

, whereas NL is neglected (Table 4). Microula sikkimen-

sis is mainly concentrated in degraded alpine steppe and

meadow, so the distribution of M. sikkimensis is associated

with altitude. It is distributed only in the altitude range

from 2,600 to 5,000 m. It is discovered to have increased

within areas of altitude 2,600-3,500 m, decreased in areas

of altitude 3,500-5,000 m, and disappeared where the al-

titude is over 5,000 m. The optimum AL value x

a1

= 3,500

m (Wang et al., 2003b).

As the growth period of M. sikkimensis is short (from

April to September) the quantity of heat needed is an

important factor. According to physiological experiment

results, M. sikkimensis is a medium cold resistant plant

that can grow at no less than 0oC (Wang et al., 1997b,

1998c,d). The heat requirement is expressed in terms of

growing-degree-days, i.e. annual .0oC accumulated tem-

perature (AT), including in the case of M. sikkimensis, for

which the annual mean temperature (from April to Sep-

tember) (AMT, in oC) is used. The cold and hot tolerances

of the species are expressed in terms of minimum and

maximum temperature of the coldest and warmest month

(NT and MT), respectively. AMT, AT, MT and NT were

significantly correlated with each other as well as with

seed yield (Table 3). Absolute value of partial correlation

coefficient between seed yield and AT was the biggest, so

AT was chosen as a key habitat factor and named x

2

while

the others were omitted (Table 4). The optimum AT value

is x

a2

= 1422oC (Wang et al., 1997b, 1998c,d, 2003b).

Annual average precipitation (AP) and soil moisture

(SM), during the M. sikkimensis growth cycle, had a

significant correlation with each other as well as being

strongly related to seed yield (Table 3), but a partial cor-

relation coefficient between seed yield and AP was larger

than that between seed yield and SM. Therefore, AP was

chosen as a key habitat factor and named x

3

(Table 4). The

field survey data shows that precipitation is beneficial for

branching and tillering when AP reaches 400 mm. When

AP is greater than 600 mm, the tillering and squaring abil-

298

Botanical Studies, Vol. 47, 2006

LIN et al. — The habitat niche-fitness of

Microula sikkimensis

299

ity of M. sikkimensis is limited; when AP is above 800

mm, M. sikkimensis becomes waterlogged, and its growth

is restricted, and it may even die. AP is the habitat factor

affecting the moisture utilization and competition ability

of M. sikkimensis. The optimum AP value is x

a3

= 400 mm

(Sun and Wang, 1998; Wang et al., 1998d,e).

Soil acidity (pH) was negatively correlated with seed

yield (Table 3). According to results of the field survey,

when pH is within 4.8-7, the distribution of M. sikkimensis

is greatest. From 7-8.5, the distribution decreases gradu-

ally. In acid or alkaline soil with pH below 4.8 or above

8.5, there is no distribution. It has been reported that M.

sikkimensis is very sensitive to the presence of dissolved

salts in its root zone. Soil salinity (EC) in each site was

far below the limit of salinity (Niu, 1997), an electrical

conductivity of 16 dS/m (Table 2). Partial correlation co-

efficients between seed yield and pH were higher than that

of EC (Table 4). Hence, pH was chosen as a key habitat

factor and named x

4

while EC was omitted. According to

physiochemical experiment results, M. sikkimensis is suit-

able for the somewhat acidic soils; the optimum pH value

is x

a4

= 6.5 (Niu, 1997; Wang et al., 1998d).

In Alpine steppe and meadow soil most OM derived

from dead roots can reside in soils for decades due to low-

er temperature. TK is abundant (>15 g/kg) in the surface

horizon because of a lower weathering intensity of the par-

ent material. AVK is higher than 100 mg/kg and belongs

to the top supplying potassium level. Soil clay grain con-

tent (PS) is much lower and homogeneous. The correla-

tion and partial correlation coefficients between seed yield

and TK, AVK, OM, and PS were too low to be selected as

key habitat factors (Tables 3, 4). Significant correlations

were observed between TN and AHN and between TP and

AVP. Furthermore, seed yield and TN, AHN, TP and AVP

showed significant correlations (Table 3). The partial cor-

relation coefficients between seed yield and AHN, TN,

AVP, TP are 0.70, 0.65, 0.69, and 0.66 respectively (Table

4). AHN and AVP determined the supply of soil nutrients

and were chosen as key habitat factors and named x

5

and

x

6

, respectively, while TN and TP were neglected. Based

on substrate cultivation experiments and pot experiments

(Zhang and Deng, 1998; An and Ma, 2002), the optimum

AHN and AVP values are x

5a

= 60 mg/kg and x

6a

= 24.6

mg/kg, respectively.

Plant density (PD) could be considered as a key habitat

factor and was named x

7

according to its correlation and

partial correlation coefficients to seed yield. The most

suitable PD is x

7a

= 25 plants/m

2

(Zhang and Deng, 1998,

1999a; An and Ma, 2002; Wang et al., 2003a, b).

Overall, the statistical analyses showed the importance

of seven habitat factors that had been selected as key

habitat factors from the original 19 factors, including

the following controllable factors, namely AHN, AVP,

pH, PD, which could be improved through agricultural

management, and uncontrollable factors, namely AL, AT

and AP, which could not.

Constructing habitat niche-fitness model for M.

sikkimensis

Microula sikkimensis has the characteristic of three-

base points with respect to its response to each key habitat

factor, including the upper limit, the optimum value, and

the lower limit, which was in accordance with the results

of the field survey and transplanting trials. The nearer it

approaches the region’s border, the lower the fitness will

be. So the value of HNF responded to variation in each

key habitat factor, presenting a bell-shaped curve along

its gradients. Firstly, calculating the relative degree of

similarity between an actual habitat state and optimum

habitat requirement, the HNF model can be constructed,

which is intuitively and mathematically meaningful, as

follows:

7

1

}

)

(

,

)

min{(

7

1

2

2

.

.

.





.



i

x

x

x

x

k

F

i

i

ai

ai

i

i

... (3)

In Eq. (3), the value of HNF, F represents the degree

of similarity of an actual habitat to the optimum habitat,

which reflects the demand-supply relation between plant

growth and its habitat resources. x

i

is the actual state of

key habitat factor i. In order to emphasize the median

trend, all collected data of AHN (x

5

), AVP (x

6

), pH (x

4

) and

PD (x

7

) at each study site under natural (field surveys) or

cultivation conditions were transformed into the average

over three samplings during different phonological peri-

ods before analyses; x

ai

represents the optimum value of

key habitat factor i as mentioned above. k

i

is the weight

coefficient of key habitat factor i, then calculated using the

formula:

7

1

,

)

(

7

1

.

.

.

=

+

+

=

.

=

i

b

a

b

a

k

i

i

i

i

i

i

............................. (4)

Where a

i

is the factor loading of key habitat factor i in

PCA1 (the first principal component); b

i

is the factor load-

ing of key habitat factor i in PCA2 (the second principal

component) (Table 5). The larger the contribution of the

key habitat factor is, the higher the weight of the key habi-

tat factor will be.

After calculating, the results of the weight coefficients

are: k

1

= 0.141611, k

2

= 0.149163, k

3

= 0.13804, k

4

=

0.143239, k

5

= 0.148563, k

6

= 0.141107, k

7

= 0.138277.

Thus, the specific calculation formula was obtained. The

values of HNF are shown in Table 6 according to Eq. (3).

In order to prove the validity of Eq. (3) as a measure for

the degree of similarity of an actual habitat to the optimum

habitat, and to explain its mathematical justification, the

familiar Proportional Similarity Index (PSI, Feinsinger et

al., 1981), which is commonly used in similarity measures

in many ecological studies was tested and verified as fol-

lows:

300

Botanical Studies, Vol. 47, 2006

[ ]

7

,...

2

,

1

,

,

min

5

.

0

1

7

1

7

1







.

.





i

q

p

q

p

PSI

i

i

i

i

i

i

[ ]

7

,...

2

,

1

,

,

in



i

q

p

i

i

.................................................................. (5)

Here,

.

7

,...

2

,

1

,

,

7

1



μ

μ

.

’

|

|

"

¥

w

μ

μ

.

’

|

|

"

¥



.



i

x

x

x

x

q

i i

ai

i

ai

i

μ

μ

.

’

|

|

"

¥

w

μ

μ

.

’

|

|

"

¥



.



7

1

i i

i

i

i

i

x

x

x

x

p

.

7

,...

2

,

1

,

,

7

1



μ

μ

.

’

|

|

"

¥

w

μ

μ

.

’

|

|

"

¥



.



i

x

x

x

x

q

i i

ai

i

ai

i

μ

μ

.

’

|

|

"

¥

w

μ

μ

.

’

|

|

"

¥



.



7

1

i i

i

i

i

i

x

x

x

x

p

................... (6)

Where

i

x

is the mean of key habitat factor i. The calcula-

tion results of PSI are shown in Table 6. From Table 6, we

obtain that the varied ranges of HNF F and PSI are 0.662

. F . 0 .937 and 0.848 . PSI . 0.983, respectively. Obvi-

ously, the varied range of F is more extensive than that of

PSI; hence F better reflects the varied differences of HNF

under different habitat conditions on a large-scale eco-

geographical distribution.

Seed yield—habitat niche-fitness relationship

Because of the different thermal, hydrodynamic and

soil nutrient conditions at different sites in natural condi-

tions, the value of HNF differed among different sites.

It was largest in Hongyuan County of Sichuan Province

and lowest in Maqu County of Gansu Province (Table 6).

The values of HNF under cultivation conditions increased

14.26% on average compared with the values in nature

(Table 6). It was found that the lower the value of HNF

was in nature, the more it increased under cultivation

conditions. For example, the increase in HNF in Maqu

County, Gansu Province was nearly 10 times that in Hon-

gyuan County, Sichuan Province (32.628% vs 3.675%,

respectively) (Table 6).

Due to different habitat conditions at different sites, the

seed yield was also different, and generally low in nature.

The seed yield was highest in Hongyuan county, Sichuan

Province and lowest in Maqu County, Gansu Province

Table 5. Factor loading matrix of key habitat factors.

The key habitat factor

PCA1 (a

i

) PCA2 (b

i

)

Altitude (X

1

)

0.8296 -0.18642

Annual >0°C accumulated

temperature (X

2

)

0.8915 0.1787

Annual average precipitation (X

3

) 0.8012 -0.1892

Soil acidity (X

4

)

0.0634 0.9643

Soil alkaline hydrolytic nitrogen

(X

5

)

0.2126 0.8533

Soil available phosphorus (X

6

) 0.1949 0.8175

Plant density (X

7

)

0.2982 0.6939

Eigenvalue

2.87

2.48

% of variance

47.67 41.20

Cum. % of var.

47.67 88.87

Note: The first principal component (PCA1); The second prin-

cipal compoment (PCA2).

Table 6. The values of HNF (F) and the Proportional Similarity Index (PSI).

Site

F in nature F under cultivation

conditions

Rate of increase in

HNF (%)

PSI in nature PSI under cultivation

conditions

Maqu

0.662

0.878

32.628

0.848

0.954

Hezuo

0.678

-

-

0.856

-

Haiyan

0.708

0.863

21.893

0.871

0.947

Huangzhong

0.731

-

-

0.882

-

Maerkang

0.757

-

0.895

-

Tianzhu

0.775

0.830

7.097

0.904

0.931

Gangcha

0.815

-

-

0.923

-

Aba

0.853

-

-

0.942

-

Ruoergai

0.881

-

-

0.956

-

Menyuan

0.884

0.937

5.995

0.957

0.983

Hongyuan

0.898

0.931

3.675

0.964

0.980

LIN et al. — The habitat niche-fitness of

Microula sikkimensis

301

Table 7. Observed and estimated seed yield.

Site

Estimated seed

yield value (kg/ha)

in nature

b

Observed seed

yield value in

nature (kg/ha)

a

Absolute

deviation

percent (%)

Estimated seed

yield value (kg/ha)

under cultivation

conditions

b

Observed seed yield

value (kg/ha) under

cultivation conditions

a

Absolute

deviation

percent (%)

Maqu

54.94

50

9.88

399.41

400.5

0.27

Hezuo

64.98

72.7

10.62

-

-

-

Haiyan

88.08

95.7

7.96

353.87

360.91

1.95

Huangzhong

110.25

110.05

0.18

-

-

-

Maerkang

140.93

140.88

0.04

-

-

-

Tianzhu

166.23

167.42

0.71

269.09

203.55

32.2

Gangcha

236.73

240.6

1.61

-

-

-

Aba

326.05

331.94

1.77

-

-

-

Ruoergai

409.1

418.78

2.31

-

-

-

Menyuan

418.98

428.82

2.29

630.72

710

11.17

Hongyuan

467.88

450.45

3.87

602.89

600

0.48

Note:

a

The data in this column are mean values of three samples;

b

Estimated seed yield per hectare calculated by Eq. (7).

(Table 7). When agricultural technology, especially artifi-

cial fertilizer, was put into effect in the transplanting trials,

the number of fertile tillers per unit area was enhanced

significantly, bringing an increase in seed yield. The mean

value of seed yield was 227.94 kg/ha under natural condi-

tions among study sites while under cultivation conditions

the mean value of seed yield was 454.99 kg/ha, increasing

99.61% (Table 7). This characteristic exhibited a corre-

sponding trend to that of HNF: the lower the natural seed

yield was, the more it increased under cultivation condi-

tions (Table 7).

The value of HNF reflects not only the degree of

fitness of the species to its habitat, but also restricts the

seed yield. The seed yield is remarkably interrelated with

HNF, but it is not a linear relationship. Taking HNF as the

regressor, the relationship was fitted by least squares as

log-log regressions of seed yield on HNF, as follows:

F

Y

In

025

.

7

904

.

6

In





.................................... (7)

Where Y is the estimated seed yield per hectare at F level

of HNF. R

2

, adjusted R

2

and the standard error of the

estimate reached 0.987, 0.986 and 0.094, respectively.

The estimated seed yield was calculated by Eq. (7), and

observed values are shown in Table 7. The estimated seed

yield was well consistent with the observed data, and the

average value of the absolute deviation percent was 5.46%,

demonstrating the validity of the model in predicting seed

yield.

DISCUSSION

Key habitat factors

Much of the research conducted in ecological science

is devoted to analyzing species–environment relationships

(Harper, 1977), and this has produced a lot of approaches

that relied entirely on the choice of ecological factors

(Wang, 1990; Retuerto and Carballeira, 2004). Therefore,

selecting key factors, which describe an actual habitat

state, plays a crucial role in the content of the relation-

ship between species and environment. In fact, the fol-

lowing habitat factors: sunlight, temperature, geographic

range, soil water content, and soil nutrition may be closely

related to the growth period of crops. In our study the

habitat factors associated with seed yield were condensed

into seven by Pearson Correlation and Partial Correlation

Analyses, a comparison of alternative methods to iden-

tify the most important variables influencing seed yield.

This way, the effect of interactions between different key

habitat factors can be accounted for. The seven key habitat

factors not only contained climatic factors such as AT and

AP, the site factor AL, edaphic factors such as AHN, AVP

and pH, but also plant density (PD). The components of

key habitat factors included in the HNF model have the

advantage of integrating geographic data with plant per-

formance. Moreover, they also meet the needs of the soil-

plant-atmosphere continuum theory. This suggests that the

core of construction of the HNF model should be focused

on the seven key habitat factors. The HNF model contain-

302

Botanical Studies, Vol. 47, 2006

ing these key habitat factors makes it possible to evaluate

the adaptive extent of M. sikkimensis to different habitat

conditions in a large-scale eco-geographical study.

Construction of habitat niche-fitness model

An apparent supply–demand relationship exists be-

tween a realized habitat and species’ optimum habitat

requirements during the M. sikkimensis growth cycle.

The niche theory as a kernel of modern ecology provides

a sound theoretical background for HNF (Hutchinson,

1957; Li and Lin, 1997; Lin and Li, 1998; Buggeman and

O’Nuallain, 2000). The characters of the habitats in real-

ity are different from its optimum niche in some aspects.

If an actual habitat or a modified habitat which suited the

species existed, the population would increase, exhibiting

an increase of yield; otherwise, it would decrease, result-

ing in a decrease of yield. That tallies with the concept of

HNF. Normally, when the actual habitat state changes on

a large scale, the value of HNF needs a wide distribution

on the subset [0, 1]. By taking M. sikkimensis as the ob-

ject, Hutchinson’s (1957) niche concept of n-dimensional

super-volume was extended; the HNF for M. sikkimensis

is defined as the degree of similarity between the supply

of an actual habitat and the requirement for the optimum

habitat, and is a synthesis of key habitat factors to describe

a habitat state. The mathematical model for HNF in Eq.

(3), which is the degree of similarity defined in n-dimen-

sional supervolume in essence, first calculated the relative

degree of similarity between an actual habitat state and

the optimum habitat requirement. In fact, the importance

of various key habitat factors to M. sikkimensis is differ-

ent. Considering the actual conditions of unequal weight

among the key habitat factors, the weights of key habitat

factors as HNF model parameters play a crucial role in

model usability. Then, the weight determination of key

habitat factors was integrated with factor loadings derived

from a Principal Component Analysis (PCA) in this study.

This method as a measure for the degree of similarity of

an actual habitat to the optimum habitat is a new improve-

ment compared with the Proportional Similarity Index

(Feinsinger et al., 1981), which is commonly used in simi-

larity measures in many ecological studies. The values of

HNF obtained from Eq. (3) have wider distribution, i.e.

0.662 . F . 0.937, compared with those obtained from the

Proportional Similarity Index (PSI), 0.848 . PSI . 0.983.

Therefore, the HNF model is superior to the Proportional

Similarity Index. From the point view of agro-ecology, the

HNF is a new concept which indicates a description of the

adaptability of the species to its habitat. Our results sug-

gested that HNF be a new model to evaluate the adaptive

extent of M. sikkimensis on a large-scale eco-geographical

distribution.

The seed yield prediction model

Many yield prediction models have been developed

since the 1960s (Dahl, 1963; Duncan and Hesketh,

1968; Rosensweig, 1968; Murphy, 1970; Duncan and

Woodmansee, 1975; Seligman and Van, 1989; Wang,

1990; Scian and Bouza, 2005). Their performance

requires a lot of parameters relating to plant physiological

processes, such as photosynthesis, assimilation and

respiration (which respond to climatic conditions), soil

moisture, and fertilizer application, but determining these

parameters is quite difficult. Using HNF as a surrogate for

composite environment factors to establish the seed yield

prediction model is a new approach and has proved to be

effective. Our results show that the simulation agrees well

with observed seed yield, which means that the model

has a high predictive power for seed yield. Compared

with traditional yield prediction models, the model based

on HNF provides a new approach to predict seed yield

accurately.

Upper limit of HNF and seed yield, threshold

value of HNF and limit seed yield

Studying HNF variation in response to uncontrollable

and controllable key habitat factors provides a good es-

timate of fitness variance. While each controllable key

habitat factor alone can be experimentally manipulated to

study its influence on fitness, the relative importance and

potential impacts of different factors are very hard to in-

vestigate experimentally (Retuerto and Carballeira, 2004).

Here, we attempted to include seven key habitat factors

simultaneously in an analysis. The effects of controllable

key habitat factors on the survival and growth of M. sik-

kimensis were examined under cultivation condition at

five sites selected from eleven field sites. In this study, the

trials under cultivation indicated that agricultural mea-

sures significantly increased the value of HNF and seed

yield, which implies that agricultural measures debugged

controllable key habitat factors to approach their opti-

mum values resulting in habitat modification. In general,

the key habitat factors explaining variability in HNF are

technological change, including moderate fertilization, im-

proved management practices and disease control, as well

as other human interventions aimed at increasing seed

yield. However, even if technological innovations in cul-

tivation are optimized, the increase in HNF cannot be in-

finite. Therefore, each site has its upper limit of HNF. Eq.

(3) is used to forecast this limit when all controllable key

habitat factors reach their optimum values, and the results

are shown in Table 8. The seed yield corresponding to the

upper limit of HNF for each study site is forecasted using

Eq. (7); results are listed in Table 8 as upper limit of seed

yield. The maximal upper limit of seed yield is predicted

to be 821.98 kg/ha at Haiyan County, Qinghai Province,

with the minimal upper limit of seed yield being 396.22

kg/ha at Maerkang County, Sichuan Province. These re-

sults suggest that Haiyan County, Qinghai Province has

the best productive potential among the eleven study sites.

Understanding the control of HNF by cultivation technol-

ogy would result in better adaptation for plants to habitat

conditions and a greater seed yield for M. sikkimensis.

The upper limit of seed yield should be obtainable under

LIN et al. — The habitat niche-fitness of

Microula sikkimensis

303

Table 8. Upper limit of HNF and seed yield (kg/ha) for M.

sikkimensis among study sites.

Site

Upper limit of HNF Upper limit of

seed yield (kg/ha)

Maerkang

0.877

396.22

Huangzhong

0.896

460.61

Aba

0.907

501.83

Hezuo

0.917

542.02

Maqu

0.943

659.65

Ruoergai

0.944

664.58

Hongyuan

0.949

689.71

Menyuan

0.956

726.25

Gangcha

0.965

775.66

Tianzhu

0.969

798.53

Haiyan

0.973

821.98

optimized agricultural technology when the effects of the

relationship between HNF and seed yield are understood

and managed accordingly.

When F = 0.75, a prediction interval with a 95% confi-

dence interval for seed yield (kg/ha) by Eq. (7) is [113.91,

153.12] where the seed yield is acceptable because the

seed yield per kg was priced at 3.63 dollars. The member-

ship of a certain habitat, i.e. whether or not a certain area

is favorable for M. sikkimensis, is determined by a thresh-

old value defined as F=0.75. If this so-called cutpoint is

exceeded, the habitat shows an acceptable seed yield and

vice versa (Table 9). Therefore, computing HNF can be-

come a decision making tool to tell the farmers whether M.

sikkimensis can grow at a target site or not. An appropriate

HNF value will make the plants thrive, but an excessively

low value would result in physiological malfunction to the

plants and even make them die off. Considering the com-

mercial and biological seed yield requirements of M. sikki-

mensis, a HNF membership grade matrix was constructed

to specify the distribution patterns of M. sikkimensis in

different HNF intervals (Table 9).

From the theoretical point of view, when all values

of the key habitat factors reach the optimum, then value

of HNF attains the maximum, that is F=1, and the cor-

responding seed yield goes by the name of the limit seed

yield. This reaches 996.25 kg/ha as forecast by our Eq. (7).

This result coincides with the theoretical seed yield of 984

kg/ha extrapolated by An and Ma (2002) from the average

seed yield per plant in pot experiments at 2000 in Menyu-

an County, Qinhai Province.

The aims of the HNF model and seed yield prediction

model are to help farmers correctly select suitable target

sites and create meaningful summaries of site-specific

management information. They can provide good

answers to farmers’ questions relating to utilization and

development of M. sikkimensis in the following aspects:

whether introduction of M. sikkimensis in a particular area

is suitable or not, judged by the HNF (>0.75) and further

judged by the seed yield in nature and the potential seed

yield under cultivation conditions. Seed yield is tightly

related to the value of HNF. Therefore, we can enhance

HNF through rational agricultural technology and then

improve seed yield. For example, we can attain the goal

of improving the HNF for M. sikkimensis by forecasting

each index of uncontrollable key factors, ensuring rational

cultivation technology, and ultimately increasing seed

yield. Further improvements in these models, such as

combining them with a geographic information system

(GIS) database to compile maps on HNF and seed yield

prediction, are envisaged to improve ease of application.

Table 9. The grades of HNF for M. sikkimensis and distribution patterns in different HNF interval.

Grades HNF interval Predictive seed yield

interval (kg/ha)

Distribution zone Suitability and biological performance

1

0.9-1

476.20-998.25 The core area

Optimal. It thrives and has high reproduction.

2 0.75-0.9 132.30-476.20 Appropriate eco-environmental

zone

Suitable. It grows and reproduces satisfactorily.

3 0.55-0.75 14.97-132.30 Restricted zone

Quasi suitable. Its growth is restrained and

reproductive capacity declines.

4 0.40-0.55

1.60-14.97 Marginal zone

Adverse. Its growth and survivor are abnormal, and

reproduction is impossible.

5 <0.40

<1.60

Survival forbidden zone

Unsuitable survival.

304

Botanical Studies, Vol. 47, 2006

Acknowledgements. The work described in this paper

was substantially supported by the Western Key Project of

the Science Foundation of China (No. 90102011), a proj-

ect of the Gansu Provincial Natural Science Foundation

(No. 3ZS041-A25-006) and the Foundation of the Key

Laboratory of Grassland Agro-ecosystems. The authors

are very appreciative to Margaret Cargill, Adelaide Gradu-

ate Centre, the University of Adelaide, Australia, whose

critical and thorough comments markedly improved the

content and tone of this manuscript. The authors are also

grateful to the anonymous reviewers and editors for their

constructive suggestions and comments on the original

manuscript.

lITeRATURe CITeD

An, C.G. and R.D. Zheng. 1996. GC/FTIR determination of un-

saturated FA components in seed oil of Microula sikkimen-

sis. Chinese Lipid. 21: 46-47.

An, H.M. and Y.S. Ma. 2002. Effects of different fertilization

conditions on transplanting pot experiments of Microula

sikkimensis in Qinghai-Tibet Plateau. Qinhai Agric. For-

estry Sci. Technol. 2: 12-14.

Black, C.A. 1979. Methods of soil analysis. Amer. Soc. Agron.

2: 771-1572.

Buggeman, J. and B. O’Nuallain. 2000. A niche width model of

optimal specialization. Computational & Mathematical Or-

ganization Theory 6: 161-170.

Dahl, B.E. 1963. Soil moisture as a predictive index to for-

age yield for Sandhills range type. J. Range Manage. 16:

128-132.

Duncan, D.A. and R.G. Woodmansee. 1975. Forecasting forage

yield from precipitation in California’s annual rangeland. J.

Range Manage. 28: 327-329.

Duncan, W.G. and J.D. Hesketh. 1968. Net photosynthesis rates,

relative leaf growth rates, and leaf numbers of 22 maize

grown at eight temperatures. Crop Sci. 8: 670-674.

Feinsinger, P., E.E. Spears, and R.W. Poole. 1981. A simple

measure of niche breadth. Ecology 62: 27-32.

Fu, H., Q. Wang, Z.Y. Zhou, S.Z. Zheng, and J.C. Meng. 1997.

Analysis of fatty acids of seed oil of Microula sikkimensis

in Tianzhu by GC/MS. Acta Agrestia Sin. 5: 206-208.

Fu, H., Z.Y. Zhou, and Q. Wang. 1999. Study on the nutrient

component in seeds of Microula sikkimensis. Pratacultural

Sci. 16: 18-20.

Gee, G.W. and J.W. Bauder. 1986. Particle-size analysis. In. A.

Klute (eds.), Methods of Soil Analysis. Part I, 2nd Edition,

Agronomics Monograph, Vol. 9. ASA and SSSA, Madison,

WI, pp. 383-409.

Harper, J.L. 1977. Population Biology of Plants. Academic

Press, London.

Honda, C. 1962. Procedure for determination of nitrogen in soil

by Kjeldahl method. J. Sci. Soil Manure. 33: 195-200.

Hutchinson, G.E. 1957. Concluding remarks. Cold Spring Har-

bor Symp. Quant. Biol. 22: 415-427.

J iang, H.Y. and W.X. Wang. 2004. Application of principal

component analysis in synthetic appraisal for multi-objects

decision-making. J. Wuhan Univ. Technol. 28: 467-470.

John, M.K. 1970. Colorimetic determination of phosphorus in

soil and plant material with ascorbic acid. Soil Sci. 109:

214-220.

Li, M.Y., L.C. He, and Z.C. Wu. 1999a. Preventing and treating

effect of total oil of Microula sikkimensis on experimental

hyperlipemia of rats. J. Chinese Med. 24:106-108.

Li, M.Y., L.C. He, Z.C. Wu, and Q. Wang. 1999b. Effect of total

oil of Microula sikkimensis on rats blood changing. J. Chi-

nese Med. 24:135-140.

Li, Z.Q. and J.Z. Ren. 1997. The models of suitability degree of

grassland organism and their application. Chinese J. Ecol.

16: 71-75.

Li, Z.Z. and H. Lin. 1997. The niche-fitness model of crop popu-

lation and its application. Ecol. Modell. 104:199-203.

Lin, H. and Z.Z. Li. 1998. The niche-fitness model of crop in

semi-arid regions and the quantitive analysis of the results

of water-fertilizer regulation experiment. J. Lanzhou uni-

versity (Natural Sciences) 34: 100-105.

Liu, R.T., H.L. Lin, and R. Wang. 2000. A study on ecological

adaptation of saiga tatarica—Principal component analysis

of main ecological factors. Acta Ecol. Sin. 20: 184-189.

Lu, Z.G. and Y. Yang. 1997. Review on application and utiliza-

tion of nitrogen, phosphorus fertilizers and their pollution

on agricultural ecological environment in alpine regions of

Qinghai-Tibet Plateau. Agro-Environment & Develop. 14:

30-48.

Murphy, A.H. 1970. Predicted forage yield based on fall precipi-

tation in California annual grasslands. J. Range Manage.

23: 363-365.

Niu, J.L. 1997. The studies on salt tolerance of Microula sikki-

mensis. J. Gansu Agricul. Univ. 32: 376-380.

Olsen, S.R. and L.E. Sommers. 1982. Phosphorus. In. A.L. Page

et al. (eds.), Methods of Soil Analysis. Part II, 2nd Edition.

Agronomics Monograph, Vol. 9. ASA and SSSA, Madison,

WI, pp. 403-427.

Ren, J.Z. and H.L. Lin. 2005. Assumed plan on grassland eco-

logical reconstruction in the source region of Yangtze River,

Yellow River and Lantsang River. Acta Prataculturae Sin.

14: 1-8.

Retuerto, R. and A. Carballeira. 2004. Estimating plant respons-

es to climate by direct gradient analysis and geographic

distribution analysis. Plant Ecol. 170: 185-202.

Rosensweig, M.L. 1968. Net primary productivity of terrestrial

communities: prediction from climatological data. Ameri-

can Naturalist. 102: 67-74.

SAS Institute 2000. SAS User’s Guide. North Carolina, USA..

Statistical Analysis System Institute Inc.

S cian, B.V. and M.E. Bouza. 2005. Environmental variables

related to wheat yields in the semiarid pampa region of Ar-

gentina. J. Arid Environ. 61: 669-679.

LIN et al. — The habitat niche-fitness of

Microula sikkimensis

305

Seligman, N.G. and H. van Keulen. 1989. Herbage production

of a Mediterranean grassland in relation to soil depth, rain-

fall and nitrogen nutrition: A simulation study. Ecol. Mod-

ell. 47: 303-311.

SPSS Institute 1997. SP SS for Windows Base System User’s

Guide Release 7.5. Marija J.Norusis, USA. Statistics Pack-

age for Social Science Institute Inc.

Steel, R.D. and J.H. Torrie. 1980. Principles and Procedures of

Statis tics: A Biometrical Approach. McGraw-Hill Book

Co., New York.

Sun, J.H. and Q. Wang. 1998. Studies on the germination per-

formance of Microula sikkimensis seeds. Acta Agrestia Sin.

6:26-32.

Wang, F.T. 1990. A study on the weather-yield simulation and

model. China Acta Meteorological Sin. 4: 479-502.

Wang, J.Q., Q. Wang, and J.H. Yang. 1997a. Germplasm re-

sources of Microula of Gansu. China Pratacultural Sci. 14:

9-13.

Wang, Q., H. Fu, and Y. Han. 1997b. Study on hot damage to

leaf tissue of Microula sikkimensis. J. Gansu Agricul. Univ.

32: 370-375.

Wang, Q., K.Z. Shang, and C.L. Xu. 1998a. A study on the

distribution and population characteristics of Microula sik-

kimensis. Pratacultural Sci. 15: 18-26.

Wang, Q., Y.F. Xie, and C.L. Xu. 1998b. Study on the space

regularity and feature of Microula sikkimensis. Grassland of

China 4:21-27.

Wang, Q., K.Z. Shang, and W.L. Cheng. 1998c. A study of cold

injury on Microula sikkimensis leaf cell. Pratacultural Sci.

15: 23-26.

Wang, Q., H. Fu, and Y. Han. 1998d. A study on the stress resis-

tance of Microula sikkimensis root tuber. Pratacultural Sci.

15: 16-18.

Wang, Q., H. Fu, and Z.Y. Zhou. 1998e. A study on response of

water stress of Microula sikkimensis. Acta Agrestia Sin. 6:

179-184.

Wa ng, Q., R.F. Lian, M.C. Wang, H.M. An, and Z.X. Guo.

2003a. Study on the technique of planting Microula sikki-

mensis. Acta Agrestia Sin. 11: 321-324.

Wang, Q., J.Z. Ren, Z.X. Guo, and H.M. An. 2003b. The value

and characteristics of the Microula sikkimensis. J. Natural

Resources 18:247-251.

Zhang, J.C. and Y.C. Deng. 1998. Substrate cultivation experi-

mental studies on cultivation of Microula sikkimensis. J.

Sichan Grassland 4: 18-21.

Zhang, J.C. and Y.C. Deng. 1999a. The growing and developing

differences of Microula sikkimensis under the same culti-

vating conditions. J. Sichan Grassland 5: 8-10.

Zhang, J.C. and Y.C. Deng. 1999b. Studies on fodder processing

by using stalks of Microula sikkimensis. J. Sichan Grass-

land 5: 61-62.

Zhen, S.Z., J.C. Meng, D.Y. Wang, H. Fu, and Q. Wang. 1997.

Two new limonoids from the seed of Microula sikkimensis.

Planta Med. 63: 379-380.

306

Botanical Studies, Vol. 47, 2006

微孔草（Microula sikkimensis）生境生態棲位適宜度與草籽產

量關係的研究

林慧龍　任繼周　王 欽

蘭州大學草地農業科技學院 甘肅草原生態研究所

　　微孔草（M. sikkimensis）是隸屬於紫草科微孔草屬的兩年生草本植物，主要分佈于青藏高原東½

的退化草甸和草原，作為新興的在地油料作物種，其在醫藥、保健、飼草料等方面均具有多方面的新

應用，因此，在青藏高原日益退化的高山草甸和草原放牧生態系統中，利用微孔草草籽生產，建立可

持續的生產生態系統，不失為一種既取得生產效益又利於生態恢復的理性選擇之一。自 1994 年 4 月至

2004 年 9 月，先後在青藏高原東½的 3 省 11 縣進行了大½圍的野外½查和移栽試驗，收集了各研究

位點的大量生境參數，通過相關與偏相關分析篩選出描述生境的七個關鍵生境因子。引入生態棲位理

½，定義微孔草生境生態棲位適宜度是現實生境供給與微孔草對生境的最適需求之間的相似程度，在

對各關鍵生境因子對生境生態棲位適宜度的權重進行界定的基礎上，構建了微孔草生境生態棲位適宜

度模型，對野外½查和移栽試驗的計算結果表明：生境生態棲位適宜度較好地反映了不同研究位點的

生境差異。在推薦的農藝措施下，因改善了生境促進了微孔草生長，各移栽試驗點的生境生態棲位適

宜度都有不同程度地提升，平均提升率達 14.26%，亦促使草籽產量平均增長 99.61%；生境生態棲位適

宜度作為½多環境因子的綜合指標與草籽產量間存在顯著的雙對數回歸關係，經實驗數據驗證，觀測

記錄的草籽產量與該回歸關係模擬估計值之間的平均絕對離差百分數是 5.46%，意味著該回歸方程在大

尺度預測草籽產量上的可用性，並且以生境生態棲位適宜度為自變量建立草籽產量預測模型的方法，

比之於傳統產量預測模型而言，具簡便準½之利，是一種新的嘗試。通常農藝措施通過½控可控關鍵

生境因子逼近最優值，實現生境改善，從而提高了生境生態棲位適宜度值和草籽產量，然而這種提升

力並不是無限的，因此每個位點都有其生境生態棲位適宜度上限值和相應的草籽產量上限。生境生態

棲位適宜度模型、草籽產量預測模型及其研究結果對選擇微孔草適宜生長區，進行栽培、生產、管理

具有指導意義。

關鍵詞：生境生態棲位適宜度 (HNF)；微孔草 (Microula sikkimensis)；草籽產量；青藏高原。