Phylogeny explains why less therapeutically redundant plant species are not necessarily facing greater use pressure
Cite this dataset
Coe, Michael (2021). Phylogeny explains why less therapeutically redundant plant species are not necessarily facing greater use pressure [Dataset]. Dryad. https://doi.org/10.5061/dryad.69p8cz91x
Understanding which factors influence medicinal plant species selection and harvest or use pressure can provide valuable insights for sustainable management of natural resources and conservation efforts. The utilitarian redundancy model, a theoretical framework in ethnobotany, suggests that species that are therapeutically redundant or fulfill similar therapeutic functions within traditional ethnomedicine are less likely to be under greater use pressure. However, species’ evolutionary relatedness and the preference of certain species over others to treat a given illness can directly affect how use pressure is diffused across several groups of species. These factors may alter the strength of the therapeutic redundancy-use pressure relationship.
Medicinal plant species that fulfill the same therapeutic functions may experience greater use pressure despite their level of therapeutic redundancy because they are preferred—where most people select these species preferably over other species that are equally available for a given treatment. Further, species that are closely related evolutionarily may be more likely to be harvested not because they are therapeutically unique but because they share evolutionary traits such as secondary chemistry with other medicinally important species which may make them more prone to being harvested.
We investigate the effects of species therapeutic redundancy, use value, preference, and evolutionary relatedness on species use pressure in the Shipibo-Konibo community of Paoyhan in the Peruvian Amazon region. We used phylogenetic generalized least squares models to identify significant predictors of species use pressure for 62 medicinal plant species cited by 30 participants and fulfilling 31 therapeutic functions in Shipibo-Konibo ethnomedicine.
Our model controlling for species’ shared evolutionary history indicated that therapeutically redundant medicinal plants experienced greater levels of use pressure. However, as preference increased, the effect of therapeutic redundancy on species use-pressure became less positive. Contrary to predictions, species preference by local people alone did not predict use pressure. Further, when we control for species’ shared evolutionary history, the effect of preference on species use pressure was dependent on therapeutic redundancy.
Our study illustrates the importance of controlling for evolutionary relatedness between species in studying plant-human interactions and the complexity involved in employing the utilitarian redundancy model to inform management and conservation efforts.
We used semi-structured interviews and free listing, an elucidation technique commonly used in the social sciences that seeks to identify specific information on a given cultural domain of the investigated community (Albuquerque et al., 2014). As such, each participant was asked to list the medicinal plants they know and their uses for treating illnesses in ethnomedicinal contexts according to the participants’ emic perspective. Semi-structured interviews were used because they allow for greater flexibility during the interview process compared to structured interviews and they allow for guided or specific data collection on an given cultural domain compared to unstructured interviews. Further, this approach allowed for more precise data collection as we aimed to get an in-depth understanding of medicinal plant use by knowledgeable participants without having to interview the same participant more than once, which, is not always possible (Albuquerque et al., 2014; Bernard, 2017). These approaches were coupled with focus group discussions supplemented by participant observations and walk in the woods (Albuquerque et al., 2014) to collect data that were used to estimate species therapeutic redundancy, use values and preference and to test their effects on species use pressure of medicinal plants used by the Shipibo-Konibo for healing. The use of such triangulation of methods recommended in ethnobiology seeks to ensure the reliability of data collected where one method is used to verify or cross-reference the responses obtained from another elicitation technique (Albuquerque et al., 2014).
To test the effects of species use preference, redundancy and use values, on species use pressure, we used general linear models (GLM) and phylogenetic generalized least squares (PGLS) in R 3.4.3 (R Development Core Team, 2019). Because our response variable, use pressure, was estimated as a measurement (biomass) we used a gaussian error structure for both models (Crawley, 2013). Prior to running our models, we tested for a correlation between predictor variables and excluded any that were significant (McGarigal et al., 2013). The PGLS model in addition to testing the effects of preference, redundancy and use values, also controlled for phylogenetic relatedness between medicinal plant species. We compared outcomes of both models to understand role of species shared evolutionary history (Heinrich & Verpoorte, 2012) on the prediction of species use pressure. Controlling for species evolutionary relatedness in our models allows us to account for non-independence between observations due to phylogenetic history between species (Mundry, 2014). This allowed us to provide a direct test of our predictor variables on medicinal plant species use pressure.
To develop the PGLS model, we built a phylogeny of medicinal species cited by participants using the S. PhyloMakerfunction in R (Qian & Jin, 2016; Jin & Qian, 2019). Our phylogenetic tree was built by pruning a mega-tree comprised of molecular data from GenBank, phylogenetic data from the Open Tree of Life and fossil records. It includes all plant families, approximately 10, 000 genera and 70,000 species of vascular plants in the world. As such, our phylogenetic tree generated by the S. PhyloMaker function consists of plant species cited by participants where all plant families and the majority of genera were resolved (Jin & Qian, 2019). For both GLM and PGLS models, we started with a saturated model that included all the three predictors (preference, redundancy and use value), and developed subsequent nested models’ candidates by sequentially removing one of the predictors. To select the best fitting models, we estimated the ∆ AIC (Akaike information criterion) for each model (Crawley, 2013) as the difference in the AIC between each model and the model with the lowest AIC. We then selected models with ∆ AIC< 2.
Missouri Botanical Garden
University of Hawaiʻi at Mānoa
University of Tennessee at Knoxville