67--Intent to Award - Camera Trap Systems

ing camera deployment, all images will be processed using AI software trained on images derived exclusively from our camera trap model. Once images have been classified density and productivity estimates can be derived and compared to those derived from existing monitoring frameworks to determine how accurately current monitoring programs reflect wild turkey population dynamics. Therefore, it intends to make a direct award to Reconyx, Inc. on a sole source basis under the authority of FAR Overhaul (RFO) 12.102(a). This is not a request for competitive quotations or proposals but rather an intent to sole source notification. Questions or concerns may be addressed to the Contract Specialist, Jonathan Wyatt jonathan_wyatt@ios.doi.gov no later than 8/28/2026. The U.S. Geological Survey (USGS) considers employee signatures to be Personal Protected Information, and therefore all employee signatures have been redacted. By posting this document, the Contracting Officer attests all appropriate signatures have been obtained. 1. Contracting Agency and Activity The Department of the Interior, United States Geological Survey, Ecosystem Mission Area, Cooperative Research Units Program 2. Nature of Action Being Approved. The U.S. Geological Survey (Cooperative Research Units, Indiana Cooperative Fish and Wildlife Research Unit) requests award of a non-competitive procurement to Reconyx. The delivery date is 60 days from the receipt of purchase order. 3. Description of Supplies or Services. Procurement of materials required for 50 camera traps for monitoring system for wildlife abundance/density estimation. This consists of 50 imaging set ups each containing a camera trap, camera security cable and combination lock, and memory card. All these materials will be purchased from the camera trap company Reconyx. 4. Statutory Authority The proposed action may be awarded on a sole source basis under the authority of FAR Overhaul (RFO) 12.102(a). 5. Rationale Supporting Use of Citation in No. 5. For reasons of data integration with existing camera trapping efforts, this project must continue buying and using Reconyx camera systems. This is because different trail camera models all have unique photo trigger and recording characteristics. Because of this, using multiple trail camera models within a single study is strongly discouraged as doing so can negatively influence results and inferences; it is the best scientific practice to maintain trail camera models during a study. Modern statistical approaches, such as space-to-event (STE) and time-to-event (TTE) models, for estimating density using data from trail cameras are highly sensitive to sampling effort (e.g. McMurray et al. 2022,). Because of the large volume of photos collected in studies using camera traps, image classification (identification of animals in photos) is often accomplished using statistical software or AI algorithms (Bothmann et al. 2023, https://doi.org/10.1016/j.ecoinf.2023.102231). Because individual trail camera models have unique optics, in addition to unique infrared sensors, this means that the images produced from different camera models cannot be processed and interpreted in the same way. Because one of our objectives is to use these cameras in combination with existing trail camera arrays and data collection procedures, maintaining continuity in trail camera brand is critical to ensure that photo classification is consistent. 6. The Efforts to Identify Additional Sources Including the Market Research Conducted. Market research consisted of conducting internet searches and reviewing scientific literature concerning camera traps. As discussed in several of the papers cited in section 6 (Apps and McNutt 2018; Kelly and Holub 2008; Palencia et al. 2022), different trail camera models result in substantially different data in terms of infrared triggers, to say nothing of differences in image quality. To date, no widely accepted alternatives to the use of consistent camera models within a study have been identified. Moreover, because the difference among camera models manifests in a myriad of ways (trigger distance, infrared sensor sensitivity, field-of-view, image quality, etc.), it seems unlikely that a unified approach to dealing with these differences is feasible. Literature searches (Google Scholar, ScienceDirect) failed to yield any published examples of unified approaches for dealing with discontinuity in data among camera models. Although other camera models have reported manufacturer specifications suggesting they are like Reconyx, field tests of these models continue to yield substantial differences among models when directly compared (Swearingen et al. 2023, https://doi.org/10.1002/wsb.1422). 7. Future Plans to Permit Competition. Once we have completed our project and provided wildlife density estimates to partner agencies, at the scale of individual management areas, we can work towards optimizing our sampling framework to provide density estimates at even larger spatial scales (e.g. county, management zone) or possibly compare our results to those derived from different camera models, however these are both currently beyond the immediate scope of this project.