Earth's rich botanical diversity now exists in the Anthropocene. Human influence in the biosphere is so prevalent that there is hardly any pristine wilderness left on Earth. Human-induced increases in atmospheric CO2 concentrations and the linked climate changes affect all localities. Other anthropogenic ecological changes such as defaunation and introductions of non-native species are also pervasive. As a result, the numbers of a large proportion of plant species are declining, and many risk extinction (Willis et al., 2017). These pressures are likely to increase in strength, and we risk losing up to 50% of all vascular plant species during this century (Pimm and Raven, 2017). Thus, there is a strong need to improve our understanding of human-influenced vegetation in terms of not only risks to Earth's plant diversity, but also in terms of its biodiversity capacity (i.e., its ability to harbor biodiversity, notably in terms of species) and how to maximize it. In the words of environmental writer Emma Marris (2011, pp. 2–3), “We must temper our romantic notion of untrammeled wilderness and find room next to it for the more nuanced notion of a global, half-wild rambunctious garden, tended by us… Rambunctious gardening is proactive and optimistic; it creates more and more nature as it goes, rather than just building walls around the nature we have left.” With this essay, I point to some areas where research can contribute to proactive conservation and restoration of botanical diversity in the Anthropocene. I note that I here use restoration in its future-oriented sense (Choi, 2007)—with a focus on establishing ecosystems that are sustainable in current and future environments and accepting moving, somewhat unpredictable endpoints. Novel ecosystems are a key issue here. They are defined by novelty (new species combinations relative to a historical base-line) and human agency (their novel state should be a result of human action, but not dependent on continued human intervention for its maintenance) (Hobbs et al., 2006). While the concept is much debated, there is no doubt that large and increasing parts of Earth's natural and semi-natural areas harbor novel ecosystems (e.g., Fig. 1). These, in large part arise from the spread of non-native species, but may also result from changes to climate (Ordonez et al., 2016) or other aspects of the abiotic environment, with synergistic effects between the two, e.g., spread of introduced evergreen broadleaved species in response to climatic warming in some temperate areas (Walther et al., 2007). The conventional approach to novel ecosystems is to remove the non-natives or, if this approach is unsuccessful or impossible, to view these systems as of reduced value. However, under realistic scenarios, novel ecosystems will become increasingly prevalent given growing globalization and climate change. Importantly, removal approaches to keep a certain status quo species composition will become increasingly unfeasible in many areas. Further, research suggests that plant communities are often not saturated with species and could harbor more species than they do, meaning that non-native species need not by definition be a risk to native species (e.g., Stohlgren et al., 2008). Although the literature on perceived invasive plants is rich, it tends to focus on problem cases, and we still do not have a good general understanding of the biodiversity capacity of novel ecosystems. In some cases, native species are indeed suppressed or excluded, but in other cases, the outcome is rich mixes of native and non-native species, including both plants and associated animals, as seen in the case of Puerto Rico and other Caribbean islands, where widespread species-rich secondary forests comprising native and introduced species have developed (Lugo, 2004; Lugo et al., 2012). Further, dominance by non-natives may vary across the landscape and with time, e.g., potentially with biotically mediated declines with time after invasion through accumulation of enemies, plant–soil feedbacks, and evolutionary responses in native species (Flory and D'Antonio, 2015), so local problem cases are not necessarily instructive. We need to develop a broad research agenda aimed at understanding and enhancing the diversity capacity of novel ecosystems, in terms of plants, but also other organisms. I emphasize that the above argument is not meant to suggest that native species and ecosystems dominated by natives should not remain a key focus for restoration and conservation or that non-native species are not sometimes a serious threat to native biodiversity (even if boosting small-scale species richness). Instead, it is a call for not automatically assuming the latter and to look at non-native species and novel ecosystems in a more nuanced manner. Climate change is another key issue for Anthropocene botanical research. Plants are already affected by anthropogenic climate change, e.g., with strong upslope range shifts across the world (e.g., Morueta-Holme et al., 2015), leading to accelerating community changes on mountain tops (Steinbauer et al., 2018), but also diebacks of keystone plant species in some areas (Breshears et al., 2005). However, we still do not have a solid general understanding of to what extent current vegetation dynamics are shaped by climate change. Importantly, the dynamics are clearly complex and involve disequilibrium dynamics with strong time lags on immigration, local population build-up (after initial immigration), and extinction (Svenning and Sandel, 2013). The emergence of novel climatic conditions makes it even more difficult to predict biodiversity outcomes, e.g., what will happen in warming tropical lowlands (Colwell et al., 2008). Climate-driven dynamics are likely to become much stronger in the near future and could become a strong risk to plant diversity (Zhang et al., 2017). Importantly, we have to expect ongoing human-driven climate change for centuries or more, meaning that we will need to tackle continually changing conditions. Hence, we need to improve our understanding not just of how plant communities will react, but also how to manage and restore (in a forward-looking manner) natural areas to decrease diversity losses and exploit any potential for diversity benefits. Some of the key issues to address include (1) improving our understanding of the characteristics and dynamics of plant diversity and vegetation functioning under strong climate change and in novel climates, (2) effective use of assisted migration to minimize dispersal limitation (Gallagher et al., 2015), (3) handling and reducing time lags in the build-up of functional communities and even individuals (e.g., time needed for development of mature or senescent trees), (4) the role of non-natives in the assembly of diverse plant communities in a changed climate, as well as the role of these novel occurrences in to safeguarding plant species at risk from climate change in their native area, (5) maximizing delays in climate-driven extinction, in integration with the adaptive approaches 2–4. A third point that I want to emphasize is the need for developing the botanical foundations for rewilding, here defined as ecological restoration to promote self-regulating, biodiverse ecosystems. I see rewilding as a highly important idea for developing proactive approaches to not just safeguard, but also promote biodiversity in the Anthropocene. Our diversity of plants and other organisms has evolved in wide expanses of nature, but a large proportion of species have experienced strong human-driven reductions in available habitat, i.e., comparing habitat availability at evolutionary time scales to the present. This is true in a relative sense even for the many species that have small natural distributions. Considering this and the fundamental importance of area and connectivity for biodiversity, in the long-term, focusing on isolated protected areas as safe havens of most species will not be sustainable for Earth's biodiversity. We will need to be much more ambitious and work toward having wide areas available for nature. As human populations are increasing, such wide nature areas will have to be in the context of a human-occupied world. Human disturbance of many kinds will be widely occurring, new nature areas will often be located in areas with a history of intense land-use or closely intermixed with intensely used areas, and, on top of this, all nature areas will be continually affected by climate change. Considering competition for economic resources, it seems infeasible to base broad-scale ecological restoration on intense investments or ongoing human management. Given this instability and resource competition, to ensure durable ecosystems, it would seem wise to focus our “rambunctious gardening” on approaches favoring biodiversity with little or no sustained management (cf. Dudley, 2011). Rewilding, with its active design toward self-management, clearly is central here and also contributes to social wishes and needs for self-willed nature. Approaches involving facilitation of spontaneous social activities with biodiversity benefits could also contribute. Given the importance of large vertebrates (megafauna), and in particular large herbivores, in shaping ecosystems, notably via top-down trophic processes, these have been a key focus in rewilding, leading to the definition of trophic rewilding, as species introductions to restore top-down trophic interactions and associated trophic cascades to promote self-regulating, biodiverse ecosystems (Svenning et al., 2016). Key effects of such large-vertebrate re-establishment include promoting heterogeneity in vegetation structure and abiotic conditions (e.g., soil factors), carbon pool diversification (e.g., production of dung and carcasses), as well as their roles as propagule dispersers, with likely positive effects on the local and landscape capacity of ecosystems for biodiversity, including for plants (Fig. 2). Rewilding via re- establishment of megafauna has been controversial, notably in terms of species and historical baselines. Noting that an evolutionary time-scale baseline seems most meaningful in terms of long-term, broad-scale biodiversity maintenance, implementation would always have to consider its Anthropocene context, notably societal acceptance and human–wildlife conflicts. Rewilding in the broader sense goes beyond trophic restoration and also includes, e.g., abiotic factors such as self-sustaining fire and hydrological dynamics, which may also contribute to promoting self-managing, biodiverse Anthropocene ecosystems. However, there is a strong need for research on rewilding, as empirical studies remains scarce (Svenning et al., 2016). Notably, there is a dearth of knowledge on rewilding impacts on plant diversity and on how to design implementation to maximize diversity results. Further key issues include how rewilding can be applied to improve biodiversity outcomes in contexts such as novel ecosystems, spontaneous reforestation in response to land abandonment, and climate-change-driven ecosystem dynamics, and, from a societal perspective, how rewilding can be implemented in ways that have long-term socioecological sustainability in a human-dominated world. The coming decades will offer huge challenges for safeguarding Earth's plant diversity. We will need to make best use of all possibilities to reduce negative impacts and promote the development of approaches that favor plant diversity. A key priority is to protect remaining natural areas and the critically endangered species that we are in immediate risk of losing. However, we also need to be much more ambitious, expanding and improving space for nature and biodiversity within intensely used and occupied regions, even in small urban green spaces. Such an ambitious approach will be needed to avoid strong diversity losses for plants as well as other organism groups, but will also benefit people's quality of life, by protecting or enhancing local nature and biodiversity. As a further challenge, we will have to deal with changing abiotic and biotic conditions everywhere. I see improving the scientific basis for the nuanced management of novel ecosystems, adaptive responses to climate change, and the restoration of self-regulating, biodiverse ecosystems as important components toward promoting a “rambunctious garden” rich in plant species and other organisms for the Anthropocene. I thank Editor-in-Chief Pamela Diggle and two anonymous reviewers for their helpful comments. This work is a contribution to my VILLUM Investigator project “Biodiversity Dynamics in a Changing World” funded by VILLUM FONDEN (grant 16549), my Carlsberg Foundation Semper Ardens project MegaPast2Future (grant CF16-0005), and the TREECHANGE project funded by the Danish Council for Independent Research | Natural Sciences (grant 6108-00078B).