(taken from Foley et al. 2012. GnosisGIS in press)
Martins et al. (1978) provided extensive collection records for South and Central American sand flies gleaned from the literature and from the author’s own collections. These records are also the basis for many of the records and maps in Young and Duncan (1994), however, location data were given as town and state, and maps cannot reliably be used to obtain georeferenced occurrence data. Therefore, we scanned a copy of Martins et al. (1978) and enabled optical character read (OCR) within the PDF, to allow copying and pasting of relevant text into Microsoft ® Excel. Georeferences were obtained with the online gazetteer exploration tool, Biogeomancer Version 1.2.4 ( http://bg.berkeley.edu/latest/ ). Search strings were arranged with as much geographic information as possible to minimize errors, i.e. precise location, and state, and country. Where only one record for a populated place was obtained this was assumed to be the correct location. In such cases, the gazetteer details, the geocoordinates, and the estimate of spatial error were copied from Biogeomancer. Spatial error in Biogeomancer is estimated by the point radius method (see http://manisnet.org/GeorefGuide.html ). The point radius method (Wieczorek et al. 2004) portrays error as a radius around a geocoordinate. Having an error estimate for a collection georeference allows the user to judge the spatial resolution of remotely sensed data that are appropriate (Foley et al. 2009).
In other cases, no records were identified, only non-populated place names were identified, or more than one populated place was identified. Where no record was identified, this was often a result of errors in the spelling or the wrong current name for the state, for instance Pôrto Feliz rather than Porto Fleiz, or Mato Grosso rather than Mato Grosso do Sul. Internet searches were useful to resolve spelling and locations in many cases. Where more than one populated place was obtained these locations were either clustered or far apart. When possible locations were close to one another (e.g. less than 25 km apart), an attempt was made to give one location to encompass all possibilities. For example, two locations for Aracruz, Espirito Santo, Brazil were obtained that were 20 km apart. In that case the Global Administrative Areas (GADM) administrative category that they occurred in was chosen to represent the location, but in other cases one representative location was chosen and the estimate of spatial error manually increased to cover the extent of all locations. When more than one populated place was identified that were far apart (e.g. greater than 25 km apart), these records were set aside for future georeferencing, when more information may become available. For example, “Esperanza, Belize” resulted in 4 population centers, one archeological site and one man-made feature. In all cases, information about georeferencing decisions was associated with the collection record in SandflyMap.
Species names were those from the WRBU Catalog of Subfamily Phlebotominae ( http://www.sandflycatalog.org/ ), which incorporates the classification of Galati (2003).
Martins AV, Williams P, Falcao AL, 1978. American sand flies (Diptera: Psychodidae, Phlebotominae). Academia Brasileira de Ciencias, Rio de Janeiro.
Return to top