Biogeographers and conservation biologists have long been interested in identifying and characterizing geographic regions containing a higher concentration of biodiversity and derived natural resources than surrounding areas, ranging from within- and among-species diversity through to ecosystem services [1
], at different spatial, temporal and taxonomic scales [4
]. Centres (also known as hotspots) and peripheries (coldspots) of plant diversity have been shown to be unevenly distributed and to play a fundamental role in shaping ecosystems and delivering associated benefits to humans and other species [5
]. Mapping efforts contribute to a better fundamental understanding of both biodiversity (e.g., species extinction, diversification and co-existence) [7
], and the interaction between people and nature, including the resulting socio-economic benefits and threats [8
]. This is a particularly urgent endeavour in the current context of a rapidly growing global human population with increasing consumer expectations, posing a serious threat for both plant diversity and its long-term contributions to people [12
]. Factors, such as land use and climate change, pollution, direct exploitation of species, and biological invasions have direct and indirect impacts on plant diversity [13
], potentially undermining current and unrealised plant-based adaptive solutions [15
] and traditional knowledge associated with plant uses [17
]. Documenting the distribution of plant diversity and its uses are, therefore, critical steps towards developing the transformative changes required to achieve socio-economic sustainability, while preserving life on Earth.
Conservationists, constrained by finite resources, have used the concept of biodiversity hotspots to advocate for the allocation of international efforts and resources in regions of the world containing exceptionally diverse, unique and threatened biodiversity [18
]. However, biodiversity hotspots often fail to capture the multi-faceted nature of biodiversity. For example, hotspot designations may consider a narrow range of organisms [19
], and miss non-terrestrial habitats [3
], phylogenetic and functional diversity (but see [20
]). The seminal definition of a biodiversity hotspot, which was based solely on plants from tropical forests, highlights this shortcoming [21
]. An additional limitation of biodiversity hotspots, as currently defined, is that they rarely consider anthropogenic interactions as anything other than a threat [22
]. However, the sustainable and responsible use of nature, as demonstrated by traditional small-scale livelihoods and indigenous communities around the world, presents opportunities for resolving the current biodiversity crisis and global challenges facing humanity [12
]. Finally, although often considered implicitly [13
], nature’s contributions to people are rarely considered when setting global conservation priorities, except for a few regulating ecosystem services [2
Agro-biodiversity is a sub-component of biodiversity that accounts for the variety of life that contributes to food and agriculture [25
]. In the broadest sense, it can be broken down into two components: (a) Planned agro-biodiversity, which refers to the diversity within and across species (domesticated or undomesticated) that are used by people, and (b) associated agro-biodiversity, which refers to species that surround and/or enhance planned agro-biodiversity [26
]. In order to identify species and forms of high potential for improving and sustaining agriculture, research efforts have generally focused on mapping centres of origin and diversity of major domesticated plant species (mostly food crops) and associated wild relatives [27
]. While the distribution of plants of highest importance for commodity production and human nutrition is now widely studied [6
], much uncertainty remains about the large fraction of neglected and underutilized species that contribute to a wide array of provisioning, support, regulation and cultural services [30
]. Moreover, although recent international treaties incentivized the preservation of plant genetic resources in the face of major global challenges [30
], the distribution of useful plant genetic diversity is mainly studied through proxies (e.g., taxonomy, geographic distribution and environmental data) [28
]. Likewise, the drivers and spatial patterns of decline in agro-biodiversity remain poorly understood [15
Here we explore the relationship between hotspots of wild plant diversity and regions containing high levels of agro-biodiversity by (i) reviewing the history of the two concepts, (ii) characterizing the degree to which they overlap in space and the human and biological drivers of these geographic (in-) congruencies, and (iii) considering how to better integrate the multi-faceted aspects of useful plant diversity, including genetic and phylogenetic diversity. Finally, we propose a new general framework and discuss future avenues for obtaining an improved understanding and preservation of global hotspots of useful plant diversity [31
3. From Biodiversity to Agro-Biodiversity Hotspots: A Geographic Continuum
The geographic distribution of biodiversity and agro-diversity hotspots has long been investigated, but rarely together. Very little is known about their spatial intricacies despite recent calls for their integration [31
]. Several areas of high plant diversity do not include primary regions of agro-biodiversity (e.g., California, Caribbean, Brazilian Cerrado, South Africa, Madagascar, Pacific islands) and a few agro-biodiversity regions are not recognized as diversity hotspots (e.g., large parts of Eastern Asia and India) (Figure 1
). Although human activities are altering this pattern [12
], global plant species richness generally decreases with increasing latitude [5
], reflecting past environmental changes, land configuration and the evolutionary histories of species [81
]. On the other hand, the history of global human migrations, civilisations, economy and cultural preferences have been profoundly intertwined with the distribution and availability of natural resources (species richness, abundance and properties) to shape regions of agro-biodiversity [9
]. Here, we present a new set of analyses to explore, illustrate and discuss the spatial congruence between biodiversity and agro-biodiversity hotspots, and its relationship with two important processes—human selection of species and species evolutionary history.
3.1. From Popularity to Anonymity
Vavilov provided an early spatial representation of the origins of cultivated plants by mapping the distribution of major food species. However, by restricting (justifiably) his focus mainly to selected food plants encountered in his field experiences, Vavilov was unable to produce a comprehensive assessment of the distribution of all plant species selected for use by humans across the world. Identifying which plant species are most selected by humans for a broader range of uses in addition to food, and characterizing their geographic distribution is, thus, key for defining agro-biodiversity hotspots and for relating their variation to the wider context of biodiversity.
In our currently globalized world and big data era, it is now possible to investigate human preferences for species more extensively. Here we assessed the popularity of most vascular plants on the free online encyclopaedia Wikipedia (www.wikipedia.org
), using search data as a proxy for cultural preference, knowledge of plant species, use by humans and domestication intensity. As Wikipedia can be edited by anyone at any given time, it cannot be considered a reliable source of information without critical evaluation. However, our analysis remains independent from the quality of Wikipedia articles as it only examines the interaction between users and the web platform by quantifying numbers of page views. Species names and occupied geographic regions were retrieved for 339,924 species from the World Checklist of Vascular Plants (WCVP) [51
]. Wikipedia uses the taxonomic backbone of the WCVP, so no additional name matching was performed. Geographic regions were retrieved at the national or sub-national level (finest level three) of the World Geographical Scheme for Recording Plant Distribution (WGSRPD), which was developed by the International Working Group on Taxonomic Databases for Plant Sciences [83
]. We retrieved species popularity on Wikipedia as measured by the number of page views over the period 1 January 2016 to 1 January 2020 using the R package pageviews. This information relates to English Wikipedia pages only as it is the language with the highest number of pages overall and often the source of translations into other languages. We ranked these data according to three categories: (1) The 1000 most popular species; this includes plants used as food, medicine, timber, ornamentals and cosmetics (Supplementary Materials
), (2) the remaining species covered by Wikipedia (i.e., those with an available page), (3) the remaining species not documented in Wikipedia (i.e., those without a page).
The global richness distribution of the 1000 most popular species is strikingly similar to the Vavilov primary centres with particularly high richness in subtropical regions of the Northern hemisphere (Figure 2
a). There are also expansions towards temperate areas (i.e., Eastern North America, Europe and Central Asia), while the Northern Andes, Eastern Africa and the Indo-Malayan regions have relatively low crop richness at a global scale, but high richness within their respective continents. Richness generally increases towards the tropics for species that are documented in Wikipedia but fall outside of the 1000 most popular category, with particularly high concentrations in the Mediterranean and subtropical regions that are characterized by high plant species endemism but low richness in major crops (e.g., Western North America, South Africa, Australia) [5
] (Figure 2
b). In contrast, richness in species not documented in Wikipedia tends to follow a latitudinal gradient similar to that observed for total plant diversity and most similar to biodiversity hotspots (Figure 2
]. These findings highlight the relationship between the spatial structure of biodiversity and agro-biodiversity, and people’s knowledge, perception and use of nature. Indeed, we observe the existence of a geographic continuum between popular plant diversity that may be more intensively used by humanity (based on the existence of a Wikipedia page; 2A) and anonymous plant diversity that may be less heavily used (based on the lack of a Wikipedia page; 2C). One artefactual limitation in our assessment of the distribution of plant popularity is the general over-representation of English-speaking regions (e.g., United States, Canada, Australia) given that our data extraction came from English Wikipedia pages only. This also likely explains relatively low values of popular species richness in regions, such as the Andes or Ethiopia (Figure 2
a), although those are still visible at the continental scale.
3.2. From Domesticates to Wild Relatives
Alongside major crops, Vavilov was also interested in documenting, mapping, collecting, using, and preserving wild ancestors and closely related species [27
]. By considering less known and undomesticated (or less domesticated) parent and sister species, Vavilov directly connected his definition of regions of origin/diversity with wild plant diversity. Recent studies illustrate this link between biodiversity and agro-biodiversity hotspots; these assessed the distribution of the closest and more distant relatives of major crops and found high species richness in plant diversity hotspots, such as the Brazilian Cerrado and Atlantic Forest or South-East Asia [28
Here we assessed the current distribution of 222 major international crops and 2,731 of their wild relative species using comprehensive lists from the USDA ARS GRIN-Global Taxonomy and geographic data from the World Checklist of Vascular Plants, again retrieved at the national and sub-national level (level three) of the WGSRPD [51
]. Crop wild relatives are classified across three gene pools based on both their relatedness (using phylogenetics and systematics) and crossing ability with the crop [86
]: gene pool one comprises the most closely related (even conspecific) wild species that are generally fully interfertile with the crop; gene pool two includes more distant relatives that may be crossed to the crop with more difficulty; and gene pool three typically contains the most distantly related and least compatible species within the genus to which the crop belongs (sometimes including other genera). Here, we assess changes in geographic patterns across a gradient from cultivated species to their closest wild relatives to their more distant wild relatives, by mapping richness for crops and each associated gene pool separately. When more than one species was identified in a gene pool for a given crop, we merged their distribution to assign the same weight to each crop, thus, avoiding overrepresentation of genera with many wild relatives. Geographic data were not available at the infra-specific level (i.e., sub-species, varieties, forms) for all taxa, and so we performed analyses at the species level.
Geographic patterns in major crop species richness strongly overlap with the Vavilov centres (Figure 3
a) and are also similar to the 1000 most popular plant species on Wikipedia (Figure 2
a). Given that gene pool one is composed of the closest crop wild relatives, including progenitors and/or wild types of the crop species, the distribution of gene pool one species richness is very similar to that of the crops (Figure 3
b). Slight increases and decreases are respectively observed inside and outside Europe, which may be explained by more extensive documentation of European crop wild relatives compared to those that occur in other regions [88
]. Gene pools two and three provide a more diffuse representation of the Vavilov centres: species richness decreases in most of the Vavilov centres, but increases in surrounding regions, particularly (but not exclusively) towards the tropics and plant diversity hotspots (Figure 3
c,d). Although centres of crop diversity remain visible when mapping secondary and tertiary gene pools, the geographic signal diminishes as we move further away from the most explored and popular branches of the tree of life. In contrast, areas of high biodiversity start to emerge, which is even more striking when species richness mapping does not account for the over-representation of genera with many wild relatives [28
]. This reinforces the existence of a geographic continuum between agro-biodiversity (i.e., widely used and cultivated crops) and biodiversity (i.e., non-domesticated and less used sister species of crops) related to the strength of the interaction between humans and plants, but also plant evolutionary history.
5. Towards a Unified Concept of Agro-Biodiversity Hotspot
Hotspots of both biodiversity and agro-biodiversity have long relied on counting numbers of species (i.e., species richness; Figure 5
a) and assessing threats (Figure 5
b). While biodiversity scientists have mainly focused on numbers of rare species (including many narrowly distributed taxa) for conservation, agronomists have been interested in diversity within gene pools (i.e., numbers of domesticated species and wild relatives) [27
] and within crops (e.g., numbers of landraces) [112
]. Although taxon counts remain extremely useful, new approaches are now proposed to account for the multi-faceted nature of (agro-)biodiversity, such as functional diversity accounting for the diversity in species chemical properties and eco-/agri-system functions [113
], phylogenetic diversity as a potential proxy for functional and property diversity (Figure 5
c) or for identifying gene sources for breeding programmes (Figure 5
d). As highlighted by their seminal definition from N. Myers and recent publications [114
], biodiversity hotspots are also deeply related to the distribution of a wide range of threats, many of which are shared with agro-biodiversity (e.g., land use and climate change, pollution, biological invasions, over-harvesting), but less formally included in the geographic assessments of the latter (Figure 5
]. Although the different facets and threats of agro-biodiversity are not all expected to overlap geographically, our paper proposes to assess them jointly rather than separately.
Primary regions of agro-biodiversity have focused on relatively few important crops selected by researchers, whereas they have mainly explored wild relatives of these domesticated species, which are not always considered for other uses than crop improvement. Given the existence of a wide domestication spectrum, ranging from major global crops (i.e., highly domesticated species) to those harvested in the wild [16
], we believe that further work on regions of agro-biodiversity should expand the focus to better include the long list of plants that provide food and other cultural benefits to humanity (Figure 5
e). This will be made effective through the documentation of the tremendous diversity of neglected and under-utilized species of the world (with more than 30,000 useful plant species known to date [115
]). Many of these species occur naturally in low-income countries, including already established biodiversity hotspots, which are also often home to large human populations and cultural diversity [22
]. Understanding the drivers of the distributions of nature’s contribution to people across the domestication spectrum (from climate and land use to socio-economic factors; Figure 5
e) is also fundamental to define hotspots and design conservation and development efforts to sustain socio-environmental sustainability.
The recognition of biodiversity hotspots and agro-biodiversity (Vavilov) centres have played important roles to raise public awareness, foster research and attract political action to preserve and use natural resources sustainably. Given the urgency and magnitude of the global challenges outlined by the United Nations’ Sustainable Development Goals and the recent report on biodiversity loss by the intergovernmental science-policy platform on biodiversity and ecosystem Services (IPBES) [12
], it is more important than ever to refine, integrate and disseminate such powerful concepts. Failing to protect hotspots of natural resources, and especially agro-biodiversity, would have damaging consequences on nature and human livelihoods, both at those centres and in their peripheries. Our paper calls for the further development and integration of a range of commonly used and more recently proposed indices, while accounting for the key interaction with biodiversity, into the agro-biodiversity hotspot concept.