Application of ground-truth for classification and quantification of bird movements on migratory bird habitat initiative sites in southwest Louisiana: final report

Abstract

This project was initiated to assess migrating and wintering bird use of lands enrolled in the Natural Resources Conservation Service’s (NRCS) Migratory Bird Habitat Initiative (MBHI). The MBHI program was developed in response to the Deepwater Horizon oil spill in 2010, with the goal of improving/creating habitat for waterbirds affected by the spill. In collaboration with the University of Delaware (UDEL), we used weather surveillance radar data (Sieges 2014), portable marine radar data, thermal infrared images, and visual observations to assess bird use of MBHI easements. Migrating and wintering birds routinely make synchronous flights near dusk (e.g., departure during migration, feeding flights during winter). Weather radars readily detect birds at the onset of these flights and have proven to be useful remote sensing tools for assessing bird-habitat relations during migration and determining the response of wintering waterfowl to wetland restoration (e.g., Wetlands Reserve Program lands). However, ground-truthing is required to identify radar echoes to species or species group. We designed a field study to ground-truth a larger-scale, weather radar assessment of bird use of MBHI sites in southwest Louisiana. We examined seasonal bird use of MBHI fields in fall, winter, and spring of 2011-2012. To assess diurnal use, we conducted total area surveys of MBHI sites in the afternoon, collecting data on bird species composition, abundance, behavior, and habitat use. In the evenings, we quantified bird activity at the MBHI easements and described flight behavior (i.e., birds landing in, departing from, circling, or flying over the MBHI tract). Our field sampling captured the onset of evening flights and spanned the period of collection of the weather radar data analyzed. Pre- and post-dusk surveys were conducted using a portable radar system and a thermal infrared camera. Landbirds, shorebirds, and wading birds were commonly found on MBHI fields during diurnal surveys in the fall. Ducks (breeding and early migrating species) were also detected on diurnal surveys, but were less abundant than the previously mentioned taxa. Wading birds were the most abundant taxa observed during evening surveys up to 5 min before dusk when their numbers declined and duck densities increased. Ducks accounted for 64.0% of all birds detected from 0-5 min before dusk. Most ducks observed at that time were flyovers (71.4%), but circling (9.2%), departing (12.1%), and landing birds (7.4%) were also detected. In fall, the portable radar system detected two peaks in bird movement: one shortly before sunset and a second shortly after dusk. The later movement began just before dusk, peaked approximately 9 min after dusk, and concluded within 20 min after dusk. The flight headings of birds changed in relation to time from dusk. In general, the majority of targets flew towards the southwest before dusk and towards the northeast after dusk. The change in flight direction pre- and post-dusk may be related to movements dominated by migratory versus local flight. In winter, ducks, shorebirds, wading birds, and landbirds were the most abundant taxa in diurnal surveys. Geese were abundant at times, but their frequency of occurrence and densities were highly variable. The majority of ducks, shorebirds, and wading birds were observed feeding in MBHI fields. Landbirds and geese were more commonly seen resting. Overwintering ducks and geese dominated the movements near dusk (95.9% of all birds ≤ 5 min pre-dusk). Ducks were more frequently observed landing in (40.8%) and flying over (33.5%) MBHI fields while geese were mainly observed circling (54.7%) and flying over (38.9%) sites. Most of the shorebirds detected < 5 min before dusk (74.6% of all shorebirds) were departing the MBHI fields. Portable radar and thermal infrared camera data indicate that large northeastward movements of waterfowl (99.9% of birds identified to taxa) occurred after dusk (~10 min post-dusk). Most birds observed on radar during this peak were flyovers and did not use the MBHI fields (78.9%); however, birds were detected landing in (10.9%) and departing from (2.9%) MBHI fields. The post-dusk movements may have been waterfowl feeding flights that routinely occur in southwest Louisiana between roost sites in coastal marsh and foraging sites in agricultural fields to the north. After the conclusion of these movements ca. 30 min post-dusk, portable radar data showed little activity through the night until approximately 0.5 to 1.5 hr pre-dawn. Radar data within 30 min pre-dawn indicate that most birds departed MBHI fields on flight headings toward the southwest. The pre-dawn movements were likely waterfowl departing from their foraging sites and returning to roosting areas in coastal marshes to the south. Shorebirds, ducks, and wading birds were the most abundant taxa during diurnal surveys of MBHI fields in spring, and the majority of individuals were observed actively foraging rather than resting. Breeding, overwintering, and transient migrant species were all detected on MBHI fields. Near dusk, the majority of birds in flight were ducks (67.7% of all birds) that were flying over (38.2%), departing from (34.2%), or landing in (22.9%) MBHI fields. These results contrast with our winter observations when 40.8% of ducks landed in MBHI fields and 9.1% departed from fields. Portable radar and thermal camera data documented a peak in bird movements shortly after dusk, however, the peak was of lower magnitude than observed in the winter. Thermal camera data identified the birds as mostly shorebirds (57.3%) and waterfowl (40.4%). Flight headings were more variable than winter and lacked an undirectional flow. After the post-dusk movement had concluded, bird activity remained low throughout the night until approximately 30 min before dawn when a small peck in activity was observed. Flight headings during the pre-dawn were variable and multidirectional. We compared bird abundance data collected by each of our three sampling techniques (portable radar, thermal infrared camera, and direct visual observation) for the 45-min observation period immediately preceding dusk; the period when all three survey methods were used simultaneously. Abundance data from the three methods were significantly correlated at P &le; 0.05. We documented diurnal and nocturnal bird use of MBHI fields. Most observations near dusk in winter, when weather radar data were sampled, were of ducks and geese, and in spring, shorebirds and ducks. Our winter observations show large synchronous movements of waterfowl occurring near dusk. These birds were moving to the NE and feeding in agricultural fields at night. Portable radar data suggest that birds stay in these fields through the night and make return flights near dawn.