Uncovering Spatial Variation in Acoustic Environments Using Sound Mapping

Abstract

Animals select and use habitats based on environmental features relevant to their ecology and behavior. For animals that use acoustic communication, the sound environment itself may be a critical feature, yet acoustic characteristics are not commonly measured when describing habitats and as a result, how habitats vary acoustically over space and time is poorly known. Such considerations are timely, given worldwide increases in anthropogenic noise combined with rapidly accumulating evidence that noise hampers the ability of animals to detect and interpret natural sounds. Here, we used microphone arrays to record the sound environment in three terrestrial habitats (forest, prairie, and urban) under ambient conditions and during experimental noise introductions. We mapped sound pressure levels (SPLs) over spatial scales relevant to diverse taxa to explore spatial variation in acoustic habitats and to evaluate the number of microphones needed within arrays to capture this variation under both ambient and noisy conditions. Even at small spatial scales and over relatively short time spans, SPLs varied considerably, especially in forest and urban habitats, suggesting that quantifying and mapping acoustic features could improve habitat descriptions. Subset maps based on input from 4, 8, 12 and 16 microphones differed slightly (< 2 dBA/pixel) from those based on full arrays of 24 microphones under ambient conditions across habitats. Map differences were more pronounced with noise introductions, particularly in forests; maps made from only 4-microphones differed more (> 4 dBA/pixel) from full maps than the remaining subset maps, but maps with input from eight microphones resulted in smaller differences. Thus, acoustic environments varied over small spatial scales and variation could be mapped with input from 4-8 microphones. Mapping sound in different environments will improve understanding of acoustic environments and allow us to explore the influence of spatial variation in sound on animal ecology and behavior.

Fig 1. Layouts of microphone arrays used to record the sound environment.

(A) Layout of full 24-microphone array used to record the ambient environment and (B) nested microphone subset arrays (● = 4, ■ = 8, ▲ = 12, and * = 16) used during noise introductions. As examples, the 8-microphone subset array included four central microphones (■) and four corner microphones (●), and the 12-microphone array included these 8 microphones plus four more (▲). ‘N’ represents a hypothetical location for noise introduction and each ‘M’ represents one of four additional microphones surrounding the noise source. The overall spatial extent is 60 m x 60 m for the entire plot and 6 m x 6 m for internal squares.

Fig 2. Sound maps of ambient conditions across each of five sites in each of three habitats.

Maps were generated using full arrays of 24 microphones and show overall sound pressure levels (SPL, dBA). Note that the color scale is the same for all maps, but the actual SPLs reflected by the scale differ across maps and are indicated by the range of values below each map; we presented each with a unique scale to best visualize spatial variation in SPLs within a site. Maps within each habitat are ordered from the smallest to the largest range of SPLs within a given habitat type.

Fig 3. Full and subset sound maps of ambient conditions in three habitats.

Full maps were generated using 24 microphones, with subset maps generated by sequentially dropping groups of microphones from full arrays. Each map illustrates the same randomly selected time point within a habitat. Note that the range of SPLs mapped is similar (5–6 dBA), but that the absolute values of SPLs differ among maps for the three habitats.

Fig 4. Map differences of ambient conditions varied with microphone number (A) but not across habitats (B).

Mean pixel difference is the average difference in overall sound pressure levels (dBA) calculated by subtracting a subset map from the corresponding full array map and averaging overall all pixels. The asterisk denotes a significant difference as tested using Tukey HSD (P < 0.05). Error bars represent standard error.

Fig 5. Examples of sound maps illustrating overall SPLs (dBA) during noise introductions (■) in three habitats.

Maps show noise introduction within arrays without (A) and with (B) additional microphones around noise source, and noise introduction at the edge of arrays without (C) and with (D) the additional microphones. Within habitat, maps in (A) and (B) illustrate the same randomly selected locations and time points, as do (C) and (D). Arrays are the same as those shown in Fig 3. Note that the ranges and values of SPLs differ among maps.

Fig 6. Map differences during noise introductions varied with microphone number (A) and across habitats (B).

Mean pixel difference is the average difference in overall sound pressure levels (dBA) calculated by subtracting a subset map from the corresponding full array map and averaging overall all pixels. The asterisk denotes a significant difference as tested using Tukey HSD (P < 0.05). Error bars represent standard error.

Fig 7. Effect of additional microphones on differences between maps depended on location of noise introductions.

Mean pixel difference is the average difference in overall sound pressure levels (dBA) calculated by subtracting a subset map from the corresponding full array map and averaging overall all pixels. The asterisk denotes a significant difference as tested using Tukey HSD (P < 0.05). Error bars represent standard error.