The most frequently mentioned species were well-known medicinal plants of central Mexico. Except for Laelia autumnalis, Dioscorea galeottiana and Salvia melissodora, all are included in the compilation of the medicinal plants of the Balsas River Basin [29]. The species with the highest use value have also been reported in other studies [4, 29, 35, 46].

An unexpected result was the existence of a kind of informal apprenticeship system in the region for medicinal plant use. This has not been reported previously for other regions in Mexico, though it probably exists.

The most-used species were either weedy, or woody plants of the deciduous tropical forest; native non-weedy herbs and shrubs/lianas were much less useful. The predominance of these plant types as sources of medicinal plants has been noted before [1, 2].

This study tentatively supports the ecological apparency hypothesis (more common plant species are also more commonly used) for medicinal species. It also appears to apply to subgroups based on life forms, in our case for herbs and shrubs/lianas. The lack of confirmation of the apparency hypothesis for trees may be due to the high value of the material obtained from some of them, which could have led to population decline. We do not have measured data on this, but our informants said that Amphipterygium adstringens, Crescentia alata, Ternstroemia lineata and Eysenhardtia polystachya (the latter also widely used as firewood) used to be more common and are collected commercially. These species are highly demanded in the national market, and there are reports of overcollection from other parts of central Mexico [4, 47]. Persea americana populations, which were included in the list because they occurred in apparently natural places but may have been remnants of previous human occupation, were in decline because of a disease introduced by modern varieties. Salix humboldtiana, a riparian species, was removed from its natural habitat to establish orchards and sugarcane plantations. However, statistical caveats apply to these results, and the subject should be explored with a larger dataset of species.

Previous work on this subject has also found mixed results, often with relatively weak, but significant correlations. Comparisons are further complicated by other works studying mainly to woody plants, and various use domains. Particularly timber trees, as mentioned in the introduction, often [26, 48], but not always [49] conform to the apparency hypothesis. Medicinal plants have been little studied [7, 8], and they used different criteria for apparency, but they did find some relationship between abundance and importance values [4].

However, their correlations are weaker than in our study; their R2 was 0.62, 0.59 [48] and 0.21 [50] for the life forms. They interpret this to mean that apparency is important, but not the only factor that influences plant selection. Other multidomain studies found few or negative correlations for medicinal plants [50, 51]. We suggest that our higher correlations could be due to a better dataset for this domain, and interviews with specialists, whereas the other studies interviewed the general population and had relatively few data for medicinal plants; higher correlations were also found by Christo et al. (2012) [49], who interviewed specialists on timber use. It is also possible that different vegetation types have different dynamics; perhaps the more arid vegetation types with their higher proportion of medicinals conform to the hypothesis, and more humid types do not; however, Guerra et al. [51] did not confirm the hypothesis in the Caatinga, which is similar to our dry tropical forests.

Our data can be interpreted to support both of the possible main causes of medicinal plant use, abundance and intrinsic usefulness. People use those plants that are more common and easily learned. However, there are also differences in intrinsic usefulness of ecological groups of species due to selection pressures. So, people use mainly the most common, accessible species of inherently useful groups.

We suggest that support for the apparency hypothesis, particularly for wild-growing medicinal plants, may be dynamic. This could explain the various and somewhat contradictory results obtained in different regions and vegetation types. In primary vegetation with only local uses (or in inaccessible or more distant sites), the relationship between apparency and use may be difficult to document when use is infrequent and more idiosyncratic; also, it is often a byproduct of other activities such as hunting (with some exceptions, e.g., timber use for trees). Once a vegetation type becomes accessible, medicinal plant collection is efficient, and there is an external market for some of its products, apparency should play a large role—up to a point. Once overcollection reduces the populations, the apparency hypothesis again would not be supported. So, studies should separate native and exotic species, weeds and non-weeds, commercial and local use and consider distance to the populations, collection intensity and past history [17].

Life form, just like botanical family, predicts some characteristics relevant for medicinal uses. Differing defense mechanisms leads to differences in the type of biochemical content; recovery and regeneration after damage vary substantially between herbs and trees. Also, apparency for humans depends not only on frequency, but also on size and visibility, so differences should be expected between herbs, shrubs and trees, which is shown here and should be integrated into theory. Finally, we propose that the validity of the apparency hypothesis may depend on the degree of collection pressure and thus vary over time.

Our data show that local collectors have relatively good estimations of population size and dynamics. This means that interviews, particularly of several knowledgeable informants, can deliver useful data for management without more costly vegetation surveys, as long as interviewees are confident that their information will not be abused [27, 52, 53].

Importance—the relative value that different species have for humans—is one of the basic metrics for analyzing the relationship between humans and their vegetation. A large number of methods and indices have been employed over time [12, 22, 54]. Our data allow for comparison of some widely used indicators.

Use value, frequency of mention, cultural significance and recent use were all relatively closely related. Also, people in our study cited mostly species that were in actual use, and not historical ones (which was expected, of course, as we interviewed active healers).

However, support was somewhat weaker for the common assumption that the number of uses of a medicinal plant is correlated with its importance [12, 28]. Some species, such as Tagetes lucida, Tagetes micrantha and Barkleyanthus salicifolius, did have high importance values and several uses. However, others, such as Selaginella lepidophylla, Amphipterygium adstringens or Equisetum hyemale, had very specific uses and were also considered important. Thus, number of uses as an importance indicator and part of indices should be used with some caution as it may lead to underestimation of the importance of species with only one or two uses. The relationship between number of uses and importance, particularly of wild-growing medicinals, should be explored further in other contexts.

Our data show that frequency of mention, relatively easy to obtain, is a good overall indicator of importance, and that generally, importance measured in various ways is also correlated with current use. This may be due to our species selection: We considered only native species that were obtained from natural vegetation, not the cultivated exotics that often have multiple uses [7, 55] and whose inclusion would probably have strengthened the statistical correlation between number of uses and other indicators.

This study has some limitations. We only included 52 species—because of constraints on the patience and attention of interviewees—which made our life form categories, and especially lianas/shrubs, rather small. The data may be susceptible to statistical error, outliers and non-normal datasets, which did occur in some data subsets. Also, the vegetation surveys were not made for this study, but rather for a vegetation description; the survey plots were placed preferentially and mostly in forested areas. Also, many other factors influence medicinal plant use—such as prevalent illnesses, accessibility and available alternatives. Plant populations can vary under different collection pressures, which may also change over time for other reasons.

However, we consider the relatively large number of vegetation plots and interviewed healers, repeated interviews and a close relationship with the interviewees, as well as the floristic knowledge of the first author compensate for some of these shortcomings. Despite the multiple factors that influence medicinal plant use, we still found some signal supporting the relationship between frequency and importance in the data. Thus, we hope to encourage other studies to look at apparency within life forms to confirm or refute the thesis that medicinal plant use is also—among other factors—influenced by the commonness of the species.