Global pattern of plant utilization across different organisms: Does plant apparency or plant phylogeny matter?

Abstract The present study is the first to consider human and nonhuman consumers together to reveal several general patterns of plant utilization. We provide evidence that at a global scale, plant apparency and phylogenetic isolation can be important predictors of plant utilization and consumer diversity. Using the number of species or genera or the distribution area of each plant family as the island “area” and the minimum phylogenetic distance to common plant families as the island “distance”, we fitted presence–area relationships and presence–distance relationships with a binomial GLM (generalized linear model) with a logit link. The presence–absence of consumers among each plant family strongly depended on plant apparency (family size and distribution area); the diversity of consumers increased with plant apparency but decreased with phylogenetic isolation. When consumers extended their host breadth, unapparent plants became more likely to be used. Common uses occurred more often on common plants and their relatives, showing higher host phylogenetic clustering than uncommon uses. On the contrary, highly specialized uses might be related to the rarity of plant chemicals and were therefore very species‐specific. In summary, our results provide a global illustration of plant–consumer combinations and reveal several general patterns of plant utilization across humans, insects and microbes. First, plant apparency and plant phylogenetic isolation generally govern plant utilization value, with uncommon and isolated plants suffering fewer parasites. Second, extension of the breadth of utilized hosts helps explain the presence of consumers on unapparent plants. Finally, the phylogenetic clustering structure of host plants is different between common uses and uncommon uses. The strength of such consistent plant utilization patterns across a diverse set of usage types suggests that the persistence and accumulation of consumer diversity and use value for plant species are determined by similar ecological and evolutionary processes.

ency has also been extended to ethnobiology, where humans are treated as foragers or consumers, similar to nonhuman herbivores (Lozano, Araújo, Medeiros, & Albuquerque, 2014;de Lucena, de Lima Araújo, & de Albuquerque, 2007). As a variant of the plant apparency hypothesis, the ecological apparency hypothesis implies that humans tend to collect and use apparent plants, similar to insects and other organisms (Phillips & Gentry, 1993;Ribeiro et al., 2014;Soldati et al., 2016). Apparent plants are more likely to be found by parasites, natural enemies, pollinators and humans (Feeny, 1976;Phillips & Gentry, 1993;Schlinkert et al., 2015), while rare plants are difficult to find or become profitless and therefore escape enemies (i.e., the rare species advantage hypothesis) (Bachelot & Kobe, 2013;Chew & Courtney, 1991;Parker et al., 2015). According to the optimal foraging strategy, more available species should be preferred because they are easier to discover and should therefore reduce time and energy costs (Gonçalves, Albuquerque, & de Medeiros, 2016).
Hence, apparency indicators can be divided into two major categories (Table 1): (1) quantitative availability, which increases the random encounter rate between consumers and plants, related to either random searching or active searching by parasites (e.g., the abundance, spatiotemporal distribution, or biomass of a given plant species); and (2) qualitative detectability, which makes plants visually or chemically distinct from their background and is related to consumers' perceptual abilities and feeding habits (Castagneyrol, Giffard, Péré, & Jactel, 2013;Courtney, 1982;Schlinkert et al., 2015;Strauss et al., 2015;Wiklund, 1984) (e.g., the odor, color, plant composition or background environment of a given plant species). The ecological apparency hypothesis in ethnobiology (Phillips & Gentry, 1993) focuses more on quantitative apparency, while the plant apparency hypothesis in insect ecology (Feeny, 1976) focuses more on qualitative apparency and plant chemical defenses. Quantitative apparency will generally increase attacks from enemies, while the roles of qualitative apparency are complicated. For example, specific plant defensive compounds may reduce visiting and feeding by most insects but attract some herbivore specialists (Smilanich et al., 2016), and red leaf color is a warning signal for many animals, but there are some exceptions (Stutz et al., 2016).
However, the total apparency of one plant is the combination of all quantitative and qualitative apparency indicators and functions as a whole in relation to enemies.
In the present study, we focused on only quantitative measurements of plant spatiotemporal availability that are objective and "ultimate," without reference to the detective abilities of relevant organisms for particular hosts (Courtney, 1985;Rhoades, 1979) or the degree of differences in the searching environments (Strauss et al., 2015).

| Parasite patterns on different host plant islands
If we consider host plants to be analogous to islands (Janzen, 1968;Joy & Crespi, 2012;Miller, 2012), species-area relationships or species-distance relationships can be adopted to describe the incidence or richness of parasites on hosts. The "area" in species-area relationships can refer to any quantitative apparency indicator, such as the number of individuals, distribution range, body size, species number, or total biomass of a host taxon (Feeny, 1976;Joy & Crespi, 2012;Kamiya, O'Dwyer, Nakagawa, & Poulin, 2014;Miller, 2012). Higher apparency is associated with more host-parasite encounters (random placement hypothesis) and more niches for parasites (habitat diversity hypothesis) (Miller, 2012;Strona & Fattorini, 2014).
Studies on whether a certain plant is used by humans or other organisms are generally performed at local or regional scales (Brändle & Brandl, 2001;Guèze et al., 2014). In contrast, global-level studies are scarce but are important for evaluating whether plant apparency or plant phylogeny can predict patterns of plant use across different organisms. Moreover, the present study may be the first to consider human and nonhuman consumers together in the analysis of plant utilization patterns. We will prove that at a global scale, plant apparency and phylogenetic isolation can be important predictors of plant utilization and consumer diversity.

| Data collection
The main sources of global plant utilization data used in the present study included review articles, monographs, professional databases and specialized websites addressing plant uses (Appendix S1). As some sources may be outdated or incomplete, we also employed the ISI Web of Science ™ (WoS) to obtain more plant utilization data based on keyword searches (Appendix S1). For example, to study the host plants of Agromyzid flies, we obtained an initial host plant list from the book "Host Specialization in the World Agromyzidae (Diptera)" published in 1990 and then searched WoS publications from 1990 to 2015 using the following search terms: Topic: (Agromyzidae) AND Topic: (host plant*). However, when too many hits were obtained in WoS (>500 hits), we narrowed the search terms by replacing TS (Topic) with TL (Title) and so on (this seldom occurred). We then manually checked and extracted host plant names article by article.
The deadline for all utilization data was 31 December 2015. The literature search using WoS was similar to increase the sampling effort in field investigations; however, few additional host families were identified through the WoS search and most of those families were small (Appendix S1). Thus, even without the WoS search, the general T A B L E 1 Indicators of plant apparency for susceptibility to encounter by parasites  (2013); Strauss et al. (2015) patterns observed in the present study were consistent and were confirmed by our previous data analyses.
Plant names (species, genera, families, or mixtures of the three levels) were checked and resolved using the Taxonomic Name Resolution Service, v 4.0 (Boyle et al., 2013) and were verified with The Plant List, v 1.1 (http://www.theplantlist.org/). Then, we summarized the list of matched and accepted family names for each utilization group.
We focused on angiosperm plants in this study only because many utilizers, such as pollinators and Tischeriidae, seldom use ferns and gymnosperm. The names of 420 angiosperm families (Parker et al., 2015) were obtained according to the APG III system (

| Plant families as islands
To compare the plant utilization patterns among human and nonhuman consumers, we employed presence-absence data as consumer characteristics and plant families as islands. Presence-absence data for plant utilization at the plant family level is easy to obtain, while the collection costs for abundance data for plant utilizers at a plant species level are high, making it impossible to obtain such data, especially at larger spatial scales. Presence-absence data are more appropriate for clarifying the effects of host characteristics on parasite similarity (Locke et al., 2013), especially at broad scales (i.e., continental to global scales).
If we regard a plant species as the island, the plant's genus or family is like an archipelago of species islands (Janzen, 1968). The above species-area, species-apparency and species-distance-relationships should also be true at the archipelago level. Such relationships between parasites and host plant families have been discovered in galling insects (Joy & Crespi, 2012;Price, 1977) and other insects (Lill, Marquis, & Ricklefs, 2002 & Price, 1979;Mendonça, 2007;Price, 1977;Veldtman & McGeoch, 2003;Ward & Spalding, 1993). The existence of more species in a given family corresponds to more available niches for parasites (de Araújo et al., , 2013Joy & Crespi, 2012;Milton de Souza Mendonça, 2007). However, the relationship between parasite richness and plant genus size is weaker than that for plant family size (Araújo, 2011;).

| Plant phylogeny
We constructed a phylogenetic supertree (Appendix S2) with ages for all vascular plant families based on the R2G2_20140601 super tree

| Plant apparency
In this study, we regarded the numbers of species and genera and of each plant family was calculated using ImageJ 1.48v (Schneider, Rasband, & Eliceiri, 2012) based on a pixel number-area transformation (Appendix S4). Note that the distribution maps for some plant families were merged from sketch maps of within-family groups. For some small families without a distribution map, we assigned the distribution area a small value of 1,000 km 2 .

| Data analyses
The presence-absence of utilized plant families for each utilization group was recorded as either binary data or 0-1 data. Defining the utilization probability (UP) as the probability that the plant family was utilized by one utilization group, 1-UP was the probability that

| RESULTS
We listed world plant families utilized by insects, mites, microbes, pollinators, and humans (i.e., different utilizers) based on a metaanalysis. As predicted, the utilization probability for every utilization group increased significantly with plant apparency (a > 0, p(a = 0) < .05; Figure 1, Appendices S5 and S6) and decreased significantly with plant phylogenetic distance to common plants (a < 0, p(a = 0) < .05; Appendix S7). It appears that common plants are always selected for common uses. As the plants that are primarily used as food for either insects or humans, these plants would be expected to exhibit high abundance and high accessibility (Thomas, Vandebroek, & Van Damme, 2009), and the relatives of the primary plants would also show a high probability of being targeted.
The binomial GLM models were useful for describing the presence-absence of each species on plants with different apparency ratings. This approach is similar to the species-area curves employed in island biogeography, if one regards the plants as islands, apparency as island size and phylogenetic distance as island distance. Generally, apparent and abundant plants supported more utilization groups than unapparent plants (Figure 2), while plants that were phylogenetically close to common plants presented more consumer diversity than phylogenetically distant plants (Figure 3). We refer to the latter phenomenon as an "isolated species advantage" rather than a "rare species advantage." That is, more consumer diversity is found on larger plant islands or on islands closer to the largest islands, which is consistent with many previous studies involving hosts as islands (Miller, 2012;Parker et al., 2015).
An unexpected exception was found for Orchidaceae (Figure 2), which hosted fewer herbivores and sexual systems than other common plant families. The acceleration of orchid species diversification in history is correlated with, for example, the evolution of pollinia, epiphytic habits and sophisticated insect pollination mechanisms (Givnish et al., 2015). Pollinating predatory wasps, nectary-attracting bodyguard ants and flowers showing wasp mimicry may play important roles in the protection of orchids from herbivory (Subedi et al., 2011). For example, herbivory attacks induce more extrafloral nectar exudation to recruit more natural enemies (Subedi et al., 2011). In addition, plants with epiphytic habitats are usually poor in resources, and herbivory on epiphytes, such as orchids, is therefore relatively lower than that on nonepiphytes (Winkler, Hülber, Mehltreter, Franco, & Hietz, 2005).
When we checked another epiphytic plant family, Bromeliaceae, we were surprised to find that Orchidaceae and Bromeliaceae exhibited the same utilizer ratio (UR).

| DISCUSSION
Apparent plants, which are under strong selection pressure from both specialists and generalists, can produce quantitative defenses (tannins and lignin), whereas it is difficult for parasites to specialize toward nonapparent plants, which therefore require only qualitative defenses (alkaloids and terpenoids) against generalists (Strauss et al., 2015).
Host plant chemistry may be determined by plant phylogeny (Heidel-Fischer et al., 2009), where closely related plants share similar biological and chemical defenses and, thus, can be vulnerable to the same types of parasites (Davies & Pedersen, 2008). Alternatively, apparent plants such as trees can facilitate host shifts between phylogenetically distant plants (Heidel-Fischer et al., 2009).
Similar to nonhuman foragers, humans can behave as specialists or generalists (de Albuquerque, Soldati, & Ramos, 2015). Plant apparency might play a more important role for generalists than for specialists, while the latter are more or less associated with special plant chemicals (Gonçalves et al., 2016;Soldati et al., 2016).
Regarding human utilization, the phylogenetic structure differed distinctly from random for wide uses (food, medicines, environmental uses, food additives, materials, and weeds; D < 0.886, p random < 0.1), Among tritrophic levels, the accumulation of parasitoids is determined by plant commonness, rather than herbivore richness on plants (Nascimento et al., 2014).
The phylogenetic structure of common plant sexual systems, such as hermaphroditism, dioecy, and monoecy (D < 0.99; p random < 0.2), was more clumped than that of other, uncommon plant sexual systems (D > 0.99; p random > 0.4). Plants with different sexual systems not only are associated with different pollination groups (Charlesworth, 1993) but also suffer different herbivory and pathogen pressures (Ashman, 2002 (Forister et al., 2015). Among herbivores, leaf consumers and bark consumers are more specialized than sap consumers and wood consumers. One reason for this difference might be that leaves and bark exhibit more chemical barriers than sap and wood. Highly specialized modes of utilization, such as social uses, are highly speciesspecific. The role of one plant species will not be fully replaced by other close relatives of the same genus. For example, the opium poppy (Papaver somniferum) is the only species to produce opium in Papaveraceae (Darokar et al., 2014). Once we obtain sufficient global utilization data at the plant genus or species level, we might identify similar patterns. Alternatively, we will be able to test these utilization patterns at lower taxonomic levels on a regional scale when such detailed data are available.
Plants are treated as resources in the ecological apparency hypothesis (Lozano et al., 2014;de Lucena et al., 2007). It would be interesting to extend the apparency hypothesis to animal hosts or abiotic resources. For example, the seven most abundant elements on Earth (iron, oxygen, silicon, magnesium, sulfur, nickel, and calcium) (Morgan & Anders, 1980) but not the 8th most abundant element, aluminum, are also included among the 15 richest elements in the human body.
In summary, our results provide a global illustration of plant-con-  com/c/NRW49). This manuscript was edited for English language by American Journal Experts (AJE).

CONFLICT OF INTEREST
None declared.