1 Introduction

The process of desertification is the succession of ecosystems driven by human land-use and management and natural events (NBCDMFC, 1996). Therefore, along the gradient of these driving forces, there occurs a transition of vegetation structure and function (Li, 2001), which indicate the different stages or degrees of desertification (Cross, 1994). Much has been gained in understanding the processes and development trend of desertification under various pressures in other desert areas by studying the vegetation structure and function along the succession gradient (Danin, 1996; Huenneke, 1995; Li et al., 1999). Such knowledge is vital to ecosystem management and desertification combating. However, there exists wide diversity in original ecosystem types, natural conditions and change patterns, principal driving force as well as the history of disturbance in the studied areas. Consequently, the results obtained from other regions cannot be simply applied in Chinese desert areas. In this study, we have conducted systematic field experiments on vegetation structure in terms of species composition and spatial pattern, and function (biomass production, element biogeochemistry, water transportation, etc.). Primary work in Minqin, China, focused on ecosystem carbon cycling, the most important ecosystem function process. We suppose that: (1) Along the desertification gradient, there should be a significant ecosystem function gradient, which can be indicated by ecosystem carbon metabolism; (2) In respect to biogeochemistry, there are remarkable differences in nutrient cycling strength and spatial distribution along the gradient; (3) Increase in desertification severity, with the development of sparse and specialized vegetation, matter cycling strength decreases and the spatial distribution of vegetation tends to be uniform.

2 Site conditions

The study area is located in Liujiadi (103°15′ E, 38°38′ N), about 29 km northwest of Minqin county, Gansu Province, P.R. China. Annual mean air temperature in the study area is 7.8℃, with the maximum of 38.1℃ and the minimum of -28.8℃. The annual mean precipitation is 115 mm, most of the precipitation falling from July to September. The dryness degree is 5.3, annual average potential evaporation is 2643.9 mm, which is 23 times larger than annual precipitation. The underground water table is 18 m with annual decreasing rate of 0.63 m as the result of overdrawing underground water for irrigation and the free-frost period is between 165 days. The soil is a typical wind-blown sandy soil with low organic matter and soil fertility. From the edge of an oasis in the east to mobile sand dune in the west, there exists a transitional zone for approximate 2, 000 m, along which there occurs Tamarix ledeb communities, Nitraria tongutorum communities and mobile dune, respectively. The spatial pattern of vegetation distribution along the ecotone landscape between oasis and desert was reported by Jia et al. (2002).

3 Methods 3.1 Gas exchange measurement

Leaf photosynthetic rate was measured by LI-6400P (LICOR, Inc. Superior Street, Lincoln, NE, USA) during June 18 to 20 and September 28 to 30, 2000, respectively, in the transitional zone from oasis to desert. Three leaves (in case of T. ledeb the bundle of leaves) from the healthy plant species of T. ledeb and N. tongutorum community, T. ledeb and N. tongutorum, respectively, were selected for gas exchange measurements. The sampling leaves were fully expanded and on the south-facing aspect in the upper third of the canopy. The measurements began at about 10:00 in the morning when the stomata fully opened. All the measurements were made under the artificial light source, the blue-red light source accessory of the LI-6400P system. The photosynthetic active radiation (PAR) was set at 1 600 μmol·m^-2s^-1 both in June and in September. The leaves were cut off after measurements and brought to the laboratory for dry weight determination. All other environmental parameters were recorded automatically and simultaneously by LI-6400P.

The responses of photosynthetic rate to light intensity (usually call A-PAR curve) and to CO₂ concentration (A-C_i curve) were determined by controlling PAR and CO₂ concentration in the leaf chamber. For A-PAR curve, 9 PAR intensities were applied by the sequence of 2 000, 1 500, 1 000, 500, 200, 100, 50, 20, 0 μmol·m^-2s^-1 in June and 8 PAR points, 1 600, 1 300, 1 000, 700, 400, 100, 50, 10 μmol·m^-2s^-1 in September, respectively. The CO₂ in reference chamber (C_a) was controlled to be 360 μmol·mol^-1. A-C_i curve was determined only in September with light intensity of 1 600 μmol·m^-2s^-1. The C_a concentrations changed sequentially by 360, 200, 100, 50, 150, 400, 800, 1 200 μmol·mol^-1, respectively. Temperature was not artificially controlled due to the cloudy weather and battery supply limitation. The A-PAR curve and A-C_i curve for each species were measured for 3~5 times, respectively, both in June and in September.

Leaf respiration was measured inside the leaf chamber when the light source of LI-6400P was turned off and the leaf chamber was totally dark. Dynamics of leaf chamber CO₂ concentration was recorded every 3 s whilst dark respiration rate was calculated accordingly. This process keeps up for about 3 to 5 minutes. The C_a was maintained at 360 μmol·mol^-1. The measurements of leaf respiration were made on the same leaves with photosynthetic measurements.

3.2 Soil respiration measurement

Soil respiration chamber, an accessory of LI-6400P, was used for soil respiration in this experiment. With the plastic collar, it covered an area of some 200 cm². The atmospheric CO₂ concentration was detected first and it was set as the reference point. Three cycles were performed in this dynamic chamber method for each site. The five locations in different directions and heights of the dune were selected for measurements in T. ledeb sandy dune, while in other sites three locations were randomly chosen as replica.

3.3 Data analysis

A one-way analysis of variance (ANOVA) was used to test the leaf eco-physiological parameters of T. ledeb and N. tongutorum in June and September respectively. Duncan′s test was done to four positions of T. ledeb community and five vegetation types, respectively. Difference was considered significant only at P < 0.05.

4 Results 4.1 Photosynthetic rate of T. ledeb and N. tongutorum leaves under controlled conditions

There was a marked diurnal variation in PAR in the study area. For example, the highest PAR reached 1 940 μmol·m^-2s^-1 and lowest less than 430 μmol·m^-2s^-1 during the experiment period. PAR was rather lower in September than in June due to cloudy weather. Under the experimental PAR and C_a condition, T. ledeb leaves had a much higher leaf-area-based photosynthetic rate than N. tongutorum leaves. The former was 1.55 and 1.98 times larger than the latter in June and September, respectively. However, the photosynthetic rates based on leaf dry weight were similar for the two plant species. There were no difference in LSP for the two species, but the differences occurred between June and September. LCP had the similar response as LSP did (see Tab.1).

4.2 Response of photosynthetic rate to light intensity

The leaf-area-based photosynthetic rate of T. ledeb leaves was consistently higher than that of N. tongutorum irrespective of PAR values. The higher was the radiation, the larger was the difference (Fig. 1). However, N. tongutorum leaves had higher light use efficiency than T. ledeb did in June, in particular, based on leaf-dry-weight. In addition, the two species had similar A-PAR response curves if based on leaf-dry-weight (Fig. 2). There was no remarkable difference between their photosynthetic rates along the PAR gradient, although T. ledeb still had slightly higher photosynthetic rate in most cases. The two species also had similar light saturation point and light compensation point (Tab. 1). At the end of growing season (in late September), the light saturation point and light compensation point of both species reduced. Their photosynthetic rates also decreased by approximately 50% in contrast to measurements made in June, and the light efficiency declined much more (Tab. 1).

4.3 Response of photosynthetic rate to CO₂ concentration

Fig. 3 showed that photosynthetic rate of T. ledeb leaves became leveled off when atmospheric CO₂ concentration was above 800 μmol·mol^-1. However, at this CO₂ concentration range, the photosynthetic rate of N. tongutorum leaves increased obviously linearly, indicating that the highest CO₂ concentration supplied in this experiment did not reach its CO₂ saturation point. At high CO₂ concentrations, stomatal conductance of N. tongutorum decreased gradually, and it might happen that a higher concentration would inhibit stomatal openness. It implied that N. tongutorum might have higher CO₂ saturation point than T. ledeb in late September. Carboxylation efficiency of T. ledeb leaves was also lower (Tab. 2). However, the maximum photosynthetic rate at saturation point (calculated by model) was higher for T. ledeb based on leaf area. Nevertheless, N. tongutorum leaves would have higher photosynthetic rate if calculated on leaf-dry-weight. The two species had similar CO₂ compensation point in this season.

Comparison of the photosynthetic rates at the two CO₂ concentrations, i.e. atmospheric concentration 360 μmol·mol^-1 and saturation point concentration, indicated that the photosynthetic rate of T. ledeb leaves increased by less than 50% while N. tongutorum increased by more than 150%. Therefore, N. tongutorum had higher carboxylation enzyme activity and larger response to CO₂ elevation. It was supposed that N. tongutorum would be more competitive than T. ledeb does in future atmospheric CO₂ elevation scenario with respect to carbon fixation.

4.4 Water use efficiency of the T. ledeb and N. tongutorum leaves

In this experiment, water use efficiency (WUE) of N. tongutorum was found to be consistently higher than that of T. ledeb. Although N. tongutorum had much lower photosynthetic rate, its stomatal conductance and the resultant transpiration rate were much lower. Therefore, N. tongutorum tended to transpire less and conserve more water. Elevated CO₂ concentration gradually inhibited stomatal opening of N. tongutorum leaves and led to lower transpiration rate at high CO₂ concentration (Fig. 4). However, its photosynthetic rate enhanced rapidly with increasing CO₂ concentration, leading to a remarkable increases in WUE (Fig. 5). For T. ledeb, high CO₂ had a little influence on stomatal opening or transpiration rate while the photosynthetic rate was promoted slightly.

4.5 Dark respiration rate of T. ledeb and N. tongutorum leaves

With the fast auto-recording property of the LI-6400P system, it is possible to distinguish the dark respiration in the field. In June, T. ledeb leaves had lower dark respiration rate compared to N. tongutorum. If based on leaf dry weight, however, we would get the reverse result that N. tongutorum leaves had lower dark respiration rate. The leaf-dry-weight-based dark respiration rate of T. ledeb leaves was about 1.5 times than that of N. tongutorum leaves. In late September, the leaf-area-based dark respiration of T. ledeb leaves was higher than that of N. tongutorum. Compared to the measurements in June, the dark respiration reduced by 28% and 50% for T. ledeb and N. tongutorum leaves in September, respectively.

4.6 Soil respiration along the vegetation gradient

Soil respiration was usually the most important part of total ecosystem respiration, in some systems it accounted for 70%~95% (Cui et al., 2001). In this study, the total aboveground respiration was not estimated due to the lack of respiration measurements of branches and stems.

There were significant spatial and seasonal variations in soil respiration rate and CO₂ emission quantity in the study area. Among the five locations, the lowest CO₂ emission rate usually occurred in the mobile sand area, while the lowest value was found in crust-covered surface in September (Tab. 4). Within the T. ledeb sandy dune, the largest differences among the four measured locations were 50% and 60% in June and September, respectively (Tab. 4). The seasonal differences for the four locations ranged from 22% to 152%. Of all the sampling locations, soil respiration rate was lower in September than in June, with an average of about 40% reduction and ranging from 11% to 78% (Tab. 4).

5 Discussion 5.1 Change of vegetation carbon fixation along desertification gradient

As native desert species, the photosynthetic rate of T. ledeb and N. tongutorum based on leaf dry weight was similar. This was due mainly to the greater specific leaf weight (SLW) of T. ledeb compared to N. tongutorum. Estimation of leaf area in the field might add extra error to it, too. Therefore, plant and community level carbon fixation depends on total leaf-dry-weight of a plant or community. T. ledeb plant usually has much higher leaf dry-weight than N. tongutorum plant, and it also true for comparison based on community level. Hence, the T. ledeb community has higher carbon-fixing capacity and productivity. Similar light response of the two species assures the difference of carbon fixation during the whole growth period. It implies that productivity loss resulting from desertification is due to succession of vegetation dominated by species with different carbon-fixing capacity in this region.

T. ledeb is characterized by high transpiration rate and low WUE; its distribution largely relies on shallow groundwater or irrigation in nearby croplands. Rapid decrease in groundwater and lack of available irrigation water resources (Wang, 1999) may lead to reduction in distribution area of T. ledeb community. Future drier climate is likely to worsen this situation. On the contrary, N. tongutorum use water more economically. This maybe partly contributed to its distribution along the vegetation gradient - it was farther away than T. ledeb community from the agriculture area in the oasis, where side penetration of irrigation supplies water to surface soil of nearby T. ledeb community. In another word, T. ledeb had higher water competitive ability. It had much higher transpiration rate and stomatal conductance, and this enable T. ledeb to use soil water quickly and competitively. As the result, other species will be depressed by available water coming from the same location. When a drought year occurs, however, T. ledeb may suffer much more than N. tongutorum does. Such water and light use strategies directly influence their distribution and community dynamics and desertification process as well. In the future with increasing atmospheric CO₂ concentration and decreasing precipitation, N. tongutorum community is likely to shift into the area where T. ledeb community occurs. In addition, decrease in area of T. ledeb community will be larger than that of N. tongutorum community. Even the photosynthetic rate of N. tongutorum is also promoted by elevated CO₂, less carbon-fixing apparatus in an area unit will reduce total productivity of the region, and consequently, severe desertification is expected.

5.2 Carbon loss and carbon balance along desertification gradient

At community level, T. ledeb plants usually have much higher leaf area or leaf dry weight and it can fix more carbon and produce more biomass than N. tongutorum do. It also loses large amount of carbon through respiration although the percentage in apparent photosynthesis was lower. This is because T. ledeb leaves have higher dark respiration rate than N. tongutorum leaves.

Soil respiration rate reflects soil biological activities and strength of biogeochemical processes (Luizao et al., 1998). The lowest respiration rate in mobile sand area is due to low content of root, organic matter, and microorganisms. In the middle of June, favorable air and soil temperature improve the growth of above- and below-ground plant biomass. Therefore, more carbon is fixed and transported to root and more CO₂ released from growth and maintaining respiration of roots. Besides, in this season the soil water content is relatively high, which also stimulates growth of roots and soil microorganisms, as well as organic matter transformation and element movement. In September, low temperature and soil water content inhibit biological activities and matter transformation. Cold tolerant Reaumuria soongdica grows well in late September and has stable soil CO₂ emission rate.

Even though there was no obvious difference in respiration rates between soils adjacent to N. tongutorum and T. ledeb plants, the difference is supposed to occur at community scale due to various complexities of relief and community structure. In T. ledeb community, between high T. ledeb mounds there are N. tongutorum mounds, crust-covered fixed sandy dunes, and flats with growing R. soongdica and Zygophyllum xanthoxylum. Hence, there exists higher heterogeneity of soil physical, chemical and biological conditions in T. ledeb community, which infers different spatial and temporal patterns of biogeochemical process and strength.

5.3 Crust-mediated vegetation dynamics and desertification

The function of soil crust is a special characteristic in this region. The crust intercepts rain and dew water (Li et al., 2000) and exhibits high respiration rate in June. However, crust loses water easily through evaporation since the water is hold in the crust surface, and this was accompanied with rapid crust decline in biological activity as seen in the case of respiration in September. This function of crust, in one hand, dries the lower soil layers and lead to root and plant death (Li et al., 2000). In the other hand, crust decreases soil water storage and makes plant inhabitation more difficult. We suppose that favorable conditions in surface soils of N. tongutorum or T. ledeb mounds result in crust development (Li et al., 2000). Once there appears large area of crust in N. tongutorum or T. ledeb mounds, the N. tongutorum and T. ledeb plants will die. The mounds turn to be crust-covered fixed mounds without plants or with few unhealthy plants. The crust will be finally dried up and eroded by sandy wind. The mounds then transfer to mobile sandy dunes.

6 Conclusion

T. ledeb community has higher rate of photosynthesis and soil respiration, therefore, its intensity of carbon biogeochemical cycling was higher than that of N. tongutorum community. During the desertification process from T. ledeb community to N. tongutorum community and further to mobile sand, local and regional carbon cycling and nutrient turnover changes dramatically. The strength of cycling as well as temporal and spatial variation decreases.

Further measurement of above-ground growth rate and biomass of communities, plant and soil respiration, soil organic matter content and its transformation should be integrated into consideration for better understanding the community-level carbon budget and carbon biogeochemical cycling along the transitional zone between oasis and desert.