Landscape scale species distribution modeling across the Guiana Shield to inform conservation decision making in Guyana

Species distribution modeling using museum and herbarium collections has been greatly facilitated by analytical algorithms such as MaxEnt. The ability to use herbarium and museum collections to inform conservation decision-making can greatly enhance conservation efforts in biodiversity rich countries when human capacity and in-country data are limited. Guyana is used as a case study for landscape scale biodiversity assessment under such constraints. I compiled specimen records for seven taxon-groups (invertebrates, amphibians, reptiles, birds, mammals, ferns and non-seed plants, and seed plants), across the Guiana Shield, South America, to assess landscape scale biodiversity richness. Collector and taxonomic bias were addressed a priori in MaxEnt by generating a bias surface layer to down-weight areas of high collection intensity by smoothing the sampling distribution. I summed modeled output for each taxon-group to generate taxon-group specific floral, faunal, and all-taxa biodiversity density surfaces. These surfaces were used to (1) identify areas of relative high biodiversity density; (2) assess possible conservation areas; and (3) compare modeled areas of conservation interest with those proposed by the Government of Guyana. In addition, I compared proposed conservation sites with the location of indigenous (Amerindian) and non-indigenous settlements, and lands used for natural resource extraction. I present three conservation scenarios based on the all-taxa biodiversity surface: (i) biodiversity-only, (ii) biodiversity and available lands; and (iii) collaboration with indigenous peoples. The conservation assessment used here provides an objective basis for selecting conservation sites. Model output can also be used to focus biodiversity assessments on poorly modeled and sampled locations.