Content uploaded by David Singer
Author content
All content in this area was uploaded by David Singer on Sep 03, 2024
Content may be subject to copyright.
1
Diel and seasonal vocal activity patterns revealed by passive
1
acoustic monitoring suggest expert recommendations for
2
breeding bird surveys need adjustment
3
David Singer1,2,*, Johannes Kamp3, Hermann Hondong3, Andreas Schuldt2, Jonas Hagge1,2
4
5
1 Northwest German Forest Research Institute, Department of Forest Nature Conservation, Göttingen, Germany
6
2 University of Göttingen, Department of Forest Nature Conservation, Göttingen, Germany
7
3 University of Göttingen, Department of Conservation Biology, Göttingen, Germany
8
9
* Corresponding author. E-mail address: d.singer@posteo.de (D. Singer), phone: +49 551 69401 245
10
ORCID of the authors:
11
- David Singer: 0000-0001-9808-0747
12
- Johannes Kamp: 0000-0002-8313-6979
13
- Andreas Schuldt: 0000-0002-8761-0025
14
- Jonas Hagge: 0000-0001-8938-6680
15
Author contributions
16
David Singer: Conceptualization, Data curation, Methodology, Formal analysis, Writing - Original Draft, Project
17
administration; Johannes Kamp: Writing – Review & Editing, Supervision; Hermann Hondong: Writing - Review
18
& Editing; Andreas Schuldt: Writing - Review & Editing, Supervision; Jonas Hagge: Conceptualization, Writing
19
- Review & Editing, Project administration
20
21
2
Acknowledgements
22
We thank the team from the Forest Nature Conservation department of the Northwest German Forest Research
23
Institute for conducting the fieldwork: G. Brauer, O. Henning, A. König, A. Lindner, K. Lorenz, N. Mosel, J.
24
Milch, A. Nehrkorn, S. Nitzschke, C. Rachow, J. Wellhäuser, K. Werner and S. Wiehemeyer. We thank the Lower
25
Saxony Ministry of Food, Agriculture and Consumer Protection for the financial support (Stadt.Land.ZUKUNFT)
26
and the Niedersächsische Landesforsten for supporting the study within their forests. We also thank two
27
anonymous reviewers for their valuable and constructive comments on an earlier version of this paper. We
28
acknowledge support by the Open Access Publication Funds/transformative agreements of the Northwest German
29
Forest Research Institute.
30
3
Diel and seasonal vocal activity patterns revealed by passive
31
acoustic monitoring suggest expert recommendations for
32
breeding bird surveys need adjustment
33
Abstract
34
Species identification and recording in breeding bird surveys vastly rely on the registration of avian calls and
35
songs. Despite comprehensive expert knowledge on species-specific activity patterns, data-based analyses of vocal
36
activity patterns are lacking. Recent advances in passive acoustic monitoring allow the direct measurement of bird
37
vocal activity at very high temporal resolution. We conducted a comprehensive survey, recording 25,000 hours of
38
audio data at 256 forest sites in Lower-Saxony, Germany, to investigate vocal activity patterns of the European
39
forest bird community. Our results reveal a high degree of inter-specific variability in seasonal and diel vocal
40
activity patterns, including strong circular patterns along the day-night cycle and a significant seasonal component.
41
Comparing acoustic detectability to species-specific survey recommendations revealed critical temporal
42
discrepancies for 64.2 % of species, and standard protocols (four hours after sunrise) showed discrepancies for
43
41.5% of species. This highlights the need for temporal survey optimization to reduce imperfect detection and
44
increase accuracy and precision. Emphasis should be given to the hours before and after sunrise and also sunset
45
for sampling less detectable species. Combining observer-based surveys with passive acoustic monitoring might
46
leverage the strengths of both methods. Our results also emphasize the potential of continuous recording schedules
47
in passive acoustic monitoring to capture diverse temporal patterns. This study provides a baseline for future
48
research on vocal activity patterns across habitats, throughout the year, and regarding anthropogenic impacts. Our
49
findings may raise awareness among ornithologists about the sources of variation in acoustic detectability and its
50
implications for breeding bird surveys, highlighting potential for methodological adjustments in survey timing and
51
consequences for carful interpretation of bird surveys.
52
Keywords (4-6)
53
passive acoustic monitoring, AudioMoth, BirdNET, imperfect detection, detection probability, phenology
54
4
Statements and Declarations
55
The authors declare no conflict of interest.
56
Introduction
57
Acoustic communication plays a pivotal role in the ecology of animals, including birds (Rosenthal and Ryan
58
2000). Spatio-temporal patterns of animal vocalisations provide detailed information on site occupancy and
59
behaviour (Kershenbaum et al. 2016; Gibb et al. 2019). Birds sing to attract mates and defend their territories
60
(Catchpole and Slater 2008) and use alarm calls to alert conspecifics and other species to predators (Hollén and
61
Radford 2009; Gill and Bierema 2013). In bird surveys the spatio-temporal distribution of species songs and calls
62
are used to infer presence and abundace and to delineate territories of breeding birds (Bibby et al. 1992; Südbeck
63
et al. 2005). Particularly in ecosystems with dense vegetation, bird observations by humans predominantly result
64
from the detection of auditory cues, while bird sightings are a less important source of observations (Brewster and
65
Simons 2009). Hence, most bird surveys vastly rely on the assumption that birds vocalize regularly during a survey
66
so that their presence and abundance can be inferred by the observer.
67
Survey protocols of bird monitoring programmes set standards concerning seasonality, daytime and weather
68
conditions (Bibby et al. 1992; Greenwood et al. 1994; Jiguet et al. 2012; Sauer et al. 2017), while in the inner
69
tropics such standardization is often challenging due to the absence or irregularity of seasonality (Bennun 2000).
70
Standardized breeding bird surveys reach back to the 1960s in North America and parts of Europe (Sauer et al.
71
2017), and expert recommendation to standardize the diel and seasonal timing were developed early and have been
72
constantly reviewed and improved (Greenwood et al. 1994).
73
Germany is a country with a long tradition of breeding bird surveys, especially using the territory mapping
74
method (Südbeck et al. 2005). Survey protocols are well-established and have been widely applied for breeding
75
bird surveys since the 1960s (Flade 1994). Territory mapping is the preferred method in conservation, impact
76
assessment and environmental planning since the 1990s. However, most survey recommendations regarding
77
phenological recommendations are based on expert knowledge (Südbeck et al. 2005), as observational studies
78
quantifying diel and seasonal activity patterns are scarce (but see Robbins 1981; Morelli et al. 2022). Although
79
5
recommendations for the phenological aspects of bird surveys can be based on standardized data (e.g. Strebel et
80
al. 2014), changes in seasonal activity due to climate change have rendered some older recommendations outdated
81
(Rubolini et al. 2007; Romano et al. 2023)
82
As already Zimmer (1919) recognized, detailed studies on temporal activity patterns conducted by human
83
observers are inherently limited in terms of covered species, sites, seasonal period, time of day and is complicated
84
by uncertainties arising from the varying expertise of human observers. Recent advancements in passive acoustic
85
monitoring (PAM) now make it possible to directly measure vocal activity of entire bird assemblages in high
86
temporal resolution over extended periods, with simultaneously surveying multiple sampling sites without human
87
observer bias (Lellouch et al. 2014; Krause and Farina 2016; Thompson et al. 2017; Roark and Gaul 2021). With
88
the combination of energy-efficient autonomous recording units (Hill et al. 2018) and powerful species detection
89
algorithms (Kahl et al. 2021) that are capable to identify avian vocalisations of entire species assemblages with
90
high precision (Singer et al. 2024; Funosas et al. 2024), it is even possible to acquire data on fine-scale diel activity
91
patterns. Beyond enhancing our understanding of bird ecology, such detailed insights into species-specific vocal
92
activity patterns may have practical implications for the improvement of survey protocols, particularly in the
93
context of phenological shifts under climate change (Balantic and Donovan 2019). Species-specific knowledge on
94
(acoustic) detectability may also improve the understanding of variability, and the assessment of uncertainty, in
95
breeding bird surveys and long-term bird monitoring (Strebel et al. 2014; Balantic and Donovan 2019). Better
96
quantification of temporal activity patterns could inform species conservation (Day et al. 2015; Mariton et al.
97
2023).
98
In this study, we present high-resolution diel and seasonal vocal activity patterns of European forest bird species
99
at the landscape scale, using passive acoustic monitoring across 256 study sites, covering an area of over 45,000
100
km2. We aim to adress three key research objectives:
101
(1) To quantify the diel and seasonal diversity of vocal activity patterns in European forest birds at 10-min
102
intervals;
103
(2) To compare these patterns to those used in breeding bird survey recommendations based on expert
104
knowledge to identify mismatches between data-driven and expert-knowledge-based phenologies;
105
(3) To illustrate, how recommendations for breeding birds could be improved by data-driven approaches such
106
as passive acoustic monitoring.
107
6
Materials and methods
108
Data collection and pre-processing
109
We recorded audio data at 256 forest sites in Lower-Saxony, Germany (Fig. 8). The study region covers parts
110
of the atlantic and continental biogeographic regions. Sites covered a broad range of forest habitats, including old-
111
growth deciduous stands left unmanaged for more than 50 years, and managed deciduous, mixed and coniferous
112
forest stands of various age. All study sites were randomly chosen within state forest areas, owned by the
113
‘Niedersächsische Landesforsten’. Regarding mean annual temperature and precipitation of the period 1991-2020
114
(DWD 2022), the study sites were representative for the climatic conditions of German forests expect for the alpine
115
region (Fig. 9). Elevation of the study sites ranges between 0-660 m asl.
116
128 sites were studied in spring 2022, the other 128 sites in spring 2023. Due to battery failures at 24 sites in
117
2022/2023 these sites were sampled again in spring 2024 to obtain complete time-series for all 256 sites. To record
118
audio data autonomously, one AudioMoth (Hill et al. 2018) (versions 1.0.0/1.1.0 that are technically equal) per
119
site was attached to a tree trunk in waterproof IPX7 cases at ca. 1.8 m height. Devices were programmed to record
120
at a sampling rate of 32 kHz for 30 s every 10 min from 1st March to 21st May in each year, covering the main
121
part of the breeding period of most Central European forest birds. We analysed 97 hours of audio data per plot,
122
totalling up to 2.9 years (25.190 hours) of audio data. All audio data were analysed using the artificial neural
123
network BirdNET Analyzer 2.2 (Kahl et al. 2021), with default settings (min_conf = 0.1, sensitivity = 1, spp = 1,
124
overlap = 0). BirdNET provided a list of species and corresponding confidence scores for ten 3-s intervals for each
125
30 s audio file. We deactivated the use of eBird species distribution data (Sullivan et al. 2009) within the BirdNET
126
analysis, as most bird watchers in Germany contribute their observations to the platform ‘ornitho.de’ instead of
127
eBird (Hertzog et al. 2021), hence, the eBird-data may be less complete regarding species coverage. We
128
To minimize false-positive species detections in the raw BirdNET classifications before analysing species-
129
specific vocal activity patterns, we first applied the species-specific thresholding approach of Singer et al. (2024).
130
This approach makes use of aggregated time-series features (e.g. the mean confidence score of adjacent acoustic
131
samples) to improve the differentiation of true- and false-positive detections in automated species classification
132
data. The first author validated 225 BirdNET detections per species by listening to all audio files across the
133
confidence score range, and derived species-specific threshold rules using conditional inference trees to reduce
134
false positive detections in the raw BirdNET data (Singer et al. 2024).
135
7
Analysis of activity time series
136
We included only time-series of a minimum of five days with at least two detections of a species, aiming to
137
exclude recordings including potentially false positive BirdNET detections. . We transformed the remaining time-
138
series of raw BirdNET detections into binary activity data (1 for activity, 0 for no activity) per 30-second file and
139
calculated the moving average including five time steps (5 time steps of 10 min = 50 min) of these time-series as
140
a robust measure of detection probability. Finally, we averaged the detection probabilities across sites per species
141
to receive averaged time-series across the study region (Fig. 1). Hence, relative detectability reaches a value of
142
one at the time when a species is detected at the highest number of sites simultaneously. Compared to simply
143
summing species detections across sites per time step, this measure of detection probability accounts for spatial
144
variation in the absolute number of detections to avoid that activity patterns are dominated by single bird
145
individuals at certain sites. Consequently, the resulting activity patterns are a robust measure of spatial synchrony
146
in detectability of the species at landscape level. Only species with time-series from at least five sites were
147
included. We visualised the resulting time-series of relative detectability as ‘diel-seasonal heatmaps’, where the x-
148
axis represents the day of the year, and the y-axis represents the time of the day (Fig. 3). Compared to classical
149
two-dimensional time-series visualisations, this approach is well-suited for illustrating the diel circularity of
150
species activity, capturing the day-and-night patterns in the temperate zone.
151
152
Fig. 1 Data preparation scheme from the BirdNET detections of all sample sites to one averaged detectability time-series per bird species.
153
To infer similarity in diel-seasonal activity pattern between species, we calculated the Pearson’ correlation
154
coefficients of the relative detectability time series (81 days with 10 min time-steps) comparing all species and
155
conducted an agglomerative hierarchical cluster analysis on the correlation matrix, using the ‘ward.D2’ method.
156
This method minimizes the total within-cluster variance at each step of the clustering process (Warren Liao 2005).
157
8
Visualisations (Fig. 4) were done with the R packages corrplot (Wei and Simko 2021) and ggplot2 (Wickham
158
2016).
159
Comparison to expert-knowledge survey recommendations
160
We aimed to assess the additional information content of PAM-derived phenological data when compared to
161
expert-derived recommendations provided in the survey manual for Germany of Südbeck et al. (2005). Here, the
162
breeding season is divided into thirds of months (beginning (B), mid (M), end (E)) and favourable seasonal survey
163
period (coinciding with maximum detectability based in expert opinion) are suggested for all species, based on
164
consultation of species experts (Südbeck et al. 2005). Species accounts of the manual also contain suggestions of
165
the favourable daytime to survey a given species. We translated this expert knowledge on seasonal and diel
166
favourable survey periods into a machine-readable dataset. Semi-quantitative estimates of best survey time were
167
translated into four categories: early morning 9:00 CET, morning: 10:00 CET, late morning 11:00 CET, noon:
168
12:00 CET. Dawn and dusk were defined as civil dawn/dusk and calculated using the suncalc R package
169
(Thieurmel and Elmarhraoui 2022) for the centroid of all study sites (Fig. 8). To compare the species-specific,
170
relative acoustic detectability from the PAM data to the expert-knowledge based survey recommendations, we
171
averaged the diel activity time-series within the thirds of months and visualised the resulting distribution of relative
172
detectability per third of month.
173
We compare maxima of acoustic detectability from the PAM to those based on expert-knowledge in the species-
174
specific survey recommendations. We calculated the temporal completeness as the proportion of high acoustic
175
detectability (≥ 0.5) covered by the survey recommendations and the temporal specificity as the proportion of
176
survey recommendation time period covering periods of actually high acoustic detectability. We evaluated the
177
temporal alignment between species-specific high acoustic detectability and the community scale standard survey
178
methodology for breeding bird surveys in the temperate zone, which comprises four hours after sunrise (Jiguet et
179
al. 2012; Kamp et al. 2021). We categorised the temporal alignment of survey recommendations and species
180
detectability according to the risk of imperfect detection due to sampling at periods of low detectability (Fig. 2).
181
9
182
Fig. 2 Conceptual visualisation of four cases of temporal alignment between high relative detectability of species (as revealed by passive
183
acoustic monitoring) and expert-knowledge based survey recommendations as measured by temporal specificity and temporal completeness
184
of the survey recommendations.
185
To evaluate the usefulness of this standard survey recommendation at species community level, we further
186
analysed hourly detectability of all species relative to sunrise, using the moving average as introduced above.
187
Specifically, we calculated the hourly percentage of species with high detectability (≥ 0.5) for each third of months.
188
Species with short detectability peaks (≤ 2 h) were outlined additionally (Fig. 7). Furthermore, we assessed the
189
hourly dissimilarity in the set of species with high detectability, distinguishing turnover (species exchange) and
190
nestedness (gain/loss of species) components of temporal beta-diversity following Baselga (2010) using the R
191
package betapart (Baselga and Orme 2012).
192
193
10
Results
194
Species-specific vocal activity patterns
195
Our high-resolution data from passive acoustic monitoring revealed distinct diel-seasonal activity patterns of 53
196
European forest bird species (Fig. 3). Species occupancy, i.e. the presence of a species detected by PAM, ranged
197
from five (which was our minimum criterion) to 256 sites. The number of detections per species ranged from 885
198
(Grey-headed woodpecker) to 1,029,028 (European Robin, Fig. 10). The hierarchical clustering of activity time
199
series revealed four clearly distinct activity types, initially separating a diverse group of diurnal species with peak
200
activity in March/April from three others with very distinct activity patterns, including nocturnal, crepuscular, and
201
migratory (diurnal) species (Fig. 5). The first large group of diurnal species could further be separated into species
202
with early spring and mid-spring activity; however dissimilarity was comparably low here (average linkage height
203
of diurnal species cluster: 0.497, average linkage height of nocturnal, crepuscular and migratory (diurnal peak
204
May) cluster: 0.660, average linkage height of all species: 0.603).
205
11
53
206
Fig. 3 Diel-seasonal heatmaps of relative acoustic detectability of 53 European forest bird species, sorted according to similarity of activity
207
patterns, revealed by hierarchical clustering (Fig. 4). Relative detectability derived with a moving average approach (cf. methods section) is
208
12
scaled per species, meaning that relative detectability is 1 at the time when the species was detected at the maximum of sites simultaneously.
209
Black lines show sunrise and sunset times. n indicates the number of sites included for each species.
210
211
Fig. 4 Time-series correlations (Pearson correlations of the 82 day time series shown in Fig. 3) between the relative detectability of species.
212
Rectangles show six clusters of species with similar activity patterns. Top panels show averaged diel-seasonal vocal activity heatmaps (cf.
213
Fig. 3) per cluster, bottom dendrogram shows the results of the agglomerative hierarchical clustering (see Methods for details).
214
215
13
Comparison to expert-knowledge based survey recommendations
216
Diel activity patterns varied markedly between species, encompassing bimodal, unimodal, and uniform
217
distributions with different skewness and kurtosis. Diel patterns and also changed over the season in many
218
species (species examples in Fig. 5, for figures of all 53 forest bird species see Appendix Fig. 11).
219
220
Fig. 5 Diel activity patterns of selected bird species per third of month (B = beginning, M = mid, E = end, 3 = March, 4 = April, 5 = May)
221
including at least one example species per cluster (Fig. 4). Figures for all 53 species can be found in the Appendix (Fig. 11). Dark blue
222
bars mark the expert knowledge-based temporal survey recommendations from Südbeck et al. (2005), light blue bars mark their extended
223
survey recommendations. The dashed line marks a relative detectability of 0.5, a threshold that here defines high detectability. Times of
224
14
the day with high detectability are marked in dark orange. Night times are marked in grey, the time between civil dawn and sunset/civil
225
dusk and sunset with a lighter grey.
226
Temporal completeness and specificity of the species-specific survey recommendations varied widely across
227
species, encompassing all possible combinations of high and low completeness and specificity (Fig. 6a). The same
228
applied to the standard protocol for breeding bird surveys; however, temporal specificity was generally higher
229
compared to species-specific survey recommendations, except for species with nocturnal and crepuscular activity
230
species which are not covered with by the standard protocol at all (Fig. 6b).
231
Species-specific survey recommendations demonstrated a critical temporal discrepancy (categories C1 or C2)
232
for 64.2 % of the species. For 26.4 % of the species survey recommendations did only cover less than 50 % of the
233
period of high detectability but were specific (category B) while species-specific survey recommendations were
234
adequately aligned with species detectability for 9.4 % of the species (category A). Temporal alignment was
235
generally less critical for the standard protocols with critical discrepancy (categories C1 or C2) for 41.5 % of the
236
species. For 32.1 % of the species standard protocols did not reach high completeness but high specificity (category
237
B) while standard protocols were adequately aligned for 26.4 % of the species (category A).
238
239
Fig. 6 Temporal alignment of (A) species-specific survey recommendations (Südbeck et al. 2005) and (B) standard protocols for breeding
240
bird surveys in Europe (first 4 hours after sunrise, (e.g. Jiguet et al. 2012; Kamp et al. 2021)) with high relative species detectability (≥0.5),
241
expressed as temporal specificity and completeness of the survey recommendations/standard protocol. Background color scale outlines
242
the density of species. Detailed values for all species can be found in Table 1 and a summary in Table 2.
243
244
15
The analysis of hourly detectability across species revealed a seasonal pattern. While in the first half of the study
245
period (B3-M4) the percentage of species with high detectability was at maximum in the first two hours after
246
sunrise and decreased afterwards, the pattern was more constant from end of April onwards (E3-M5) and did not
247
decline strongly beyond four hours after sunrise (Fig. 7). Hourly dissimilarity also followed a clear pattern with a
248
distinct peak of turnover at sunrise and maximum nestedness before sunrise, here representing a gain of species
249
with high detectability during dawn phase (Fig. 7).
250
251
Fig. 7 Percentage of species with high detectability and hourly dissimilarity of the set of species with high detectability, decomposed into
252
turnover (exchange of species) and nestedness (gain/loss of species) (Baselga 2010), in relation to sunrise throughout the breeding
253
season. Panels show month thirds: B = beginning, M = mid, E = end, 3 = March, 4 = April, 5 = May. Percentages of species are calculated
254
in relation to the total number of species per month third. Dashed vertical lines mark the 4 hours after sunrise, which is the standard
255
recommendation for breeding bird monitoring programmes in Europe (Jiguet et al. 2012; Kamp et al. 2021). Species are separated
256
according to the duration of high detectability, outlining species with short period of high detectability in orange and others in blue.
257
258
16
Discussion
259
Our study provides detailed information on vocal activity for 53 European forest bird species, extending our
260
understanding of their diel and seasonal phenology fundamentally. We found a high inter-specific variability of
261
seasonal and diel vocal activity patterns and identified distinct species clusters of vocal activity patterns (i.e.,
262
diurnal, nocturnal and crepuscular species) within a European forest bird community. Further, our results provide
263
detailed insights into the acoustic detectability of birds in temperate forests relevant for breeding bird surveys and
264
bird monitoring in general. Our analysis of the temporal alignment between periods of high acoustic detectability
265
and existing expert-knowledge based survey recommendations revealed temporal mismatches for 64.2 % of the
266
species for species-specific recommendation (Südbeck et al. 2005) and 41.5 % of the species for standard protocols
267
of breeding bird community surveys (4 hours after sunrise), suggesting survey methods for observer-based
268
breeding bird surveys might benefit from adjustment.
269
Two-dimensional temporal partitioning of vocal activity
270
We found strong circular patterns of vocal activity within the European forest bird community along the day-
271
night-cycle, encompassing nocturnal, diurnal and crepuscular vocal activity patterns. While the ‘dawn chorus’ - a
272
phase of high vocal activity around sunrise - has received much attention in previous studies (Bruni et al. 2014;
273
Gil and Llusia 2020; Puswal et al. 2022), complete diel activity patterns remain poorly studied. Hence, our results
274
provide basic ecological knowledge on diel activity patterns of 53 European forest bird species in the temperate
275
zone and may serve as valuable baseline for future studies (e.g. studying temporal shifts of activity due to climate
276
change, urbanisation or habitat alteration). Even though diel circularity of bird vocal activity is long known (Allen
277
1913; Robbins 1981; Morelli et al. 2022), our study quantifies fine-scale differences in species acoustic
278
detectability throughout the diel cycle for a comprehensive set of bird species at very high temporal resolution.
279
Some diel activity patterns might come as a surprise to the experienced bird surveyor: For example, Eurasian
280
Blackbirds revealed a distinctly higher activity around dusk compared to dawn while Song Thrushes were similarly
281
active in the morning and evening, suggesting that standard surveys focusing on the early-morning hours yield
282
abundance estimates biased low for Blackbird. Some species like the Common Cuckoo and Wood Lark revealed
283
partial nocturnality while Tawny Owl demonstrated rather continuous activity throughout the night. Species like
284
Common Chiffchaff, Eurasian Blue Tit or Short-toed Treecreeper were constantly active throughout the day.
285
17
We also found a strong seasonal component in the vocal activity patterns during the breeding season. All species
286
within the cluster ‘diurnal - peak May’ (Fig. 4) are mid- to long-distance migratory species and show a negative
287
correlation with detectability of most of the other diurnal species. However, also the other diurnal species reveal a
288
gradual seasonality, with many of the Tit species, both Treecreeper species and Black Woodpecker revealing high
289
detectability at the very beginning of March while other species like Common Chaffinch, Dunnock or the other
290
Woodpecker species revealed maximum detectability end of March/beginning of April. Such seasonal
291
differentiation of acoustic bird communities is generally well documented (Thompson et al. 2017; Vokurková et
292
al. 2018; Puswal et al. 2022; Wu et al. 2023) and especially driven by seasonal migration (Mason 1995; Krishnan
293
2019).
294
The temporal niche partitioning illustrated here is known from activity studies of mammals (Bennie et al. 2014),
295
and is seen as a mechanism that facilitates coexistence of sympatric species (Frey et al. 2017). Mammals alter their
296
temporal activity patterns to reduce predation risk (Veldhuis et al. 2020). This strategy may also explain the
297
observed crepuscular activity pattern of the Eurasian Pygmy Owl, regarded as a strategy to avoid predators such
298
as the larger owl species. However, vocal activity of birds is not a comprehensive measure of all activities that
299
birds do throughout the day (or night), and thus only partly explains overall detectability in surveys. Eurasian
300
Blackbirds showed a strong crepuscular pattern in vocal activity (Fig. 3), but diurnal activity can be high and
301
constant as revealed by camera trapping (Schlindwein et al. 2024). Hence, concepts that may explain temporal
302
activity partitioning of species in camera trapping studies are not directly applicable to vocal activity.
303
Implications for bird surveys
304
The analysis of the temporal alignment between periods of high acoustic detectability and species-specific
305
survey recommendations (Südbeck et al. 2005) revealed critical temporal discrepancy for a majority of the studied
306
species. Regarding standard protocols (sampling for 4 hours after sunrise), temporal alignment was also critical in
307
many species (41.5 %). Even though the resulting imperfect detection may and should be accounted for statistically
308
(Kéry and Schmidt 2008; Strebel et al. 2014; Kellner and Swihart 2014), to our knowledge this is almost never
309
done in the context of environmental planning or conservation programmes. Hence, temporal survey optimisation
310
through data-based, regularly updated knowledge on temporal patterns of detectability has the potential to increase
311
accuracy and precision of surveys. Species-specific surveys should be pinpointed to periods with constant high
312
detectability (category A, Fig. 2), as surveys with low temporal specificity may underestimate absolute population
313
sizes (Tomiałojć and Lontkowski 1989). Even though effects of low specificity on the estimation of relative
314
18
population trends may be less critical (as the bias is systematic), surveys at periods of low detectability might fail
315
to detect small population change (Wood et al. 2019). Thus, for the 64.2 % of species revealing critical
316
discrepancies (categories C1 and C2) we recommend adapting the species-specific survey recommendations to the
317
periods of high detectability. Low temporal completeness but high specificity (category B) may not critically effect
318
estimations of population sizes or trends for species that are well detectable throughout the whole day, however
319
for species with high detectability limited to short time periods like crepuscular species (e.g. Eurasian Blackbird,
320
Song Thrush, Eurasian Robin), survey recommendations should not only be temporally specific but also
321
temporally complete. For the species of category B an extension of the recommended survey periods to the periods
322
of high detectability would also increase economic efficiency of field surveys, as observers may stay longer in the
323
field per day and consequently need less survey days to cover a study area.
324
Optimisation of survey timing is feasible in studies or monitoring programs focusing on single bird species,
325
nevertheless when sampling complete species assemblages timing of surveys has to be a compromise to cover as
326
many as possible species, or species of interest. It is therefore promising that temporal specificity was generally
327
higher for standard protocols (4 hours after sunrise) compared to the species-specific recommendations from
328
Südbeck et al. (2005). However, we still identified a number of rather common species whose detectability
329
revealed a critical mismatch with the standard protocol used in monitoring programmes of common breeding birds.
330
As the peak of turnover in the set of species with high detectability at sunrise demonstrates (Fig. 7), sunrise is a
331
critical turning point regarding the detectability of species. Also the detectability of the species with a short phase
332
of high detectability accumulates around sunrise (Fig. 7), demonstrating crucial importance of the two hours
333
around sunrise for sampling critical species at high detectability. It is common sense to conduct extra surveys for
334
nocturnal species in territory mapping (Südbeck et al. 2005), however the magnitude of changes in detectability
335
of species with high detectability around sunrise has not been considered in the standard protocols yet (Jiguet et
336
al. 2012; Kamp et al. 2021). Regarding our results, standard protocols for common breeding bird monitoring should
337
optimally include an extra sampling with focus on a limited set of species (e.g. Blackbird, Robin, Goldcrest, Lesser
338
Spotted Woodpecker, Eurasian Treecreeper, Willow Tit) starting one hour before sunrise. As the species set with
339
high detectability remains rather constant throughout the morning (Fig. 7), the standard round of sampling may
340
follow, starting one hour after sunrise. However, contrary to our expectation, the detectability of species to decline
341
more pronounced at the beginning of the season but stayed rather constant at the end (Fig. 7), hence surveys in
342
temporal forests may be extended beyond four hours after sunrise from end of April onwards. Optimally
343
Blackbirds, as one of the most common bird species, would we surveyed in a specific sampling in the evening
344
shortly before sunset.
345
19
Synergies of acoustic and observer-based surveys
346
Species-specific surveys as well as standard survey protocols should be designed in a way that suggests surveying
347
in periods of the highest high acoustic detectability of species. Nevertheless, adaptations face logistical restrictions
348
in observer-based real-world surveys. Therefore, combining observer-based surveys with PAM can leverage the
349
strengths of both methods. Substantial synergistic potential lies in the targeted use of PAM at certain locations
350
within a study area in addition to territory mapping, to increase detectability of rare species or species of low
351
detectability, and to gauge diel and seasonal detectability that might vary with region and year(Gaylord et al. 2023;
352
Baroni et al. 2023).PAM is particularly beneficial for sampling forest bird species with short phases of high
353
detectability (e.g. Eurasian Wood Cock, Blackbird, Eurasian Treecreeper, Willow Tit, Eurasian Pygmy Owl) as
354
well as crepuscular species in other habitats (e.g. Grey Partridge in agricultural landscapes). Furthermore,
355
observer-based playback surveys are the proposed method to survey e.g. woodpeckers or owls due to their low
356
spontaneous vocalization rates (Südbeck et al. 2005). However, playback of songs or even alarm calls can induce
357
unnecessary stress for birds and may lead to an overestimation of population size due to the attraction effects,
358
hence using PAM to study species with low vocalisation rates should be preferred over playback surveys whenever
359
feasible.
360
Implications for schedules of passive acoustic monitoring
361
As our results demonstrated, temporal patterns within temperate acoustic bird communities are diverse and
362
encompass distinct seasonal and diel activity peaks. Hence, in PAM continuous recording schedules without
363
limitation to certain times of the day are worth the effort, as the periods of highest detectability of all occurring
364
species are included. This does not imply a necessity for continuous full-time recordings but rather suggests that
365
audio samples should be evenly distributed across the 24-hour cycle to optimally capture all species present. Our
366
schedule with 5% temporal coverage - recording for 30 sec every 10 minutes – seem to be a suitable schedule for
367
studying activity patterns of complete species assemblages. Studying activity patterns of 12 subtropical montane
368
forest bird species, Wu et al. (2023) also found that short but regularly distributed recordings closely resembled
369
the activity patterns observed in continuous recordings. Similar findings were made for a tropical bird community,
370
where species richness accumulated more rapidly with short audio samples recorded at high temporal resolution
371
(Metcalf et al., 2022). However, previous PAM studies focussing on birds often used unevenly distributed
372
recording schedules, often focussing on the dawn period (Sugai et al. 2019). Generally, terrestrial soundscape
373
20
studies were found to have the lowest coverage of the complete diel cycle compared to aquatic or marine studies
374
(Darras et al. 2024). With recent technological developments, it becomes achievable to record audio data at high
375
temporal resolution schedules or even with constant recoding, as autonomous recording units are energy efficient
376
(Hill et al. 2018) and species classification algorithms are fast (Kahl et al. 2021).
377
Future perspectives
378
Despite the advantages of passive acoustic monitoring to study vocal activity patterns, there are still
379
methodological questions to address in future studies: Do species detection algorithms perform constantly
380
throughout the daily cycle or is there e.g. a reduced detectability of certain species within acoustically complex
381
soundscapes during the dawn chorus? How do weather conditions shape vocal activity patterns? How pronounced
382
is inter-annual variability of vocal activity patterns? Furthermore, BirdNET currently does not differentiate
383
between calls and songs. However, as different types of vocalizations have different ecological functions (Gill and
384
Bierema 2013; Gil and Llusia 2020), they might also vary over the day, e.g. depending on the activity patterns of
385
predators. Recent developments of BirdNET and other algorithms will allow to differentiate songs and calls soon
386
(McGinn et al. 2023). More broadly, research on passive acoustic monitoring should focus on understanding how
387
species-specific activity patterns relate to habitat use. For example, determining which activity patterns indicate
388
breeding activity in different species.
389
Our study was limited to forest sites, however vocal activity patterns may differ between different habitat types.
390
Thus, future work should gather comparable high-resolution acoustic data in other habitat types, e.g. urban or
391
agricultural areas. Species-specific activity patterns may also differ with habitat. Also for some of the rare forest
392
species, our data is still limited as only a few of the study sites were occupied. Hence, targeted sampling for species
393
like Boreal Owl, Eurasian Pygmy Owl or Grey-headed Woodpecker would improve the reliability of the described
394
vocal activity patterns. Furthermore, our study was limited to the months March to May, however the breeding
395
season of some species extends beyond this period. Vocal activity peaks of some early species may have been
396
missed as we did not cover February. Sampling the complete annual cycle would reveal further interesting vocal
397
activity patterns throughout the year, even before and after the breeding season. Hence, future work could even
398
dive deeper into temporal patterns of song and call behaviour of bird species assemblages across habitats and
399
throughout the year. Large potential lies within global cooperation of soundscape ecologists for future meta-
400
analyses (Darras et al. 2024).
401
21
Apart from identifying temporal variation in vocal activity, anthropogenic impacts on diel-seasonal acoustic
402
activity patterns including noise pollution, artificial light at night and climate change could be studied in future
403
PAM studies. Anthropogenic noise and artificial light at night are both known to interfere with bird vocal activity
404
(Fuller et al. 2007; Kempenaers et al. 2010; Dominoni et al. 2016; Cretois et al. 2024), while climate change can
405
cause phenological shifts of vocal activity. However, cumulative effects and interactions of anthropogenic impacts
406
on vocal activity patterns throughout the year remain unknown. PAM has proven to serve as a valuable tool for
407
such studies (Balantic and Donovan 2019; Roark and Gaul 2021).
408
409
410
22
Conclusions
411
We were able to gain fundamental insights into the diverse vocal activity patterns within the European forest bird
412
community. Our comparison of these patterns to the expert-based survey recommendations and standard protocols
413
for breeding bird surveys revealed a critical temporal mismatch between high vocal activity and the recommended
414
survey timing for a large proportion of species. Species-specific survey recommendations and possibly also the
415
standard protocols should be re-evaluated to enhance the accuracy and precision of bird surveys. Beyond
416
adjustments to survey timing, it is crucial to raise awareness among ornithologists about the variability of acoustic
417
detectability across temporal bird communities and the implications for the interpretations of breeding bird
418
surveys.
419
Data accessibility
420
Data are available on email request by the corresponding author.
421
422
23
References
423
Allen FH (1913) More Notes on the Morning Awakening. The Auk 30:229–235. https://doi.org/10.2307/4071548
424
Balantic C, Donovan T (2019) Temporally adaptive acoustic sampling to maximize detection across a suite of
425
focal wildlife species. Ecology and Evolution 9:10582–10600. https://doi.org/10.1002/ece3.5579
426
Baroni D, Hanzelka J, Raimondi T, et al (2023) Passive acoustic survey reveals the abundance of a low-density
427
predator and its dependency on mature forests. Landsc Ecol. https://doi.org/10.1007/s10980-023-01667-
428
1
429
Baselga A (2010) Partitioning the turnover and nestedness components of beta diversity. Global Ecology and
430
Biogeography 19:134–143. https://doi.org/10.1111/j.1466-8238.2009.00490.x
431
Baselga A, Orme CDL (2012) betapart: an R package for the study of beta diversity. Methods in Ecology and
432
Evolution 3:808–812. https://doi.org/10.1111/j.2041-210X.2012.00224.x
433
Bennie JJ, Duffy JP, Inger R, Gaston KJ (2014) Biogeography of time partitioning in mammals. Proceedings of
434
the National Academy of Sciences 111:13727–13732. https://doi.org/10.1073/pnas.1216063110
435
Bennun L (2000) Assessing and monitoring bird populations in Africa: an overview. Ostrich 71:214–215.
436
https://doi.org/10.1080/00306525.2000.9639915
437
Bibby CJ, Burgess ND, Hill DA (1992) Bird census techniques. San Diego Academic Press, London
438
Brewster JP, Simons TR (2009) Testing the importance of auditory detections in avian point counts. Journal of
439
Field Ornithology 80:178–182. https://doi.org/10.1111/j.1557-9263.2009.00220.x
440
Bruni A, Mennill DJ, Foote JR (2014) Dawn chorus start time variation in a temperate bird community:
441
relationships with seasonality, weather, and ambient light. J Ornithol 155:877–890.
442
https://doi.org/10.1007/s10336-014-1071-7
443
Catchpole C, Slater PJB (2008) Bird song: biological themes and variations, 2nd ed. Cambridge University Press,
444
Cambridge [England] ; New York
445
Cretois B, Bick IA, Balantic C, et al (2024) Snowmobile noise alters bird vocalization patterns during winter and
446
pre-breeding season. Journal of Applied Ecology 61:340–350. https://doi.org/10.1111/1365-2664.14564
447
Darras KF, Rountree R, Van Wilgenburg S, et al (2024) Worldwide soundscape ecology patterns across realms
448
Day J, Baker J, Schofield H, et al (2015) Part-night lighting: implications for bat conservation: Part-night lighting
449
and bats. Anim Conserv 18:512–516. https://doi.org/10.1111/acv.12200
450
Dominoni DM, Greif S, Nemeth E, Brumm H (2016) Airport noise predicts song timing of European birds.
451
Ecology and Evolution 6:6151–6159. https://doi.org/10.1002/ece3.2357
452
DWD (2022) Climate observation data - Open Data Server
453
Flade M (1994) Die Brutvogelgemeinschaften Mittel-und Norddeutschlands: Grundlagen für den Gebrauch
454
vogelkundlicher Daten in der Landschaftsplanung. IHW-Verlag
455
Frey S, Fisher JT, Burton AC, Volpe JP (2017) Investigating animal activity patterns and temporal niche
456
partitioning using camera-trap data: challenges and opportunities. Remote Sensing in Ecology and
457
Conservation 3:123–132. https://doi.org/10.1002/rse2.60
458
Fuller RA, Warren PH, Gaston KJ (2007) Daytime noise predicts nocturnal singing in urban robins. Biol Lett
459
3:368–370. https://doi.org/10.1098/rsbl.2007.0134
460
24
Funosas D, Barbaro L, Schillé L, et al (2024) Assessing the potential of BirdNET to infer European bird
461
communities from large-scale ecoacoustic data. Ecological Indicators 164:112146.
462
https://doi.org/10.1016/j.ecolind.2024.112146
463
Gaylord M, Duarte A, McComb B, Ratliff J (2023) Passive acoustic recorders increase White-headed Woodpecker
464
detectability in the Blue Mountains. JFO 94:art1. https://doi.org/10.5751/JFO-00330-940401
465
Gibb R, Browning E, Glover‐Kapfer P, Jones KE (2019) Emerging opportunities and challenges for passive
466
acoustics in ecological assessment and monitoring. Methods in Ecology and Evolution 10:169–185.
467
https://doi.org/10.1111/2041-210X.13101
468
Gil D, Llusia D (2020) The Bird Dawn Chorus Revisited. In: Aubin T, Mathevon N (eds) Coding Strategies in
469
Vertebrate Acoustic Communication. Springer International Publishing, Cham, pp 45–90
470
Gill SA, Bierema AM-K (2013) On the Meaning of Alarm Calls: A Review of Functional Reference in Avian
471
Alarm Calling. Ethology 119:449–461. https://doi.org/10.1111/eth.12097
472
Greenwood JJD, Baillie SR, Gregory RD, et al (1994) Some new approaches to conservation monitoring of British
473
breeding birds. Ibis 137:S16–S28. https://doi.org/10.1111/j.1474-919X.1995.tb08437.x
474
Hertzog LR, Frank C, Klimek S, et al (2021) Model‐based integration of citizen science data from disparate sources
475
increases the precision of bird population trends. Divers Distrib 27:1106–1119.
476
https://doi.org/10.1111/ddi.13259
477
Hill AP, Prince P, Piña Covarrubias E, et al (2018) AudioMoth: Evaluation of a smart open acoustic device for
478
monitoring biodiversity and the environment. Methods in Ecology and Evolution 9:1199–1211.
479
https://doi.org/10.1111/2041-210X.12955
480
Hollén LI, Radford AN (2009) The development of alarm call behaviour in mammals and birds. Animal Behaviour
481
78:791–800. https://doi.org/10.1016/j.anbehav.2009.07.021
482
Jiguet F, Devictor V, Julliard R, Couvet D (2012) French citizens monitoring ordinary birds provide tools for
483
conservation and ecological sciences. Acta Oecologica 44:58–66.
484
https://doi.org/10.1016/j.actao.2011.05.003
485
Kahl S, Wood CM, Eibl M, Klinck H (2021) BirdNET: A deep learning solution for avian diversity monitoring.
486
Ecological Informatics 101236. https://doi.org/10.1016/j.ecoinf.2021.101236
487
Kamp J, Frank C, Trautmann S, et al (2021) Population trends of common breeding birds in Germany 1990–2018.
488
J Ornithol 162:1–15. https://doi.org/10.1007/s10336-020-01830-4
489
Kellner KF, Swihart RK (2014) Accounting for Imperfect Detection in Ecology: A Quantitative Review. PLOS
490
ONE 9:e111436. https://doi.org/10/gd882s
491
Kempenaers B, Borgström P, Loës P, et al (2010) Artificial night lighting affects dawn song, extra-pair siring
492
success, and lay date in songbirds. Curr Biol 20:1735–1739. https://doi.org/10.1016/j.cub.2010.08.028
493
Kershenbaum A, Blumstein DT, Roch MA, et al (2016) Acoustic sequences in non-human animals: a tutorial
494
review and prospectus. Biological Reviews 91:13–52. https://doi.org/10.1111/brv.12160
495
Kéry M, Schmidt B (2008) Imperfect detection and its consequences for monitoring for conservation. Community
496
Ecology 9:207–216. https://doi.org/10.1556/ComEc.9.2008.2.10
497
Krause B, Farina A (2016) Using ecoacoustic methods to survey the impacts of climate change on biodiversity.
498
Biological Conservation 195:245–254. https://doi.org/10.1016/j.biocon.2016.01.013
499
Krishnan A (2019) Acoustic community structure and seasonal turnover in tropical South Asian birds. Behavioral
500
Ecology 30:1364–1374. https://doi.org/10.1093/beheco/arz087
501
Lellouch L, Pavoine S, Jiguet F, et al (2014) Monitoring temporal change of bird communities with dissimilarity
502
acoustic indices. Methods Ecol Evol 5:495–505. https://doi.org/10.1111/2041-210X.12178
503
25
Mariton L, Le Viol I, Bas Y, Kerbiriou C (2023) Characterising diel activity patterns to design conservation
504
measures: Case study of European bat species. Biological Conservation 277:109852.
505
https://doi.org/10.1016/j.biocon.2022.109852
506
Mason CF (1995) Long-term trends in the arrival dates of spring migrants. Bird Study.
507
https://doi.org/10.1080/00063659509477167
508
McGinn K, Kahl S, Peery MZ, et al (2023) Feature embeddings from the BirdNET algorithm provide insights into
509
avian ecology. Ecological Informatics 74:101995. https://doi.org/10.1016/j.ecoinf.2023.101995
510
Morelli F, Brlík V, Benedetti Y, et al (2022) Detection Rate of Bird Species and What It Depends on: Tips for
511
Field Surveys. Front Ecol Evol 9:. https://doi.org/10.3389/fevo.2021.671492
512
Puswal SM, Mei J, Wang M, Liu F (2022) Daily and Seasonal Patterns in the Singing Activity of Birds in East
513
China. Ardea 110:. https://doi.org/10.5253/arde.v110i1.a4
514
Roark E, Gaul W (2021) Monitoring migration timing in remote habitats: assessing the value of extended duration
515
audio recording. ACE 16:art21. https://doi.org/10.5751/ACE-01885-160121
516
Robbins CS (1981) Effect of time of day on bird activity. Studies in Avian Biology 6:275–286
517
Romano A, Garamszegi LZ, Rubolini D, Ambrosini R (2023) Temporal shifts in avian phenology across the
518
circannual cycle in a rapidly changing climate: A global meta-analysis. Ecological Monographs 93:e1552.
519
https://doi.org/10.1002/ecm.1552
520
Rosenthal GG, Ryan MJ (2000) Visual and acoustic communication in non-human animals: a comparison. J Biosci
521
25:285–290. https://doi.org/10.1007/BF02703937
522
Rubolini D, Møller AP, Rainio K, Lehikoinen E (2007) Intraspecific consistency and geographic variability in
523
temporal trends of spring migration phenology among European bird species. Climate Research 35:135–
524
146. https://doi.org/10.3354/cr00720
525
Sauer JR, Pardieck KL, Ziolkowski DJ Jr, et al (2017) The first 50 years of the North American Breeding Bird
526
Survey. The Condor 119:576–593. https://doi.org/10.1650/CONDOR-17-83.1
527
Schlindwein X, Randler C, Kalb N, et al (2024) Seasonal variation in the diurnal activity pattern of Eurasian
528
blackbirds (Turdus merula) in the forest. J Ornithol 165:137–146. https://doi.org/10.1007/s10336-023-
529
02096-2
530
Singer D, Hagge J, Kamp J, et al (2024) Aggregated time-series features boost species-specific differentiation of
531
true and false positives in passive acoustic monitoring of bird assemblages. Remote Sensing in Ecology
532
and Conservation n/a: https://doi.org/10.1002/rse2.385
533
Strebel N, Kéry M, Schaub M, Schmid H (2014) Studying phenology by flexible modelling of seasonal
534
detectability peaks. Methods Ecol Evol 5:483–490. https://doi.org/10.1111/2041-210X.12175
535
Südbeck P, Andretzke S, Fischer S, et al (eds) (2005) Methodenstandards zur Erfassung der Brutvögel
536
Deutschlands, 1st edn. Mugler, Radolfzell
537
Sugai LSM, Silva TSF, Ribeiro JW, Llusia D (2019) Terrestrial Passive Acoustic Monitoring: Review and
538
Perspectives. BioScience 69:15–25. https://doi.org/10.1093/biosci/biy147
539
Sullivan BL, Wood CL, Iliff MJ, et al (2009) eBird: A citizen-based bird observation network in the biological
540
sciences. Biological Conservation 142:2282–2292. https://doi.org/10.1016/j.biocon.2009.05.006
541
Thieurmel B, Elmarhraoui A (2022) suncalc: Compute Sun Position, Sunlight Phases, Moon Position and Lunar
542
Phase
543
Thompson SJ, Handel CM, Mcnew LB (2017) Autonomous acoustic recorders reveal complex patterns in avian
544
detection probability. The Journal of Wildlife Management 81:1228–1241.
545
https://doi.org/10.1002/jwmg.21285
546
26
Tomiałojć L, Lontkowski J (1989) A technique for censusing territorial song thrushes Turdus philomelos. Annales
547
Zoologici Fennici 26:235–244
548
Veldhuis MP, Hofmeester TR, Balme G, et al (2020) Predation risk constrains herbivores’ adaptive capacity to
549
warming. Nat Ecol Evol 4:1069–1074. https://doi.org/10.1038/s41559-020-1218-2
550
Vokurková J, Motombi FN, Ferenc M, et al (2018) Seasonality of vocal activity of a bird community in an
551
Afrotropical lowland rain forest. J Trop Ecol 34:53–64. https://doi.org/10.1017/S0266467418000056
552
Warren Liao T (2005) Clustering of time series data—a survey. Pattern Recognition 38:1857–1874.
553
https://doi.org/10.1016/j.patcog.2005.01.025
554
Wei T, Simko V (2021) R package “corrplot”: Visualization of a correlation matrix
555
Wickham H (2016) ggplot2: Elegant graphics for data analysis. Springer-Verlag New York
556
Wood CM, Popescu VD, Klinck H, et al (2019) Detecting small changes in populations at landscape scales: a
557
bioacoustic site-occupancy framework. Ecological Indicators 98:492–507.
558
https://doi.org/10.1016/j.ecolind.2018.11.018
559
Wu S-H, Ko JC-J, Lin R-S, et al (2023) Evaluating community-wide temporal sampling in passive acoustic
560
monitoring: A comprehensive study of avian vocal patterns in subtropical montane forests. F1000Res
561
12:1299. https://doi.org/10.12688/f1000research.141951.1
562
Zimmer C (1919) Der Beginn des Vogelgesanges in der Frühdämmerung. Verh orn Ges Bay 14:152–180
563
564
565
27
Appendix
566
567
Fig. 8 Map of the study sites that were stratified randomly distributed in managed and unmanaged state forest area across Lower-Saxony,
568
Germany. The cross marks the centroid of all study sites for which the average sunrise/sunset times were calculated, used in the analyses.
569
570
Fig. 9 Climatic conditions at the study sites in Lower-Saxony, compared to the conditions at 1000 random forest sites in Germany, based on
571
multi-annual climate raster data 1991-2020 (DWD 2022).
572
573
28
574
Fig. 10 a) Number of BirdNET detections per species and b) number of sites per species included in the analyses.
575
576
29
Fig. 11 Diel activity patterns of 53 European bird species per third of month (B = beginning, M = mid, E = end, 3 = March, 4 = April, 5 =
577
May. Dark blue rectangles mark the expert-knowledge based temporal survey recommendations from Südbeck et al. (2005), light blue
578
rectangles mark their extended survey recommendations. The dashed line marks a relative detectability of 0.5, here defined as high
579
detectability. Times of the day with high detectability are marked in dark orange. Night times are marked in grey, the time between civil
580
dawn and sunset/civil dusk and sunset with a lighter grey.
581
582
30
583
31
584
32
585
33
586
587
34
Table 1 Temporal alignment of high detectability of 53 European forest bird species (revealed by passive acoustic monitoring) and species-
588
specific survey recommendations (Südbeck et al. 2005) and standard protocols for breeding bird surveys in Europe (Jiguet et al. 2012;
589
Kamp et al. 2021), categories: A = Adequate temporal alignment, B = Benign temporal discrepancy, C1/C2 = Critical temporal discrepancy
590
(cf. Fig. 2), for details see methodology.
591
Order
Scientific name
English name
Completenes
Südbeck et al.
Specificity
Südbeck et al.
Category
Südbeck et al.
Completeness
standard protocol
Specificity
Standard protocol
Category
standard protocol
CUCULIFORMES
Cuculus canorus
Common
Cuckoo
0.98
0.23
C1
0.54
0.52
A
COLUMBIFORMES
Columba oenas
Stock Dove
0.51
0.74
A
0.74
0.90
A
COLUMBIFORMES
Columba
palumbus
C. Wood-
Pigeon
0.47
0.73
B
0.68
0.80
A
GRUIFORMES
Grus grus
Common
Crane
0.38
0.15
C2
0.72
0.38
C1
CHARADRIIFORMES
Scolopax
rusticola
Eur.
Woodcock
0.20
0.31
C2
0.00
0.00
C2
STRIGIFORMES
Aegolius
funereus
Boreal Owl
0.30
0.06
C2
0.00
0.00
C2
STRIGIFORMES
Glaucidium
passerinum
Eur. Pygmy-
Owl
0.83
0.04
C1
0.00
0.00
C2
STRIGIFORMES
Bubo bubo
Eur. Eagle-
Owl
0.59
0.11
C1
0.00
0.00
C2
STRIGIFORMES
Strix aluco
Tawny Owl
0.45
0.74
B
0.00
0.00
C2
PICIFORMES
Dendrocoptes
medius
Mid. Sp.
Woodp.
0.97
0.49
C1
0.67
0.60
A
PICIFORMES
Dryobates minor
Less. Sp.
Woodp.
0.92
0.10
C1
0.92
0.23
C1
PICIFORMES
Dendrocopos
major
Great Sp.
Woodp.
0.91
0.27
C1
0.91
0.49
C1
PICIFORMES
Dryocopus
martius
Black
Woodp.
0.55
0.50
C1
0.48
0.64
B
PICIFORMES
Picus viridis
Eur. Green
Woodp.
0.87
0.30
C1
0.75
0.47
C1
PICIFORMES
Picus canus
Gray-h.
Woodp.
0.89
0.08
C1
0.89
0.24
C1
PASSERIFORMES
Oriolus oriolus
Eur. Gold.
Oriole
0.48
0.43
C2
0.57
0.65
A
PASSERIFORMES
Garrulus
glandarius
Eur. Jay
0.62
0.39
C1
0.86
0.68
A
PASSERIFORMES
Corvus
monedula
Eur.
Jackdaw
0.83
0.41
C1
0.83
0.65
A
PASSERIFORMES
Corvus corone
Carrion
Crow
0.34
0.12
C2
0.74
0.66
A
PASSERIFORMES
Corvus corax
Common
Raven
0.66
0.62
A
0.39
0.67
B
PASSERIFORMES
Periparus ater
Coal Tit
0.12
0.52
B
0.30
0.70
B
PASSERIFORMES
Lophophanes
cristatus
Crested Tit
0.39
0.50
C2
0.50
0.83
B
PASSERIFORMES
Poecile palustris
Marsh Tit
0.30
0.03
C2
0.30
0.04
C2
PASSERIFORMES
Poecile
montanus
Willow Tit
0.36
0.06
C2
0.36
0.05
C2
PASSERIFORMES
Cyanistes
caeruleus
Eur. Blue Tit
0.18
0.46
C2
0.22
0.62
B
PASSERIFORMES
Parus major
Great Tit
0.06
0.28
C2
0.20
0.46
C2
PASSERIFORMES
Lullula arborea
Wood Lark
0.77
0.28
C1
0.77
0.35
C1
PASSERIFORMES
Aegithalos
caudatus
Long-tailed
Tit
0.55
0.22
C1
0.30
0.71
B
PASSERIFORMES
Phylloscopus
sibilatrix
Wood
Warbler
0.37
1.00
B
0.37
1.00
B
PASSERIFORMES
Phylloscopus
trochilus
Willow
Warbler
0.22
0.96
B
0.29
0.97
B
PASSERIFORMES
Phylloscopus
collybita
Com.
Chiffchaff
0.13
0.98
B
0.35
0.99
B
PASSERIFORMES
Sylvia atricapilla
Eur.
Blackcap
0.20
0.51
B
0.28
0.69
B
35
PASSERIFORMES
Sylvia borin
Garden
Warbler
0.27
0.94
B
0.37
0.96
B
PASSERIFORMES
Regulus
ignicapilla
Com.
Firecrest
0.17
1.00
B
0.33
1.00
B
PASSERIFORMES
Regulus regulus
Goldcrest
0.23
0.63
B
0.06
0.02
C2
PASSERIFORMES
Troglodytes
troglodytes
Eur. Wren
0.77
0.90
A
0.64
0.70
A
PASSERIFORMES
Sitta europaea
Eur.
Nuthatch
0.51
0.60
A
0.51
0.97
A
PASSERIFORMES
Certhia familiaris
Eur.
Treecreeper
0.30
0.05
C2
0.42
0.08
C2
PASSERIFORMES
Certhia
brachydactyla
Short-toed
Treecr.
0.39
0.99
B
0.33
0.86
B
PASSERIFORMES
Sturnus vulgaris
Europ.
Starling
0.39
0.34
C2
0.77
0.66
A
PASSERIFORMES
Turdus
philomelos
Song Thrush
0.51
0.25
C1
0.20
0.09
C2
PASSERIFORMES
Turdus viscivorus
Mistle
Thrush
0.89
0.29
C1
0.89
0.53
A
PASSERIFORMES
Turdus merula
Eur.
Blackbird
0.03
0.02
C2
0.00
0.00
C2
PASSERIFORMES
Muscicapa
striata
Spotted Flyc.
0.46
1.00
B
0.62
0.67
A
PASSERIFORMES
Erithacus
rubecula
Europ. Robin
0.17
0.27
C2
0.00
0.00
C2
PASSERIFORMES
Ficedula
hypoleuca
Eur. Pied
Flycatch.
0.42
0.48
C2
0.28
0.85
B
PASSERIFORMES
Phoenicurus
phoenicurus
Com.
Redstart
0.14
0.38
C2
0.16
0.27
C2
PASSERIFORMES
Prunella
modularis
Dunnock
0.68
0.76
A
0.75
0.95
A
PASSERIFORMES
Anthus trivialis
Tree Pipit
0.46
0.92
B
0.46
0.92
B
PASSERIFORMES
Fringilla coelebs
Com.
Chaffinch
0.07
0.71
B
0.38
0.96
B
PASSERIFORMES
Coccothraustes
coccothraustes
Hawfinch
0.41
0.32
C2
0.47
0.74
B
PASSERIFORMES
Pyrrhula pyrrhula
Eur.
Bullfinch
0.04
0.01
C2
0.71
0.35
C1
PASSERIFORMES
Loxia curvirostra
Red
Crossbill
0.88
0.20
C1
0.70
0.49
C1
592
Table 2 Summary of the temporal alignment of high detectability of 53 European forest bird species (revealed by passive acoustic monitoring) and species-
593
specific survey recommendations (Südbeck et al. 2005) and standard protocols for breeding bird surveys in Europe (Jiguet et al. 2012; Kamp et al. 2021),
594
categories: A = Adequate temporal alignment, B = Benign temporal discrepancy, C1/C2 = Critical temporal discrepancy (cf. Fig. 2), for details see
595
methodology.
596
Species-specific recommendations (Südbeck et al. 2005)
Standard protocols (4 hours after sunrise)
Temporal alignment category
Number of species
Proportion of species
Number of species
Proportion of species
A | Adequate alignment
5
0.09
14
0.26
B | Benign discrepancy
14
0.26
17
0.32
C1 | Critical discrepancy
16
0.30
8
0.15
C2 | Critical discrepancy
18
0.34
14
0.26
597
