Plants contribute to the spatial structure of urban systems not only through their presence in parks and reserves, but also via their occurrence throughout the entire urban mosaic. Understanding today’s urban biodiversity will help us to predict future biodiversity and the ecological consequences of modern biodiversity changes (Cadotte et al., 2008). Urban green space has become one of the most extensively-researched topics in urban ecology because of the environmental, social and economic benefits of green space (Peckham et al., 2013). Urban vegetation acts as an effective noise and dust barrier, removes large quantities of atmospheric carbon dioxide, saves energy, and mitigates the Urban Heat Island (UHI) effect (i.e. when a metropolitan area is significantly warmer than its surrounding areas); in addition, urban vegetation has aesthetic and cultural values (McPherson et al., 1997; Tyrväinen et al., 2005). Nowadays, because the urban environment in developing countries is deteriorating (WHO, 2014), people living there frequently look toward urban green space for improving their living environment. For example, people living in Beijing often complain about air pollution, which has become especially serious in recent times (Fig. 1A,1B; Wong, 2014). Increasing the number of green areas in Beijing as well as improving their quality may help to alleviate the problems derived from air pollution and other environmental risks to health. Recently some proposals, such as the greening of Beijing rooftops (there are ca. 100 million square metres of roof space suitable for setting up green roofs; Zhang, 2013) and the establishment of green “air corridors” (Wu & Zheng, 2014) have been put on the table.

Since the mid-2000s, we have explored the urban vegetation and plant diversity changes in different urban structural units (e.g. universities or similar work units) in order to promote urban vegetation ecological service functions. Taking Beijing as a case study (Fig. 2), we have surveyed urban vegetation/plant diversity at two different levels: firstly, we explored the effects of urbanization on urban green space/plant diversity at the landscape level (greening percentage, fragmentation degree; Fig. 3); secondly, we examined these effects at the plant species level (structure, composition, and origin) and their response mechanism (Fig. 4). Our final aim is to achieve a better incorporation of plant diversity and ecosystem services in Beijing’s future urban planning and design.

Beijing has been the capital city of China for 800 years and has a three-millennium history as a human settlement. An eminently rural city until the middle of the 20th century (Haw, 2007), it has recently experienced an impressive demographic and economic growth. Today, Beijing is one of the largest cities in the world, with an estimated population of ca. 21 million inhabitants in the whole Municipality of Beijing, a number that represents a nearly sevenfold increase since the 1950s. The Municipality of Beijing occupies 16,807.8 km², and the central urban area confined within the fifth ring road (i.e. the region that can be considered the “downtown”; Fig. 2) is 652.2 km². In our studies, remote sensing images (Satellite Pour l’Observation de la Terre, SPOT-5, in 2002) of the fifth ring of Beijing were divided into 264 2 × 2 km grids (Fig. 2); 1–4 urban structural units (USUs, i.e. universities or similar work units) were selected in each grid based on uniformity and importance. Properties of an USU were defined by the designations in Gill et al. (2008) as well as by the Urban Forest Effects (UFORE) model. Each unit boundary was delineated using remote sensing images, and land cover was divided into impermeable ground (built-up land), forest land, grassland, mixed forest and grassland, agricultural land, and water bodies (water-covered areas). During our field work, we investigated at least one sampling plot in each USU with stratified sampling. In each sampling plot, we recorded height, diameter at breast height (DBH), and origin for the tree species; height, cover, and origin for the shrubs; and cover, abundance, and origin for the herbs. We also included other biophysical information (e.g. longitude, latitude, and distance to urban center).

At the landscape level, we have identified a total of 453 USUs belonging to 12 primary USU types and to 38 different secondary USU types (see Wang et al., 2013, for a detailed explanation on the types of USUs). The percentage of built-up, green, and water-covered areas varied across the different urban structural units (of the 38 secondary USU types, only 22 were used for our analyses; Fig. 3). The percentage of built-up area was highest in those regions with hotels (84.9%), and lowest in areas with primary and middle schools (16.4%). The percentage of woodland area was highest in primary and middle schools (69.4%), and lowest in entertainment plazas (7.2%). The percentage of grassland area was highest in farmland or orchards (22.7%), and lowest in siheyuan (0.4%; siheyuan are Beijing’s historical courtyards; Fig. 1C). The percentage of mixed woodland and grassland was highest in public squares (16.1%), and lowest again in siheyuan (0.05%). The percentage of water-covered areas was highest in areas containing government properties (23.1%), although water-covered areas were totally absent in several types of urban structural units (Fig. 1). The percentage of agricultural land was highest in business/enterprise zones (6.1%), and totally absent in several types of urban structural units (Fig. 3). Finally, the percentage of green space (considering “green space” as the sum of woodland, grassland, and mixed woodland–grassland) was highest in primary and middle schools (82.8%), and lowest in areas with hotels (15.1%). There was no significant correlation between the percentage of green space and construction period in colleges/universities, parks and high-density residential areas (the USU types for which we had enough sample sizes to carry out statistical analysis). However, there was an “inverted-U curve” relationship between the housing price and green space percentage in residential areas, which can be described by the equation: f = 17736.45 + 348.21x – 4.15x ², p = 0.0022 < 0.05. This relationship implies that socio-economic factors like housing prices may have a role in determining the pattern of urban ecosystems in Beijing (Wang et al., 2013).

At the plant species level, within the fifth ring road we found 551 plant species of 313 genera and 103 families, of which 118 were trees, 99 shrubs, and 296 herbs. Nearly half (48.3%) of the total plant species were aliens, and these predominated in the shrub/tree layers (68.4%) but not in the herb layer (31.1%). Only 7.7 and 1.9% of the plant genera found in Beijing urban areas were endemic to East Asia and to China, respectively. Species richness and density were positively correlated both for tree/shrub and herb taxa. This correlation may indicate that, although some species predominate in our study area (such as some tree species widely planted along the city streets, e.g. Populus tomentosa Carrière var. tomentosa, Magnolia denudata Desr., and Juniperus chinensis L.), there is an important array of species in relation to their densities. Most of the socio-economic variables studied were related to at least one of the four plant diversity variables (herb richness, herb density, tree/shrub richness, and tree/shrub density). Land use showed a significant relationship in all four cases, which generally had lower values in cultural and educational institute-dominated areas (areas that in Beijing are generally characterized by large extensions of urbanized land). Notably, tree/shrub density values were significantly higher in residential areas, which is at least partly due to traditional beliefs: for example, some trees such as Juniperus chinensis, Pinus tabuliformis Carrière var. tabuliformis, and Platycladus orientalis (L.) Franco are commonly planted in siheyuan areas because of their traditional association with longevity (Profous, 1992). The year of a given area’s establishment was also largely related to plant diversity, with higher diversity indices in more recently developed areas. This result was expected given the dynamics of the urban development of Beijing during the last 60 years: intense “urban sprawling” (peaking during the 1980 and the 1990s), followed by more environmentally-friendly urbanization practices during the 2000s related to the greening of the city for the 2008 Olympic Games (Fig. 1D; Wang et al., 2012).

Also at the plant species level, we have made a detailed study of the patterns of biological invasions in Beijing, which was carried out not only for the city center but also throughout the entire Municipality. We compiled a comprehensive list of naturalized and invasive species and their associate traits (distribution, habit, life form, habitat, geographic origin, and mode of introduction). One hundred and twelve naturalized (including 48 invasive) plants were identified within the Beijing Municipality. Although these numbers can be regarded as “modest” compared to other large Asian cities such as Shanghai, Hong Kong, or Singapore (see Table 3 in Wang et al., 2011), they represent a significant fraction of the alien species richness reported for the entire country: namely, up to one-quarter of the total naturalized flora of China (420 to 861 species; Wu et al., 2010; Jiang et al., 2011) and one-fifth of the invasive flora of China (270 to 488 plant species; Weber et al., 2008; Xu et al., 2012). Moreover, finding 112 naturalized species represents a ca. 23% increase from the previous report of Liu et al. (2002), suggesting that biological invasions in Beijing have increased over the last decade. Some of the invasive species observed by us are among the most noxious invaders in China, such as Alternanthera philoxeroides (Mart.) Griseb., Ambrosia trifida L., Lolium temulentum L., Solidago canadensis L., and Sorghum halepense (L.) Pers. Most of the naturalized and invasive plants belonged to six families (Asteraceae, Poaceae, Amaranthaceae, Euphorbiaceae, Brassicaceae, and Fabaceae) and are annual herbs that preferentially grow in disturbed sites (e.g. roadsides and waste places). North and South America were the main sources of the naturalized and invasive flora of Beijing (with almost half of the species having this origin; Fig. 1G). Beijing districts that have recently experienced the highest human population growth, the greatest urban expansion (to cope with the huge demand for new housing, as the inner districts have been completely urbanized since the 1960s), and the largest economic growth were also those with the highest number of naturalized and invasive species. Managed green spaces in urban districts (i.e. street trees, lawns, urban and residential gardens, and parks) would have a major role for the growing number of naturalized plants in Beijing. Green areas substantially increased in number during recent years to prepare the city to host the 2008 Olympic Games (during the period 2001–2007, over 700 spaces were built in the downtown—an increase of nearly 9000 ha; UNEP, 2009).

After a decade of uninterrupted study of Beijing’s urban flora, we have two main concerns about China’s future urban green space and plant diversity research. First, as a subdivision of modern ecology, urban ecology is also focused on the relationships between biological organisms and their surrounding environment. In urban ecosystems, the main biological organisms are the same as those in other settings, with the notable exception of humans, who obviously play a very important role in the urban ecosystem. Conversely, the environment in the urban ecosystem is comparatively more fragmented and heterogeneous than other ecosystems such as agricultural or grassland ecosystems. One of the main obstacles that urban studies typically face is the limited access to data for some of main driving variables, especially those related to socio-economic features of urban areas (e.g. household income, race, education level, population density, etc.). In contrast, physical variables are usually readily determined, as these are typically collected within the ecosystem (e.g. rainfall, temperature, and soil properties). In American urban ecology studies, researchers can get high resolution socio-economic variables from the National Aeronautics and Space Administration (NASA) (Fink et al., 2003) or companies such as Nielsen and its PRIZM geo-demographic segments for the United States (Grove et al., 2006). However, in Chinese cities, researchers cannot access high resolution socio-economic data at present, due to a series of shortcomings on the frequency of censuses (every ten years at best), on the quality of statistics collected by government agencies, and on the reliability of official statistics. For instance, we can acquire basic data, such as information about the population of each district in Beijing from the Beijing Statistic Yearbook series, but we cannot get more useful statistics at a finer scale; e.g. there is no way to know how many people are living in a given community, how many of them are foreigners, how old they are, what their education level is, what their income is, and so on. These factors, as demonstrated elsewhere (reviewed in Table 2 of Wang et al., 2012) can affect urban vegetation/plant structure, composition and origin, among other variables. Therefore, China’s future urban ecology studies still have a long way to go to understand how human (socio-economic) drivers have molded biodiversity patterns in urban areas.

Second, urban ecology is a relatively new discipline in China. In 1978, China was a rural country (only 20% of the population lived in cities in that time); however, between 1978 and 2012 China’s urban population increased by more than a half billion people (today, the percentage of urbanization has reached 52%), and the urban population is expected to increase by about 250 million more over the next two decades (World Bank, 2014). Thus, field data are badly needed in order to reveal the patterns of urban ecology in China as well as the drivers that give rise to those patterns. Urban ecologists usually do “plant sampling” in cities in the same manner as forest ecologists do “vegetation sampling” in forests. However, the heterogeneity of the Chinese urban matrix is very high; in Beijing, for example, very ancient one-story siheyuan areas from the Ming Dynasty coexist with late-19^th century districts and some of the most modern high-rise buildings in the world; Fig. 1H). As a result, it is challenging to find enough 20 × 20 m sampling sites in urban settings. In our opinion, the position and size of plots in urban plant sampling should take the land use features into consideration, and urban ecologists should adopt flexible standards for sampling comprehensively and representatively. Moreover, urban ecology also needs long-term, continuous monitoring data to support its hypotheses and this, again, is of particularly significance in China, whose cities evolve so quickly.