Assessing roost disturbance of straw-coloured fruit bats (Eidolon helvum) through tri-axial acceleration.

The disturbance of wildlife by humans is a worldwide phenomenon that contributes to the loss of biodiversity. It can impact animals' behaviour and physiology, and this can lead to changes in species distribution and richness. Wildlife disturbance has mostly been assessed through direct observation. However, advances in bio-logging provide a new range of sensors that may allow measuring disturbance of animals with high precision and remotely, and reducing the effects of human observers. We used tri-axial accelerometers to identify daytime flights of roosting straw-coloured fruit bats (Eidolon helvum), which were used as a proxy for roost disturbance. This bat species roosts on trees in large numbers (often reaching hundreds of thousands of animals), making them highly vulnerable to disturbance. We captured and tagged 46 straw-coloured fruit bats with dataloggers, containing a global positioning system (GPS) and an accelerometer, in five roosts in Ghana, Burkina Faso and Zambia. Daytime roost flights were identified from accelerometer signatures and modelled against our activity in the roosts during the days of trapping, as a predictor of roost disturbance, and natural stressors (solar irradiance, precipitation and wind speed). We found that daytime roost flight probability increased during days of trapping and with increasing solar irradiance (which may reflect the search for shade to prevent overheating). Our results validate the use of accelerometers to measure roost disturbance of straw-coloured fruit bats and suggest that these devices may be very useful in conservation monitoring programs for large fruit bat species.

Introduction

Wildlife disturbance from human activities is a global threat contributing to the loss of biodiversity [1]. This threat has spread to remote natural regions, and is a common problem in protected areas for wildlife conservation [2, 3]. Disturbance may lead to changes in animals’ activity patterns [4, 5], energy expenditure [6, 7], physiological parameters [7, 8], foraging behaviour [9, 10], breeding success [11] and roosting behaviour [12, 13]. Ultimately, it can drive changes in species distributions and richness [1, 14]. However, assessing the impacts of disturbance on wildlife is a challenging task, as impacts vary across species and contexts [14]. Sophisticated methods have been employed, such as measuring changes in stress hormones, cardiac response and immunocompetence [15]. However, the direct observation of changes in behaviour is still the most prevalent approach described in the literature [16].

The use of automated methods in behavioural studies of wildlife disturbance, such as infrared motion detectors [17], radio-telemetry [7, 10, 18] and global positioning system (GPS) tracking [19], has become more common in recent years. These methods provide large amounts of accurate data and reduce the influence of the observer on the behaviour of the target animals [20], although biologging methods may cause some disturbance to the animals [21]. Further innovative applications are expected in the near future from advances in animal tracking technology that make available a range of new sensors to measure behavioural parameters [22, 23]. Tri-axial accelerometers, in particular, are present in most modern tracking devices, allowing precise measurement of animals’ body motion, from which different behaviours can be discriminated [23]. However, to our knowledge, these sensors have never been used to measure animal disturbance.

The straw-coloured fruit bat (Eidolon helvum) is a large Old World fruit bat species (Pteropodidae) that occurs across sub-Saharan Africa [24]. It feeds upon a large variety of fruits and flowers, often moving tens of kilometres between the roost and the foraging areas on a daily basis [25]. These features contribute to making this species a keystone seed disperser in Africa, and critical for maintaining vegetation dynamics in fragmented tropical forests [26]. Despite its high ecological relevance, populations of straw-coloured fruit bat are declining across its range, with hunting probably being the main cause [27-29]. This species roosts on trees in large numbers, reaching hundreds of thousands of individuals [13], with an estimated five to ten million bats in the largest known colony at Kasanka National Park (Zambia) [30]. Very often roosts are located in urban areas [13, 31], making them especially vulnerable to interactions with humans [28, 29, 32].

New and exact information of disturbance levels in roosts of this species is crucial to inform conservation actions and protection regulations. This study aimed to examine the potential of tri-axial accelerometery to monitor disturbance of roosting bats. For that purpose, we tagged 46 straw-coloured fruit bats with GPS-accelerometer dataloggers in five roosts in Ghana, Burkina Faso and Zambia. We identified daytime flights in roosts from accelerometer signatures and use them as a proxy of roost disturbance. Flight is an extremely energy-consuming activity for bats, demanding up to 34 times the basal metabolic rate [33], thus we expect them to avoid flying during the roosting period. The occurrence of daytime roost flights was then modelled against our activity in the roosts during the days of trapping, as a predictor of roost disturbance, and natural stressors that may influence flight probability during the roosting period. We predicted that: (1) our presence in the roosts will lead to higher daytime flight probability; (2) bats will be more prone to fly in days of high solar irradiance to find shaded perches and avoid heat stress [34, 35]; (3) bats will fly less during daytime when it rains, as rain increases flight energy costs [36]; (4) bats will fly less at higher wind speeds, which might increase flight energy costs [37].

Materials and methods

Ethics statement

Fieldwork, including bat handling and tagging, was approved by Ghana Wildlife Division of the Forestry Commission (permit FCWD/GH-01), Zambia Wildlife Authority (permits ZAWA 421902 and ZAWA 547649), and the Director of the Parc Urbain Bangr-Weoogo (Mr Moustapha Sarr).

Study areas

Straw-coloured fruit bats were captured and tagged with dataloggers in five roosts at four different areas across continental Africa: Accra and Kibi in Ghana, Ouagadougou in Burkina Faso, and Kasanka National Park in Zambia (Fig 1).

Fig 1

Location of straw-coloured fruit bat roosts studied at Burkina Faso (top left), Ghana (bottom left), and Zambia (bottom right). Black dots represent roosts where bats were captured and tagged with tracking devices and white dots represent other roosts used by tagged bats. Background map, provided by GADM (https://gadm.org), is licensed for academic use.

In Accra, bats were captured in a roost near the city centre, in the area of the 37 Military Hospital (5.586°N, 0.185°W). Accra is one of the largest cities in West Africa with almost two million people. The area around the city still holds remnants of coastal savanna forest, but is dominated by introduced tree species. The colony varies in size across seasons: it peaks during the dry season, reaching 100,000–250,000 individuals, and only a few thousand individuals are present during the wet season [25, 31].

In Kibi, bats were captured in two neighbouring roosts (ca. 19 km apart), one in Old Tafo (6.235°N, 0.394°W) and the other at Kibi Palace (6.165°N, 0.555°W). Kibi is a rural area with ca. 168,000 people, covered by moist semi-deciduous forests, farmlands and degraded forests [31]. The colonies peak during the dry season with a total of 40,000–50,000 individuals and the numbers decline during the wet season down to a low of a few hundred individuals [31].

In Ouagadougou bats were captured in a roost located in an urban park near the city centre (Parc Urbain Bangr Weogo, 12.398°N, 1.489°W). This city has ca. 1.5 million people and is located in the savanna biome [38]. During monthly counts undertaken in 2013 and 2014, this colony peaked during the wet season with 70,000–125,000 individuals while the roost was vacated during the dry season [38].

In Kasanka National Park, bats were captured in the largest colony known for this species, with an estimated peak of ten million individuals [30]. The park covers an area of 420 km2, dominated by Miombo forests [30], and the roost site is located in a patch of Mushitu swamp forest near the Fibwe Campsite (12.587°S, 30.242°E). In this roost, bats are present only from October to December (wet season) [30]. The region has a low population density (14 people per km2) with about 85% of the population living in rural areas.

Bat capture and tracking

We tagged bats between 2009 and 2014 during different years for each study area and both wet and dry seasons in Kibi and Accra (Table 1). We netted bats in the morning (3:00 to 06:00), as they returned from foraging. We weighted the captured bats, determined their age and sex and measured the length of the forearm. Dataloggers (20–24 g e-obs GmbH, Munich, Germany) were fitted only on large individuals (239 to 321 g) to minimize effects of the extra load on their behaviour. Most tagged bats were adult males (43 individuals), but we also tagged two young males and one adult female (S1 Table of S1 File). We attached dataloggers with glue (Sauer Hautkleber, Manfred Sauer GmbH) to the back of the bat (for 12 individuals) or with a neck collar made of goat leather and closed with degradable suture thread (for 34 individuals, S1 Table of S1 File, [25, 31, 39]). The weight of the datalogger and collar (when used) ranged from 6.9 to 10.5% of the bats’ body mass (mean: 8.5%). Dataloggers recorded GPS locations only during the night (18:00 to 6:00, at least every 30 min), but tri-axial acceleration was recorded around-the-clock in bursts of 13 or 14s per min at 20 or 18.74 Hz depending on the logger generation (S1 Table of S1 File). Data were retrieved using a base station connected to a directional high-gain antenna. Further details on field procedures and the tracking devices can be found in earlier studies that used data from the same bats [25, 26, 31, 37, 40].

Table 1

Summary of the periods of data collection and the number of bats tracked in each study area.

Area	Year	Season	Dates of collection	Number of bats
Accra	2009	Wet	26/08–31/08	6
	2011	Dry	02/02–09/02	4
Kibi Palace	2011	Wet	27/08–31/08	2
Kibi Old Tafo	2012	Wet	28/08–16/09	4
	2013	Dry	25/01–01/02	3
	2013	Wet	20/09–24/09	1
Ouagadougou	2013	Wet	19/08–31/08	4
	2014	Wet	17/06–24/06	6
Kasanka	2013	Wet	04/12–11/12	3
	2014	Wet	29/11–08/11	13

Roost disturbance estimation

We used flight events during daytime (7:00 to 17:00) as a proxy of roost disturbance. Daytime roost flights were detected from acceleration readings with high variation in heave compared with surge and sway (Fig 2). During flight, the body of the bat shows regular vertical oscillation of high amplitude and lower variation on the lateral and longitudinal planes (Fig 2, [23]). Specifically, acceleration bursts were classified as “flying” if they matched the following criteria: (1) the mean of heave amplitudes calculated for each second was higher than 10 m/s2, this value being higher than the corresponding values of surge and sway; (2) the mean of heave values was lower than those of surge and sway. Criterion (1) ensured that the datalogger had oscillation of high amplitude in the vertical axis, but not as much in the other axes (see Fig 2 top right plot). Criterion (2) ensured that the datalogger was upright in case the bat was flying. An upside-down placement of the datalogger would increase heave values by 19.6 m/s2 (i.e. 2 g units), making them higher than those of surge and sway (see Fig 2 bottom right plot as an example). We also considered as “flying” the acceleration bursts that matched the above criteria only at their beginning or ending sections, which we interpreted as landing or departure events. Cases of dubious oscillation patterns, suggesting that the bat was flying with the datalogger wrongly positioned were excluded from analyses. We applied classification criteria to the data with an R [41] routine but we validated visually all acceleration signatures classified as “flying” and those with high oscillation of any axes that were classified as “not flying”. We also applied this classification to a subsample of data that included GPS locations (i.e. collected after 18:00) to confirm that bats with acceleration signatures classified as “flying” were indeed moving. Although flight behaviour was classified every minute, we aggregated classifications each day because of the small number of bursts classified as “flying”. Thus, the variable used in the analysis was binary, representing the occurrence of daytime roost flights for each animal in each day of tracking. We excluded data from bats that moved more than 500 m during the day (identified by comparing the morning and evening GPS fixes), as we could not accurately define which roost they spent the day. We also excluded data from the first day of tracking for each animal, as we expected its behaviour to be affected by the recent capture and handling.

Fig 2

Representation of tri-axial accelerometer attached to a bat and respective axes (z–heave, x–surge, y–sway) and acceleration signatures of different behaviours.

Illustration by Sara Gomes based on a photograph of Mark Carwardine.

Predictors of roost disturbance and natural stressors

We tested the effects of our presence in the roosts during the days of trapping and a set of environmental variables on the probably of bats to fly during the day. Although there was no time overlap between the data used for analyses and our visits to the roosts (i.e. we left one hour before data were collected for analyses), we assumed that our presence had a lasting disturbance effect. Bats were captured with mist nets set at the level of the canopy, which disturbed animals that were roosting in the nearby trees, and we expected their escape flights to spread throughout the roost affecting bats tagged in previous trapping sessions. We also assumed that solar irradiance, precipitation and wind speed could potentially influence daytime flight behaviour of bats in the roosts based on earlier studies [34-37]. These variables were obtained from open access weather databases (http://www.sasscalweathernet.org for Zambia roosts and https://globalweather.tamu.edu for all the others) with a temporal resolution of one day. For roosts located close together, the data was retrieved from the same weather stations. This was the case of smaller roosts located around the main roosts where bats were captured (white dots in Fig 1), and also for the two main roosts located in Kibi (Fig 1). We did not include temperature as predictor in our models because daily mean values were not available for all study areas.

Modelling procedures

We evaluated the effects of our presence in the roosts during trapping days and weather variables on the probability of bats to undertake diurnal flights at the roost with two Generalized Linear Mixed Models (GLMMs). The first included the full data set, the second excluded the data from trapping days. For both models, the occurrence of daytime roost flights was included as the dependent variable, and individual ID and roost ID were included as random intercept factors. The first model included trapping day, solar irradiance, precipitation and wind speed as fixed effects. Trapping day was binary (trapping days vs regular days) and the remaining variables were continuous. The second model used all variables but trapping day as fixed effects. Models were fitted with the function glmer of the R-package lme4 [42]. Marginal and conditional R squared were calculated with the function r.squaredGLMM of the MuMIn R-package [43]. Temporal autocorrelations of model residuals were generally low, not requiring further corrections (S1 Fig of S1 File).

Data accessibility

The full tracking dataset is available at the Movebank Data Repository [44].

Results

We retrieved daytime flight data from 46 individuals during one to seven days, providing 167 observations (S1 Table of S1 File). Among these, 129 were recorded at roosts where captures took place and 38 at other roosts. Daytime roost flights were identified from acceleration data in 24 cases.

The model including the full dataset showed significant effects of our presence in the roosts during trapping days and solar irradiation on the probability of bats to exhibit diurnal flights in the roost (Table 2). During the days of trapping bats were more likely to fly in the roost than on regular days (Fig 3, Table 2). Daytime roost flight probability also increased with solar irradiance, with a more pronounced effect when solar irradiance exceeded 20 MJ/m2 (Fig 3, Table 2). The remaining variables did not affect the diurnal flight of bats at the roost (Table 2).

Table 2

Summary of binomial GLMMs testing the effects of environmental variables on the probability of straw-coloured fruit bats to fly at their roosts during the day.

The response variable was assigned as 1 for the days when the bats flew in the roost and 0 otherwise. The first model included days when we trapped bats with mist nets in the roosts, therefore we included trapping day as a binary model predictor (trapping days vs regular days). Both models included individual ID and roost ID as random intercept factors. Marginal and conditional R2 were calculated with the function r.squaredGLMM of the MuMIn R-package [43]. Significant relationships are shown in bold and are plotted in Fig 2. Units of parameter range: Solar irradiance—MJ/m2; Precipitation—mm/day; Wind speed—m/s.

Model	Parameter	Range	Estimate	SE	Z	P-value	R2 cond./marg
With trapping days	Intercept	-	-7.164	2.005	-3.57	>0.001	0.192/0.175
	Trapping day	0–1	1.634	0.543	3.01	0.003
	Solar irradiance	2.5–30.7	0.236	0.089	2.66	0.008
	Precipitation	0.0–87.0	0.044	0.041	1.08	0.280
	Wind speed	0.5–4.0	-0.335	0.368	-0.91	0.363
Without trapping days	Intercept	-	-6.008	1.831	-3.28	0.001	0.06/0.06
	Solar irradiance	2.5–30.7	0.175	0.089	1.96	0.050
	Precipitation	0.0–87.0	0.034	0.042	0.82	0.410
	Wind speed	0.5–4.0	-0.144	0.429	-0.34	0.738

Fig 3

Model partial effects of our presence in the roosts during trapping days and solar irradiation on the probability of bats to undertake diurnal flights at the roost.

The model is a binomial GLMM that also includes wind speed and precipitation as predictors, and individual ID and roost ID as random intercept factors (see Table 2). Error bars (left plot) and shading areas (middle and right plots) represent 95% confidence intervals.

When excluding the data recorded during the days of trapping, only solar irradiation showed a significant effect on daytime roost flight probability, with an increasing pattern similar to the first model (Table 2). The dataset used in this model contained 131 observations, 36 less than the first model.

Discussion

Our study reports and validates the use of tri-axial accelerometery as a novel approach to monitor roost disturbance of large bats. Although animal tracking devices allow precise recording of animal behaviour and reduce the influence of human observers, they have rarely been used to monitor animal disturbance [but see 7, 10, 12, 18, 19]. Among sensors in tracking devices, tri-axial accelerators are particularly effective because they record behavioural data at high frequencies and demand low battery power to operate. This contrasts with the recording of GPS fixes that require higher battery power, thus delivering coarser measurements of animal behaviour.

We showed that diurnal roost flights were more likely to happen during the days we trapped bats in the roosts (Fig 3). This result validates the use of tri-axial accelerometery to monitor bat roost disturbance by humans as our presence in the roosts during trapping days was an unequivocal cause of disturbance. Hunting may have a comparable effect in the roosts as hunters use firearms and nets to reach bats roosting high in the trees [13, 29]. Besides direct mortality, these methods cause an overall disturbance of the roost. Therefore, our method has high potential to monitor hunting in bat roosts, an activity that is assumed to cause population declines of straw-coloured fruit bats [27-29].

Among the remaining factors tested as model predictors, only solar irradiance showed a significant effect. Bats were more likely to fly within the roost during the day with increasing solar irradiance, particularly when this variable exceeded 20 MJ/m2 (Fig 3). This may reflect bats seeking shade to prevent overheating, particularly on very hot days. This thermoregulatory behaviour has been described in Australian flying fox species (Pteropus alecto and Pteropus poliocephalus) when exposed to high temperatures [35]. We had also expected bats to fly less within the roost during days with precipitation, as rainfall increases flight energy costs [36]. However, events of significant precipitation were relatively rare in our sample, only in 7% of the sampling days it rained more than 10 mm. This likely prevented the identification of an effect of precipitation in our models. We also did not find an effect of wind on the probability of diurnal flights at the roost. This shows that, against our expectations, bats did not avoid flying on windy days, at least within the range of wind speed observed during data collection (0.5 to 4 m/s).

The model that excluded data recorded during the trapping days showed a poorer fit than the previous (Table 2). In this model, solar irradiance had a significant relationship with daytime flight probability at the roost, which showed a similar pattern to that of the first model. The weakness of this model was expected given the considerable reduction in sample size (36 observations less than the original dataset).

We must emphasize that our models did not explore all factors that could contribute to the disturbance of roosting bats. Large bat colonies are likely to attract non-human predators [45], but we were unable to assess the significance of this potential disturbance factor. We also did not evaluate hunting as a factor of disturbance, although this is assumed to be one of the main causes of population declines in straw-coloured fruit bats [27-29]. The exclusion of these potentially important factors may have influenced the precision of our models (Table 2).

Although straw-coloured fruit bats are still relatively abundant across their distribution range, their role as a seed disperser may be seriously impacted by ongoing population declines [27, 29], with consequences for the maintenance and regeneration of tropical forests in Africa [26]. The fact that this species aggregates in large roosts often located in urban areas makes it particularly vulnerable to human pressures, thus the monitoring of detrimental factors in these roosts is of utmost importance. We believe that the method presented here can be an effective solution in conservation monitoring programs of large fruit bats, particularly to monitor unprotected roosts located in remote areas. It can also potentially be used for poacher detection, perhaps combined with motion capture sensors and other methods already being used [46]. Tracking devices have evolved considerably since our data collection. Most are now solar charged, extending their lifespan, and can send data remotely (by GSM or Satellite), which reduces fieldwork effort and prevents the loss of data. In addition, their costs have reduced considerably, making them accessible to conservation projects with relatively modest budgets. Thus, the method described here can certainly be implemented at better cost-efficiency in the current days and in the near future.

(PDF)

Click here for additional data file.

18 Aug 2020

PONE-D-20-15313

Assessing roost disturbance of straw-coloured fruit bats (Eidolon helvum) through tri-axial acceleration

PLOS ONE

Dear Dr. Santos,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Oct 02 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Christian Vincenot, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments:

All the reviewers agree that the study is sound and interesting. Please address all the minor suggestions and also see particularly to the insightful comment by reviewers 2 and 3 on the statistical analysis.

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that Figure 1 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

2.1.    You may seek permission from the original copyright holder of Figure 1 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

2.2.    If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

Reviewer #3: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: General comments

Anthropogenic disturbance impacts on biodiversity are of concern globally. Here the authors assessed roost disturbance of straw-coloured fruit bats (Eidolon helvum) through a novel technique of tri-axial acceleration. They have assumed that all flights during the daytime are a consequence of human disturbance. My only concern is how much bat flight occurs during the daytime at roosts because of interactions between bat individuals? Although you used daytime flights as a proxy for disturbance were you able to validate any of these? The manuscript is presented well.

Specific comments

Title:

Fine.

Abstract:

Generally fine.

L32 maybe use the common name instead of just ‘bats’.

L32 put GPS in full.

Introduction:

Good synthesis of the relevant literature.

L55 put GPS in full.

Methods:

Generally well presented.

L96 Does your University/ Institute not require you to have animal ethics permission/ clearance from them for your proposed study?

L133 delete hours.

L135 perhaps replace ‘deployed’ with ‘fitted’.

L137 insert type of glue in parentheses after ‘glue’.

L138 insert ‘neck’ before ‘collar’.

L141 delete hours.

L142 abbreviate ‘minutes’.

L143 data is a plural so change ‘was’ to ‘were'.

L151 delete hours.

L153 change to ‘compared with’.

L162 change to ‘analyses’.

L166 delete hours.

L178 reword to ‘of bats flying during the daytime’.

L179, 180 change to ‘analyses’.

L183 replace ‘propagate’ with ‘spread’.

L192 change to ‘analyses’.

L198 replace ‘due to’ with ‘because of’.

Results:

Generally fine.

L227 insert ‘a’ before ‘model’.

Discussion:

Good synthesis.

L267 insert Latin name.

References:

I have not checked if Journal format was followed.

Reviewer #2: Comments to the authors

I hereby provide comments for the manuscript Assessing roost disturbance of straw-coloured fruit bats (Eidolon helvum) through tri-axial acceleration.

The study that I reviewed has two goals: 1) test a new method for the assessment of disturbance to bat roosts and 2) identify causes of disturbance to the straw-coloured fruit bat. The manuscript is well written and structured, it is easy to follow and I think it constitutes a nice contribution. I have two main comments and a number of minor suggestions, the latter directed at improving clarity at specific points in the manuscript. I will highlight when they are just a suggestion and the change can be flexible.

Comment 1: I have some concerns regarding the structure of the models.

As you mention also in your discussion, there can be factor that affect flight activity linked to the individual colony that are not addressed. It would be maybe relevant to have ‘location’ as random factor to account for those, or at least to check this as you do for the individual level.

The other point about the models is for the use of ‘human density’. It is not clear to me if A) there are 4 unique values only (one for each location/point of capture, the black ones in Figure 1), or if B) the ‘human density’ variable is calculated around each identified roost (both black and white points in Figure 1). From one of the graphs I can think maybe option B) because the rugplot at the bottom of the graph seems to show multiple lines, but it should be made more clear. If instead it is case A), I think there might be a statistical problem – ‘location’ could be confounded with ‘human density’, and ‘human density’ might not qualify to be a continuous variable as I believe the effective sample size is 4 (correct if I’m wrong). In this case, ‘location’ could become a categorical fixed factor and conclusions about ‘human density’ could be made only at discussion level if the most ‘disturbed’ area is also the one with highest ‘human density’.

Comment 2: I think the manuscript puts too much emphasis on results from model 1 in both the discussion and, especially, abstract. Or better, should acknowledge that there are contrasting results. Yes, model 2 has a lower R^2, but both are not very high and I can’t think why removing the capture days from the dataset should reduce the impact of ‘human density’. You could also explore if conducting model selection would highlight some of the main effects.

In either cases, the manuscript could also list the testing of the method as a goal in the aim section of the introduction.

Minor comments:

Line 49. ‘species distributions’

Line 50. Suggestion: ‘…is a challenging task, as impacts vary across species and contexts.’

Line 54. the use of automated methods is has become more common – to answer the task described in the previous paragraph (assessing the impact of disturbance) or for studies in general?

Lines 55-57. They do reduce the influence of the observer, but we still know very little of how the behaviour in bats is affected by relatively heavy tags (>5% of the bat weight), although it is likely that they provide some sort of disturbance. Maybe you could revise this sentence to either mention the duality in their use or highlight different benefits connected to your study.

Line 58. ‘new sensors for’ -> ‘new sensors to measure’

Lines 69-70. worth mentioning a few more detail about the roosting habitat? (for example, if they have particular requirements in terms of forest density/type, etc.)

Line 71. Add Country name where park is located.

Line 72. ‘making thenm’

Line 77-78. A small step seems to be missing here regarding as to why daytime flights can be used a proxy for disturbance. Also ‘which where’ -> ‘and used them’

Line 83. ‘predicted’ -> ‘assumed’

Line 84. ‘the colonies that..’ -> ‘the colonies, and we predicted that this…’

Line 104. ‘Black’

Lines 108-112. Descriptions of the other study areas also briefly mention vegetation type, but here it is not described. Also, sometimes the word ‘areas’ is used and sometimes ‘locations’ (see for example line 98 and Table 1). Change for consistency.

Lines 113-117. Mention if the two sampling points in this area are regarded as one (for example, if environmental variables are considered homogeneous across the two)

Line 121. 70,000

Line 127. Suggestion: For structure consistency mention which season (wet or dry) October to December represent.

Line 133: ‘we weighted the captured bats’

Line 138. Mention glue type

Line 141. How often were GPS points recorded?

Line 154. ‘torsion’ -> ‘variation’. Is this change correct? If I understand correctly, heave, surge and sway are translational motions (the ones you can identify with the loggers), while rotational motions (torsion) would be pitch, roll and yaw.

Lines 156-157. Any reference for this method of assessing flying motion?

Lines 157-159. Please rephrase this sentence and give also a less technical explanation of why that ensures correct positioning (similar to the way you explain flight in lines 153-154).

Lines 167-168. To further improve understanding, use also the expression ‘occurrence of daytime roost flights’ here (to let the reader easily understand that you will use a binary variable).

Line 168-169. Why do you exclude bats that travel more than 500 meters? Explain please.

Line 170. Here you say you remove the first day of tracking of the animals, but isn’t this the ‘capture day’ that it is later on included in the first model version?

Line 178. ‘probably -> ‘probability’

Lines 176-193. As it can be inferred from my other comments, I think here there is some detail missing regarding the covariate used and the levels at which they are defined. Is every bat from a certain area/location going to have the same solar irradiance? Or are these variables defined with higher spatial resolution? What about human density? These details are important to evaluate the suitability of model structure. Also, for the effect of capture activity it should be said explicitly here or in the next paragraph that it is binary yes/no for each study unit (correct?).

Line 196. ‘bat captures’ -> ‘bat capture activity’

Line 200. So, your analysis unit is ‘occurrence of daytime roost flight’ defined for each bat and each day of tracking (rephrase as you wish, but I suggest to state it clearly somewhere in the text).

Line 212-214. You can also provide some basic information on how many days had flight occurrence or what are the ranges of the variables measured (for example, if human density is defined at the location density, you can mention values for each location, or similar info). This helps contextualise the results.

Line 241. Maybe write the actual number instead of the number of observations excluded.

Reviewer #3: This study assesses roost disturbance in straw-coloured fruit bats using GPS-accelerometers to measure the probability of daytime flights using the bat capture days as a proxy for human disturbance. Bat poaching is of increasing concern not only for bat conservation but also zoonotic diseases, so new methods to detect and deter poachers are of importance. I am not totally convinced by the efficacy of the method at the current time (e.g. short tracking duration, weight of devices and relatively labour intensive), but it is likely the technology will advance to a point where it becomes viable so methods making use of these results show promise for the future. The study uses bats from 5 different sites and has a good sample size from each. The only major comment I have is that the current analyses do not take into account the fact that there are five different sites and the temporal relatedness of variables (see below). I’m uncertain whether the changes to the model I have suggested below will alter the results, but it needs to be incorporated to check that the results for human density in particular are not just a result of other unmeasured differences between sites. I confident that you will be able to make these changes, and I congratulate you on an otherwise well-written and executed paper.

Major comments:

As far as I can tell, you have not accounted for the effect of colony location in your analysis. The study design is a nested repeated measures design where you have 5 colonies with several individuals tracked in each colony during different seasons and for varying numbers of days. Given the human density metric will be the same for each bat at the same colony (and to a lesser degree the weather variables as sampling occurred over numerous days), this means that your current model essentially makes the assumption that all of the differences between sites are due to human density and does not allow for any unmeasured differences between sites, which is unlikely given the large spatial and temporal distribution of sites. There are a couple of different ways to solve this problem, such as nesting random effects, but I think I would personally add the location as a categorical fixed-effect to measure if there are differences between the 5 sites (see below).

The second problem is that the temporal relatedness of the capture dates is not taken into account. Temporal autocorrelation predicts that observations taken closely together are more likely to be related to each other e.g. the bats captured on the same day are likely to be more related to each other (particularly in colonies) than the bats captured a week later. For example, bat 1607 was captured on the 4th February and tracked for 5 days, which overlaps with bat 1616 captured on the 6th February and bat 1620 captured on the 7th February. For your study, I recommend that you nest date within each site because the dates at different sites are unlikely to impact the results. For further reading on autocorrelation, see Pinheiro, J. C., & Bates, D. (2000). Mixed-effects models in S and S-PLUS; and Boyce et. al. (2010) Temporal autocorrelation functions for movement rates from global positioning system radiotelemetry data.

Last time I checked, adding temporal autocorrelation could not be done easily in lme4 but it can be done for a binomial model using a combination of nlme and MASS (or potentially a Bayesian package like brms). In nlme and MASS, your R code would look something like:

library(MASS)

library(nlme)

model <- glmmPQL(flights ~ capture + humans + solar + precipitation + wind + location, random = ~ 1 | individual, correlation=corAR1(form = ~ date| location), family = binomial, data = your_data)

(corAR1 = Autoregressive Lag-1 correlation).

This model will take some time to run as it will have to do repeated calls. Please do conduct your own research to check the model I have written here is correct as I have done this relatively quickly without your data, particularly the syntax on the autocorrelation as I have not nested that before. It goes without saying that you need to add back in your individual random effects. There is no reason to remove individual random effects on unbalanced designs.

Minor comments:

Line 55: Comma required after Ref 19 to create a clause – otherwise the tense is wrong.

68: assumed to decline – assumed to be declining? More specific would be better. Are there any studies quantifying this?

72: then – them

Methods: Please add the sample sizes into the main text either in the study area descriptions and/or Table 1.

Conclusions: The results of accelerometer studies can also be used to refine other methods for detecting poachers, such as motion capture methods. You could add a couple of sentences in the discussion about this.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: Yes: Lucy A. Taylor

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

30 Sep 2020

Editor Comments:

All the reviewers agree that the study is sound and interesting. Please address all the minor suggestions and also see particularly to the insightful comment by reviewers 2 and 3 on the statistical analysis.

Response: We agree that the suggestions of reviewers 2 and 3 on the statistical analysis were appropriated and made changes in our models in accordance.

Copyright of map used in Figure 1.

Response: The background map of Figure 1 was developed by Database of Global Administrative Areas (GADM) and its use is allowed for academic publishing (https://gadm.org/license.html). This information was added to the caption of Figure 1.

Reviewer #1: General comments

Anthropogenic disturbance impacts on biodiversity are of concern globally. Here the authors assessed roost disturbance of straw-coloured fruit bats (Eidolon helvum) through a novel technique of tri-axial acceleration. They have assumed that all flights during the daytime are a consequence of human disturbance. My only concern is how much bat flight occurs during the daytime at roosts because of interactions between bat individuals? Although you used daytime flights as a proxy for disturbance were you able to validate any of these? The manuscript is presented well.

Response: We clarified in lines 82-89 that we investigated human disturbance (caused by our presence in the colonies during trapping) but also natural stressors described in the literature to influence the probability of bats to fly during the day. We think the removal of human density, as suggested by reviewers #2 and #3, helped to make a clear distinction between human and environmental stressors.

We agree that some flights in the colonies may have occurred due to interactions between individuals. However, we have no reason to think that those interactions will influence or invalidate the relationships that emerged from our models.

With regards to the use of daytime flights as a proxy for disturbance, we added a justification and a reference in lines 80-82.

Specific comments

Title:

Fine.

Abstract:

Generally fine.

L32 maybe use the common name instead of just ‘bats’.

Response: Corrected.

L32 put GPS in full.

Response: Added.

Introduction:

Good synthesis of the relevant literature.

L55 put GPS in full.

Response: Added.

Methods:

Generally well presented.

L96 Does your University/ Institute not require you to have animal ethics permission/ clearance from them for your proposed study?

Response: The Max Planck Institute of Animal Behavior, the leading institution of this project, required only permits from authorities of the countries where fieldwork took place. Those permits accounted for national regulations for ethics in animal experimentation and the use wild animals in scientific research in each country.

L133 delete hours.

Response: Deleted.

L135 perhaps replace ‘deployed’ with ‘fitted’.

Response: Changed.

L137 insert type of glue in parentheses after ‘glue’.

Response: Added.

L138 insert ‘neck’ before ‘collar’.

Response: Added.

L141 delete hours.

Response: Deleted.

L142 abbreviate ‘minutes’.

Response: Abbreviated.

L143 data is a plural so change ‘was’ to ‘were'.

Response: Corrected.

L151 delete hours.

Response: Deleted.

L153 change to ‘compared with’.

Response: Corrected.

L162 change to ‘analyses’.

Response: Corrected.

L166 delete hours.

Response: Deleted.

L178 reword to ‘of bats flying during the daytime’.

Response: Changed.

L179, 180 change to ‘analyses’.

Response: Corrected.

L183 replace ‘propagate’ with ‘spread’.

Response: Changed.

L192 change to ‘analyses’.

Response: This sentence was removed as a result of changes described in our response to comment 1 of Reviewer #2.

L198 replace ‘due to’ with ‘because of’.

Response: This sentence was removed as a result of changes described in our response to comment 1 of Reviewer #2.

Results:

Generally fine.

L227 insert ‘a’ before ‘model’.

Response: Corrected.

Discussion:

Good synthesis.

L267 insert Latin name.

Response: Added.

References:

I have not checked if Journal format was followed.

Reviewer #2: Comments to the authors

I hereby provide comments for the manuscript Assessing roost disturbance of straw-coloured fruit bats (Eidolon helvum) through tri-axial acceleration.

The study that I reviewed has two goals: 1) test a new method for the assessment of disturbance to bat roosts and 2) identify causes of disturbance to the straw-coloured fruit bat. The manuscript is well written and structured, it is easy to follow and I think it constitutes a nice contribution. I have two main comments and a number of minor suggestions, the latter directed at improving clarity at specific points in the manuscript. I will highlight when they are just a suggestion and the change can be flexible.

Comment 1: I have some concerns regarding the structure of the models.

As you mention also in your discussion, there can be factor that affect flight activity linked to the individual colony that are not addressed. It would be maybe relevant to have ‘location’ as random factor to account for those, or at least to check this as you do for the individual level.

The other point about the models is for the use of ‘human density’. It is not clear to me if A) there are 4 unique values only (one for each location/point of capture, the black ones in Figure 1), or if B) the ‘human density’ variable is calculated around each identified roost (both black and white points in Figure 1). From one of the graphs I can think maybe option B) because the rugplot at the bottom of the graph seems to show multiple lines, but it should be made more clear. If instead it is case A), I think there might be a statistical problem – ‘location’ could be confounded with ‘human density’, and ‘human density’ might not qualify to be a continuous variable as I believe the effective sample size is 4 (correct if I’m wrong). In this case, ‘location’ could become a categorical fixed factor and conclusions about ‘human density’ could be made only at discussion level if the most ‘disturbed’ area is also the one with highest ‘human density’.

Response: We agree with the reviewer and this comment is consistent with that of reviewer #3 (major comments first paragraph). As mentioned by both reviewers, colonies may vary in several factors with potential influence on the flight behaviour of bats, with human density being among those factors. We followed the suggestions of the reviewers and re-fitted our models with colony ID and individual ID as random factors and excluding human density as predictor (Table 2 and Figure 3 were updated with the results of the new models, as well as the text in lines 204-215, 231-237, 241-244, 282-283). Compared with the models built before, the inclusion of colony ID slightly influenced the variation explained by individual ID. That is the reason why we decided to include individual ID as random factor in our models. The results are similar to the earlier models but the exclusion of human density required changes across the manuscript (see lines 35, 83-88, 201, 229, 269-270).

Comment 2: I think the manuscript puts too much emphasis on results from model 1 in both the discussion and, especially, abstract. Or better, should acknowledge that there are contrasting results. Yes, model 2 has a lower R^2, but both are not very high and I can’t think why removing the capture days from the dataset should reduce the impact of ‘human density’. You could also explore if conducting model selection would highlight some of the main effects.

Response: We think this issue is solved with the new models, for which there is no inconsistency in the significance of the common predictors. However, the new results reinforce the idea that the differences in R2 between models are mainly due to sample size (see Table 2).

Regarding the model selection suggestion, we used an information-theoretic approach, rather than frequentist, as we did not have an exhaustive collection of potential predictors of flight behaviour. Following this approach, we drew a clear hypothesis for each variable used as predictor (lines 85-89), supported by the existing literature, and used the models to test those hypotheses. We think model selection would be more appropriated in a frequentist approach.

In either cases, the manuscript could also list the testing of the method as a goal in the aim section of the introduction.

Response: Added in lines 76-77

Minor comments:

Line 49. ‘species distributions’

Response: Corrected.

Line 50. Suggestion: ‘…is a challenging task, as impacts vary across species and contexts.’

Response: Changed.

Line 54. the use of automated methods is has become more common – to answer the task described in the previous paragraph (assessing the impact of disturbance) or for studies in general?

Response: Clarified.

Lines 55-57. They do reduce the influence of the observer, but we still know very little of how the behaviour in bats is affected by relatively heavy tags (>5% of the bat weight), although it is likely that they provide some sort of disturbance. Maybe you could revise this sentence to either mention the duality in their use or highlight different benefits connected to your study.

Response: The sentence was revised and a reference of disturbance effects of biologging was added.

Line 58. ‘new sensors for’ -> ‘new sensors to measure’

Response: Corrected.

Lines 69-70. worth mentioning a few more detail about the roosting habitat? (for example, if they have particular requirements in terms of forest density/type, etc.)

Response: This species does not show particular preference for trees species or forest types as roosting habitat (see reference [13]), except that it often roosts in urban parks as specified in lines 72-74.

Line 71. Add Country name where park is located.

Response: Added.

Line 72. ‘making thenm’

Response: Corrected.

Line 77-78. A small step seems to be missing here regarding as to why daytime flights can be used a proxy for disturbance.

Response: This was also pointed out by reviewer #1. We added a justification and a reference in lines 80-82.

Also ‘which where’ -> ‘and used them’

Response: Corrected.

Line 83. ‘predicted’ -> ‘assumed’

Response: This sentence was changed to address the issues raised in comment 1 of this reviewer.

Line 84. ‘the colonies that..’ -> ‘the colonies, and we predicted that this…’

Response: This sentence was changed to address the issues raised in comment 1 of this reviewer.

Line 104. ‘Black’

Response: Corrected.

Lines 108-112. Descriptions of the other study areas also briefly mention vegetation type, but here it is not described. Also, sometimes the word ‘areas’ is used and sometimes ‘locations’ (see for example line 98 and Table 1). Change for consistency.

Response: A brief description of vegetation was added as suggested (lines 110-111). References to “site” or “location” were changed to “area” for consistency (lines 100, 132, 150, 201 and Table 1).

Lines 113-117. Mention if the two sampling points in this area are regarded as one (for example, if environmental variables are considered homogeneous across the two)

Response: All roosts were considered separately in the analysis. We think this becomes clear from Fig 1. However, weather variables were retrieved from the same stations in roosts close together. This was clarified in lines 197-199.

We also added the distance between the two main colonies in Kibi (line 114).

Line 121. 70,000

Response: Corrected.

Line 127. Suggestion: For structure consistency mention which season (wet or dry) October to December represent.

Response: Added.

Line 133: ‘we weighted the captured bats’

Response: Corrected.

Line 138. Mention glue type

Response: Added.

Line 141. How often were GPS points recorded?

Response: Added.

Line 154. ‘torsion’ -> ‘variation’. Is this change correct? If I understand correctly, heave, surge and sway are translational motions (the ones you can identify with the loggers), while rotational motions (torsion) would be pitch, roll and yaw.

Response: Corrected.

Lines 156-157. Any reference for this method of assessing flying motion?

Response: Using accelerometers to identify flight of animals is a relatively common method. For that we cited a review paper in line 156. However, the calculations described in lines 157-168 are specific for this dataset.

Lines 157-159. Please rephrase this sentence and give also a less technical explanation of why that ensures correct positioning (similar to the way you explain flight in lines 153-154).

Response: A more complete explanation is now provided in lines 163-165. Further changes were made in lines 157-162 in order to clarify the classification criteria used.

Lines 167-168. To further improve understanding, use also the expression ‘occurrence of daytime roost flights’ here (to let the reader easily understand that you will use a binary variable).

Response: A sentence as added in lines 174-176 to clarify this aspect.

Line 168-169. Why do you exclude bats that travel more than 500 meters? Explain please.

Response: Explanation added (lines 177-178).

Line 170. Here you say you remove the first day of tracking of the animals, but isn’t this the ‘capture day’ that it is later on included in the first model version?

Response: “Capture day” used across the manuscript refer to the days that we visited the colony for captures in general. Our presence was assumed to disturb the whole colony, as described in lines 186-192. We also assumed that all bats with dataloggers deployed earlier would reflect the disturbance of the colony (added in line 192). Apart from that, we expect the capture and handling of each bat would affect its behaviour for the rest of the day, thus we decided to exclude the data from the first tracking day of each bat.

We added this explanation in lines 178-179 and we also re-worded several parts of the manuscript for clarification (lines 34-35, 37, 186, 224, 225, 226, 234, 235, 241, 247, 261, 263, Table 1 model and variable names).

Line 178. ‘probably -> ‘probability’

Response: Corrected.

Lines 176-193. As it can be inferred from my other comments, I think here there is some detail missing regarding the covariate used and the levels at which they are defined. Is every bat from a certain area/location going to have the same solar irradiance? Or are these variables defined with higher spatial resolution? What about human density? These details are important to evaluate the suitability of model structure.

Response: We now explain in lines 197-199 that weather variables were retrieved from the same stations in roosts placed close together. This doesn’t mean that there is much repetition of values in our dataset as there is little time overlap between the different bats tracked (see Table S1) and weather data were updated every day. Human density was excluded from the analysis, as detailed above in our response to comment 1 of this reviewer.

Also, for the effect of capture activity it should be said explicitly here or in the next paragraph that it is binary yes/no for each study unit (correct?).

Response: Added in lines 210-211 and also in Table 2

Line 196. ‘bat captures’ -> ‘bat capture activity’

Response: We edited this sentence to address the issues described above (reviewer comment to line 70)

Line 200. So, your analysis unit is ‘occurrence of daytime roost flight’ defined for each bat and each day of tracking (rephrase as you wish, but I suggest to state it clearly somewhere in the text).

Response: We now state this clearly in lines 174-176.

Line 212-214. You can also provide some basic information on how many days had flight occurrence or what are the ranges of the variables measured (for example, if human density is defined at the location density, you can mention values for each location, or similar info). This helps contextualise the results.

Response: The number of daytime roost flight occurrences was added to the text (line 223). The range of the variables used as model predictors was included in Table 2.

Line 241. Maybe write the actual number instead of the number of observations excluded.

Response: Added

Reviewer #3: Comments to the authors

This study assesses roost disturbance in straw-coloured fruit bats using GPS-accelerometers to measure the probability of daytime flights using the bat capture days as a proxy for human disturbance. Bat poaching is of increasing concern not only for bat conservation but also zoonotic diseases, so new methods to detect and deter poachers are of importance. I am not totally convinced by the efficacy of the method at the current time (e.g. short tracking duration, weight of devices and relatively labour intensive), but it is likely the technology will advance to a point where it becomes viable so methods making use of these results show promise for the future. The study uses bats from 5 different sites and has a good sample size from each. The only major comment I have is that the current analyses do not take into account the fact that there are five different sites and the temporal relatedness of variables (see below). I’m uncertain whether the changes to the model I have suggested below will alter the results, but it needs to be incorporated to check that the results for human density in particular are not just a result of other unmeasured differences between sites. I confident that you will be able to make these changes, and I congratulate you on an otherwise well-written and executed paper.

Response: We agree that the method described will be more valuable in the near future than when the data was collected (2009-2014). Actually, if the data collection was done in the current days, we would have available similar dataloggers at much lower costs, with longer lifespan and the remote transmission of data, which would reduce the costs and fieldwork effort considerably. This idea was already in lines 302-305 but we added a complement in lines 306-307.

We describe below how we addressed the remaining issues.

Major comments:

As far as I can tell, you have not accounted for the effect of colony location in your analysis. The study design is a nested repeated measures design where you have 5 colonies with several individuals tracked in each colony during different seasons and for varying numbers of days. Given the human density metric will be the same for each bat at the same colony (and to a lesser degree the weather variables as sampling occurred over numerous days), this means that your current model essentially makes the assumption that all of the differences between sites are due to human density and does not allow for any unmeasured differences between sites, which is unlikely given the large spatial and temporal distribution of sites. There are a couple of different ways to solve this problem, such as nesting random effects, but I think I would personally add the location as a categorical fixed-effect to measure if there are differences between the 5 sites (see below).

Response: We agree with the reviewer. This issue was also pointed by reviewer #2 (major comment 1). We followed the suggestions of both reviewers and re-fitted our models with colony ID and individual ID as random factors and excluding human density as a predictor (Table 2 and Figure 3 were updated with the results of the new models, as well as the text in lines 204-215, 231-237, 241-244, 282-283). The inclusion of colony ID slightly influenced the variation explained by individual ID, therefore we decided to include individual ID as random factor in our models. The results are similar to the earlier models but the exclusion of human density required changes across the manuscript (see lines 35, 83-88, 201, 229, 269-270).

We should clarify that the roost ID included 30 different roosts (not 5, see Fig 1).

The second problem is that the temporal relatedness of the capture dates is not taken into account. Temporal autocorrelation predicts that observations taken closely together are more likely to be related to each other e.g. the bats captured on the same day are likely to be more related to each other (particularly in colonies) than the bats captured a week later. For example, bat 1607 was captured on the 4th February and tracked for 5 days, which overlaps with bat 1616 captured on the 6th February and bat 1620 captured on the 7th February. For your study, I recommend that you nest date within each site because the dates at different sites are unlikely to impact the results. For further reading on autocorrelation, see Pinheiro, J. C., & Bates, D. (2000). Mixed-effects models in S and S-PLUS; and Boyce et. al. (2010) Temporal autocorrelation functions for movement rates from global positioning system radiotelemetry data.

Last time I checked, adding temporal autocorrelation could not be done easily in lme4 but it can be done for a binomial model using a combination of nlme and MASS (or potentially a Bayesian package like brms). In nlme and MASS, your R code would look something like:

library(MASS)

library(nlme)

model <- glmmPQL(flights ~ capture + humans + solar + precipitation + wind + location, random = ~ 1 | individual, correlation=corAR1(form = ~ date| location), family = binomial, data = your_data)

(corAR1 = Autoregressive Lag-1 correlation).

This model will take some time to run as it will have to do repeated calls. Please do conduct your own research to check the model I have written here is correct as I have done this relatively quickly without your data, particularly the syntax on the autocorrelation as I have not nested that before. It goes without saying that you need to add back in your individual random effects. There is no reason to remove individual random effects on unbalanced designs.

Response: We were aware of the possible temporal autocorrelation issues mentioned, as they are common in tracking studies, particularly those using high temporal resolution data. However, the aggregation of data into daily values (described in lines 173-176) was likely to reduce or cancel temporal autocorrelation of model residuals.

We now conducted a formal analysis of temporal autocorrelation of our model residuals and present the results in Fig S1 (with an explanation added in lines 214-215). As shown in Fig S1, correlation of model residuals does not show a temporal trend, thus we think that adding a temporal correlation structure to our models would lead to unnecessary complexity. But we are very grateful for the detailed explanation, literature and R code with possible solutions to account for temporal autocorrelation.

We should emphasize that there was no time overlap between the data collected in the four different regions (shown in Tables 1 and S1). Thus, we were able to use the full dataset in our temporal autocorrelation analysis.

Minor comments:

Line 55: Comma required after Ref 19 to create a clause – otherwise the tense is wrong.

Response: Corrected.

68: assumed to decline – assumed to be declining? More specific would be better. Are there any studies quantifying this?

Response: “assumed to” was removed. The conclusions of UICN most recent evaluation (reference [27]) states that this species in indeed declining.

72: then – them

Response: Corrected.

Methods: Please add the sample sizes into the main text either in the study area descriptions and/or Table 1.

Response: Added in Table 1.

Conclusions: The results of accelerometer studies can also be used to refine other methods for detecting poachers, such as motion capture methods. You could add a couple of sentences in the discussion about this.

Response: Added in lines 300-302 and a supporting reference was also included.

Submitted filename: Response to reviewers.docx

Click here for additional data file.

5 Nov 2020

PONE-D-20-15313R1

Assessing roost disturbance of straw-coloured fruit bats (Eidolon helvum) through tri-axial acceleration

PLOS ONE

Dear Dr. Santos,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Dec 20 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Christian Vincenot, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (if provided):

The manuscript is basically fit for publication as far as I am concerned. Please just address the few remaining minor comments by reviewer 2 (esp. the supplementary data requested) and resubmit, after which I will promptly send the acceptance letter. Congratulations on the interesting study.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #3: Thank you for the revised version of your manuscript (previously Reviewer 3). As before, this study assesses roost disturbance in straw-coloured fruit bats using GPS-accelerometers to

measure the probability of daytime flights using the bat capture days as a proxy for human

disturbance. The statistics are much improved without the use of the human density metric which was the same at each colony. There are still a few question marks I have over the analysis, particularly in relation to roost ID (see below), but I do not think these will have much impact on the results. I congratulate the authors on a well-written manuscript.

Minor comments:

The authors have used roost ID rather than colony as a random effect, which means for some colonies there are multiple roosts and others just the main colony site. For example, Ouagadougou seems to have 10 roosts whereas has Accra only has the main colony site. Although designed to reduce the bias of repeated measures, this design could give additional weight to certain colonies, particularly when you consider some of the weather data for each roost is taken from the same weather station (e.g. bats at different roosts on the same day would have the same solar irradiance value). I think any impacts would be very minor, and I also think you need to take the bat biology into account (e.g. significance of different roosts), which I am not able to do as I’m not a bat expert. It would be helpful if you could run your analyses using colony location as well and/or justify your decision to use roost in the manuscript when you refer to colony in the rest of the manuscript.

L55: has = have

Table 1: Please distinguish the two Kibi colonies as you refer to 5 colonies throughout the rest of the manuscript. Given that the two Kibi colonies are so close together and that the bats do seem to move roosts further than the distance between these two colonies, I’m not totally sure whether you should be referring to 4 or 5 colonies (when considering the spatial scale between these two sites and the other 3), but I’m not a bat expert so I leave this to you.

L208: If you use Roost ID as a random effect, please add an additional supplementary table detailing the sample size at each roost (bats and days). I presume this cannot be added to Table S1 as some of the bats will have swapped roost, but it’s helpful to see the overall distribution of your sample size.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

6 Nov 2020

Reviewer #3 comments

Thank you for the revised version of your manuscript (previously Reviewer 3). As before, this study assesses roost disturbance in straw-coloured fruit bats using GPS-accelerometers to measure the probability of daytime flights using the bat capture days as a proxy for human disturbance. The statistics are much improved without the use of the human density metric which was the same at each colony. There are still a few question marks I have over the analysis, particularly in relation to roost ID (see below), but I do not think these will have much impact on the results. I congratulate the authors on a well-written manuscript.

Minor comments:

The authors have used roost ID rather than colony as a random effect, which means for some colonies there are multiple roosts and others just the main colony site. For example, Ouagadougou seems to have 10 roosts whereas has Accra only has the main colony site. Although designed to reduce the bias of repeated measures, this design could give additional weight to certain colonies, particularly when you consider some of the weather data for each roost is taken from the same weather station (e.g. bats at different roosts on the same day would have the same solar irradiance value). I think any impacts would be very minor, and I also think you need to take the bat biology into account (e.g. significance of different roosts), which I am not able to do as I’m not a bat expert. It would be helpful if you could run your analyses using colony location as well and/or justify your decision to use roost in the manuscript when you refer to colony in the rest of the manuscript.

Response: We think we should account for possible roost specific effects on bat behaviour, which would not be accounted if we use colony site as random effect instead. Please note in Fig. 1 that roosts can be as far as 90 km from each other, thus potentially under different environmental conditions. Colony sites, on the other hand, do not have as much ecological meaning as they simply mark the places where the captures were conducted.

We anyway ran the analysis with colony site as random effect, as suggested, and the results were nearly identical (see the table below; the legend is the same as for Table 2 in the manuscript).

Model Parameter Range Estimate SE Z P-value R2 cond./marg

With trapping days Intercept - -7.169 2.033 -3.53 >0.001 0.24/0.22

Trapping day 0-1 1.666 0.530 3.14 0.002

Solar irradiance 2.5-30.7 0.242 0.090 2.68 0.007

Precipitation 0.0-87.0 0.042 0.043 0.99 0.329

Wind speed 0.5-4.0 -0.320 0.377 -0.85 0.398

Without trapping days Intercept - -6.008 1.831 -3.28 0.001 0.06/0.06

Solar irradiance 2.5-30.7 0.175 0.089 1.96 0.050

Precipitation 0.0-87.0 0.034 0.042 0.82 0.410

Wind speed 0.5-4.0 -0.144 0.429 -0.34 0.738

We agree that we used the term “colony” several times when we should have used “roost”. That was corrected across the manuscript.

L55: has = have

Response: We think “has” should be used here as it refers to “The use of automated methods”.

Table 1: Please distinguish the two Kibi colonies as you refer to 5 colonies throughout the rest of the manuscript. Given that the two Kibi colonies are so close together and that the bats do seem to move roosts further than the distance between these two colonies, I’m not totally sure whether you should be referring to 4 or 5 colonies (when considering the spatial scale between these two sites and the other 3), but I’m not a bat expert so I leave this to you.

Response: We made the changes suggested in Table 1. We now refer to roosts instead of colonies (see above), thus we maintained the reference to 5, instead of 4.

L208: If you use Roost ID as a random effect, please add an additional supplementary table detailing the sample size at each roost (bats and days). I presume this cannot be added to Table S1 as some of the bats will have swapped roost, but it’s helpful to see the overall distribution of your sample size.

Response: We added a supplementary table (Table S2) containing the roost location for each bat in each day of tracking.

Submitted filename: Response to Reviewers.pdf

Click here for additional data file.

9 Nov 2020

Assessing roost disturbance of straw-coloured fruit bats (Eidolon helvum) through tri-axial acceleration

PONE-D-20-15313R2

Dear Dr. Santos,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Christian Vincenot, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Thank you for addressing promptly the remaining comments. The manuscript is now accepted. Congratulations again on the interesting work!

Reviewers' comments:

12 Nov 2020

PONE-D-20-15313R2

Assessing roost disturbance of straw-coloured fruit bats (Eidolon helvum) through tri-axial acceleration

Dear Dr. Santos:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Christian Vincenot

Academic Editor

PLOS ONE