Discussion
Time lapse cameras may allow the collation of more standardized data
than motion triggered cameras and potentially reduce the number of empty
images produced by false triggering. Our study shows that the overall
daily capture probability of taiga bean geese was higher with cameras
set to time lapse compared to motion sensitive cameras. However, there
was no significant difference in the daily number of geese captured with
the two trigger types. This difference in the results of the two models,
albeit quite subtle, could at least partly be explained by the motion
triggered cameras failing to capture the more distant animals, whereas
the time lapse method captures both near-by and distant animals. This
makes positioning of the camera traps on time lapse setting less
critical as animals can be detected even if they do not use their exact
assumed route or habitat. Indeed, in order for the animal to be captured
with the motion trigger mode, the PIR sensor must detect motion in the
trigger area while the animal is within the camera’s viewable area
(Moeller et al. 2023).
Our study on taiga bean geese in their breeding areas shows that time
lapse cameras are more useful than motion triggered cameras in
collecting data on the habitat use of animals because of the higher
capture probability of the former. In addition, required memory card
space and battery needs are predictable. This is an important finding,
because the use of time lapse cameras reduces the need to visit the
camera sites for changing batteries and memory cards, hence reducing the
disturbance during the sensitive breeding period of the geese. An
important advantage also is that time lapse cameras will produce more
reliable data of sites that are used by taiga bean geese in the breeding
areas, an information crucially important for protection of the species.
This information is difficult to obtain using human observers as the
species is highly elusive and occupies very remote breeding areas.
Choosing the triggering mode depends on the research objectives and
characteristics of the target species and its habitat. As our study
shows, time-lapse may be more suitable, for example, in studies
involving habitat use and the designation of protected areas where it is
important to distinguish between used and unused sites. Motion trigger
mode, on the other hand, may be more useful in studies on movement or
behaviour, or in cases where the density of the animals is low, as it is
possible that the time lapse method may fail to detect the target
species altogether (Moeller et al. 2023). Moreover, motion triggered
cameras have been shown to capture a higher proportion of animals than
cameras operating on time lapse, such as in a study monitoring wildlife
underpass usage (Pomezanski and Bennett 2018).
Camera trapping has gained popularity in wildlife studies over the
recent years due to its efficiency to collect opportunistic image data
without the need to dedicate numerous hours of researchers’ time present
on the study site. At the same time, large volumes of images collected
with this method continue to be one of the challenges in the subsequent
data management and analysis. For example, a single camera operating on
time lapse with a 15-minute interval, a setting used in this study,
outputs 96 photos in a 24-hour period, and accumulates a dataset
containing more than a thousand images over a study period longer than
ten weeks. On the other hand, motion sensitive triggering may produce
thousands of “empty” images in a short time period that still need to
be checked for the presence of the target species. In total, our
two-year study period produced nearly 350,000 images, which all were
gone through manually, forming the most labour intensive and costly part
of the study. In order to reduce the time allocated to this painstaking
process, machine learning techniques (automatic identification
algorithms) together with citizen science approach (e.g., Wei et al.
2020, Hilton et al. 2022, Bjerge et al. 2023) have the potential to
become some of the most important innovations in the cost-effective
identification and counting of animals from large camera trap datasets
in future studies.
We found a difference in the number of geese and in their capture
probability (i.e., presence) between the two study years. Unfortunately,
because standardized monitoring data of the annual numbers of the taiga
bean goose in the breeding areas are not available, it is not possible
to say if the between-year difference reflects a real difference of
breeding numbers or just random variation. Nevertheless, the findings of
this study and those of Nykänen et al. (2021), together covering data
from four successive breeding seasons (2018‒2021), suggest that wildlife
cameras are a feasible method for monitoring taiga bean goose during the
breeding season. First, they can reveal between-year differences both in
the numbers and occurrence of the species. Second, wildlife cameras set
at fixed sites are a cost-effective method to gather highly comparable
long-term data from remote breeding areas that are difficult to access
and cover using other survey methods. And finally, non-expert citizens
could be engaged in the monitoring based on wildlife cameras, making it
possible to cover a large number of potentially important breeding sites
(peatlands).