|Hallam, Thomas - UNIV. OF TENNESSEE|
|Raghaven, Aruna - UNIV. OF TENNESSEE|
|Kolli, Haritha - UNIV. OF TENNESSEE|
|Dimitrov, Dobromir - UNIV. OF TENNESSEE|
|Federico, Paula - UNIV. OF TENNESSEE|
|Qi, Hairong - UNIV. OF TENNESSEE|
|Mccracken, Gary - UNIV. OF TENNESSEE|
|Betke, Margrit - UNIV. OF TENNESSEE|
|Kennard, Kimberly - UNIV. OF TENNESSEE|
|Kunz, Thomas - BOSTON UNIVERSITY|
Submitted to: Ecological Complexity
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: May 19, 2009
Publication Date: June 28, 2009
Citation: Hallam, T.G., Raghaven, A., Kolli, H., Dimitrov, D.T., Federico, P., Qi, H., McCracken, G.F., Betke, M., Westbrook, J.K., Kennard, K., Kunz, T.H. 2009. Dense and sparse aggregations in complex motion: Video coupled with simulation modeling. Ecological Complexity. 7:69-75. Interpretive Summary: The capability to accurately count and track large groups of highly mobile animals, such as bats or corn earworm moths in flight, is needed for agricultural and ecological assessments of abundance and density. A framework is introduced that joins video counting and simulation modeling that can be applied to a wide range of mobile animals. An important contribution of the simulation model is that error estimates of counts are obtained to help address problems caused by animals partially or completely hidden by other animals from the perspective of two-dimensional video imaging. Two techniques for video counting are presented: the first estimates the number of bats in a video of a dense colony of Brazilian free-tailed bats (which eat corn earworm moths and other crop insect pests), and the second estimates the number of moths in a video of an agricultural area where corn earworm moths are abundant. For dense populations, an estimate of group size obtained from video images generally requires a model for error estimation because of animals hidden from view, but for sparse groups such as in our moth videos, these errors tend to be small and modeling does not appear as essential as for estimates of the dense group size. The results of this research will aid in the development of techniques that can apply video imaging to rapidly and accurately assess the group size of flying pest insects and other dispersing animals.
Technical Abstract: In censuses of aggregations composed of highly mobile animals, the link between image processing technology and simulation modeling remains relatively unexplored despite demonstrated ecological needs for abundance and density assessments. We introduce a framework that connects video censusing with simulation modeling that appears applicable to a diverse spectrum of mobile animal aggregates. An important contribution of the model is that error bounds on counts are obtained to help address problems caused by occlusions induced by video projection of three dimensional space onto a two dimensional space. Through illustrative examples, two techniques for video censusing are presented. The first provides estimates of the numbers in a video of a dense colony of Brazilian free-tailed bats (Tadarida brasiliensis), and the second provides estimates of the numbers of noctuid moths in a video of an agricultural area in south-central Texas. For dense populations, an estimate of aggregate size obtained from video images generally requires a dynamic model for error estimation because of confounding issues such as masking or occlusion whereas for sparse aggregates such in our moth videos, these errors tend to be small and modeling does not appear as essential as in the dense aggregate census estimation.