Similarity and difference in vegetation structure among three desert communities
In this study, species composition and structure differed among three communities, but all had the same trends in seasonal changes. Vegetation is usually closely related to environmental variables and understanding the relationship has been a central issue in plant ecology and synecology (Pausas and Austin, 2001; Jafari et al., 2004; Dias and Melo, 2010). Not only interspecific and intraspecific interactions but also the environmental influence together contribute to the formation and evolution of community structure and function; simultaneously, vegetation effects changes to the environment (Putman, 1994; Li, 2000). In arid areas, water and nutrients are the key factors limiting plant survival (Rundel and Gibson, 1996; Whitford, 2002; Li et al., 2008; Ward, 2009); and in rugged terrain (e.g. hills and sand-dunes), position and aspect have also been shown to be important (Parker, 1988). In the south-western area of the Gurbantunggut Desert which has a stable climate and minor human disturbance (Sun and Yang 2010), heterogeneous microhabitats developed among different communities depending on water content, soil nutrients, soil surface structure and the position on the sand-dune (Qian et al. 2009). Soil water content in layers from 10 cm to 60 cm layers at the bottom of sand-dunes is much higher than soil water content near the top (Wang et al., 2009; Zhang et al., 2010a). EDC and STC mostly grow from the bottom of sand dunes to mid-slope, whereas ASC generally grows in bare sand on the top of sand-dunes. Therefore, we can assume that in the dune system, there is more soil water available to EDC and STC than ASC. It is important also to note that BSCs which are involved in a series of processes which change physical and chemical soil properties, cover the base of dunes but are absent from the top (Zhang et al., 2006; Zhang, 2010b; Zhao et al., 2010). BSCs can improve soil nutrient content (N, P, also K) (Zhang et al., 2006), principally in the 0–5 cm layer. In this study, all eight soil factors are significantly different among the three communities (Table 2). With the exception of pH and TP, the soil parameters for EDC and STC are more than that for ASC. NMDS ordinations and Pearson correlation analysis also reveal that BSC, SOM, TN and EC are most important factors influencing community characteristics and thus the habitat of ASC is very different from those of the other two communities.
We did not investigate the seed banks in this study, but it was obvious from literature that seed banks affected community structure; different habitats also had an effect on seed banks, seed germination and formation of BSCs (Su et al., 2007; Zhao et al., 2010; Liu et al., 2011b). It is difficult for seeds to lodge on the smooth surface of bare sand on dune crests which are also more exposed to wind erosion than the fixed sandy areas at the bottom of dunes or at mid-slope. Seeds are more likely to be blown to the base or mid-slope of sand-dunes, enhancing the resources of the seed bank in those locations (Wang et al., 2003). Wind also reduces the stability of the sandy surface of dune crests making them unsuitable for BSC formation. BSCs can accelerate seed germination, seedling survival, growth and nutrient uptake for most desert species (Long and Li, 2003; Zhang and Nie, 2011), although several studies showed that BSCs inhibited the germination of some species (e.g. Malcolmia africana and Ceratocarpus arenaarius) (Nie et al., 2009). Larger seed banks can reduce the effect of inhibition by BSCs, augmenting species richness in EDC and STC.
Comparable microhabitats in EDC and STC have produced a complex community structure in EDC and STC with an abundance of species, very different from that found in ASC (Figures 3 and 5; Table 3). In all three communities, the number of life-forms change seasonally from spring (four), to summer (five) and to autumn (three) (Figure 4) and the number of species present in spring and summer exceeds the number present in autumn. Therefore, the similarity in response of all three communities to seasonal changes in vegetation structure (species richness and life-forms) implies convergence; evidence of divergence implicit in response to differences in soil nutrient content, moisture content and BSCs.
Biomass allocation pattern and nutrient return
The total live biomass of the three communities in the Gurbantunggut Desert in early autumn were relatively low (82.97 – 140.05 g m–2). Semi-fixed sand-dunes presented the most well developed landforms (dominated by Haloxylon forest) in our study area, with a biomass of 7.53 Mg hm–2 (Wang et al., 2005). Therefore, the TB in EDC, STC and ASC accounts for 17.6%, 15.2% and 11.3% of the TB in Haloxylon forest, respectively. Dead biomass accounted for a quarter of the total biomass in the three communities. Return of nutrients from plants to the soil was an important step in nutrient recycling, providing resources for further plant growth. In the Gurbantunggut Desert, atmospheric deposition accounts for a small quantity of N (Liu et al., 2011a), but most is derived from the decomposition of dead biomass. The Horqin Sandy Land has higher but varying proportions of dead biomass proportions (40.7% – 65.1%) (Li et al., 2005) than the Gurbantunggut Desert (approximately 25%). In contrast, the proportions in subtropical forest (3.31% – 8.6%) (Shen et al., 2011) and temperate deciduous forest (7.4% – 20.3%) (Fang et al., 2007), are significantly less than those of the desert, owing to the rapid nutrient cycling rate in the forests. The consistent proportions of dead biomass in the three desert communities of this study may suggest a convergent adaptation which enhances the stability of nutrient cycling in the different communities.
Studies have reported that although constituting a relatively small portion (less than 1%) of the total biomass in forest in North America, the herbaceous understory, especially for ephemeral plants, has a quantifiable significance at the ecosystem level, mediating carbon dynamics and energy flow and influencing the cycling rates of essential nutrients, including N, P, K, and Mg (Parsons and Moldenke, 1975; Gilliam, 2007). Despite that the dead biomass was not separated into shrub or herbaceous origins in this study, the herbs accounted for relatively larger proportion of the dead biomass. This indicated that herbs played an important role in soil nutrient cycling in deserts.
Consequently, although there were differences in absolute biomass, biomass allocation patterns on a community-scale among the three desert communities were either the same or very similar, presenting obvious evidence of convergence of desert shrub communities in arid environments.
Ecological significance of seasonal changes in structure of herbaceous plants
There are numerous herbaceous species in all three communities, and the seasonal changes of herbs represent similarity. The seasonal variations in life-form and species richness are closely related to the appearance and disappearance of herbaceous species (HSV and HLV), especially HSV. Central Asia is considered to the centre of origin and distribution of HSV following the disappearance of the Tethyan Ocean in the Late Neogene and Early Quaternary Period (Mao and Zhang, 1994). Distribution of HSV extended as far as the Mediterranean, Western Asia and North Africa. In China, the Junggar Basin, which includes the Gurbantunggut Desert, is regarded as the easternmost point of HSV distribution (about 205 species) (Mao and Zhang, 1994). In the Gurbantunggut Desert, HSV benefit from limited snow/rainfall in winter and early spring (Mao and Zhang, 1994; Lan and Zhang, 2008) and being to germinate in March and April in spite of low temperatures. HSV flower and mature in May/June prior to the arrival of hot, dry summer conditions.
During the growing period (April to September), the Gurbantunggut experiences very strong winds, predominantly from April to June, in other words, the growth period of HSV (Wang et al., 2003; Li et al., 2010). HSV and BSCs play an important role in reducing sandstorm frequency (Zhang et al., 2006; Zhang et al., 2010b) although sandstorms are less frequent and not as intense as those experienced in other deserts of China (Wang et al., 2003). In early spring, HSV can form dominant synusia or small communities so that some shrub communities can only thrive after HSV has died off in early summer (Zhang and Chen, 2002). Not only does HSV effectively maintain soil surface stability but also generates change in species composition of desert communities (Table 3), important factors in the preservation of biodiversity and stability of desert ecosystems.
Following the demise of HSV in late summer, HLV grew profusely to occupy the eco-niche left by HSV. However, between the disappearance of HSV and appearance of HLV there appears to be an interval during which there is less competition for resources, allowing the two “similar functional groups” to grow at different times during two seasons (summer and autumn). HLV plays an important role in the desert ecosystem until late autumn. The relationship of HSV to HLV was not germane to community type and was a feature shared by all three desert shrub communities.
The total biomass of herbaceous species was exceptionally low in all three communities (Table 4), and the proportions of total biomass of HLV to TB of total species in autumn never exceeded 8.0%. If the study is expanded to a larger scale to include the small tree Haloxylon, the proportions of total biomass of HLV to TB of total species in autumn drop to 0.6% (EDC), 1.2% (STC) and 0.55% (ASC). Thus the biomass of herbs is easily overlooked in large-scale estimation of biomass but in the Gurbantunggut Desert the ecological significance (e.g. retaining ecosystem stability and species diversity, preventing sand diffusion, animal husbandry, soil nutrient cycling) of herbs, especially HSV, cannot be disregarded.