2.2. Data collection and recording hardware

This paper contains two acoustic datasets, one collected in 2016 and the other in 2018. In 2016, we conducted sampling to compare avian vocalizations in mid‐aged and mature forest stands, replicated across Hubbard Brook and Jeffers Brook watersheds. In 2018, sampling was concentrated in Hubbard Brook forest and designed to provide high resolution within one forest area, allowing for more detailed examination of spatial patterning in vocalization activity. In both years, recorders were activated each morning for a 10‐minute period spanning 06:20–06:30 local time (UTC‐4). Depending on the date, the recordings started at 55–75 min after sunrise. The 10‐min interval for recording bouts parallels a common point count duration (Buskirk & McDonald, 1995) and was chosen to be long enough to capture most species vocalizing at that site on that morning, but short enough that we could still annotate many different mornings and compare inter‐ and intraspecific patterning of vocalizations among days (Tobias et al., 2014). For annotation, we chose a sample size of 20 dates per year as being both sufficient and manageable. We selected the dates within years such that they were distributed throughout the period of available recordings, but with longer intervals between dates later in the season when there were generally fewer vocalizations. Dates for annotation were chosen in advance of examining the sound recordings and so were not biased with respect to vocalization activity or ambient sound levels.

We used Olympus DS‐40 recorders (Olympus, Center Valley, PA, USA) deployed in plastic boxes and connected to their original microphone by a 1‐m extender cable. Each microphone was placed at a height of 2 m and was suspended below a fabric rain shield 25 cm in diameter. The recorders generated MP3 files with a sampling rate of 44.1 kHz on the “high‐quality” setting, with the manufacturer's maximum microphone sensitivity, no frequency filter, and no variable control voice actuator. The MP3 files were converted to 16 bit WAV files using Switch Plus converter (NCH software, Canberra, Australia) so that recordings could be digitally analyzed and manipulated. The compressed MP3 format discards some high‐frequency information, resulting in smaller files, but lower acoustic resolution, particularly at frequencies higher than those used by most bird species. These missing data are not recovered with the conversion to WAV format.

In 2016, three recorders were in the mature forest and two were in mid‐aged forest in both the Hubbard and Jeffers Brook forests. Within watersheds, recorders were separated by 50–200 m, minimizing the chances that an individual songbird was captured by multiple recording devices. Recorders were activated for 10 min each morning from 09 June to 21 July. In 2016, the estimated dates of leaf out of canopy trees and median initiation of first clutch by a representative locally breeding bird, the Black‐throated Blue Warbler (Setophaga caerulescens), were approximately 12 May and 31 May, respectively (Lany et al., 2016). We analyzed audio data for the following 20 dates in 2016: 09‐Jun, 11‐Jun, 15‐Jun, 16‐Jun, 18‐Jun, 20‐Jun, 24‐Jun, 26‐Jun, 28‐Jun, 30‐Jun, 01‐Jul, 03‐Jul, 04‐Jul, 06‐Jul, 07‐Jul, 08‐Jul, 14‐Jul, 16‐Jul, 19‐Jul, and 21‐Jul.

In 2018, we sampled vocalizations only within Hubbard Brook Experimental Forest. The 10 recorders were distributed across an area that has been the focus of long‐term studies of breeding songbirds (Holmes et al., 1986; Holmes, 2011; Rodenhouse & Holmes, 1992; Townsend et al., 2013). Distances between recorders ranged from 176 to 1729 m (Figure ​(Figure1),1), allowing us to test for correlations in behavior at different distances. Recorders were activated for 10 min each morning from 20 May to 15 August. The estimated dates of leaf‐out and median initiation of first clutch by Black‐throated Blue Warblers in 2018 were 10 May and 30 May. We analyzed audio data for the following 21 dates in 2018: 13‐May, 15‐May, 17‐May, 20‐May, 23‐May, 26‐May, 29‐May, 01‐Jun, 04‐Jun, 07‐Jun, 10‐Jun, 13‐Jun, 16‐Jun, 19‐Jun, 22‐Jun, 25‐Jun, 28‐Jun, 01‐Jul, 04‐Jul, 07‐Jul, 10‐Jul.

All 410 10‐minute recordings (200 from 2016 and 210 from 2018), and their associated metadata are available at Symes et al. (2021).