A toxic endophyte-infected grass helps reverse degradation and loss of biodiversity of over-grazed grasslands in northwest China

Overgrazing of China’s grasslands is increasingly causing biodiversity to decline. In degenerated grasslands of northwest China endophyte (Epichloё gansuensis) infected Achnatherum inebrians (drunken horse grass) is becoming widely distributed because of its toxicity to livestock. In this study, we investigated the ecological consequences of endophyte toxicity in this native grass, at three sites in northwest China, by comparing seed production of plant species and arthropod abundance in overgrazed grasslands with and without the presence of A. inebrians. Our findings demonstrate that the presence of endophyte infected A. inebrians reduces the loss of plant and arthropod biodiversity by providing a protected nursery free of animal grazing. Therefore, A. inebrians, typically regarded as an unwanted toxic invader by pastoralists, should be viewed as beneficial for grasslands as its presence maintains plant and arthropod biodiversity, and provides a foundation stone in the reconstruction and restoration of these grassland ecosystems.


Impact of A. inebrians on arthropods.
Most of the arthropod families identified from soil samples were found with significantly greater numbers in TA compared to NA. In Xiahe, of the 15 families of arthropods 12 were significantly greater in TA when compared with NA. Distribution of three arthropod families did not significantly differ between in the two treatments. Total quantity of individuals of all families was significantly more abundant in TA than NA (Table 2).
In Guinan, there were 11 families of arthropods that were unique to TA. No arthropods were identified in the NA pastures in this study. Nine of the 11 families had significantly greater number in TA than NA (Table 2).
In Alxa, there were 12 families of arthropods identified, seven of which were only found in TA, with 5 families being present in both treatments, only one of which was significantly increased in the TA sample. Total quantity of all arthropods was significantly greater in TA than NA ( Table 2).

Contribution of zokor and pika to the spread of A. inebrians. The establishment rate from seeds of
A. inebrians on molehills of zokor was significantly greater than in NA in Xiahe (Fig. 1a). The average density of one-year old A. inebrians plants in bare patches was significantly greater than in NA in Guinan (Fig. 1b).

Discussion
Achnatherum inebrians is likely to play a positive role in the protection and regeneration of overgrazed grassland ecosystems in northwest China since it was shown in this study to be able to colonise bare ground. The growth habit of A. inebrians is such that large tussock plants develop interspaced with other palatable pasture species which can become established due to the production by A. inebrians of endophyte toxins that deter livestock grazing. This sheltered environment becomes a nursery area where the protected plants can flourish and again produce seed. Therefore, rather than being an undesirable invader [12][13][14] , this study has revealed that A. inebrians provides a means by which the seed resources of edible plant species can be saved in the face of over-grazing.
Without bare soil A. inebrians has a poor ability to establish from seed. The bare land in which A. inebrians can establish and which is now a widespread feature of degenerated grasslands, is frequently the result of the feeding activity of burrowing animals 18 . A common viewpoint is that the bare patches caused by pika and the molehills of bare earth caused by zokor are a major cause of grasslands degeneration 6 . Zokor and pika are widely distributed in grasslands and have a high density in the Qinghai-Tibetan Plateau (QTP) 20,21 . Bare patches caused by pika accounted for 62.6% of the total area of degraded grassland in Qinghai Province, while 25% of the total area was degraded in the QTP 18 . However, it is likely that continual overgrazing by livestock is primarily responsible for the conspicuous damage that these rodents are now causing. As suggested by Harris 22 , the presence of abundant rodents may be an indicator rather than a cause of grasslands degradation. Maybe overgrazing weakens the plant cover, even leading to the eradication of some plants and making the land more prone to bare patch formation. With a reduction in the concentration of suitable plants for food 23 , it seems that these small animals have increased their intensity of burrowing 24 and also the area of burrowing activity in order to obtain sufficient food. These activities of the small animals facilitated the spread of A. inebrians, which can reach a height of 60-150 cm that is close to the height of livestock and produce about 700 small easily-shed seeds from each compact inflorescence 9 . The seed also has long awns that can adhere to livestock which may aid dispersal.
Achnatherum inebrians plants alone will not provide a stable ecosystem and this study has shown that within and between these large plants, members of the grasslands plant community also become established and have mature seeds. For a period of time after the onset of overgrazing there will remain a viable seed-bank in the soil and this may be the source of many of the new plants. However, continuous overgrazing will reduce the density and richness of forage seeds in soil [25][26][27] . Plants within and between the protective A. inebrians tussocks were shown in this study to produce more abundant seed compared with overgrazed areas. Some of the seeds may fall on the ground and so enrich the local seed-bank. In addition, some seed may move with the wind while other seed may be consumed by birds and thus get long distance dissemination 28 . These seeds will regenerate the broad-species seed-bank essential for maintaining a long-term stable ecosystem. Arthropod diversity and abundance are also enhanced by the presence of A. inebrians. Part of the reason for the increased presence of arthropods is likely to be due to the open soil structure with high plant residues and organic matter content within A. inebrians protected pasture. In contrast, soil is generally very compacted in degraded pasture. In addition, the abundance of leaves will provide a food source for herbivorous arthropods 29 . Further, the high numbers of inflorescences provides pollen that is also a food source for some arthropods 30 and so further increases arthropod biodiversity. The abundant arthropods and seeds will also become a food source of birds 28,31 .
The grasslands sites studied in the research ranged from a very dry area (Alxa) dominated by a C 4 grass, to a wet area (Guinan) dominated by a sedge species. Achnatherum inebrians had become naturally established in this diverse range of habitats. This ability to colonise under such varied conditions makes this grass an effective agent for the regeneration of diverse degraded grasslands.
A further example of ecological impacts from another toxin-producing Epichloë species (E. coenophialia) that infects tall fescue (Lolium arundinaceum) 32 has shown different effects on biodiversity. When endophyte-infected tall fescue was introduced to a pasture ecosystem by densely planting, the density and diversity of other plant species  Table 2. Comparison of number of arthropods in TA and NA. Notes: P t indicates difference of total abundance of all arthropods families in TA and NA in the three sites; " * " means P < 0.05; " ** " means P < 0.01; " *** " means P < 0.001; TA means areas with tussock of A. inebrians; NA means overgrazed areas without A. inebrians.
Scientific RepoRts | 5:18527 | DOI: 10.1038/srep18527 was reduced compared with tall fescue that is endophyte-free 33 . Endophyte-infected tall fescue also significantly reduced diversity of arthropods compared with endophyte-free plants 34,35 . However, one study suggested that there was no difference on effect on biodiversity between endophyte-infected and endophyte-free of tall fescue, but environmental condition played more important role 36 . Further research has also shown that compared with endophyte-free, endophyte-infected Achnatherum robustum and Festuca arizonica can increase biodiversity [37][38][39] , and endophyte-infected L. multiflorum strengthened interactions of resource-consumers 40 . All of above studies compared different ecological roles between endophyte-infected and endophyte-free grass, but ignored the relationship between grass-endophyte symbiosis and livestock. A common practice to recover degenerated grassland is to use wire fences to prevent grazing livestock from exacerbating grassland degradation. However, wire fences also hindered the migration of wild animals 41,42 . In contrast, our research has demonstrated that A. inebrians can function as a natural fence which allows free movement of animals whilst naturally deterring grazing from damaging livestock. As such A. inebrians can be very helpful in the recovery of grassland.
In conclusion, A. inebrians can play a positive role in the stabilization and restoration of degenerated grasslands in northwest China. In addition, the effects of burrowing animals, implicated by many as a causal factor for grasslands degradation 6 , may in fact be beneficial in the restoration of pastures by providing suitable bare land for A. inebrians to establish.

Methods
Site descriptions. We conducted this study at three grasslands in northwest China that have been overgrazed by livestock in recent decades, and where Achnatherum inebrians plants have become increasingly more abundant. Xiahe county of Gansu province (35°07′ N, 102° 26′ E), Guinan county of Qinghai province (35° 28′ N, 101° 16′ E) and Alxa Left Banner county, Inner Mongolia (38° 39′ N, 105° 46′ E). In Xiahe, the dominant grass is Elymus nutans; Guinan, a wetland dominated by Kobresia capillifolia; Alxa, an arid grassland where Pennisetum centrasiaticum, a C 4 grass, is the dominant species. Details of the three sites, including temperature and precipitation data, were presented in Supplementary Table S

Impact of A. inebrians on seed status of other plant species. In order to determine the effects of A.
inebrians on other plant species in the three sites, we employed two treatments in overgrazed grassland of each site: one area with tussocks of A. inebrians in the quadrats (5 m × 5 m), termed tussock area (hereafter TA) and the other one without A. inebrians, termed non-A. inebrians area (hereafter NA) that was still covered by short vegetation, and each treatment had four replications. The seed status of plants in each quadrat was recorded. We only investigated plants that had mature seeds. The inflorescence number for Poaceae and Cyperaceae species, because of relatively uniform inflorescences, and the plant number of plant species in other families with seeds were counted by visual observation in every quadrat in situ. Difference of vegetation between TA and NA can be seen in Fig. 2a.

Impact of A. inebrians on arthropods.
In each overgrazed grassland in Xiahe, Guinan and Alxa, 30 soil blocks (0.4 m in diameter and 0.2 m in depth, 15 each for the TA and NA quadrats) were randomly dug out and crumbled. All arthropods except for ants in one block were put in one collection bottle. The arthropod collections were brought to the laboratory and then the specimens were identified to the family level and counted. Contribution of zokor and pika to the spread of A. inebrians. A seed establishment experiment was conducted in grassland in Xiahe. Seeds of A. inebrians collected from Xiahe grassland in September 2012 were used in this study. Fifteen blocks were randomized in the overgrazed grassland, with each block having two small circular plots approximately 15 cm in diameter. One plot was on a zokor molehill and the other one was in adjacent overgrazed area (NA) that was still covered by short vegetation. These small plots were away from each other by at least 1 m. The sowing time was in April 2013, and in each plot 100 seeds were sown. The establishment rate of A. inebrians in each of the small plots was assessed in September 2013. Given the presence of A. inebrians in previously bare area produced by pikas, the ability of A. inebrians naturally colonising in these bare patches was investigated. This study was conducted in the grassland trial site of Guinan county. Only those A. inebrians plants that had established within the past year (less than one-year old plants), were investigated in this study to calculate the increase of plant number of A. inebrians in one year. Some areas covered by A. inebrians were randomly selected to compare the establishment ability of A. inebrians in bare patches and overgrazed grassland areas (NA) that were still covered by short vegetation. Those areas containing A. inebrians were not only close to bare patches but also close to NA areas. For each treatment, four quadrats (10 m × 10 m) were randomly selected in the overgrazed grassland. Calculation of the mean number of one-year old A. inebrians plants per m 2 was carried out in every quadrat. The establishment ability of A. inebrians was compared by the mean number per m 2 in the two treatments. The difference in establishment between bare land (molehill and patch) and NA can be seen in Fig. 2b-d. Data analysis. Data analysis was performed with SPSS 17.0 for Windows (USA). A nonparametric, Mann-Whitney U test was used to test: 1) the impact of A. inebrians on seed status of different plants species in two treatments; 2) the impact of A. inebrians on number of different arthropods in two treatments. Independent samples T-test was used to test: 3) the establishment rates of A. inebrians in molehill produced by zokor and NA; 4) the number of one-year old A. inebrians per m 2 in bare patch produced by pika and NA.