Weather conditions during hunting season affect the number of harvested roe deer (Capreolus capreolus)

Abstract Due to human‐induced climate and landscape changes, distribution and abundance of many ungulate species have increased worldwide. Especially in areas where natural predators are absent, hunting is the essential management tool for regulating ungulate populations. Therefore, understanding the factors associated with harvest rates is the first step toward an adaptive management approach. Weather influences hunter and ungulate behavior and thus presumably harvest, but how and which meteorological parameters are linked to harvest numbers have rarely been evaluated. We used nearly 65,000 “sit and wait” and driven hunt harvests of roe deer (Capreolus capreolus) in Bavaria, Germany, and weather data from 2008 to 2017 to test for factors affecting roe deer harvests (i.e., temperature, rain hours, wind speed, sunshine duration, snow depth, workdays vs. weekends, month) using zero‐inflated negative binomial mixed‐effect models. Our results reveal that, besides workdays, high temperatures and prolonged rain resulted in fewer harvested animals, whereas sunshine duration in summer and snow height in snow‐rich areas partially favored harvests during sitting hunts in summer and winter, respectively. The influence of wind speed varied over the course of the year. In summer and autumn, wind speed commonly had a negative effect, positively affecting harvests in winter in some regions. Daily harvest numbers decreased during the summer and autumn hunting periods (May till mid‐October), while they increased during the winter period (mid‐October till mid‐January). Interestingly, harvest success during driven hunts, which are planned well in advance and therefore take place largely independent of weather conditions, was similarly affected by the weather. This result suggests that the inferred weather influence is not only due to the hunters' decisions but also due to deer behavior. Since many ungulate populations may further benefit from climate change, building an understanding of the relationship between hunting success and weather will aid adaptive ungulate management.


| INTRODUC TI ON
Many ungulate species have expanded their distribution and increased in abundance throughout Europe and North America (Côté et al., 2004;Diekert et al., 2016;Milner et al., 2006). Factors that have facilitated these trends include increased availability of forage and suitable habitats due to phenological changes in response to climatic changes (e.g., milder and shorter winters) or agricultural and forest management, as well as reduced abundance or even absence of natural predators (Gortázar et al., 2000;Milner et al., 2006;Rickbeil et al., 2019). Like other ungulates, roe deer (Capreolus capreolus) primarily benefited from these changes. Roe deer are widespread throughout Europe (Putman et al., 2011) and are the most abundant ungulate in the federal state of Bavaria, Germany, with an increasing population trend over the last decades Due to increasingly favorable environments and the absence of natural predators, hunting is the main management tool for population regulation in many areas. State authorities commonly guide the management of game populations based on a hunting system, usually according to the age structure, sex ratio, and density. On the other hand, state authorities can influence hunting activities by restricting the hunting period, times of the day when hunting is allowed, hunting equipment, hunting methods, or the number of hunters per unit area (Mysterud et al., 2019;Stedman et al., 2004). In Bavaria, roe deer may be hunted between 5 and 9 months of the year depending on the respective sex and age class (relevant legal regulation: §19 AvBayJG), and hunters have the responsibility to fulfill the quotas set by the authorities. In contrast to many other countries, the comparatively long hunting period allows hunters to choose when to hunt to obtain their harvests freely.
The hunting success of natural predators depends, for example, on the individual (e.g., physical condition, experience), the target prey (e.g., physical condition, densities), and weather conditions (Mech et al., 2003). For hunters, the weather is also an important factor influencing their decision to go hunting, along with factors such as attitude, social norms, preference for target animals, knowledge, or time availability (Bhandari et al., 2006;Darimont & Child, 2014;Diekert et al., 2016;Kennan et al., 2008;Rivrud et al., 2014). Relationships between weather and hunting success have been shown for other (deer) species (Curtis, 1971;Fobes, 1945;Hansen et al., 1986;Leorna et al., 2020). For instance, for Virginia, USA, Curtis (1971) reported a negative relationship between white-tailed deer (Odocoileus virginianus) seen and the mean daily temperature, but a positive one with total daily precipitation.
Upon such presumed weather effects, climate change could influence harvest numbers via altered prey activity patterns in response to weather and hunter decisions as well as environmental changes triggered by climate change (e.g., prolonged vegetation cover, favorable habitats resulting from natural disturbances). Here, we combined daily roe deer harvest and weather data covering ten years to test which weather factors affected the daily harvest rate of roe deer in seven regions across Bavaria. Such empirically based insights can be the first step toward an adaptive management approach.
We hypothesized that both hunters and roe deer alter their behavior in response to the weather. More specifically, the overall weather influence on successful hunts may include the combined probability (see scheme in Figure 1) of (a) a hunter going to the field (weather-dependent decision to go hunting, especially true for "sit and wait" hunts, herein sitting hunts); (b) encountering roe deer during a hunt (weather-dependent likelihood of sighting deer); and (c) actual hunting success during a hunt. While other parameters may also influence all three events, these other factors should have a stronger effect on b and c than on a (e.g., a: time available to a hunter for hunting, hunters' motivation; b: hunters' skills and experience in tracking animals, knowledge about the hunting area and of how deer may react to certain weather conditions, the influence of external disturbances; c: attitude, preferences, safety considerations, shooting skills).
Regarding weather, we expected a negative influence by wind and long-lasting precipitation on harvest numbers as previously shown for roe deer, white-tailed deer, or alpine chamois (Rupicapra rupicapra) (Brivio et al., 2016;Curtis, 1971;Hansen et al., 1986). With higher wind speeds and, therefore, more noise, it is more difficult for roe deer to identify potential threats, leading to increased vigilance and preferred use of safe habitats where deer are difficult to detect (Lone et al., 2014;Mysterud & Østbye, 1999). We also predicted lower harvest numbers with heavy or long-lasting rain due to lower motivation of hunters to go hunting (Rivrud et al., 2014) and less visible deer because they use shelter and closed stands more frequently (König, 1987;Mysterud & Østbye, 1999). For temperature effects on harvest numbers of roe deer (Curtis, 1971;Progulske & Duerre, 1964), we predicted fewer harvests on hot summer days, F I G U R E 1 Scheme of hypothesized combined effects of weather and other factors on hunted roe deer numbers. (a) Decision of the hunter to go hunting; (b) sighting of roe deer; (c) successful hunt. Only if a, b, and c occur, a harvest date (=1) is recorded. We assume weather to influence both a and b, while other factors acting at all steps associated with intense solar radiation, and cold winter days, because of less visibility due to the preferred use of cover for thermoregulation (Mysterud & Østbye, 1999). In winter, we expected a positive influence of the first snow due to a higher detection probability of deer by hunters (Zagata & Haugen, 1974), but an adverse effect with increasing snow depth as deer are likely to prefer forest stands with lower snow levels and thermal protection (Courbin et al., 2017). Due to a presumed higher motivation to go hunting on weekends (A), we expected a higher number of culled deer on Saturdays and Sundays than on workdays (Mysterud et al., 2019;Rivrud et al., 2014). Finally, to address the effect of weather on the hunters' decision to go hunting, an additional model was run exclusively for driven hunts.
These hunts take place in winter, often on Fridays and Saturdays, and are planned far in advance (e.g., in summer), and concurrently, there should be no weather effects on the decision to participate in the hunt. Consequently, we expected more pronounced effects of weather-related factors on the number of hunted roe deer for the sitting hunt models than for the driven hunt models.

| Roe deer hunting data
Our study was conducted in Bavaria, the most southeastern federal state of Germany (Figure 2), located in the warm-moderate climate zone with an average temperature of 7.8°C and 933 mm annual precipitation (period 1971-2000)  where the decision to go hunting is made is likely closely related to the time and place where hunt and harvest occur (Figure 1a-c). In contrast, for driven hunts, which commonly take place during the day in winter, the decision to go hunting by the individual hunter ( Figure 1a) should be negligible since these hunts are planned and consequently signed up for well in advance. In remote alpine areas, a hunting session may require more time. Thus, taking a day off during the week for single hunting may be a common phenomenon to avoid disturbance due to those seeking relaxation on the weekends (personal communication, state hunting authority, and local heads of offices for Food, Agriculture and Forestry).
We analyzed daily roe deer harvest data from seven out of 41 BaySF management units (18.2% of the area) from May 1, 2008, to December 31, 2017 ( Figure 3). Hereafter, we will call these seven forest management units "regions" for simplicity. Two regions are located in the northwest of Bavaria (Heigenbrücken, Rothenbuch) characterized by above-average temperatures, the least number of snow days, and high percentages of deciduous forests (50% and 75%, respectively) ( Table 1). Two regions are located in the coniferdominated region in the east (Roding, Burglengenfeld with 70% and Game management may slightly differ between regions due to climate, landscape, and species composition ( assume that the overall harvest strategy did not vary between regions except that professional hunters hunt in parts of the alpine regions Sonthofen and Ruhpolding in addition to recreational hunters. This also applies for parts of the Munich region (~10% of its area).
The other regions have no additionally employed full-time hunters.
Daily harvest data were available at the district level (mean size:

ha). Each region comprises 36 (Rothenbuch) to 90 (Ruhpolding)
districts. Road-kill and cases of natural mortality were excluded from the data set. In total, 60,269 harvests were included in the analysis for the sitting hunt models and 4,121 for driven hunt models. Driven hunts were mainly conducted between October and January in all regions and years by recreational and professional hunters. Harvest data were assigned to the nearest weather station (see below), and the data set was divided into days with harvests (number of harvested deer aggregated per climate station) and other days without harvests. Data on unsuccessful hunts were not available.

| Weather data
For the study period, daily weather data of 97 climate stations were available from the German Meteorological Service (Deutscher Wetterdienst, 2020aWetterdienst, , 2020c. We chose the nearest weather station for each district and used 29 stations (two to seven per region, see Figure 2). Wind speed was only available for ten and sunshine duration for 15 stations, so we used the closest station with this measurement alternatively. We also calculated rain hours per day as hours with precipitation ≥1 mm (Deutscher Wetterdienst, 2020b).
The final set of explanatory weather factors comprised mean daily temperature (°C), sunshine duration (h/day), rain hours (h/day), and average wind speed per day (m/s) as well as daily snow depth (cm) for the winter months (October 15-January 15).

| Statistical analyses
First, we tested for correlated explanatory variables at a Pearson rank correlation threshold of |r| ≥ 0.7. Only the variables mean precipitation and rain hours were highly correlated. Because the diurnal cycle of the amount of precipitation is more pronounced than the duration of precipitation in Germany (Ghada et al., 2019), we preferred rain hours/day over mean precipitation to be included in the models.
We modeled the number of shot roe deer per day for three specific periods of the hunting season separately to account for potential seasonal effects in hunting effort, deer behavior, environmental conditions, and, notably, the particular hunting seasons (see before).
Specifically, we differentiated between "summer" from May 1 <>till August 31 on bucks and yearlings; "autumn" from September 1 till October 15 on bucks, yearlings, females, and fawns; and "winter" from October 16 till January 15 on females and fawns. The terms "summer," "autumn," and "winter" models thus relate to the hunting periods, not to the meteorological seasons.
To consider the excess zeros due to the assumed zero-harvesting days in the models, we used a zero-inflated negative binomial model (ZINB) with the glmmTMB package (Magnusson et al., 2019) in the statistical software R (R Core Team, 2019). ZINB was preferred over zero-inflated Poisson (ZIP), generalized linear (GLM), or negative binominal (NB) models as indicated by a significant Vuong test (Vuong, 1989, p < .05). In general, glmmTMB has been shown by Brooks et al. (2017) to be more flexible for estimating those models via maximum likelihood estimation and faster than packages that use Markov chain Monte Carlo sampling.
To account for potential interannual variations, we included year as a random effect. The factors workday versus weekend and month were implemented to test for possible variation during the week (workday = Monday till Friday, weekend = Saturday, Sunday) and during the hunting periods (reference categories are May in summer, September in autumn, and October in winter models). Workday and month combinations may also account for the factor time availability (gun light hours not overlapping with standard working hours, see Figure 1a).
We calculated 21 separate sitting hunt models for the seven regions and three hunting periods to reveal possible differences between the management units. Furthermore, for all regions and periods of the hunting season, an overall model including region as a random effect was calculated. Driven hunts were limited to the winter hunting season, and we did not have enough data to fit robust models for the individual regions. Therefore, only one model for all studied regions in Bavaria was built and compared to the respective overall sitting hunt model in terms of (significant) variables and effect magnitudes.
We applied an ordered-backward stepwise selection approach using the buildmer package (Voeten,

| RE SULTS
The total number of harvested roe deer ranged from 4,285 to 11,731 per region and from 3,434 (August) to 13,170 (May) per month over the 10 years analyzed. Hunters shot most roe deer on May 1 when the hunting season started, while numbers were lower in the following months of the hunting season ( Figure 3). In sum-

| Weather effects on harvest numbers
The relative risk (RR) of roe deer harvest was considerably driven by weather and calendar variables, summarized in Table 2  In the winter models, snow depth had a consistent but small negative influence in Heigenbrücken, Rothenbuch, Roding, and Sonthofen (RR 0.98-0.99), no effect in Burglengenfeld and the overall model, and only in Ruhpolding (RR 1.04) and Munich (RR 1.01) did higher snow depths lead to higher harvest numbers.

| Driven hunts in winter
We compared the RR of roe deer to be shot in winter for driven hunts (Table 3) to the respective overall sitting hunt model ( contrast to our prediction, the weather influence did not decrease for the preplanned and organized driven hunts, for which the weather influence on hunters should be negligible (see Figure 1a).

| Effect size of weather variables
The ΔRR values, which indicate the effect size, for example, for temperature, the difference between a medium warm and a medium cold day (75%-25% quantile), allow a comparison of TA B L E 2 Relative risks for roe deer being harvested per region and the overall model  Figure 4 and

| D ISCUSS I ON
The number of harvested roe deer in seven regions of Bavaria was linked to meteorological parameters and could thus also be influenced by climate change in the long run. The traditionally anchored assumptions of inhibiting influences of rain and wind on harvest success could be confirmed (or partially in the case of wind), whereas the results on favoring influences by sun and snow were less clear.
Sun and wind were seasonally restricted to summer and winter, respectively, and may require more intensive investigations on larger data sets. Higher temperatures were most importantly associated  Significance: significant at 0.1, * significant at 0.05; ** significant at 0.01; *** significant at 0.001.
Relative risks are given for an increase of one unit per selected variable from the parsimonious negative binominal zero-inflated models per region and hunting period (summer = May, June, July, August; autumn = September, October 15, winter = October 16, November, December, January 15). Month effects are compared to May (summer), September (autumn), and October (winter hunting period).

TA B L E 2 (Continued)
with pronounced lower harvest numbers across all hunting periods and regions, both for sitting and for driven hunt models. In Bavaria's largely recreational hunter-dependent hunting system, daily harvests were greater on weekends than on workdays and exhibited changes as the hunting periods progressed. Following a general discussion on the effect pathway of the weather influence, we will discuss the effects of each weather variable as well as calendar effects to derive management options from these findings.

| Self-fulfilling prophecy or independent weather effect?
Our data represent the combined effect of hunter effort, skill, behavior, and external circumstances such as weather, although the modeling approach could only take total weather effects and calendar variables into account. Workday-weekend differences clearly indicate a hunter availability issue, much like that reported by Mysterud et al. (2019) for recreational hunters in Norway. This latter study also lists other nonmeteorological factors, such as hunter skills and attitude, potential disturbance by visitors, safety considerations regarding the shot, which we could not disentangle in our study.
Weather could theoretically act at three nodes (see Figure 1 (Mysterud & Østbye, 1999). Thus, movement rates are low, and animals are less visible to hunters (Lone et al., 2014

| Temperature effects
Among the meteorological variables, the temperature had the strongest overall effect on harvest rates. It was always (i.e., for all regions and hunting periods) included in the most parsimonious models, uniformly fewer roe deer were harvested at higher temperatures, and in relative terms, the temperature had the strongest absolute influence. It is known from the literature that spring temperatures positively influence roe deer activity (Pagon et al., 2013). Compared to the May-July period, harvest numbers in August were generally lower, most likely due to a behavioral change during the rut (Krop-Benesch et al., 2013;Picardi et al., 2019), but also likely due to lower hunting effort during summer vacation. While the temperature was the main factor explaining the white-tailed deer detectability during spotlight surveys in summer (Progulske & Duerre, 1964), the effect of high temperatures on harvest rates is ambiguous in the literature. Significance: significant * at 0.05; ** at 0.01; and *** significant at 0.001 level.
The overall model relative risk calculations (RR) are given for an increase of one unit per selected variable in the most parsimonious negative binominal zero-inflated model. Month effects compared to October.
but other studies have also reported contrary results (Fobes, 1945).
In general, people prefer outdoor activities at higher temperatures up to ~27°C (Prettenthaler et al., 2015). If recreational behavior can be assumed for hunters, a higher number of hunters should be active at higher temperatures, leading, for example, to more deer seen (Curtis, 1971). However, perhaps there is a trade-off between hunting and pursuing alternative recreational activities when temperatures are ideal. The temperature influence on hunting success varied with season and land cover type in other studies (Rivrud et al., 2014), but in our study, the influence of temperature was uniform throughout the year. In winter, hunters typically assume higher chances of harvesting game at lower temperatures because bait sites are visited more frequently (Ossi et al., 2017;Ossi et al., 2020). Consequently, hunters may increase their effort during these periods.

| Rain and wind effects
Rain hours often decreased harvest success in the summer and autumn hunting periods and partly in winter, although their influence was comparatively small and significant only in ~2/3 of the cases.
This effect was noticeable in the southern alpine areas with high rainfall for all seasons. In the comparably drier parts of northern Bavaria, rain even had a positive effect on roe deer harvest numbers in the winter model. The influence of precipitation is controversially discussed in the literature, ranging from no effect on the likelihood to shoot a red deer (Cervus elaphus) (Diekert et al., 2016) to a negative effect on daily harvest of white-tailed deer (Hansen et al., 1986) and Dall sheep (Leorna et al., 2020), or even increased harvest rates with more frequent precipitation days (Fobes, 1945). The latter result was explained by better and calmer hunting conditions when the forest floor was moist. Rain only slightly influenced whitetailed deer movements (Webb et al., 2010) or had no effect (Beier & McCullough, 1990). Additionally, precipitation may also influence hunter behavior. In general, fewer outdoor activities are conducted during rainy periods (Spinney & Millward, 2011), and in a survey of hunters, precipitation was cited as one of the worst conditions for hunting (Curtis, 1971).
Like temperature, the wind speed was selected in all models and was significant in half of them, but its effect direction and magnitude

| Sunshine in summer and snow in winter
Sunshine duration only positively influenced the number of harvests five times in the summer model, once a negative and twice no influence. In addition, these effects were rather small. According to the literature, white-tailed deer showed higher activity rates during cloud-free and cold days and selected open areas to benefit from higher solar radiation (thermoregulation) (Beier & McCullough, 1990).
Our study cannot separate whether the influence of sunshine is related to the potential thermoregulation that roe deer may seek or hunter preferences. However, we suppose that sunshine hours rather have a stronger effect on the hunters because closed forests dominate most regions in this study, and cover for thermoregulation is provided nearly everywhere.
The effect of snow depth in the winter hunting season varied in magnitude and direction among the regions. In 50% of the models' snow slightly decreased harvest rates. In the presence of snow, higher hunting success is assumed due to better tracking possibilities (Fobes, 1945) and easier and more reliable identification of animals due to the higher contrasts with the surrounding (Mysterud et al., 2019). In contrast, there is evidence that deer decrease activity rates even at low snow depths due to locomotive constraints or lower food availability (Beier & McCullough, 1990;Gaudry et al., 2015). Even in our most southern alpine study regions with more snowfall, the influence of snow was only strong in Ruhpolding but not in Sonthofen. We assume that snow cover is most variable at the onset of winter, whereas, during the peak of winter and in alpine areas, high snow cover generally inhibits hunting activities.

| Calendar effects
Workday versus weekend undeniably made a difference because, in 15 of 18 models for nonalpine regions, harvest rates were greatly reduced on workdays. This most likely reflects different frequencies in human hunting activity (Ciuti et al., 2012) instead of weather effects.
For example, the workday was one of the strongest predictors of hunting effort and success on red deer in Norway, with higher culling rates on weekends (Mysterud et al., 2019;Rivrud et al., 2014), and also elk (Cervus canadensis) in Idaho, USA, were more frequently killed on weekends (Gratson & Whitman, 2000). This reducing influence of workdays was found for regions without professional hunters (Heigenbrücken, Rothenbuch, Roding, Burglengenfeld, and Munich for 90% of the area), where recreational hunting permit holders are predominantly active in addition to other BaySF employees. Interestingly, the variable workday was not selected for the two alpine regions (Sonthofen and Ruhpolding). This discrepancy could be because touristic activities hinder hunting in the Alps on weekends and change animal behavior, known as the "weekend effect" (Nix et al., 2018). Thus, professional hunters might concentrate their activities on workdays, and even recreational hunters may prefer to take one day off during the week to reach their remote hunting areas. Yet, in general, hunters who hunt during workdays are considered more effective than casual ones (Rivrud et al., 2014), an observation that we cannot evaluate due to a lack of data on hunter effort (e.g., number of hours spent hunting).
We also observed temporal patterns within hunting periods, namely decreasing harvests over time in summer and autumn and increasing during the winter hunting period, except in the snow-rich southern regions. It is reasonable to assume that hunter effort decreases as the hunting season progresses on specific target animals (e.g., the hunting season on bucks and yearlings starting on May 1 yielded the maximum daily harvest) and may again increase when the annual hunting season closes in January to meet quotas (Diekert et al., 2016). A human-induced pattern of shooting more reindeer early in the season and on weekends was also reported by Mysterud et al. (2019) for Norway. However, an indirect effect of hunting on habitat selection by roe deer is possible as they may alter their use of space with the start of the hunting season (Bonnot et al., 2013) to avoid risky habitats (Padié et al., 2015).

| CON CLUS ION
A next step to consolidate and improve our findings would be to include new data on hunter behavior and unsuccessful hunts, which were not available for our study. Since we did not have information on hunting effort per unit time, the number of hunters or deer sighted, it was impossible to account for these confounding factors, and our results show the combined effect of all possible influences.
Since large, systematic data sets on successful and unsuccessful hunts are not currently available, experimental approaches could help disentangle the effects between animals, humans, and weather.
Similarly, our analysis could not account for other confounding factors, such as (also weather-driven) human or other disturbances on wildlife and hunting activities, for example, recreational activities.
Although we are aware of this shortcoming, we were interested in understanding the net weather effect on the harvests.
We fitted models for each hunting period, assuming constant hunter effort across them. However, harvests in individual seasons are not independent. A deer can only be killed once. Thus, high success at the beginning of the hunting season lowers game densities at the end and vice versa. Especially at the start of the respective hunting season, careless and less experienced animals may be shot and/or are easier to hunt. Moreover, strictly mathematically, the maximum hunting quotas would have to be considered as censored data. However, we believe it is reasonable to assume that hunting effort is intensified at the end of winter (as suggested by our models) if it is foreseeable that the annual quota will not be met (Diekert et al., 2016).
Increasing ungulate densities and concurrent climatic changes can modify the landscape in which both wildlife and hunters operate and live. For example, because ungulate densities can alter plant species richness and forest stand composition and regeneration, that is, factors that will also contribute to the resilience of forests to climatic changes, research-driven adaptation is needed to adjust ungulate management. Study results such as those presented here can contribute to our understanding of wildlife management under changing climatic conditions. Specifically, in the hunting system we investigated, this could be implemented through regionally adapted spatio-temporal hunting strategies. For example, an interval hunting system with varying intensity of hunting pressure to create spatiotemporally structured risk landscapes (Norum et al., 2015) would be one possibility. Hunting could be stopped completely during the more unsuccessful summer months to compensate for more intensified hunting pressure when success rates are higher. Also, the end of the hunting season for bucks in October could be extended. Hunters should be encouraged to conduct hunts, preferably in suitable weather conditions, and on workdays rather than concentrate their hunting activities on weekends. Research on how climatic changes will affect roe deer populations and especially spatio-temporal patterns of habitat use at multiple scales will provide further information for adaptive management strategies. Overall, ungulate management strategies need to be adapted to local environmental conditions, ungulate densities, and the specific hunting system.

ACK N OWLED G M ENTS
We kindly acknowledge the provision of daily hunting data for seven management units through the BaySF (Bayerische Staatsforsten AöR) as well as the use of meteorological data of the German Meteorological Service (DWD). We are grateful for intensive discussions with Reinhard Menzel on all issues related to hunting in Bavaria. We thank two anonymous reviewers and Benedikt Gehr for constructive comments on an earlier version of the manuscript.

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
Climate data are freely accessible from the German Meteorological Service data server: http://opend ata.dwd.de/clima te_envir onmen t/CDC/obser vatio ns_germa ny/clima te/. The hunting data set is not freely downloadable for reasons of traceability and privacy regulations of the hunters. The data owner is an institution under public law (BaySF AöR) that assures long-term archiving of data and provisioning of data, following a specific law of public information on environmental data.