1 Introduction

Wetlands are threatened all over the world by land reclamation, pollution, and water shortages, and this is especially true in arid areas (Millennium Ecosystem Assessment, 2005; Minckley et al., 2013). However, wetlands are essential for a wide range of ecosystem services provided to humans, such as provisioning (e.g., food and water resources), regulating (e.g., water purification and storage, climate regulation), supporting (habitat, genetic diversity), and cultural (e.g., recreation and tourism) (Millennium Ecosystem Assessment, 2005). Wetland areas in China amount for a total of 65.95 million hectares, of which 40% are under threat due to pollution, overexploitation, and reclamation for urban development (He and Zhang, 2001). According to a recent report (Gu, 2015), the wetland area in China has decreased to 53.60 million hectares, accounting for 5.58% of the total surface area of the country.

Wuliangsuhai Lake in Inner Mongolia, northern China, is an example of a threatened wetland that is located in a dryland region. Depending almost exclusively on water runoff from the Hetao Irrigation Area, the lake is subject to water shortages and high loads of nutrients from nearby farms (Tetsiya, 2010). The latter leads to dense growth of aquatic vegetation and thus to rapid terrestrialization of the lake due to the accumulation of dead biomass (Yu et al., 2004). Large areas of the dominant emergent wetland plant common reed (Phragmites australis (Cav.) Trin. ex Steud.), hereafter "reed"), are harvested annually during winter for paper production, which reduces the deposition of biomass and helps to maintain the lake (Kñbbing et al., 2015).

When water passes through wetlands, it is purified and its nutrient contents are treated in complex and different processes such as sedimentation, decomposition (nitrification and denitrification), and uptake by plants. Therefore, all around the world wetlands are used to treat sewage water (Kang et al., 2002; Liu et al., 2009; Moreno-Mateo et al., 2009; Li et al., 2010). Emergent and submerged aquatic plants play a key role in this process by the uptake and transformation of nutrients (Burton et al., 1978; Steffenhagen et al., 2011; Zak et al., 2014). However, plant decomposition after the growing season consumes oxygen and brings the risk of anaerobic conditions and algae blooms. The regular harvesting of (semi-)aquatic plants can prevent or slow down the decomposition and subsequent processes (Steward, 1970; Burton et al., 1978; Nichols, 1991; Jiang et al., 2007). Such harvesting bears potential risks by pumping up nutrients from the sediment, which can lead to algae blooms (Burton et al., 1978; Huhta, 2009). However, vascular plants buffer against eutrophication by lowering the nutrient availability of phytoplankton (Huhta, 2009). Thus, plant harvesting can help to reduce eutrophication if the macrophyte density is high and the phosphorus input is limited (Burton et al., 1978; Nichols, 1991).

Common reed, one of the dominant plants in wetlands, is highly tolerant against water-level fluctuation, eutrophication, and salinity, and can grow on land and in water (Ostendorp et al., 1995; Ostendorp, 1999; Dienst et al., 2004). Its high adaptability and growth rate also render it an invasive species in some parts of the world (Chambers et al., 2003; Kiviat, 2010). However, due to its high biomass yield, up to 30 t/ha (Allirand and Gosse, 1995), it is also a valuable biotic resource for local peoples in many regions around the world (Mmopelwa, 2006; Tarr, 2006; Kñbbing et al., 2015).

Reed has been proved to be beneficial for lowering the nutrient contents of water as it takes up nitrogen (N) and phosphorus (P) through its rhizomes during the growing season (Kvet and Ostry, 1988; Schulz et al., 2011), which exceeds the N released during decomposition four-to seven-fold (Asaeda et al., 2002). Reed cutting can in some cases remove the N and P from the wetland permanently (Huhta, 2007, 2009; Jiang et al., 2007; Fogli et al., 2014). Some scientists doubt this effect or even think that reed cutting can accelerate eutrophication by pumping up nutrients from the sediment (Liu et al., 2007; Huhta, 2009), but others consider the nutrients released by macrophyte decay to be much more important (Graneli and Solander, 1988). In fact, the accumulation of organic materials in the anaerobic substrates, and nutrient storage in litter, transform reed stands into P sinks (Graneli and Solander, 1988; Asaeda et al., 2002; Zak et al., 2014). Moreover, reed cutting in winter is beneficial to the vitality of the reed bed; it improves its ability to absorb nutrients and leads to higher productivity in the following year (Hansson and Graneli, 1984; Huhta, 2009).

The general process in reed beds is that the nutrients are taken up from the sediment, stored in the large rhizome systems of the reed plants, and transferred to the plant stems and the leaves in the growing season, especially during spring time (Huhta, 2009; Schulz et al., 2011). Additionally, oxygen is transferred through the plant stems to the roots where it enhances the nitrification and denitrification processes and can lead to the retention of P through oxygenation (Huhta, 2009; Li et al., 2010).

Every year, the aboveground part of the reed dies in winter and eventually collapses, leading to biomass accumulation and great oxygen consumption through decomposition processes (Huhta, 2009). The amount and composition of the decomposed biomass has a huge influence on the water and sediment quality. The removal of aboveground biomass (AGB) can significantly improve water quality, leading to less collapsed biomass in the system and the capture of the nutrients inside the biomass (Ostendorp, 1999; Huhta, 2007; Haslam, 2009; Li et al., 2010; Fogli et al., 2014).

The amount of nutrients removed by harvesting the biomass is mainly determined by the cutting time (Schulz et al., 2011). Highest AGB nutrient content is usually reached in summer, with some regional differences (Graneli, 1990; Huhta, 2007; Schulz et al., 2011). Other important factors which determine the harvested biomass and the nutrient quantity are the size and location of the cutting area, the total size of the affected area, the frequency, and the cutting height (Huhta, 2009). In most cases, the AGB is cut in winter when the wet ground is frozen and harvesting is relatively easy. However, the amounts of nutrient extraction and the biomass are lowest at that time, being in February only 20% of the peak time in August (Graneli, 1990; Huhta, 2009). Data from Graneli (1990) for southern Sweden show that the biomass extracted through winter harvesting (standing biomass 7.4 t/ha, harvestable 5 t/ha) is only half of that in summer (10 t/ha). However, winter harvesting will significantly improve the growing conditions and lead to an AGB mass up to twice that of unharvested sites, under the premise that enough nutrients (in particular N) are available (Graneli, 1990; Huhta, 2009; Fogli et al., 2014; Zak et al., 2014).

The use of reeds in treating polluted water has been proven in other eutrophic lakes in China (Li, 2000; Jiang et al., 2007; Tian et al., 2009). Jiang et al. (2007) proved that 15 kg/t and 1.9 kg/t of N and P, respectively, are removed by harvesting P. australis in October. However, this water quality remediation is a long-term process, as the nutrients are not removed from the water directly, but from the sediments where they accumulate over time (Graneli and Solander, 1988). Summer harvest can significantly increase the amount of nutrients removed (Hansson and Fredriksson, 2004), due to larger AGB and nutrient concentration, but can potentially harm the reed plants, hinder regrowth, and release nutrients into the water (Asaeda et al., 2006; Huhta, 2009). Continuous early and mid-summer (April to July) mowing is even considered a way to combat reed stands that have become invasive; underwater cutting and burning of dry reed beds are also effective in this regard (Weisner and Graneli, 1989; Huhta, 2007, 2009). Therefore, a late-summer harvest, usually in August, is considered the best time to ensure high AGB and nutrient removal as well as good regrowth in the following year, because some of the nutrients are already relocated to the reed rhizomes (Bjñrndahl, 1985; Weisner and Graneli, 1989; Huhta, 2009).

The functions of reed in wastewater-treatment plants are well known. However, to our knowledge, no study has yet determined the specific nutrient quantities removed by biomass harvest. This study investigates the impact of winter harvest of reeds (December and January) on a lake nutrient budget, based on our own biomass data. By using available data about reed nutrient content in winter and summer, the current and potential nutrient extraction through harvesting is estimated. By comparing it to the lake nutrient influx and outflux, the effects of reed harvesting on the lake wetland ecosystem are evaluated. Our hypothesis is that the removal of nutrients contributes to the maintenance of that ecosystem, and if the removed nutrients exceed the net nutrient input, the reed harvest can contribute significantly to lake restoration.

2 Study site

With a water capacity of 250–300 million m³, Wuliangsuhai Lake is one of the largest freshwater lakes in northern China and is the largest wetland area in this semi-arid region in the Yellow River drainage basin (Fejes et al., 2008). Because the lake is very shallow with an average depth of 1 m, reed covers about half of its area. The lake is linked to the Hetao Irrigation Area in the west, to the Wula Mountains in the east, and is located at the eastern tip of the Hetao Irrigation Area (Figure 1).

Water for irrigation is diverted from the Yellow River at the western tip of the Hetao Irrigation Area. In the early 1970s, these irrigation canals became, almost exclusively, the water source for Wuliangsuhai Lake. Any excess water from the lake is drained back into the Yellow River. The Hetao Irrigation Area covers an area of 5, 800 km², with wheat, corn, and sunflower as its major crops (Tetsiya, 2010). The annual crop production in Hetao increased from 0.6 million t in 1985 to 2.7 million t in 2008 (Xi, 2010), requiring increasingly more fertilizer to be used in this region. This causes a serious increase in nutrient influx to the lake reflected by, for example, a major bloom of toxic phytoplankton (blue-green algae) in 2008 (Faafeng et al., 2008; Zhao et al., 2010). With high evapotranspiration and low precipitation, the salinization problem is also serious in the lake.

Between 1990 and 2008, about 5 billion m³ water reached the Hetao Irrigation Area, of which about one-fifth ended up in Wuliangsuhai Lake (Xi, 2010). The lake releases about 0.06 billion m³ to the Yellow River and evaporates 0.40 billion m³ (Xi, 2010). From 1998 to 2008, on average more than 2, 100 t of total N and 220 t of total P from agriculture fertilizer reached the lake (Xi, 2010).

3 Synthesis of data

The data basis for this study is a combination of available literature data and our own investigations. Some of the literature cited here originates from the same research project as this work. Because the available literature references are very sparse and scattered, the time frames for much of the data had to be aligned, meaning that the same time scales for reed harvest times and areas had to be used. As a result, some data are rather old, even if more recent data for certain elements are available.

3.1 Past literature data from Wuliangsuhai Lake 3.1.1 Lake cover and biomass The stated size of the lake fluctuates in the different literature sources, but all show a decreasing trend from the 1980s to 2010. Simultaneously, the reed area increased, thus leading to less open water (Cao et al., 2010). The open water area was extremely small between 1996 and 2000 (Table 1), then it increased to 87 km² in 2002. According to Ratnaweera et al. (2008), serious water pollution (excess nutrients from fertilizer) in the 1980s and 1990s led to the decrease of the open water area through massive vegetation growth between 1996 and 2000. Then, remediation projects around 2000 restored some of the open water area. In Wuliangsuhai Lake, the dominant emergent species in water ranging from 0.5 to 1.5 m water depth is the common reed (Phragmites australis; Barton, 2005). With the terrestrialization of the lake and a high nutrient influx from nearby agriculture, the growing conditions for reed became highly suitable. The reed area increased from 17 km² in 1975 (Ratnaweera et al., 2008) to 130 km² (total lake area 310 km²) in 2004 (Yu et al., 2008); simultaneously, the harvested reed biomass rose from 23, 000 to 97, 979 t DW (Fejes et al., 2008; Li, 2011. Usually the winter-cut reed has a moisture content of 12%; dry mass (DM) has a moisture content of zero) (Figure 2; Table 2). The average winter yield (December–February) from 1995 to 2004 was 95, 620 DW t per year (Li, 2011). Pushed by the high demand for reed in paper production, areas along the shore were converted by farmers into reed areas (artificial reed). This recent increase in harvested biomass is based on higher yields per hectare (Table 2) and also on expanding the harvest into new, previously un-harvested sites. Before 1996, the increasing yields were mainly based on the naturally expanding reed beds. Water, which is not covered by reed or cattail (Typha latifolia L.), mainly contains submerged vegetation, mostly pondweed (primarily Potamogeton pectinus) and hornwort (Ceratophyllum spp.) (Li, 2002; Ratnaweera et al., 2008). These species have increased in open water areas from 40% in 1999 to 95% in 2003 (Ratnaweera et al., 2008). The production is about 12.85 t/(ha·a), which means an annual biomass growth of about 1, 474, 000 t wet weight per year or 295, 999 DW per year (Ratnaweera et al., 2008). These plants are not harvested and therefore collapse and decompose every year (Fejes et al., 2008). The consequence is high oxygen consumption during decomposition and sedimentation. Such submerged vegetation is mainly found at depths between 1.0 m and 2.0 m; below 2.5 m water depth it cannot survive due to the lack of light (Faafeng et al., 2008). Concerns arose that the vast amount of submerged biomass would increase the risk of toxic green algae blooms (Fejes et al., 2008). 3.1.2 Soil and reed nutrient interfering At Wuliangsuhai Lake, the AGB was sampled in winter and summer by different authors (Hedelin, 2001; Patuzzi et al., 2012; Li et al., 2014). Hedelin (2001) found in 25 reed samples a nitrogen content of 3.6 kg/t (September) and 0.14 kg/t (stored winter reed) respectively, and a phosphorus content of 0.86 kg/t (September) in 2000. In 2011, Patuzzi et al. (2012) found comparable results taking 13 samples of reed growing inside and outside the lake with an average content of 3.6 kg/t nitrogen and 0.14 kg/t phosphorus (both October). Finally Li et al. (2014) sampled 0.65 kg/t in summer (August) reed growing in Wuliangsuhai Lake (Table 3). 3.1.3 Water quantity and quality The water flow into the Wuliangsuhai Lake showed a decreasing trend, falling from a peak of 733 million m³ in 1995 to around 448 million m³ in 2008 (Table 4; Xi, 2010). The drainage water from the Hetao Irrigation Area carries nutrients from the nearby farm fields and leads to the eutrophication of the lake (Table 5). Besides the pollution from agriculture, there is pollution from industries as well. With 97.1% of the water usage, agriculture is responsible for 54.7% of chemical oxygen demand (COD), 67% of total nitrogen (TN), and 31% of total phosphorus (TP). With 1.4% of the water usage, industries are responsible for 27% of COD, 6% of TN, and 23% of TP of the total pollution (data 1998–2008; Xi, 2010). In 2009, the water in the lake carried 1.74 mg/L N and 0.07 mg/L P, both far exceeding international standards (Chinese Research Academy of Environmental Sciences, 2010). The salinity in the northern and southern parts of the lake was 1.4 g/kg and 2.4 g/kg in 2011, respectively, which were twice the amounts measured in 1987-1988 (Faafeng et al., 2008). The pH value of the lake water was between 7.73 and 9.19 in 2001 (Shang et al., 2003c). 3.2 New fieldwork data 3.2.1 Field sampling We measured AGB in September 2011 and October 2012. The vegetation plots were marked in the field by Global Positioning Systems (GPS). In total, 17 samples from non-submerged (on land) and 15 from submerged (in lake water) reed stands were collected (Figure 3). The sampling plots were selected randomly but were limited to those accessible by boat, foot, or car. Following the approach of Thevs et al. (2007), the AGB was calculated using the stem density and the average total weight of five sampled shoots from four different subplots at every site. In addition, for each harvested plant, the stem length and the basal stem diameter were measured and the number of leaves were counted. Four subplots (squares with side length of 0.5 m) were chosen randomly, all of them in the same distance from a fixed central point. On each subplot, the stems were counted, and the five stems closest to the diagonal were cut. The stems were weighed, oven dried at 105 °C for 48 h, and again weighed to determine the moisture content and dry biomass per hectare. 3.2.2 Reed biomass On all of the non-submerged plots, an average standing biomass of 10.89 and 9.93 t DM/ha was measured in October 2011 and September 2012, respectively (Table 6). In the lake water plots, the standing biomass was significantly higher, at 42.70 and 16.40 t DM/ha, respectively. The average dry biomass over both years for submerged and non-submerged plots was 19.98 t DM/ha. 4 Results and discussion 4.1 Reed biomass

The biomass in the submerged plots was considerably higher compared to the non-submerged plots on the shore (Table 2). The reasons were better growing conditions due to permanent water availability as well as fewer anthropogenic disturbances such as grazing, cutting, ploughing, or burning.

The measured average standing reed biomass of about 22 t DW/ha (Table 6) was significantly higher than the calculated value in Table 2 (7.47 t DW/ha), but was similar to the results of Hedelin (2001) and Duan (2004), who measured a higher average reed dry AGB with 15.63 t DM/ha in September 2001 and 17~30 t DM/hain 2002 for Wuliangsuhai Lake, respectively. However, it has to be considered that not all available reed beds were harvested, as was anticipated for the calculated yields per hectare in Table 2. Reed harvesting is prohibited at the bird refuge on the east side of the lake, and other areas might be unattractive due to less accessibility or low yields. Zhu et al. (2014) even stated that only 4% of the total lake area (24 km²) was harvested in 2009. Calculated with the winter reed yields of 114, 669 t DW (2009; Figure 2) and a reed area of 127 km² (average 1996–2004; Table 2), the potential yield would be more than 1 million t DW of reed per year, which seems to be unrealistic. Our own measurements would, if based on the total reed area, produce a potential annual harvest of 286, 506 t DW (moisture content 12%) (Table 7).

The measured P. australis AGB is also in line with results from other wetlands around China, such as the Liaohe Delta with 14 t DM/ha (Xiao and Li, 2004), and on rewetted peatlands in Germany with 12.5 to 23.8 t DM/ha (Zerbe et al., 2013).

4.2 Nutrient budget and removal

To determine the potential contribution of reed for water purification, data about reed total yield, nutrient contents, and nutrient influx into the lake are required. In order to be a feasible restoration measure, the nutrient removal must exceed the nutrient input.

The average winter N and P contents in the AGB were 3.6 kg/t and 0.014 kg/t, respectively (Table 3). Multiplied by the average measured annual reed winter biomass yield of 100, 253 t DW, the current harvesting practice removes an average of 362 t of N and 14 t of P (1995–2010). This corresponds to 16% and 8% of the total N and P influx, respectively (Table 7), according to nutrient influx values of Table 5.

Results from Hedelin (2001), who did an analysis of P removal by reed winter harvesting at Wuliangsuhai Lake, show comparable results, with a removal of 61 t DM P (20% of the influx). In the same study, the potential removal by cutting earlier in the year, i.e., autumn, was estimated to be 196 t P.

A summer cutting of the current harvesting area would significantly increase the nutrient amount within the removed reeds, and also double the amount of AGB. Up to 106% of P could be removed, thus proving to be an effective management measure (Table 7).

The results show that not all potential reeds are harvested right now. Hypothetically, all available reed beds could be harvested to increase the nutrient removal. Table 7 presents the impact on the nutrient budget for winter (Scenario I) and summer (Scenario II) harvesting. The numbers indicate that harvesting all available reed beds could significantly increase the nutrient outtake and would, at least for one summer harvest, far exceed the input.

Some authors (e.g., Haslam, 2009) have raised concerns that regular harvesting leads to an eventual nutrient shortage, and possible fertilization is required to ensure a stable reed yield. This is not true for this case study because the irrigation system upstream ensures a steady influx of nutrients (see Table 5).

4.3 Harvesting management

Worldwide, winter reed cutting is usually done where the biomass is utilized. If the water is frozen, the harvesting is comparably easy, the biomass regrowth is enhanced, and the impact on wildlife is low or even positive (Graneli, 1990; Tanneberger et al., 2009). Also, the effect on the nutrient budget is negligible as most of the nutrients are stored in the plant rhizomes (Huhta, 2009).

A summer harvest cutting potentially increases the nutrient removal but also risks nutrient release into the water and disturbance of wildlife (Huhta, 2009). Moreover, it disturbs the growing conditions of the plants, lowers their energy reserves, and thus reduces possible yields the next year by 30%–40% (if cut in July), as shown in the UK (Haslam, 2009; Huhta, 2009). Therefore, this strategy can be used to eliminate reeds but its effect on regrowth depends highly on the timing. Overall, early summer harvesting cannot be recommended (Graneli, 1990; Huhta, 2009). One option could be summer cuttings on rotating reed areas from year to year. However, the benefit to the nutrient budget in terms of total removed nutrients has to be considered carefully.

To achieve synergies between commercial reed use and water quality improvement, the timing of the cutting seems to be the most important factor. In Europe, cutting in June and July has been shown to significantly reduce the biomass in the following year, but no growth reduction in the next year was detected if reed was harvested in August (Weisner and Graneli, 1989; Asaeda et al., 2006; Haslam, 2009; Huhta, 2009). In Japan, even cutting in July seems not to affect biomass growth next year (Asaeda et al., 2006).

4.4 Restoration effect

The nutrient influx at Wuliangsuhai Lake is very high, so reed cutting can be an essential part of nutrient removal. Also, the effect on lake restoration can only be long-term because so much of the nutrients are stored in the sediment, and will be released from the sediment over many years.

Liu et al. (2007) showed that the sediment below emergent vegetation (such as reed) and submerged vegetation in Wuliangsuhai Lake has increased inorganic P content by 44 and 107 μm/g, respectively, compared to open water. Similar trends can also be shown for organic and total P. Phosphorus is regarded as the most important factor determining the trophic status of the lake (Schindler, 1978). Rooted macrophytes take up P directly from the sediment, but also increase the P retention (up to 25%) by reducing velocity and acting as a filter (Liu et al., 2007). On the bottom of Wuliangsuhai Lake, a plant litter layer of 0.2~0.4 m has been accumulated so far, with a current terrestrialization rate of 11.7 mm/a(Liu et al., 2007; Yu et al., 2008).

Table 7 shows that the current reed-cutting pattern removes 8% of the total P. Considering the outflow every year (Table 5), more than 90% of the P will remain in the lake. Therefore, harvesting of submerged vegetation as well as reeds has been considered as a restoration tool by several local and international scientists (Shang et al., 2003a; Fejes et al., 2008; Zak et al., 2014). Steffenhagen et al. (2011) showed that different submersed macrophytes contain nutrient contents of 28~44 kg N/haand 8~12 kg P/ha. Shang et al. (2003b) calculated that harvesting 50, 000 t wet weight of submerged vegetation (mainly Potamogeton pectinatus) would annually remove 825 t of N and 85 t of P from Wuliangsuhai Lake.

Cutting of all available reed areas for summer or winter harvest seems to be unrealistic, considering the effects on wildlife and reed vitality. Given that reed summer harvesting is recommended to restore the lake, questions must be raised about the effect on the local economy. As described by Kñbbing et al. (2015) and Kñbbing et al. (2014), the local reed resources are almost exclusively utilized for paper production. Local farmers and fishers are engaged in the reed harvesting, representing an important additional income in winter times. A potential summer harvest would mean the paper production would have to be given up because it requires dry material, and instead new utilization possibilities for the cut reed would have to be found. Kñbbing et al. (2013) showed that most of the current uses of reed, i.e., for paper production, construction material, or combustion, depend on dry winter reed. Only a few products, such as biogas or fodder, are made from spring or summer reed. Also, summer harvest by boat is usually much more complicated and expensive.

5 Conclusion

Wuliangsuhai Lake can be seen as an outstanding example how ecological biomass harvesting can help to maintain and restore threatened wetlands. The fact that the restoration effect has until now been almost totally neglected is especially interesting. Reed harvesting is seen as potential income source for local people, and sometimes aquatic plants are considered as a reason for the high terrestrialization rate, even if it is just a response to high nutrient availability.

The reed beds at Wuliangsuhai Lake can positively and negatively impact the water quality. Their extensive growth is a direct consequence of high nutrient loads and rapid terrestrialization of the lake over the last decades. Reed takes up a high percentage of the nutrients in the below-and aboveground biomass from the soil and transforms them by mineralization. However, in winter the AGB of reed dies and the decay process results in high oxygen consumption and accelerates the lake terrestrialization.

The proper management of reed harvesting, as shown above, can significantly improve the effect of reed cutting on the lake nutrient budget, bearing in mind that nutrients are not removed from the sediment directly, but are uptaken from the sediment. In this respect, reed harvesting is valuable from a purely economic market standpoint and is also an effective tool to maintain the lake ecosystem. A financial loss from reed selling could potentially be compensated by the government for the natural wastewater treatment provided by the reed beds.

If the nutrient amount harvested is larger than the nutrient influx minus the outflow amount, the total mass of nutrients within the lake decreases. The lake becomes less nutrient-rich and also less eutrophic. As shown, this could be achieved by increasing the harvested area or time of harvesting. However, it also becomes clear that all possible cutting scenarios have certain constraints and must be evaluated carefully before they can be applied.

Acknowledgment:

The authors are grateful for the financial support of the project "Sustainable Water Management and Wetland Restoration in Settlements of Continental-arid Central Asia" (SuWaRest) by the Kurt-Eberhard-Bode Foundation within the Stifterverband für die Deutsche Wissenschaft.