Effective environmental factors in the distribution of vegetation types in Poshtkouh rangelands of Yazd Province (Iran)

ARTICLE IN PRESS Journal of Arid Environments Journal of Arid Environments 56 (2004) 627–641 www.elsevier.com/locate/jnlabr/yjare

Effective environmental factors in the distribution of vegetation types in Poshtkouh rangelands of Yazd Province (Iran) M. Jafaria,*, M.A. Zare Chahoukib, A. Tavilib, H. Azarnivandb, Gh. Zahedi Amirib a

Karaj Natural Resources Faculty, Tehran University, P.O. Box, Tehran 31585-4314, Iran b Natural Resources Faculty, Tehran University, Iran Received 20 June 2002; accepted 29 April 2003

Abstract The objective of this research was to study the relationships between environmental factors and vegetation in order to find the most effective factors in the separation of the vegetation types in Poshtkou rangelands of Yazd province. Sampling of soil and vegetation were performed with randomized-systematic method. Vegetation data including density and cover percentage were estimated quantitatively within each quadrat, and using the two-way indicator species analysis (TWINSPAN), and vegetation was classified into different groups. The topographic conditions were recorded in quadrat locations. Soil samples were taken in 0–30 and 30–60 cm depths in each quadrat. The measured soil variables included texture, lime, saturation moisture, gypsum, acidity (pH), electrical conductivity, sodium absorption ratio,
2
and soluble ions (Na+, K+, Mg2+, Cl
, CO2
3 , HCO3 and SO4 ). Multivariate techniques including principal component analysis (PCA) and canonical correspondence analysis (CCA) were used to analyse the collected data. The results showed that the vegetation distribution pattern was mainly related to soil characteristics such as salinity, texture, soluble potassium, gypsum, and lime. Totally, considering the habitat conditions, ecological needs and tolerance range each plant species has a significant relation with soil properties. r 2003 Elsevier Ltd. All rights reserved. Keywords: Classification; Iran; Multivariate analysis; Ordination; Poshtkouh rangelands; Soil characteristics

*Corresponding author. E-mail address: [email protected] (M. Jafari). 0140-1963/03/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0140-1963(03)00077-6

ARTICLE IN PRESS 628

M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

1. Introduction In order to better understand and manage rangeland ecosystems, it is important to study the relationship between environmental factors and plants in these ecosystems. One of the main components of rangelands is vegetation, the absence and presence of which is controlled by environmental variables such as climate, soil and topography (Leonard et al., 1984). Among different environmental factors, soil is of high importance in plant growth, and is a function of climate, organisms, topography, parent material and time (Hoveizeh, 1997). Topography (elevation, slope, and aspect) affects soil and climate, in addition to affecting temperature and evapo-transpiration (as elements of climate), makes deeper soil and higher content of organic matter, that result in intensive vegetation in the northern aspects in comparison to the southern ones (Jenny, 1980). Effects of environmental factors on plant communities have been the subject of many ecological studies in recent years. Leonard et al. (1984) found that vegetation cover had strong relationship with temperature and soil moisture. Other soil characteristics, directly or indirectly, influence the two mentioned parameters. Rooting depth, soil water potential, absorption, and distribution of nutrients are influenced by the amount and availability of soil moisture. Soil moisture is influenced by some soil characteristics such as texture, clay type, structure, soluble salts, gravel, depth and temperature (Leonard et al., 1984). Also, topography, microtopography, and amount of litter on the soil surface are other characteristics that affect soil moisture (Walker, 1979). Study of rangelands production in Kavir-e-Phino, located in the Hormozgan province, showed that some variables, such as slope, aspect, saturation moisture percentage and soil depth had the most effective role in the yield of plant species (Zare, 1998). Makarenkov and Legendre (2002) investigated the effects of water content and reflection of soil radiation on the vegetation cover percentage of Calmagrostis epigejos and Corynephrous canscens using multivariate analysis such as CCA, RDA and non-linear regression. They found that Calmagrostis epigejos is the indicator of wet sites while Corynephrous canscens indicates dry sites. Habitat condition of the three range species, namely, Festuca ovina, Cachrys ferulacea,and Bromus tomentellus were studied in relation to edaphic, climatic, and topographic conditions using ordinaion methods in Vahargan river catchment. Results indicated that edaphic factors were the most effective one in the separation of the three habitats (Iravani et al., 2002). Determining which factors control the presence, number, identity, and relative abundance of plant species remains a central goal in ecology. The main purpose of this research was to study the relationship between topographic and edaphic factors with plant species to determine the most strong factors affecting the separation of vegetation types. Understanding relationships between ecological variables in a given ecosystem helps us to apply these findings in management, reclamation, and development of similar regions.

ARTICLE IN PRESS M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

629

2. Materials and methods 2.1. Study area Poshtkouh rangelands are located in the southern slopes of the Shirkouh mountainous region of the Yazd province in center of Iran (31 20 N, 53 450 E to 31 340 N, 54 330 E). The maximum elevation of the region is 3970 m in Shirkouh Mountain and the minimum elevation is 1450 m in the margin of Kavir-e-Abarkouh. The climate varies from cold steppic to semi-desert. Average annual precipitation of the study area ranges from 250 mm in Shirkouh Mountain to 80 mm in the margin of Kavir-e-Abarkouh. Minimum temperature is recorded in December (
8 C) while the highest temperature touches +45 C in June (Fattahi, 1998). Meteorological data were calculated for a ten-year period. 2.2. Data collection Based on field surveys, vegetation types were identified. In each vegetation type, soil and vegetative attributes were described within quadrats located along two 250 m transverse transects. Quadrat size was determined for each vegetation type using the minimal area method. Considering variation of vegetation and environmental factors, ten quadrats with a distance of 50 m from each other were established in each vegetation type. Sampling method was randomized systematic. Floristic list, density and canopy cover percentage were determined in each quadrat. Vegetation cover data were recorded using ordinal scale of Van-der-Marrel (1979). Soil samples were taken between 0–30 and 30–60 cm layers. Measured soil factors included texture (determined by Bouyoucos hydromerter), saturation moisture (weighting method), organic carbon (determined using Walkely and Black rapid titration, Black, 1979), pH in saturation extract (determined by pH meter), electrical conductivity (ECe) (determined by conductivity meter), lime (determined using 1 n HCl, Jackson, 1967), soluble calcium and magnesium (determined by titration with solution EDTA method), soluble chlorine (determined by titration with AgNO3), soluble carbonate and bicarbonate (determined by titration with H2SO4 using methylorange and phenolphthalein, respectively) and soluble sodium and potassium (determined by flame photometry method). In quadrat locations, elevation and slope (using compass) and aspect were also recorded. All sampling sites were located in the southern aspect. 2.3. Data analysis methods Data matrix of environmental factors and vegetation type was made. The Windows (Ver. 3.0) of PC-ORD (McCune and Mefford, 1997) was used for classification and ordination of vegetation types in gradient of environmental factors. Data were analysed by a series of multivariate techniques such as the twoway indicator species analysis (TWINSPAN), principal component analysis (PCA) and canonical correspondence analysis (CCA).

ARTICLE IN PRESS 630

M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

Due to lack of statistical analysis, understanding the structure of plant species is associated with considerable mistakes, therefore, in the first step, vegetation of the study area was classified using TWINSPAN analysis. To use this analysis, the cover data transformed using an eight-point scale ((0–1=0.5, 1–2.5=1.75, 2.5–5=3.75, 5–7.5=6.25, 7.5–12.5=10, 12.5–17.5=15, 17.5–22.5=20, 22.5–27.5=25, >27.5=30) (Scale Van-der-Marrel, 1979)). TWINSPAN analysis is a numerical method for classification of vegetation belonging to similar groups. This allows the investigator to recognize the homogenous groups. After classification of the vegetation, relationships between environmental factors (topography and soil) and vegetation were studied using PCA and CCA. PCA is the ordination technique that constructs the theoretical variable that minimizes the total residual sum of squares after fitting straight lines to the species data. PCA does so by choosing best values for the sites. To apply PCA, data standardization is necessary if we are analysing variables that are measured in different units. Also, species with high variance, often the abundant ones, therefore dominate the PCA solution, whereas species with low variance, often the rare ones, have only minor influence on the solution. These may be reasons for applying the standardized PCA, in which all species receive equal weight (Jongman et al., 1987). Therefore, data was centered and standardized by standard deviation. Eigenvalues for each principal component was compared to a broken-stick eigenvalue to determine if the captured variance summarized more information than expected by chance. Broken-stick eigenvalues have been shown to be a robust method for selection of non-trivial components in PCA (Jackson, 1993). Principal components are considered useful, or non-trivial, if their eigenvalue exceeds that of their broken-stick counterpart (Legendre and Legendre, 1998). CCA is the new technique that selects the linear combination of environmental variables that maximizes the description of the species scores. On the other hand, CCA chooses the best weights for the environmental variables (Zahedi, 1998). This gives the first CCA axis. In CCA, composite gradients are linear combinations of environmental variables, giving a much simpler analysis, and the non-linearity enters the model through a unimodal model for a few composite gradients, taken care of in CCA by weighted averaging. Canonical ordination is easier to apply and requires less data than regression. It provides a summary of the species–environment relations (Jongman et al., 1987). Significance of species–environment correlation was tested by the distribution-free Monte Carlo test (1000 permutations). In the Monte Carlo test, the distribution of the test statistics under the null hypothesis is generated by random permutations of cases in the environmental data.

3. Results 3.1. TWINSPAN TWINSPAN was performed for vegetation analysis in 90 plots using ordinal scale of Van-der-Marrel (1979). The results of TWINSPAN classification are presented in

ARTICLE IN PRESS M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

631

Table 1 TWINSPAN of the vegetation cover in 90 quadrats and 11 species 44454444445556555555 1222222222311112111113333343333666666666777777777788888888889 2680347915159023467812578934603512468790356701248923457016891456792380136924578012334567980 7 Ep.st 8 Co.mo 9 Ha.ap 10 Se.ro 1 Ar.au 2 As.sp 3 Ar.si 4 Sa.ri 5 St.ba 6 Dor.am 11 Ta.ra

2122444367---------------------------------------------------------------------------------------------------------------------------------------------------------------4443222222-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------4445463233--------------------------------------------------------------------------------------------------------------------------------------25644123221--------------------------------------------5643642122------------------------------------------------------------------------------------------------------------------------223--3225----------------------------------------------------------------------------------------------------------------------------------------1-2233344543423776657444442211------------------------------------------------------------------------------------------5454545324----------------------------------------------------------------------------------------------------------------------21------------------------------------------------------------------------------------------------------------------------------------------------------------3232223322 --------------------------------------------------------------------------------------------------------------------------------------------------------------------------4433334455

00 00 010 010 011 011 011 011 011 011 1

000000000000000000000000000000000000000000000000000000000000000000000000000000001111111111 000000000000000000001111111111111111111111111111111111111111111111111111111111110000000011 00000000001111111111000000000000000000000000000000000000000011111111111111111111 00000000110000111111000000000011111111111111111111111111111100000000001111111111 00001111 000000111100000000001111111111111111111100000011110000111111 001111111100000000000011111111

For abbreviations and units, see Appendix A.

Table 1 and Fig. 1. According to the above-mentioned table, figure, and also eigenvalue of each division, vegetation of the study area was classified into seven main types. Each type differs from the other in terms of its environmental needs. These types are as follows: 1. 2. 3. 4. 5. 6. 7.

Ephedra strobilacea, Cornulaca monocantha, Artemisia aucheri–Astragalus sp., Artemisia sieberi, Haloxylon aphyllum, Seidlitzia rosmarinus, and Tamarix ramosissima.

In addition, A. sieberi type includes subtypes such as A. sieberi–Dorema ammoniacum, A. sieberi–Salsola rigida. According to the results of vegetation classification, quadrats were classified into different groups. PCA and CCA were used to determine the most effective environmental factors in the separation of vegetation types and to find the relationships between the existing plants and environmental factors. 3.2. PCA To determine the most effective variables on the separation of vegetation types, PCA was performed for 38 factors in seven sites. The results of the PCA ordination are presented in Table 2 and Fig. 2. Broken-stick eigenvalues for data set indicate

ARTICLE IN PRESS M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

632

N=90

Eigen=1

N=80

N=10

Ta.ra

Eigen1

N=20

N=60

Eigen=1

N=10

Ep.st

Eigen=1

N=10

N=40

Co.mo

N=20

Eigen=1

N=10

Ar.au-As.sp

Eigen= 0.93

N=30

N=10

N=10

Ar.si

Ha.ap

Se.ro

Fig. 1. Dendrogram of TWINSPAN for vegetation in the study area. For species abbreviations, see Appendix A.

that the first two principal components (PC1 and PC2) resolutely captured more variance than expected by chance. The first two principal components together accounted for 86.03% of the total variance in data set. Therefore, 72.34% and 13.69% variance were accounted for by the first and second principal components, respectively. This means that the first principal component is by far the most important for representing the variation of the seven vegetation types. Considering the correlations between variables and components, the first principal component includes environmental factors such as calcium, magnesium, bicarbonate and gypsum in the first layer (0–30 cm), clay and sand in the second layer (30–60 cm), and saturation moisture, ECe, sodium absorption ratio (SAR), soluble ions (Na+, 2
Cl
, CO2
3 , SO4 ) in both layers. Axis 1 seems to represent essentially soil salinity, texture, gypsum and potassium gradient, while axis 2 is reflecting a gradient of elevation, slope, pH and lime in the first layer and silt in the second layer. As mentioned above, PC1 accounted for 72.34% of the total variance, which is related to soil properties. Therefore, among all environmental factors, soil characteristics such as salinity, texture, gypsum and potassium are the most effective factors in the distribution of vegetation types.

ARTICLE IN PRESS M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

633

Table 2 PCA applied to the correlation matrix of the environmental factors in the study area Axis

Eigenvalue

% of variance

Cum. % of var.

Broken-stick eigenvalue

1 2 3 4 5 6 7 8 9 10

27.490 5.202 2.466 1.346 1.172 0.324 0.0 0.0 0.0 0.0

72.341 13.690 4.488 3.542 3.085 0.854 0.0 0.0 0.0 0.0

72.341 86.031 92.519 96.061 99.146 100.000 100.000 100.000 100.000 100.000

Factor

PC*1

PC*2

PC3

PC4

PC5

PC6

Principal components Elevation Slope Gravel1 Gravel2 Clay1 Clay2 Silt1 Silt2 Sand1 Sand2 Sm1 Sm2 Gypsum1 Gypsum2 Lime1 Lime2 pH1 pH2 EC1 EC2 Na1 Na2 K1 K2 Ca1 Ca2 Mg1 Mg2 Cl1 Cl2 CO1 CO2 HC1 HC2 SO1 SO2 SAR1 SAR2

0.0815 0.0743 0.1365 0.1615
0.1849
0.1824
0.1650
0.1723 0.1817 0.1819
0.1888
0.1886
0.1888
0.0071 0.0111 0.0020
0.0950
0.0908
0.1896
0.1900
0.1893
0.1896 0.0311
0.1849
0.1887
0.1228
0.1899
0.1612
0.1894
0.1897
0.1892
0.1888
0.1892
0.1174
0.1897
0.1902
0.1896
0.1890


0.3926
0.3608
0.2348
0.0742
0.0265
0.0652
0.1795
0.1842 0.0838 0.1073
0.0558
0.0294
0.0342
0.0273 0.3812 0.3687 0.3585 0.3099
0.0345
0.0043
0.0409
0.0307
0.0496 0.0216
0.0237 0.0710
0.0036 0.0878
0.0402
0.0268
0.0440
0.0305
0.0090 0.1608
0.0345
0.0143
0.0297
0.0121

0.0380
0.0040 0.1008 0.2134 0.0238 0.1274
0.0183
0.0219
0.0090
0.0843
0.0070
0.0200
0.0658
0.5446
0.1024 0.0246 0.1113 0.2230 0.0292 0.0017 0.0354 0.0372 0.3170 0.0928
0.0852
0.4635
0.0146
0.3023 0.0342 0.0313 0.0375 0.0586 0.0693 0.3153 0.0256
0.0188 0.0364 0.0604


0.0289 0.3255 0.3518 0.1582 0.0173
0.0138
0.1528 0.0212 0.0455
0.0020 0.0087
0.0038 0.0015
0.3771 0.2844 0.2878 0.0393
0.1415 0.0471 0.0687 0.0412 0.0521
0.5337 0.0777
0.0049
0.0431 0.0700 0.0821 0.0408 0.0562 0.0353 0.0337 0.0222
0.2368 0.0484 0.0512 0.0553 0.0545

0.0506
0.1313 0.0069
0.1056 0.1531 0.1406 0.1971 0.0514
0.1720
0.1040
0.0302 0.0805
0.0127
0.1884 0.2870 0.3915
0.0934
0.3201
0.0105
0.0058
0.0137
0.0137 0.5258
0.1454
0.0161 0.0369 0.0106 0.0427
0.0150
0.0100
0.0192
0.0650
0.0525
0.3759
0.0067
0.0063
0.0122
0.0553

0.1582 0.1573 0.1680 0.5426 0.2873 0.0224 0.1429
0.0887
0.2404 0.0096
0.0903
0.1653 0.0997 0.3068 0.1701 0.0048 0.3534 0.1138
0.01257 0.0606
0.0313
0.0024 0.1274 0.1095 0.0125 0.2803
0.0742
0.1664
0.0355
0.0039
0.0413
0.0231 0.0272 0.0198
0.0085 0.0413
0.0009 0.0249

4.228 3.228 2.728 2.395 2.145 1.945 1.778 1.635 1.510 1.399

For abbreviation and units, see Appendix A. *Non-trivial principal component as based on broken-stick eigenvalues.

ARTICLE IN PRESS M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

634

Axi s2

High lime and soil pH Low elevation, slope and silt

Se.ro

3

Ha.ap Como-Cac 1 Low EC and clay

High EC and clay

-15

-10

-5

Ar,si

0

Ta.ra

-1

5

Axis 1

Epst-Zy.at

-3 Ar.au-As.sp Low lime and soil pH High elevation, slope and silt

-5

Fig. 2. PCA–ordination diagram of the vegetation types related to the environmental factors in the study area. For vegetation types abbreviations, see Appendix A.

Fig. 2 shows a plot of the seven vegetation types against their values for axes 1 and 2. For the interpretation of the diagram and the vegetation types’ spatial distribution, in addition to the soil characteristics (Table 3) the following points should also be noted: 1. In the diagram, the distance between the indicator points of the vegetation types show the degree of similarity and dissimilarity in the environmental factors. 2. Considering PC1, all coefficients of the environmental factors, except for sand in the second layer, are negative. Therefore, those plant sites that are lying in the positive direction of axis 1, have inverse relationship with PC1 factors, except for sand in the second layer. In the second principal component, coefficient factors of silt, elevation and slope are negative, but in lime and acidity are positive. 3. The distance between the indicator points of the vegetation types from axes is representative of the relationship power in the explanation of variations. Whenever the length of vector loading (as indicator of the vegetation types) is bigger, the angle between vectors and axes is smaller. Therefore, the correlation between vegetation types with axes and relation power is more. H. aphyllum, S. rosmarinus, and C. monocantha types comforted in the first quarter of the co-ordinate axes. These types have inverse relation with PC1 factors, except for sand in the second layer, and PC2 factors, except for lime and acidity. Relation power depends on the length of vector loading and the angles between vectors and axes. In addition, in H. aphylum and S. rosmarinus types environmental characteristics are approximately similar in axes 1 and 2.

Depth Gravel Clay Silt

Sand S.M. Gypsum Lime pH

Ar. au–As. sp.

0–30 30–60 0–30 30–60 0–30 30–60 0–30 30–60 0–30 30–60 0–30 30–60 0–30 30–60

80.48 79.04 72.7 75.66 77.92 87.31 88.39 87.26 89.45 91.09 76.64 86.59 28.05 19.75

Ar. si Ep. st Co. mo Ha. ap Se. r. Ta. ra

25.6 26.66 19.22 17.93 8.17 13.03 10.30 11.47 9.73 22.39 11.27 17.79 0.0 0.0

8.12 8.64 13.27 16.35 9.72 3.58 5.08 5.88 6.88 5.47 15.50 8.42 44.43 52.23

11.4 11.67 14.04 7.99 12.36 9.11 6.53 6.86 3.60 2.64 7.86 4.99 27.52 28.02

20.37 0.0 20.99 0.0 21.34 0.0 19.95 0.023 22.27 1.04 25.41 35.58 21.11 0.01 27.05 3.39 19.63 0.01 20.06 0.15 19.75 0.64 25.09 6.09 49.54 6.04 64.49 6.75

0.49 0.11 12.32 14.20 10.86 3.10 24.66 22.49 21.94 17.91 44.35 36.61 13.58 12.88

EC

Na+

7.35 0.24 0.32 7.30 0.13 0.14 8.01 0.36 0.69 8.11 0.37 1.95 7.88 1.26 0.97 7.90 2.35 1.26 8.05 0.91 2.49 8.08 1.25 1.75 8.79 1.70 14.42 9.14 2.42 21.61 8.65 3.95 31.16 8.20 5.21 34.46 8.71 98.61 224.03 8.78 36.64 790.98

For vegetation types and variables abbreviations and soil characteristics units, see Appendix A.

K+ Ca2+ Mg2+ Cl
0.36 0.33 0.73 0.27 0.40 0.31 0.62 0.29 0.88 1.12 2.31 0.62 6.69 3.68

0.88 1.4 0.68 1.34 1.53 1.51 0.72 1.24 10.76 2.18 24.44 5.60 4.28 2.55 8.56 5.04 1.84 1.88 2.08 2.18 6.68 2.86 18.44 5.60 48.79 12.41 26.32 10.28

2
CO2
HCO
3 3 SO4

2.05 0.0 1.14 1.30 0.0 0.78 2.20 0.0 2.08 2.40 0.0 1.81 5.15 0.0 1.60 5.15 0.0 1.24 6.90 0.0 1.48 6.50 0.0 1.36 10.4 0.06 3.44 14.20 0.40 2.96 17.00 0.0 2.26 26.40 0.0 1.06 1024.80 11.30 19.22 436.10 5.44 2.93

SAR

0.28 0.29 0.53 0.14 0.48 0.58 1.21 1.57 7.00 0.45 20.66 0.33 2.77 1.42 8.38 1.20 5.70 11.15 9.38 18.54 22.02 19.55 28.18 10.50 507.06 406.73 226.52 178.73

ARTICLE IN PRESS

Vegetation Type

M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

Table 3 Soil physical and chemical characteristics of study area (in different vegetation types)

635

ARTICLE IN PRESS 636

M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

In the study area, environmental conditions in T. ramosissima type differ from the others. With attention to the position of this type in the third quarter of the diagram, it has a high correlation with the first axis. Therefore, this type has the most relation with variables of the first axis (ECe, gypsum, potassium and clay). Because of the bigger distance of T. ramosissima type from the second axis, this type has a weak relation with factors such as elevation, slope, acidity and lime. E. strobilacea, A. sieberi and A. aucheri–As. sp. types have inverse relation with indicator environmental characteristics of the first and second axes except for elevation, slope, sand and silt in the second layer. A. aucheri–As. sp. type has more relation with indicator characteristics of the first and second axes. Indicator environmental factors of the first and second axes in E. strobilacea and A. sieberi types are approximately similar. A. aucheri–As. sp. type has a direct relationship with elevation and slope, and an inverse relationship with lime and pH, while S. rosmarinus and H. aphyllum types have a direct relationship with lime and pH, and are inversely related to elevation and slope. Also, in terms of ECe and soluble ions rate A. aucheri–As. sp. type shows a different behavior in comparison to T. ramosissima type. 3.3. CCA CCA is a kind of technique that shows non-linear relations between species with environmental factors and chooses the best weights for environmental variables. According to Tables 4 and 5, first axis (eigenvalue=0.552) accounted for 66.4% variation in environmental factors data. Correlation between the first axis and species–environmental variables was 0.95 and Monte Carlo permutation test for the first axis was highly significant ðP ¼ 0:01Þ: The second axis (eigenvalue=0.182) explained 21.9% variation in data set. Correlation between the second axis and species–environmental variables was 0.99. In addition, the Monte Carlo test for the second axis was highly significant ðP ¼ 0:01Þ: The results of CCA ordination are presented in Fig. 3. Three groups were determined in relation to the environmental factors. Each environmental factor is an indicator of the specific habitat. A. sieberi type has non-linear relation with soluble potassium in the first layer, that is, soluble potassium is the indicator of habitat of this type.

Table 4 Canonical correspondence analysis for environmental data

Eigenvalue Variance in species data % of variance explained Cumulative % explained

Axis 1

Axis 2

Axis 3

0.552

0.182

0.012

66.4 66.4

21.9 88.3

1.5 89.7

ARTICLE IN PRESS M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

637

Table 5 Monte Carlo test for species–environment correlations Axis

Spp–envt Corr

Mean

Minimum

Maximum

P

1 2 3

0.948 0.990 0.852

0.725 0.469 0.356

0.418 0.147 0.096

0.924 0.920 0.812

0.0100 0.0100 0.0100

Axis 2

P=Proportion of randomized runs with species–environment correlation greater than or equal to the observed species–environment correlation; i.e., P=(1+no. permutations*oobserved)/(1+no. permutation).

3.0

Arsi k1

2.0

1.0

Cl1 Na1 Ec1 co1 So1 Cl2SAR2 gy1 co2 SAR1 Na2 Ec2 So2 -1.5

clay2

Tara

-2.5

Hc1 Ca1

clay1 0.0 k2 Mg1 lo2 -0.5 Mg2 Ca2

lo1

Axis 1

sm2

pH2 Hc2 0.5 li2 abs gr2 sm1 sa1 gr1 pH1 li1 sa2 slope gy2

Epst -1.0

1.5

Sero Haap Como Arau-Assp

Fig. 3. CCA–ordination diagram of the environmental data. For vegetation types and variables abbreviations, see Appendix A. (*) is the representative of the vegetation types. (K) is the representative of the environmental factors.

ARTICLE IN PRESS 638

M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

A. aucheri–A. sp., E. strobilacea, S. rosmarinus and C. monocantha types have nonlinear relation with slope, elevation, gypsum, gravel, sand, lime, acidity, calcium and soluble magnesium. Relation power depends on the relative distance between indicator points of soil characteristics and vegetation types. The above-mentioned types represent soils with light texture. A. aucheri–As. sp. type occurs in the soils with light texture and low soluble content, while the other mentioned types are associated with moderate-to-high soluble content soils. Soils with heavy texture, high salinity and soluble ions rate (particularly sodium and chlorine) indicate T. ramosissima habitat.

4. Conclusion The results showed that in the study area, among different environmental factors (topographic and edaphic variables), the distribution of vegetation types was most strongly correlated with some soil characteristics such as salinity, texture, potassium, lime and gypsum. In arid and semi-arid regions, the relation between species distribution and salinity gradient has been reported by many investigators (Ungar, 1968; Flowers, 1975; Kassas, 1975; Jafari, 1989; Moghimi, 1989; Zahran et al., 1989; Asri, 1993; Caballero et al., 1994; Maryam et al., 1995). Abu-Ziada (1980) also showed strong relationships between vegetation pattern and soil moisture–salinity gradient in the Kharga and Dakhla Oases. Soil texture controls distribution of plant species by affecting moisture availability, ventilation and distribution of plant roots. The role of soil moisture, as a key element in the distribution of the plant species, is described by Zohary and Orshan (1949) in the Dead Sea region of Israel and El-Sheikh and Yousef (1981) in Al-Kharg springs. Also, in the study area, potassium is one of the effective factors in the distribution of vegetation types. Jensen et al. (1998) suggested that K/Mg ratio is an indicator of grasslands and shrublands separation. High K/Mg ratio is suitable for shrub growth. In the present study, combination of PCA and CCA results showed that potassium indicates A. sieberi type. T. ramosissima has strong relationship with soil salinity and heavy texture. This species showed a trend to high soluble rate, salinity and clay percentage. A. aucheri–As. sp., H. aphyllum, E. strobilacea, C. monocantha and S. rosmarinus types indicate hillslope lands and also soils with light texture and changeable solubles, and hence A. aucheri–As. sp. site is located in the soils with low soluble content, while the others are in the soils with average-to-high soluble content. E. strobilacea type is reflecting the soils with considerable amount of gypsum. The types of S. rosmarinus and H. phyllum are directly related to pH and lime percentage, while A. aucheri–As. sp. type shows an inverse relation with these factors. Totally, each plant species has specific relations with environmental variables. These relations are because of habitat condition, plant ecological needs and tolerance range. Understanding the indicator of environmental factors of a given site leads us to recommend adaptable species for reclamation and improvement of that site and similar sites.

ARTICLE IN PRESS M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

639

The other results of using PCA and CCA are as follows. Since these methods are of high accuracy and have different abilities, they could be used for habitat analysis and determination of effective ecological factors. Analyzing ecological data using ordination methods makes simpler understanding of the complex relationship between plants and environmental gradients. In addition, these methods prevent presence of ineffective factors and data complexity from affecting ecological models.

Acknowledgements All costs of current research has been provided by National Plan through Scientific Research Organization of Iran and the University of Tehran. The authors express their gratitude to the Research Vice Chanceller of Tehran University and Mr. A. Nazarzadeh for their assistance and encouragement during the course of this study.

Appendix A Units and abbreviations of the vegetation types and environmental factors in the figures and tables. Artemisia acheri–Astragalus sp. Artemisia sieberi Ephedra strobilacea Cornulaca monocantha Haloxylon aphyllum Seidlitzia rosmarinus Tamaerix ramosissima Dorema ammoniacum Stipa barbata Salsola rigida Eigenvalue Elevation (m) Slope (%) Gravel (%) Clay (%) Silt (%) Sand (%) Saturation moisture (%) Gypsum (%) Lime (%) pH (acidity) ECe (ds/m) Sodium ion (Na+) (meq/l)

Ar.au–As. sp. Ar. si Ep. st Co. mo Ha. ap Se. ro Ta. ra Do. am St. ba Sa. ri Eign abs slope gr clay Lo sa sm gy Li pH EC Na

ARTICLE IN PRESS 640

M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

Potassium ion (K+) (meq/l) K Calcium ion (Ca2+) (meq/lit) Ca Magnesium (Mg2+) (meq/l) Mg Chlorine (Cl
) (meq/l) Cl Carbonate (CO2
CO 3 ) (meq/l) Bicarbonate (HCO
HC 3 ) (meq/l) Sulfate (SO2
SO 4 ) (meq/l) Sodium absorption ratio (SAR) SAR Code 1 is related to the soil characteristics were measured in the first layer (0–30 cm) Code 2 is related to the soil characteristics were measured in the second layer (30–60 cm)

References Abu-Ziada, M.E.A., 1980. Ecological studies on the flora of Kharga and Dakhla Oases of the Western Desert of Egypt. Ph.D. Thesis, Faculty of Science, Mansoura University, 342pp. Asri, Y., 1993. Some ecological characteristics of halophyte plant associations in western margin of Uromieh Lake. Journal of Research and Construction 18 (1), 21–25 (in Persian). Black, C.A., 1979. Methods of soil analysis. American Society of Agronomy 2, 771–1572. Caballero, J.M., Esteve, M.A., Calvo, J.F., Pujol, J.A., 1994. Structure of the vegetation of salt steppes of Guadelenitin (Murcia, Spain). Studies in Oecologia 10-11, 171–183. El-Sheikh, A.M., Yousef, M.M., 1981. Halophytic and xerophytic vegetation near Al-Kharg springs. Journal of Colloid Science University Riyadh 12 (1), 5–12. Fattahi, M., 1998. Survey of qualitative and quantitative trend in vegetation variation of Poshtkouh rangelands in the period of 1986–1988. MSc. Thesis, Natural Resources College of Tehran University (in Persian). Flowers, T.J., 1975. Halophytes. In: Barker, D.A., Hall, J.L. (Eds.), Ion Transport in Cells and Tissues. North-Holland, Amesterdam, pp. 309–334. Iravani, M., Khajoddin, S.J., Basiri, M., 2002. Determination of the most effective environmental factors on three habitats of range species in catchment of Vahargan river (Iran). Second National Congress on Range and Range Management, Karaj, Iran (in Persian). Jackson, M.L., 1967. Soil Chemical Analysis. Prentice-Hall of India, New Dehli, 498pp. Jackson, D.A., 1993. Stopping rules in principal component analysis—a comparison of heuristic and statistical approaches. Ecology 74 (8), 2204–2214. Jafari, M., 1989. Survey of relationship between salinity agents and distribution of plants in Damghan region. MSc. Thesis, Natural Resources College of Tarbiat modarres University (in Persian). Jenny, H., 1980. The Soil Resource Origin and Behavior. Springer-Verlag, New York, Heidelberg, Berlin, pp. 279–286. Jensen, M.E., Peck, L.S., Wilson, M.V., 1998. A Sagebrush community type classification for mountainous northeastern Nevada rangelands. Great Basin Naturalist 48, 422–433. Jongman, R.H.G., Ter. Break, C.J.F., Van Tongeren, O.F.R., 1987. Data Analysis in Community and Landscape Ecology. Center Fire Agricaltural Publishing and Documentation, Wageningen. Hoveizeh, H., 1997. Study of the vegetation cover and ecological characteristics in saline habitats of Hoor-e-Shadegan. Journal of Research and Construction 34 (1), 27–31 (in Pesian). Kassas, M., 1975. On the ecology of the Red Sea coastal land. Journal of Ecology 45, 187–203. Legendre, P., Legendre, L., 1998. Numerical Ecology, Amesterdam, Second English Edition: Developments in Environmental Modeling, Vol. 20. Elsevier Science, New York, USA, 853pp. Leonard, S.G., Miles, R.L., Burkhartdt, J.W., 1984. Comparison of soil properties associated with basin wildrye and black greaswood in the Great Basin region.

ARTICLE IN PRESS M. Jafari et al. / Journal of Arid Environments 56 (2004) 627–641

641

Makarenkov, V., Legendre, P., 2002. Non-linear redundancy analysis and canonical correspondence analysis based on polynomial regression. Journal of Ecology 83 (4), 1146–1161. Maryam, H., Ismail, S., Alaa, F., Ahmed, R., 1995. Studies on growth and salt regulation in some halophytes as influenced by edaphic and climatic conditions. Pakistani Journal of Botany 27, 151–163. McCune, B., Mefford, M.J., 1997. PC-ORD. Multivariate Analysis of Ecological Data Version 3.0. MjM Software Design. Gleneden Beach, OR. Moghimi, J., 1989. Survey of relationship between vegetation, salinity and water ground table in Hoz-eSoltan playa in Qom province, MSc. Thesis, Natural Resources College of Tehran University (in Persian). Ungar, I., 1968. Species–soil relationships on the great salt plains of northern Oklahoma. American Midland Naturalist 80, 392–406. Van-der-Marrel, E., 1979. Transformation of cover-abundance values in phytosociology, its effects on community similarity. Vegetatio 39, 97–114. Walker, B.H., 1979. Management of Semi Arid Ecosystem. (Management for Semi Arid Ecosystem). Elsevier Scientific Publishing Co., Amesterdam–Oxford–New York. Zahedi, Gh., 1998. Relation between ground vegetation and soil characteristics in a mixed hardwood stand. Ph.D. Thesis, University of Ghent Belgium, Academic Press, London, 319pp. Zahran, M.A., Abuziada, M.E., El-Demerdash, M.A., Khed, A.A., 1989. A note on the vegetation on islands in lake Manzala, Egypt. Vegetation 85, 83–88. Zare, A., 1998. Survey of relationships between topography, soil and vegetation in Kavir-e-Phino of Hormozgan provinc in Iran. MSc. Thesis, Natural Resources College of Gorgan University (in Persian). Zohary, M., Orshan, G., 1949. Structure and ecology of the vegetation in the a dead Sea region of Palestine. Palestine Journal of Botany 4, 177–206.