Expedient Metrics to Describe Plant Community Change Across Gradients of Anthropogenic Influence
2014
Marcelino, José A. P. | Weber, Everett | Silva, Luis | Garcia, Patrícia V. | Soares, António O.
Human influence associated with land use may cause considerable biodiversity losses, namely in oceanic islands such as the Azores. Our goal was to identify plant indicator species across two gradients of increasing anthropogenic influence and management (arborescent and herbaceous communities) and determine similarity between plant communities of uncategorized vegetation plots to those in reference gradients using metrics derived from R programming. We intend to test and provide an expedient way to determine the conservation value of a given uncategorized vegetation plot based on the number of native, endemic, introduced, and invasive indicator species present. Using the metric IndVal, plant taxa with a significant indicator value for each community type in the two anthropogenic gradients were determined. A new metric, ComVal, was developed to assess the similarity of an uncategorized vegetation plot toward a reference community type, based on (i) the percentage of pre-defined indicator species from reference communities present in the vegetation plots, and (ii) the percentage of indicator species, specific to a given reference community type, present in the vegetation plot. Using a data resampling approach, the communities were randomly used as training or validation sets to classify vegetation plots based on ComVal. The percentage match with reference community types ranged from 77 to 100 % and from 79 to 100 %, for herbaceous and arborescent vegetation plots, respectively. Both IndVal and ComVal are part of a suite of useful tools characterizing plant communities and plant community change along gradients of anthropogenic influence without a priori knowledge of their biology and ecology.
Show more [+] Less [-]AGROVOC Keywords
Bibliographic information
This bibliographic record has been provided by National Agricultural Library