Ornamental grasses are ubiquitous in landscape design in American, European, African countries, and also some countries in Asia. More than 100 grass species were advocated as garden plants in Karl Foerster’s book published in 1957, while Greenlee (1992) detailed 250 ornamental grasses suitable for horticultural application; Glen (2002) compiled 125 species for garden applications in southern Africa and Cullen et al. (2011) listed 187 species of Gramineae for European gardens. This popularity has been reflected by developments in the horticultural industry yielding increasing numbers of grass and grass-like plants cultivated as ornamentals (Darke and Griffiths 1994). Due to their high degree of novelty and attractive aesthetic value, are much favored ornamentals by Chinese gardening designers (Gao et al. 2008; Wu et al. 2008; Zhang and Xia 2008). Consistently, the majority of the introduced ornamental grasses originated from America and Europe, and regarded as alien with a potential risk of ecological harm developing to become as invasive species in China. The intensive introduction of alien species, chosen for aesthetic values, has caused changes of urban greening choice of plant species composition, as reported in Beijing (Zhao et al. 2010). Thus, an appropriate system of invasive risk assessment, adapted to the use and features of ornamental grasses, is well needed for China. Such a system would have value for other countries or regions with similar introductions and resource exchange, especially in an era of increased globalization and economic integration.
Results presented herein demonstrate that some or majority of ornamental grasses studied present high risk of invasion based on both of the risk assessment protocols used. About half of the ornamental grasses (48.8%) were considered high and moderate risk under WG-WRA, while assessment using the WRA protocol scored 91.5% of the studied grasses to pose invasion risk following a diverse criterion. According to the related study, the accuracy for WG-WRA was c. 65% in central Europe (Weber and Gut 2004); but for WRA, it was reported that 94% of the invasive species were considered to be rejected for their risk in the Mediterranean (Gasso et al. 2010), and it also correctly rejected all major and 86% of minor weeds in Canada (McClay et al. 2010) and a similar result in China that it correctly rejected 84% of major weeds and did not wrongly accept a major weed (He et al. 2018). Thus, WRA risk assessment protocol is more acceptable to predict invasion risk for its assessment routine compared with WG-WRA. Besides, results showed more native grass species presented high and moderate risk (81.1%) than did alien species (22.2%) under WG-WRA in the study, which is not that coincident with the common sense. Moreover, there was a relatively weaker relationship between our field observations and the risk assessment for WG-WRA compared with WRA on aspect of both biological and ecological characteristics and ornamental value as showed in the results (Figs. 2 & 3).In conclusion, it suggested that WRA is a more appropriate risk protocol for ornamental grasses, in agreement with He et al., (2018) who investigated alien plant risks in China. That is say, most of the studied ornamental grasses present high risk of invasion in view of the routine of WRA.
Based on field observations, most of the 82 ornamental grasses studied did not pose a much serious invasion risk at current time. Yet, this result was not well consistent with our assumptions on grasses’ invasiveness, as we considered ornamental grasses to have high invasion risk based on their wide adaptability and strong fecundity, with effective establishment and dispersal properties. As reported, many gramineous species are considered serious weeds or invaders (Hammer and Khoshbakht 2015). Whilst more than four-fifths (84.1%) of the plants referred to above were classified as Gramineae compared with 15.9% for other families to the result in the study, which have a trend to cause risk and become invasive. There are documented instances of invasion by ornamental grasses into native vegetation, causing harmful effect to local ecosystems. For example, crimson fountain grass (P. setaceum), has become widely naturalized or invasive in Namibia (Joubert and Cunningham 2002), South Africa (Rahlao et al. 2010) and Hawaii (Goergen and Daehler 2002; Adkins et al. 2011). Miscanthus, originating from Asia, has been reported to spread from planted ornamental settings into natural areas in the US and Australia, with a potential invasive risk in several European countries (Meyer 2004; Jørgensen 2010; Meyer et al. 2010; Quinn et al. 2012). However, in northern China this risk may not be serious currently, as ornamental grasses have been used in greening projects only since 2000 (Chen et al. 2015b). For a newly-introduced plant, it may need some time to achieve its full strength. As reported, introduced alien plants may need 50 (Wan et al. 2009) or even 100 years (Low 1997) to become invasive. Anyhow, thing may be quite different when taking time course into account. Research on fountain grass (P. alopecuroides) showed that it was reported to pose a relative low risk of spreading based its seed germination (Lv et al. 2007). Yet, fountain grass was considered to present high risk under both of the two risk assessment protocols and was also found to establish its community in an adjacent habitat in environmental-restoration regions. Anyhow, this is not a reason for complacency as the accelerated development of the horticulture industry and the passage of time may lead to a cumulative increase in invasion risk for certain grasses.
Given that ornamental grasses have been used for only two decades in China, this highlights potential limitations of the assessment systems used. For example, the low invasion potential of purple love grass (E. spectabilis) under the WG-WRA may underestimate of its true potential as it was estimated as high risk based on WRA protocols; for the underestimation of its huge amounts of seeds production (> 10000 seeds per plant) and the time elapsed since its introduction affected its risk score, which was similar to another two outliers, little bluestem (A. scoparius) and Mexican feather grass (S. tenuissima). Anyhow, there are some interesting findings on the two protocols. Results showed the correlation of the two risk assessment protocols was much higher for species than that of well-cultivated varieties; and what’s more for these cultivars, they normally attained much higher scores under WG-WRA than WRA. They both indicated there were some neglection or imperfection of the assessment systems, especially the risk assessment on selected breed. According to our study, there was also a relatively weak relationship between our field observations and the risk assessment, though they were well-related on biological and ecological characteristics, it overestimated the invasion risk of cultivars by details of fecundity and adaptability for the ornamental grasses. Thus, the two risk assessment protocols may have some limitations which need improvement on predicting the risk of well-cultivated varieties (of ornamental grasses) compared with species. The risk protocols should be well developed for local systems and updated data, also further research is needed for both trial plot survey and a larger area of application for their adaptability. Nevertheless, when taken the comparability of the two sequences of scores into account, there was a positive correlation between scores derived from the two risk protocols, though there were some outliers like purple love grass as mentioned above. Especially for the ornamental plants with high risk, the correlation was much positively related. That suggests both of the two risk assessment protocols are still available for these plants with high risk.
On the other hand, as inherent peculiarities of a species, biological and ecological characteristics are well related to invasive risk. In details, fecundity and adaptability consistently contributed to high level of risk in the study. Plants with large amount of seeds or having a strong ability of asexual reproduction all had a much high predicted risk of invasion under both of the two risk assessment protocols, e.g. common cord-grass (Spartina anglica), foxtail barley (Hordeum jubatum), reed canary grass (Phalaris arundinacea), fountain grass (Pennisetum alopecuroides), Amur silvergrass (Miscanthus sacchariflorus), common reed (Phragmites australis) and bulrush (Typha orientalis). In addition of the scored risk, there were also some evidences they could establish and spread themselves to a nearby habitat on field surveys, though it was reported that some ornamental grasses had low invasion risk based on the analysis of their ecological adaptability and reproductive characteristics (Lv et al. 2007; Zhao et al. 2012).Anyhow, some researches did pose the risk of some of these grasses, like fountain grass and foxtail barley (Chen et al. 2016, 2017b) and common cord-grass (An et al. 2020). Thus, biological and ecological characteristics should not be omitted at any risk assessment of the routine. On the other hand, ornamental value is an important determinant of also of risk the economic success of a grass as a horticultural commodity. Aesthetic features such as variegated leaves (Darke and Griffiths 1994) and, growth and inflorescence form (Hammer and Khoshbakht 2015) are favoured by garden designers and the public. But high aesthetic value promotes translocation and increases the risk of invasion. Here, however, ornamental value had only a weak positive correlation with invasion risk. Perhaps this is due to a selection effect enacted by grass breeders, whereby undesirable traits related to invasion were eliminated by breeders during the process of cultivation, e.g. cultivars of Miscanthus sinensis, Pennisetum alopecuroides and Panicum virgatum, they all have much lower risk compared to their protospecies in the study. This selection may also explain higher risk ecological characteristics determined for native grasses with short time of cultivation and breeding. Alien ornamental grasses had mostly undergone extensive domestication, whereas cultivation time was short for the native grasses included in this study. All of this contributed to the inaccuracy of the risk assessment protocols and required further studies under a fast developing horticultural industry.
China has a very broad range of ecotones and five climate regions ranging from tropical to cold temperate zone. Thus, invasion risk for southern provinces or cities (Gao et al. 2008) may be quite different from that determined here for northern China, where limited overwintering may present a check on invasive activity. Previous research has shown that invasion risk in China varies with latitude and longitude, decreasing to the north and the west; and risks were greater in economically developed regions due to higher frequency of exchange (Wan et al. 2009; Chen et al. 2017a). Based on field observations, most of the 82 grasses studied did not pose a much serious invasion risk at current time. However, previous reports have indicated some grass species, including Natalgrass (Rhynchelytrum repens) (Ding and Chen 2016) and vetiver (Vetiveria zizanioides) (Li et al. 2014), to present a much higher risk in southern China but relatively lower risk in northern China. By extension, species with lower seed yields and weaker vegetative propagation, especially well-cultivated grasses, may be more suitable for horticultural applications and resource exchange. Even for some native grass species, e.g. hardy sugar cane (S. arundinaceum) and Acorus gramineus var. soland, they also present good candidates for ornamental grass breeding, given that they are well adapted to the local environmental conditions and the consequences of spread may be less severe.