COVID‐19 lockdown measures impacted citizen science hedgehog observation numbers in Bavaria, Germany

Abstract The COVID‐19 pandemic has led to temporary changes in human–animal interactions due to changes in human activities. Here, we report on a surge in hedgehog observations during the first COVID‐19 lockdown in Germany in 2020, on the citizen science Web portal “Igel in Bayern” (Hedgehogs in Bavaria) in Germany. This increase in comparison with previous years was attributed to an increase in the number of people reporting hedgehog observations, rather than an increase in the number of hedgehog observations made by each observer. Additionally, in contrast to other studies on the effects of a COVID‐19 lockdown on observations recorded by citizen science projects, the share of observations made in more urbanized areas during the lockdown time was not higher than the change observed in less urbanized areas. This is possibly a result of the differences in COVID‐19 measures between Germany and other countries where preceding studies were carried out, in particular the lack of measures limiting traveling outdoor activities for citizens.

more present in areas with less traffic and fewer humans, but there was no lasting increase in population density (Gordo et al., 2021).
There are also reports of negative effects on wildlife originating from changes in human behavior during the COVID-19 pandemic (Gilby et al., 2021;Hiemstra et al., 2021).
In Germany, the first major COVID-19 lockdown lasted from March 22, 202022, , to May 4, 2020, and entailed restrictions on movement and contacts among people (Bayerische Staatskanzlei, 2020). Many recreational amenities were closed during the lockdown, but people were allowed to do outdoor activities, either alone or with another person from their own household, and there were no restrictions on where people could go to within the country.
Here, we report on a COVID-19 effect on observations of hedgehogs made through a citizen science portal in Germany. The NGO "Landesbund für Vogelschutz" (LBV) in Bavaria uses an app and a Web portal to monitor the occurrence of hedgehogs in Bavaria ("Igel in Bayern," www.igel-in-bayern.de). Observations started in 2015.
We use data collected by the portal to ask the following questions: 1. Did the first major lockdown in Germany lead to an increase in the number of hedgehog observations on the LBV app and portal?
2. Is a change in the number of observations due to an increased number of people making observations, or to an increased number of observations made by each observer?
3. Are there differences between urban and nonurban areas in any changes in the number of observations?

| ME THODS
Reporting of hedgehogs in the LBV app can be made with or without registering as a user. Observations of people that did not register contained the same data-fields as those of people that did register, except for the observer-ID. We obtained 107,440 observations of hedgehogs from the LBV database. Duplicate observations, as well as observations made outside of Bavaria, were eliminated from the dataset (Appendix S1). The year 2015 was also omitted from the analysis because it was the first year of the citizen science project and the year was not complete. After cleaning, 83,008 observations remained that were summarized per week. The years 2016-2019 were grouped together so that the 14,261 observations made in 2020 could be compared with expected values based on the preceding 4 years. To compare the number of hedgehog observations in 2020 with those of 2016-2019, we used generalized additive models and a spline smoothing function for the weekly observations (Appendix S1).
Confidence intervals were used for the comparison between years.
The same analysis was made to compare the weekly numbers of registered users that did observations in 2020 with those of 2016-2019.
To analyze the mean number of observations made by a person each week and compare this between years, we divided the number of observations made by registered users per week by the number of registered users in the same week and used these numbers in our analysis. Generalized additive models constructed conducted with the "mgcv" (Wood, 2021) package, and outputs were compared with "itsadug" (Van Rij et al., 2020)  would not be lost, so small islands of imperviousness would not result in small green inner-city area being classified as less urbanized.
Observations were then divided into increments of 20% impervious surface density. The percentage of observations and observers during the weeks of the first COVID-19 lockdown in every imperviousness class was compared across the years of interest as follows: similar to Crimmins et al. (2021), linear models were built between 2016 and 2019 for every imperviousness class and these were used to create an expected value with a 95% prediction interval for 2020 (Appendix S1). These were used as a measure of significance by comparing the realized value of 2020 to the prediction interval.

| Increase in the number of registered users during the COVID-19 lockdown
The number of registered users in 2020 was higher than the number

| No significant changes in the percentage of observations across different levels of urbanization
No significant differences were registered between the realized and predicted percentage of observations in different levels of urbanization in 2020. There was, however, a small but significant difference in the percentage of observers that did observations at the lowest level of imperviousness, compared to what would be expected from the prediction based on preceding years. That value was 25.16%, 2.81% less than the predicted value of 27.97% (Figure 2; Appendix S1).

| DISCUSS ION
We found a statistically significant increase in the number of hedgehog observations during the first COVID-19 lockdown period 2020 compared with the same time-period of preceding years. This result mirrors results of other studies that investigated the effect of the COVID-19 lockdown on animal observations made in citizen science projects (Basile et al., 2021;Crimmins et al., 2021;Manenti et al., 2020). Additionally, we found a pattern that to our knowledge was Error bars indicate 95% prediction intervals of the predicted values of linear models (Appendix S1). Stars* indicate realized values in 2020. Realized values falling outside of their respective modeled prediction's 95% prediction interval are treated as significantly different from expectations (significant deviation was only found for observers in the 0-20 imperviousness class). The full model output, including values for the preceding years, can be found in Appendix S1, Figures A1 and A2 not reported before, namely that this increase was primarily due to an increase in the number of observers, and not an increase in the number of observations per observer. In contrast to other studies, we found that the increases in both the number of observations and the number of observers during the lockdown period occurred equally in both more urbanized and less urbanized areas.
An increase in the number of hedgehog observation can be due to a higher observation activity of humans, or a higher activity of hedgehogs. It is very likely that the increase in hedgehog observations during the first German COVID-19 lockdown is attributable to the increase in the number of human observers, and not to an increase in hedgehog numbers or hedgehog activity. This is corroborated by the finding that there was no difference in the mean number of hedgehogs that participants reported during the lockdown period vs. the preceding years (Figure 1: 1.C and 2.C).
After the lockdown period, the number of observations and observers quickly returned to normal. There was even a short significant decrease in the mean number of hedgehogs participants reported a few weeks after the lockdown compared with the preceding years. It can be argued that this is due to a loss of interest in looking for hedgehogs because other temporarily unavailable potentially competing activities had resumed. Competing activities could be, for example, activities pertaining the hospitality industry and cultural activities such as visiting museums, activities which were not possible during the first COVID-19 lockdown in Germany.
While other studies have found that there was an increase in the share of urban observations as a total of all observations made during the lockdown, we did not find such a difference between more and less urbanized areas. A possible reason for the differences in the share of observations in more versus less urbanized areas in our study compared to others (Basile et al., 2021;Crimmins et al., 2021;Kishimoto & Kobori, 2021;Manenti et al., 2020;Rose et al., 2020;Sánchez-Clavijo et al., 2021) is the differences in measures between Germany and the countries investigated in other studies (Hirsch, 2020). During the first lockdown in Germany, many recreational amenities were closed. While people were allowed to do outdoor activities, either alone or with another person from their own household, there were no restrictions on where people could go to within the country. This is different from, for example, France and Italy, where inhabitants were only allowed to go outside alone and near their home, or the United Kingdom, where nonessential movements were banned and inhabitants were only allowed to go outside once per day, alone or with people from the same household (Hirsch, 2020). A report by the German hiking institute (Deutsches Wanderinstitut) indicated that people in Germany went on hikes more often than normally in April and May of 2020, and those responsible for the hiking routes indicated that there were often more hikers on the hiking routes (Smolka et al., 2021). Nonetheless, there was a decrease in physical activity of people, as shown, for example, for the German state Bavaria (Huber et al., 2020). We do not know where the observers live and how far they travelled to the observation point, as people did not register with their address on the portal.
It may be that more observations were made near their house, or that they went further away. We cannot distinguish between a situation in which more people left the house during the lockdown and a situation in which the same number of people went outside during the lockdown but a higher fraction of them reported hedgehogs. In both cases, the number of observers and the number of observations would increase, and we think that the first scenario is more likely. The difference in restrictions and subsequent response in activity patterns between Germany and the countries investigated in other studies could nonetheless explain why there was, in contrast to studies in other countries, no noticeable increase in the share of hedgehog observations in more urbanized areas, compared with less urbanized areas in our study while the absolute number of observations increased.

| CON CLUS IONS
Our study showed that the first COVID-19 lockdown in Germany led to an increased number of people reporting hedgehog sightings on a citizen science portal, resulting in a higher number of hedgehog observations in this period. When the lockdown was over, the number of observers dropped back, so that there was not significant difference anymore to preceding years. Our results therefore suggest that there is great potential to increase animal observations at times when people have time for such activities, but also that reporting animal observations to a citizen science portal is limited by the presence of other potential activities. Our results also suggest that the way in which human movement patterns were restricted during the first COVID-19 lockdown influenced the reporting of animal observations in citizen science portals. In the case of Germany, humans were not confined to stay in areas with higher human population density, and hence, reporting of hedgehogs also occurred in areas that were less urbanized where people could go for outdoor activities. Thus, considering the circumstances under which citizen science data were collected can help to interpret observed changes in reporting patterns.

ACK N OWLED G M ENTS
We thank the Landesbund für Vogelschutz e.V. (LBV) for supplying us with their observation data. We thank Norbert Schäffer for critical discussions. Open access funding enabled and organized by Projekt DEAL.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

A PPE N D I X 1 A D U PLI C ATE O B S E RVATI O N S
Because observations could be made by both registered and nonregistered participants, of which only the observations of registered participants could be connected to an anonymized participant ID, while the non-registered participants by default got a Participant ID of "0," these two groups had to be cleaned individually.
For preparation, the data were split into two documents: one with registered participants and one with non-registered participants.
After removal of duplicates, the final datasets were then again merged. Here, we describe the steps to remove duplicate observations, followed by a reproducible ArcGIS-specific description.
First, a shapefile (1) with 50m buffers around each datapoint was created. The shapefile with these buffers was then split into multiple shapefiles (2): one for each day in the dataset.
For each of these shapefiles, that is, for every day of the calendar day where there were observations, the observation points were pairwise dissolved, so that two observations where the buffers overlapped to any extend and that had the same data on them would be considered as identical observations. If the observations were considered to be identical, that is, referring to the same hedgehog seen by the same person, the two observation buffers would be merged, creating a larger buffer made from the combined dissolved buffers. These new buffers would gain the observation-ID of the first if the observations out of which the dissolved buffers were made.
After this was made for each created shapefile created in (2), that is, for each day, these shapefiles (2)

DATA S E T
Many other types of information that were additionally collected in the Igel in Bayern platform were not used in this analysis, because they were not relevant to the question asked. This information included, but was not limited to, the living or dead status of the hedgehog, information on the habitat/place where the hedgehog was observed, and if potential death was caused by traffic. A full overview can be found on the website (https://www.igel-in-bayern.de/).

CO M PA R I N G O B S E RVATI O N S ACROSS Y E A R S
To were then compared with the realized values for 2020, and if the realized value fell outside of the 95% prediction interval for 2020 this was treated as a significant deviation. This method has the benefit of accounting for ongoing changes throughout the preceding yearsthrough the slope of the linear model from 2016-2019 for each imperviousness class-something that methods such as a chi-square would not do. In the main text, only the comparisons between predicted and observed values for 2020 are shown. Here, the model outputs and the extrapolated values for 2020 will be shown. Figure   A1 shows that for all the imperviousness classes the realized value for 2020 does not significantly deviate from the expected value, that is, there is no significant difference between the realized and expected values in the proportional number of observations in each imperviousness class in 2020. Figure A2 shows that only in the lowest imperviousness the realized number of observers significantly deviates from the expected number of observers. There is no significant difference between the realized and expected values in the proportional number of observers in the other classes in 2020.
R-code for this analysis can be found from lines 288 onward in the attached RMarkdown file.