Long-term population trends of the lesser horseshoe bat Rhinolophus hipposideros and the greater mouse-eared bat Myotis myotis in Poland

Bats are particularly susceptible to environmental changes because of their low reproductive rate, longevity, and high metabolic rates, which lead to relatively high food requirements. Thus, bat populations take a relatively long time to recover from increased mortality rates, and monitoring schemes should cover long time periods. In this work we analyzed the population trajectories of two bat species, Rhinolophus hipposideros and Myotis myotis , the most numerous in five caves in southern Poland, which are known as important bat hibernacula on a continental scale. Data were collected by regular counts in 1985–2001, depending on the particular cave; in addition, previous data on the number of hibernating bats in these caves, available since 1951, were taken from existing publications. We analyzed time-series data using average locality indices and TRIM methods, and both produced similar results. Generally, the populations of the two studied bat species showed recent increasing trends, especially visible as an effect of recovery after years of decline. The situation recorded in southern Poland is very similar to that described in other places in Europe, where recoveries of bat populations have also been observed in the last decades. Although it is difficult to present results from formal analyses, because of the lack of good data, at least some factors— less exposure to contaminants (pesticides, heavy metals), improving food availability due to climate change, and a lower predation rate (including human pressure), both in the breeding season and during wintering—positively affected both species.


Introduction
Bats are particularly susceptible to environmental changes because of their low reproductive rate, longevity, and high metabolic rates, which lead to relatively high food requirements 1 . Thus, bat populations take a relatively long time to recover from increased mortality rates 2,3 . Moreover, bat populations show responses to environmental stressors, ranging from alterations in habitat quality to climate change, as well as direct exploitation [4][5][6] , and bats are thus recognized as excellent indicators of anthropogenic changes in the environment [7][8][9] . Important for methodological, statistical and conservation purposes are the existing long-term series, coming mainly from winter bat censuses, especially in caves 10,11 .
For example, bats have been counted since 1944 in the Schenkgroeve, an artificial limestone cave in south Limburg in the Netherlands 12 , since 1946 in Hermann's cave in Lower Austria 13 , and since 1957 in some caves in the Moravian Karst, Czech Republic 14 .
Thanks to the results obtained, it was discovered that European bat populations, particularly the lesser horseshoe bat Rhinolophus hipposideros and the greater mouseeared bat Myotis myotis, declined dramatically in the second half of the 20th century [15][16][17][18] .
After the period of decline, since the 1990s bat populations have begun to recover 19 . An increase in the number of some species of hibernating bats has been reported from many European countries: Austria 13 , Belgium 20 , the Czech Republic 21,22 , the UK 23 , Italy 24 , Ireland 25 , the Netherlands 12 , Poland 26 , Slovakia 18 , Spain 27 , Sweden 28 and Switzerland 17 .
The reasons for these changes in population trends have not been conclusively identified 17,[29][30][31] . It is believed that the bat population declines and subsequent increases may be caused by a combination of various factors, such as the spread of chemical pollutants, habitat destruction, changes in landscape structure, disturbance and destruction of roost sites, climate change, declines in insect prey, competition for prey, genetic inbreeding, and diseases 17,19,[32][33][34] .
Both species selected for this study, the lesser horseshoe bat Rhinolophus hipposideros (hereafter: RHH) and the greater mouse-eared bat Myotis myotis (MYM) have similar preferences for shelters. In winter both species hibernate in caves, mines and other cave-like structures. They prefer places with high humidity (over 80%) and stable temperatures of 6-9 °C. In summer the females form maternity colonies in caves (Southern Europe) or in buildings with spacious roofs such as church attics and castles (Central Europe), where they give birth and nurse their offspring [35][36][37] . Both species are insectivorous, but they differ slightly in their manner of foraging and their diet. RHH forages exclusively in woodlands, preferentially in dense areas, capturing its prey using echolocation in flight. It preys mainly on moths and Diptera. MYM preys on large, grounddwelling arthropods such as beetles, crickets, and spiders, gleaning them from the ground 38,39 . The two species are the most numerous hibernating species in the caves of southern Poland, an important hibernaculum on a European scale 11,16,40 . These species are excellent for monitoring population trends, as they are easy to recognize and are relatively easy to count, because they do not hide in crevices 37,39 .
The aim of the study was to determine long-term population trends of the lesser horseshoe bat and the greater mouse-eared bat and the probable causes of changes in the numbers of hibernating bats of these two species.

Population size
The detected number of individual bats between 1950 and 2020 was very variable. The

Analysis based on average locality indices
For both species the β-coefficients for linear and quadratic functions were significant (Table 1). However, for both species a lack-of-fit test showed that a quadratic function was better than a linear one ( Table 1). Calculation of the extreme point of the function for the lesser horseshoe bat showed that the population decreased up to the year 1979, after which it increased (Fig. 1). In the case of the greater mouse-eared bat, the extreme point occurred in 1980 (Fig. 1). We did not find any significant differences between the two values (chi-square = 0.98, p = 0.86).

TRIM
All of the tests showed the same slope for both species. RHH and MYM are stable and show a moderate increase. Models using the five caves as covariates have higher AIC, smaller Wald statistics and higher standard deviation than models without them ( Table 2). The The overall slope of the linear trend model for MYM shows an increase (p < 0.01).
The indices show a negative trend between 1951 and 1953, followed by a moderate increase to 1981. The first peak of the increase in 1991 is mainly driven by the population dynamics in the Nietoperzowa cave.
Despite the differences between the species, their numbers (expressed as a TRIM index of year-to-year changes) were moderately correlated (Fig. 3). We also found a positive relationship between the average annual temperature and the numerical change in the TRIM index for both species, while no such significant relationship was found for precipitation or the number of days with rainfall in a particular year (Fig. 3).   We found a significant positive correlation between the population trend of both species (RHH and MYM) and the average annual temperature in 1951-2020, but we did not find such a correlation with precipitation or with the number of days with rainfall in particular years (Fig. 3). Numerous earlier studies have demonstrated the impact of climatic conditions on the activity, survival, and reproductive success of bats 50 64 . In the vicinity of the study area the only tree pest whose numbers increased was the pine sawfly (Acantholyda posticalis), and in the years 1971-2018 the fluctuations in its numbers were very small 65 . There is no information supporting the hypothesis that a shortage of insects could be the main cause of bat population changes. In Switzerland, Bontadina, et al. 66 found that changes in prey abundance are unlikely to explain the demography of the lesser horseshoe bat. However, the same factors that affected bat numbers may also have affected the number of insects.
Bats have been considered to have a particularly effective immune system, but numerous bacteria and viruses apparently remain non-pathogenic in bats, likely due to a long process of co-evolution 67 72 . Predation is therefore a marginal factor with little impact on bat mortality.
On the other hand, the Ojców National Park is exposed to relatively high tourist pressure, due to its small area (2146 ha Both studied species showed positive trends in population size over the long time period (1951-2020). Because the study has a correlational character, and because there was no access to detailed spatial and temporal environmental (and other) data, we discuss the main potential factors affecting both bat species according to the proposal of Bontadina, et al. 17 , and we rank their influence ( Predation, including human disturbance expert evaluation medium - Table 3. Assessment of factors that may have caused changes in long-term population trends of the lesser horseshoe bat and the greater mouse-eared bat. List of factors after Bontadina, et al. 17 . Factor assessment: "+++" -most important factor, "++" -might be important, "+" -might play some role, "-" -not relevant.

Conclusions
Both studied species, the lesser horseshoe bat and the greater mouse-eared bat, have shown a significant increase in wintering population size over the last 70 years. We noted two directions of change: until the 1980s the population of both species was decreasing, and after that time it was increasing. Similar trends have been observed throughout Although the search for factors affecting population size has only a correlative character, we must note that reduced exposure to contamination was probably the most important factor in the long-term changes in the populations of both of these bat species.
However, other factors, including climate change, food shortage and diseases, may also play some role in changes in bat populations.

Study area
The five studied caves (Table 4) Table 4. Characteristics of the studied caves (NP -cave located in the national park, G -gate at the entrance).

TRIM
The ALI method does not provide test statistics significant for population change, nor does it provide standard errors and 95% confidence limits. Thus, as a second approach to analysis of the population trends we used TRIM (TRends & Indices for Monitoring data method) 80 , implemented in the rtrim library for R. This procedure makes better use of the available data, especially when some data for the years are absent-a common issue in long-term time series (in our case in the years 1950-1980)-calculating standard errors and confidence limits and offering various test statistics; it also takes into account overdispersion and serial correlation of data 80 . TRIM is also capable of categorizing data by covariates and testing their influence on the observed changes, using Wald tests.
TRIM fits log-linear models and indices that represent the effect of change between years, which indicates the relative variation of the total population size. Two types of model-based indices) and, therefore, a lack of fit of the statistical model applied. In the next step indices are used to estimate a mean annual change rate 80 . This technique has been widely employed for the analysis of temporal series in bird populations [81][82][83] and also bat populations 18,19,24,27,29 . We developed models with and without covariates (five caves).
The best-fit models were selected according to goodness-of-fit tests (the Likelihood Ratio (LR) and Chi-squared tests) and the Akaike information criterion (AIC). A significance value for a model greater than 0.05 indicates that the data fit a Poisson distribution and, therefore, that the model can be accepted. Indices, overall slope and Wald tests remain reliable in case of lack of fit 80 . In case of overdispersion or serial correlation (default TRIM thresholds: > 3.0 and > 0.4 respectively) the Wald test for the significance of slope was employed 80 .
All calculations were performed in the language R 4.0.2 using the stats, rtrim, psych and ggcorrplot libraries 84 .

Data accessibility
Dataset available on request.