Acoustic monitoring to evaluate the effect of anthropogenic noise within a park Giovanni Zambon 1 ; Andrea Potenza 2 ; Chiara Confalonieri 3 ; Alessandro Bisceglie 4 ; Claudia Cane- doli 5 ; Emilio Padoa Schioppa 6 ; Roberto Benocci 7 University of Milan Bicocca - Department of Earth and Environmental Sciences Piazza della Scienza 1, 20126 Milano, Italy

ABSTRACT The aim of this paper is to propose the use of passive acoustic monitoring (PAM) as a non- invasive method to investigate the state of communities and ecosystems. PAM operates through the study and characterization of the soundscape of an area. One of the three components of the soundscape (beside geophony and biophony) is anthrophony, which is the collection of sounds produced by human activities. This kind of sounds can have effects on natural environ- ments and natural population. In this study, recording instruments and sampling techniques have been used to acquire and collect sound data for long periods (two weeks) in a natural terrestrial ecosystem (Ticino Park) which is affected by road and rail traffic noise. The analysis conducted studied the trends of the eco-acoustic indices belonging to three measurement sites to detect the presence of characteristic trends and to evaluate the influence of the two anthro- pogenic noise sources at different distances.

1. INTRODUCTION

Eco-acoustics, the study of environmental sounds (soundscape), and its linked discipline bioacous- tics, the study of sounds produced by or affecting living organism, have grown in importance as non- invasive techniques for ecological monitoring. The term “soundscape” indicates the collection of sounds which characterize a landscape and are divided in geophony (wind, rain, waves), biophony (animal vocalizations) and anthrophony (mechanical sounds) [1-5].

1 giovanni.zambon@unimib.it

2 a.potenza@campus.unimib.it

3 c.confalonieri12@campus.unimib.it

4 alessandro.bisceglie@unimib.it

5 claudia.canedoli@unimib.it

6 emilio.padoaschioppa@unimib.it

7 roberto.benocci@unimib.it

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Studying the soundscape can help quantify the impact of anthropic noise on terrestrial and marine fauna which can, mask communication signals or cause modifications in population size, density and demography [6-9]. Its study relies on passive acoustic monitoring (PAM), which consists in placing recording devices in natural habitat [10-12], and eco-acoustic indices which analyze different sound characteristic from the recordings such as wave amplitude, frequency modulation and compare spec- tral and temporal trends.

Recently, an analysis regarding the soundscape of an urban park of Milan affected by a motorway has been conducted [13-14]. Differently to previous works [15-16], this paper focuses the study to a more natural environment such as the Ticino Park with the aim to define the response of eco-acoustic indices to the effect of biophonic and anthrophonic sounds on a long period (two weeks) and at dif- ferent distance from the vehicular infrastructures.

2. MATERIALS AND METHODS

2.1. Study site

The study was conducted at Ticino Park (45°27’ N, 8°48’ E) near Bernate Ticino, which is located in the western part of Lombardy, approximately 30 km west of Milan (Figure 1). The park experiences a temperate climate. The minimum temperature registered in the period under study was 10.0◦C, and the maximum was 29.3◦C. Sunrise and sunset vary rather little during the investigation period in our study site (two weeks). The dawn time varies from ca. 06:10 a.m. to 05:40 a.m., while the dusk time is ca. 20:30–21:20 p.m.

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 1: Study area (part of Ticino Park); the red points represent the measurement sites. This natural terrestrial ecosystem is characterized by the presence of highly impacting anthropogenic noise sources, such as the A4 motorway and the high-speed train line located at North (Figure 1) and the passage of several overflights to the nearby Malpensa airport.

2.2. Acoustic data

Recordings were collected from 26 May 2021 to 10 June 2021. For the measurements we used two SET (Soundscape Explorer (Terrestrial)) (Figure 2). The Soundscape Explorer (SET) is equipped with environmental sensors (humidity, temperature, light and atmospheric pressure), two microphones (one for low frequencies (up to 48 kHz) and one for higher frequencies (up to 192 kHz)), and two SD card slots that allow storing a maximum of 64 GB of data. In addition, the SET is fully programmable using stand-alone software, from 1 minute to 59 minutes in a 24-hour period. Different set-up options are available, such as changing the sampling rate, defining the recording schedule

using one of the two microphones, changing the time setting, changing the microphone gain, etc. Its small dimensions (20 cm x 25 cm) encourage displacement in wild as well as in urban areas. Powered by rechargeable lithium-ion batteries, autonomy is maximized by the extremely low consumption of the internal electronics. Moreover, the SET can be hung on any support using the two provided metallic brackets, such as tree branches. Aluminum black anodized hinges and level retainers assure a complete closure of the waterproof plastic containing box.

Two SET have been positioned on trees, at the height of about 4 meters (Figure 2), along a transept perpendicular to the two main anthropogenic sources (motorway and railway) (Figure 1). Choosing a height of 4 m, we have maximized the soundscape diversity.

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 2: SET closeup (left) and located on a tree at the height of about 4 meters (red circle) (right). The selected sites of measurement were three and the monitoring period was of about one week for each site.

We have programmed the instruments to record for 1 min every 5 min at a sampling rate of 48 kHz and 16-bit WAV coding form.

Because only two devices were used to record at three different sites, the instrument used in the first site (Figure 1) was moved to the middle one after one week. Therefore, we have selected for the analysis two continuous periods of about one week: the first one from 26 May 2021 (started at 13:00) to 2 June 2021 (ended at 23:54) for the first site (the nearest to the vehicoular infrastructures) and for the third site (the most distant); the second one from 3 June 2021 (started at 13:00) to 10 June 2021 (ended at 23:54) for the middle point and for the third point. Overall, we analyzed 1790 1-minute recordings for each measurement point and 240 recordings per day. For each 1-minute recording, we calculated seven eco-acoustic indices using the correspondent functions in the “Soundecology 1.3.3” package of the software R-3.6.0 [17]. The Acoustic Complexity Index (ACI) [18] and the Dynamic Spectral Centroid (DSC) [19] were calculated across the frequency range of 0.1–12 kHz and with an FFT window size of 1024. Both the Acoustic Diversity (ADI) and Acoustic Evenness (AEI) indices [20] were calculated across the frequency range of 0–12 kHz using 10 steps and a decibel threshold of − 50. Bioacoustic Index (BI) [21] was calculated across the frequency range of 2–12 kHz and with an FFT window size = 1024. Acoustic entropy index (H) [22] was calculated across the frequency range of 0–12 kHz using default parameters. Normalized Difference Soundscape Index (NDSI) [23] was calculated by setting the frequency range between 100 and 2000 Hz as an expression of “anthropic noise”, and the frequency range between 2000 and 12000 Hz as an expression of biophony, with an FFT window size of 1024.

2.3. Meteorological and physical data

Three meteorological parameters were acquired from the ARPA’s (Regional Agency for Environ- mental Protection) weather station at Magenta. The dataset of 60-minute averaged sensor readings was summarized to obtain the following daily values: hourly average air temperature (°C), hourly average rainfall (mm) and hourly average wind speed (m/s). The dawn time and the dusk time were obtained from the website www.calendariando.it [23], which provides the sunrise and sunset times for every day of the year in Bernate Ticino.

In the measurement period neither an hourly average wind speed higher than 5 m/s nor an average precipitation value higher than 2 mm/h was recorded, so it was not necessary to delete any audio recording.

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

2.4. Analysis

The analysis was performed in R 3.6.0 [17] environment. First, we have visualized the time series of the eco-acoustic indices and meteorological variables to evaluate if it was necessary to delate any recordings. Then, we compared the daily trends of the eco-acoustic indices belonging to the three measurement sites to detect the presence of characteristic trends and to study the influence of the two anthropogenic noise sources on the biophonic activity of avian species at different distances. The comparison was done on the daytime period (from 05:15 AM to 09:30 PM otherwise from a quarter of an hour before sunrise to a quarter of an hour after sunset), on the night period (from 09:30 PM to 05:15 AM otherwise from a quarter of an hour after sunset to a quarter of an hour before sunrise) and on the entire period. 3. RESULTS AND DISCUSSION

3.1 Daily and entire period profile of the eco-acoustic indices

The first analysis consisted, for each measurement site, in the comparison of the temporal profiles of the seven eco-acoustic indicators over two different time periods: a day of measurement (Figure 3) and the entire period of the campaign (Figure 4). In the graphs reported below, the three measure- ment sites are indicated as first point (the nearest to the vehicular infrastructures), middle point and third point (the farthest from the infrastructures).

ly profile Fist point, —— ADL Mie pois. —— A ADI_Dail Thied point Hour

AEI_Daily profile

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 3: Daily profiles of the eco-acoustic indices for the three measurement sites. In Figure 3 it is possible to infer a similar trend for some eco-acoustic indices in the three meas- urement points during the day. In particular, ADI, H and BI show an analogous trend, with the highest values of the indices registered in the middle site and the lowest values registered in the third site. The highest value of the middle site can be explained by considering that it is still disturbed by the infrastructures, which is weaker moving towards the farthest site, while presenting higher biophony emissions than the nearest site. Furthermore, it is possible to notice higher indices values during the daytime period caused by an increase in the singing activity of the avian community; in fact, birds sing mainly during daylight while the noise produced by the vehicular infrastructures remains almost unchanged.

H_Daily profile Hour DSC_Daily profile —$_05 First point. —— DsC_Middle point. —— OSC_Thied point BI_Daily profile 140 BI_Middle point 120 100 BL First point BLThidd pont NDSI_Daily profile nos Fest point NOSI_Middie pont —— NOSI_Théd point

As regards the remaining indices, AEI is complementary of ADI thus shows an opposite trend. ACI shows an increase during the day underling a higher modulation of sound on the third site. DSC and NDVI show a very similar trend; in the daytime the higher values belonging to the third site and the lower values are recorded in the first site while in the night period the values are equal and low in all sites. This is due to the fact that, during the day, the presence of biophonies increases with the distance from the anthropogenic sources leading to an increase in the values of these indices, while their reduced presence during the night leads the indices towards extremely low values.

ly profile ACI_Dail point ——xiFit "ACL Middle point A Tid point

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 4: Entire period profiles of the indices ADI and NDSI for the three measurement sites. Figure 4 reports the eco-acoustic indices (ACI and NDSI) profile on the entire period of measure- ment. The instrument (SET) used in the first site was moved to the middle one after one week because only two devices were used to record at three different sites. Therefore, what we have is a continuous measurement of almost two weeks on the third point (where the same instrument was left for the entire period) and two continuous measurements of one week each for the remaining two points.

Analyzing the profiles over the entire measurement period, it is possible to notice a similar trend in the three measurement points to the one observed on the single day, even if in this case the differ- ence between the first and middle site compared to those found in the third is even more marked. What emerges clearly from these graphs is the presence of a daily trend that can be found in the profiles of the various indices and which constantly repeats day after day.

ADL Thid point period profile wo Fis pont) —— 201 Middle point ADI_Entire lav 00:90:90 oo:zEzt oot 00:77:00 00:08:90 00:9¢:2T o0:2v:8t 00°80:00 00:75:90 o0:00:eT 00:90:90 Coaad oo'scst 00:72:00 00:06:90 00:9€:2T o0:2n:8T 00:80:00 00:75:90 o0:o0:et Hour

3.2 Analysis of eco-acoustic indicators between day and night period

1 pro AlDSL Entire period profile os mrpore 19:36:00

Considering that a radical difference emerged from the analysis of the time profiles of the eco- acoustic indices between the night and the day period, we calculated these two periods separately. The following figures (Figure 5-6-7) show the results of this analysis. Specifically, they report the boxplots for some indices (ADI, DSC and NDSI) for the three sites on three different periods: entire measurement period, night and day. Notice that the recording of the third site (two weeks) was divided in two periods corresponding to the length of the recordings made at the first and middle site (one week).

lot of ADI Boxpl

7 H T 3 Fsipoit ——MdepetTrd pot st ook) Fistpareight Mie portage “dpe second week) oe Mee pot ay “Wid pot secon woe) dy

Entire pe- riod

Night

Day

Third point (second

Third point (first

First point

Third point (first

week)

week) night

day

week) day

Figure 5: Boxplot of ADI for the three sites on three different periods: 24h, Night and Day.

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

moc eee | se

— : : H i i | i I = — <= Fsipoit Ido port Thedpat ist week) “dpe second week) oe

Entire pe

Night

Day

Third point (first

Figure 6: Boxplot of DSC for the three sites on three different periods: 24h, Night and Day.

Entire pe-

Night

Day

Boxplot of se

Figure 7: Boxplot of NDSI for the three sites on three different periods: 24h, Night and Day. What emerges from these graphs is how the nocturnal values of the eco-acoustic indices turn out to be lower than their diurnal values in each site. Moreover, during the night there is also a lower variability in the values (width of the boxplot) especially for DSC (Figure 6) and NDSI (Figure 7) which show very similar values between the sites. These considerations are due to a significant re- duction of the biophonic sources recorded during the night and a light reduction of traffic noise.

4. CONCLUSIONS

In this paper, we assessed the potential of eco-acoustic indices to study a natural bush on a two- week period, by analyzing their response to different anthropogenic noises and biophonic sounds throughout at three sites located at increasing distance from the vehicular infrastructures. We found out that the set of eco-acoustic indices used allows to detect the presence of a daily trend that corre- sponds to the increasing contribution of biophonic sources, such as birds, and to evaluate the variation of anthrophony and biophony along a transect. To validate these results, an aural survey on the re- cordings will be necessary. Additionally, more studies are necessary to fully understand the applica- bility and the true capacity of these indices in the analysis of the quality of a natural habitat.

5. REFERENCES

1. Farina, A. Soundscape Ecology – Principles, Patterns Methods and Application (2014). 2. Gage, S. H. & Axel, A.C. Visualization of temporal change in soundscape power of a Michigan

lake habitat over a 4-year period. Ecological Informatics , 21 , 100-109 (2014). 3. Krause, B. Bioacoustics, habitat ambience in ecological balance. Whole Earth Review . 57 , 14-18

(1987). 4. Mullet, T. C., Gage, S. H., Morton, J. M. & Huettmann, F. Temporal and spatial variation of a

winter soundscape in south-central Alaska. Landscape Ecology 31 , 1117-1137 (2016). 5. Pijanowski, B. C., Farina, A., Gage, S. H., Dumyahn, S. L. & Krause, B. L. What is soundscape

ecology? An introduction and overview of an emerging new science. Landscape Ecology 26 , 1213-1232 (2011). 6. Lohr, B., Wright, T. F. & Dooling, R. J. Detection and discrimination of natural calls in masking

noise by birds: estimating the active space of a signal. Animal Behaviour 65 , 763-777 (2003). 7. Bee M. A. & Swanson E. M. Auditory masking of anuran advertisement calls by road traffic

noise. Animal Behaviour 74 , 1765 – 1776 (2007). 8. Bayne, E. M., Habib, L. & Boutin, S. Impacts of chronic anthropogenic noise from energy-sector

activity on abundance of songbirds in the boreal forest. Conservation Biology 22(5) , 1186-1193 (2008). 9. Proppe, D. S., Sturdy, C. B. & St. Clair, C. C. Anthropogenic noise decreases urban songbird

diversity and may contribute to homogenization. Global Change Biology 19(4) , 1075-1084 (2013). 10. Ulloa, J. S., Gasc, A., Gaucher, P., Aubin, T., Réjou-Méchain, M. & Sueur, J. Screening large

audio datasets to determine the time and space distribution of Screaming Piha birds in a tropical forest. Ecological Informatics 31 , 91–99 (2016). 11. Fuller, S., Axel, A. C., Tucker, D. & Gage, S. H. Connecting soundscape to landscape: Which

acoustic index best describes landscape configuration? Ecological Indicators 58 , 207-215 (2015). 12. Gasc, A., Anso, J., Sueur, J., Jourdan, H. & Desutter-Grandcolas, L. Cricket calling communities

as an indicator of the invasive ant Wasmannia auropunctata in an insular biodiversity hotspot. Biological Invasions 20(5) , 1099–1111 (2018). 13. Benocci, R., Roman, H.E., Bisceglie, A., Angelini, F., Brambilla, G. & Zambon, G. Eco-acoustic

assessment of an urban park by statistical analysis. Sustainability 13 , 7857 (2021). 14. Benocci, R., Potenza, A., Roman, H. E., Bisceglie, A. & Zambon, G., Mapping of the acoustic

environment at an urban park in the city area of Milan, Italy, using very low-cost sensors, Ac- cepted for publication on Sensors 2022. 15. Benocci, R., Roman, H. E., Bisceglie, A., Angelini, F., Brambilla, G. & Zambon, G., Auto-cor-

relations and long time memory of environment sound: The case of an Urban Park in the city of Milan, Italy. Ecological Indicators 134 ,108492 (2022). 16. Benocci, R., Brambilla, G., Bisceglie & A., Zambon, G. Eco-acoustic indices to evaluate sound-

scape degradation due to human intrusion. Sustainability 12 , 10455 (2020). 17. R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Sta-

tistical Computing (2017). 18. Pieretti, N., et al. A new methodology to infer the singing activity of an avian community: The

Acoustic Complexity Index (ACI). Ecological Indicators 11 , 868–873 (2011). 19. Yang W, Kang J. Soundscape and Sound Preferences in Urban Squares. Journal of Urban Design

10 , 61-80 (2005).

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

20. Villanueva-Rivera, L. J., Pijanowski, B. C., Doucette, J. & Pekin B. A primer of acoustic analysis

for landscape ecologists. Landscape Ecology 26 , 1233-1246, (2011). 21. Kasten, E.P., Gage, S.H., Fox, J. & Joo W. The remote environmental assessment laboratory’s

acoustic library: An archive for studying soundscape ecology. Ecological Informatics 12 , 50-67 (2012). 22. Sueur, J., Pavoine, S., Amerlynck, O. & Duvail S. Rapid acoustic survey for biodiversity ap-

praisal. PLoS ONE 3 , p. e4065 (2008). 23. Farina, A., Lattanzi, E., Piccioli, L. & Pieretti, N., The Soundscapemeter User Manual (2012). 24. Website: www.arpalombardia.it

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW