High spatial heterogeneity in ground cover, large amounts of exposed bare soil, and modest cover from shrubs and grasses in arid and semi-arid ecosystems challenge the integration of field observations of phenology and remotely sensed data to monitor changes in land surface phenology. This research conducted at the Jornada Basin Long-Term Ecological Research (LTER) site in southern New Mexico capitalizes on object-based classification of sub-decimeter (4 cm) aerial photography to examine species-specific contributions to vegetation index values calculated across a range of grain sizes. Drawing on established field protocols for reproductive phenology, sub-decimeter imagery (4 cm), and object-based image analysis, we explore the relationship between field phenology and vegetation index values and quantify the contribution of individual species to spectral vegetation index values derived from 4 cm imagery aggregated incrementally to 30 m spatial resolution. Color-infrared imagery from a digital mapping camera was collected June 2007 across 15 LTER study sites that transect five distinct vegetation communities along a continuum of grass to shrub dominance. Object-based image analysis of 4 cm imagery provides a detailed depiction of ground cover and allows us to extract species-specific contributions to spectral vegetation indices. The ability to discern species- or functional-group contributions to remotely sensed signals of vegetation greenness can greatly enhance the design of field sampling protocols for phenological research. Furthermore, imagery from unmanned aerial vehicles (UAV) is a cost-effective and increasingly available resource and generation of UAV mosaics has been accomplished so that larger study areas can be addressed. This technology can provide a robust basis for scaling relationships for phenology-based research applications.