Atypical for northern ungulates, energy metabolism is lowest during summer in female wild boars (Sus scrofa)

Ethical statement

The present study was discussed and approved by the ethics and animals’ welfare committee of the University of Veterinary Medicine, Vienna, Austria, in accordance with good scientific practice and national legislation (GZ: BMWFW-68.205/0151-WF/V/3b/2016 and GZ: BMWFW-68.205/0224-WF/V/3b/2016). All methods were carried out in accordance with relevant guidelines and regulations. We confirm that the study was carried out in compliance with the ARRIVE guidelines. No plants or plant parts were used in this study.

Animals and study area

The study animals were kept in an outdoor enclosure (~ 55 ha, for details see “Supplementary Material”). The study enclosure was covered with a deciduous forest, mainly Turkey oak (Quercus cerris) and pubescent oak (Quercus pubescens) and included only few meadow patches. For the present study ten adult females, were used. We concentrated on females only because the live capture and handling of males are hampered by the large size and ferocity of boars. Also, due to competition and high levels of aggression between males during rut, the stocking of the enclosure was strongly female biased. During the study period (12/2016–01/2019), the animal density was ~ 1 adult female/ha plus up to 20 males (total) of different ages. Due to this relatively high density, animals were supplemented with 1–1.5 kg corn/individual once a day (at 2:00–14:00 h) at two feeding areas, each ~ 40 × 20 m. The enclosure was part of a game reserve, which was enclosed by 2.5 m high, solid, non-transparent fencing and was closed for the public. Thus, the study site provided an environment without disturbances due to hikers, bikers or straying dogs. There were no battue hunts or other disturbances due to hunting or forest management activities during the study period in the enclosure.

Animals were trapped once a year in autumn within the feeding sites to collect data on reproductive success and body condition of females and to separate some of them for implantation/explantation of loggers. While feeding, we closed the access gates and released the boars one by one trough a wooden corridor back into the enclosure. While in the wooden corridor we recorded the body mass of each individual (Gallagher SmartScale® 500, Groningen, Netherlands). Due to management reasons the juveniles (born in spring) were removed from the enclosure during this procedure.

Implantation of temperature and heart rate loggers

We implanted a heart rate logger (DST centi-HRT, Star-Oddi, Gardabaer, Iceland) and two custom-built temperature loggers in each of ten female wild boars in October/November 2016 and 2017 (age 5 and 6 years). All details about surgery techniques and anaesthesia protocols are provided in the “Supplementary Material”. Explantations were carried out approximately one year after implantations. The last explanation was carried out in January 2019. One female was implanted in two consecutive years. Mean body mass at date of implantation for all females was 71.8 ± 15.5 kg.

The heart rate logger was adjusted to record data at a time interval of 12 min to cover one year of data recording. To remove outliers, all initial data from these recorders were subjected to a running median over five consecutive values. The HR recorder was positioned subcutaneously, in proximity to the heart on the lateral rib cage, behind the moving area of the elbow, to avoid rubbing, or inserted and tethered into the ventral subperitoneal space caudal of the xiphoid process of the sternum.

The self-built temperature loggers were covered with inert surgical wax and had a weight of ~ 8 g. Time interval of recording was 4 min, the accuracy 0.01 °C. One of the two temperature loggers had an especially flat shape (3.4 × 1.9 × 0.5 cm) to fit smoothly into the subcutaneous neck region. The second temperature logger was placed into the intraperitoneal cavity, tethered at the Linea alba (diameter = 2.1 cm, height = 1.2 cm). For details on surgery, see “Supplement”.

We collected and evaluated a mean of 227.45 ± 160.69 days of heart rate recording per individual (SD, n = 11: 33 days, 58 days, 79 days, 89 days, 143 days, 189 days, 272 days, 345 days, 412 days, 421 days, 461 days), and a mean of 382.00 ± 100.17 days (SD), of subcutaneous logger recording per individual (n = 8: 143 days, 363 days, 411 days, 414 days, 419 days, 421 days, 424 days, 461 days). From the loggers implanted in the abdominal cavity we collected 338.71 ± 117.01 days (SD) per individual (n = 10: 140 days, 143 days, 363 days, 364 days, 411 days, 419 days, 421 days, 421 days, 424 days, 461 days). The hourly means of monitored heart rates of each animal over the course of the year are shown in Supplementary Fig. S1.

Activity data

To record the activity of animals, a telemetry system (Smartbow System, Zoetis, New Jersey, USA) was installed around the two neighbouring feeding areas and two close water ponds in the enclosure. The system consisted of a central solar power and computing station and ten receivers located at the height of 2–3 m. Part of the system were ear-tags (34 g; 52 mm × 36 mm × 17 mm, for details see “Supplementary Material”). The accelerometer (located inside ear-tags) measured triaxial acceleration (x, y, z). As an estimate of locomotor activity (ACT), we computed the total acceleration vector from sqrt (x2 + y2 + z2).

Climate and mast

The study site in Eastern Austria (altitude 130 m) is generally characterised by a Pannonian climate. According to long-term climate records, the mean annual temperature is 10 °C in combination with a mean precipitation of 600–700 mm and 1898 h of sunshine per year (ZAMG, 1971–2000).

We recorded ambient temperature (Ta) and black bulb temperature (Tab) at 2 m height directly at the study site (Vantage Pro 2 with black bulb extension, Davis Instruments, Hayward, USA).

To assess the extent of the acorn mast, each autumn seven nets, 4 × 4 m, were set up to collect acorns at random locations. The nets were regularly emptied between Sept. and Nov. each year, and the collected acorns were dried and weighed. In the autumns prior to the study (2016) and during both full study years (2017/2018) there was seeding of at least part of the oaks. Over ~ 90 days in each autumn we collected 52.4 g/m2, 134.8 g/m2, and 37.5 g/m2 acorn in 2016, 2017, and 2018, respectively. Thus, 2017 was a full mast year but there were acorns available in autumn throughout the study period.

Data analysis

To facilitate handling of data and to reduce autocorrelation we compiled and evaluated hourly means for all data, i.e., heart rates (HR; see Suppl. Fig. S1), intraperitoneal and subcutaneous body temperature (Tbip and Tbsc, respectively) and activity (ACT), as well as ambient air temperature (Ta) and black-bulb temperature (Tab). We further tested for effects of day of year (DOY) and hour of day (HOUR). We did not assess the influence of environmental conditions in different years, because due to logger-failures and thus scarcity of heart rates, all data were pooled for different years (with similarly warm conditions and food available year-round). Also, we did not further evaluate daily rhythms, because animals were always fed in the early afternoon, which may have influenced their timing.

We investigated the effects of season (DOY), hour of day (HOUR), and Ta on the response variables HR, Tbip, Tbsc, and ACT. We additionally used Tbip, Tbsc, and ACT as predictors for HR. As many of the relationships between these were non-linear, we used general additive mixed models (GAMMs), as implemented in package mgcv60 in R61. This function fits non-linear splines to the data, which are penalized for their “wiggliness”, i.e., the number of turning points in the fit. Because the data were repeated measurements, we calculated for all response variables mixed models with an intercept for each animal ID as a random factor (using s (ID, bs = ”re”)). Hence, these mixed models allowed for differences in the mean level of heart rates, temperatures and activities, between individuals. All residuals of models were approximately normally distributed, as inspected by normal quantile–quantile plots. Hourly means of the response variables contained various degrees of autocorrelation. This was corrected by including autoregressive order 1 (AR1) error models in GAMM-functions, which successfully reduced the autocorrelation at lag 1 to nonsignificant levels. This was confirmed by comparing the autocorrelation function of model residuals (ACF) before and after their correction. To illustrate the effects of independent variables, we show population-level predictions from GAMMs. These graphs contain rug plots to illustrate the distribution of independent variables. Because these plots were too dense for all original data (resulting in black bars), we show uniform random samples (n = 1000) from each independent predictor variable.

Because hourly mean data consisted of ~ 117,000 observations we used the mgcv function “bam”, which uses numerical methods designed for large datasets. To fit non-linear functions to predictors, we used the default thin plate splines. Only the cyclic variables DOY and HOUR were modelled using cubic cyclic splines, which are guaranteed to have identical start- and endpoints (e.g., at Jan 1 and Dec 31). GAMMs were always fitted using method REML. As Tbip and Tbsc were only moderately correlated (r = 0.30), both were entered simultaneously as independent variables in the model on heart rate.

We did not use partial regression plots from multiple regressions that included activity. This is because activity could only be recorded partly, in the vicinity of telemetry receivers. Thus, models that include ACT as well as all other predictors simultaneously, were restricted to ~ 7% of the data. However, we still used a full multiple regression model HR for the purpose of assessing relative variable importance (of DOY, HOUR, Ta, Tbip, Tbsc, and ACT). F-values from this model provide an indication of the importance of different predictors.

To model a possible role of solar radiation and basking we computed the difference between Tab and Ta, called Tdiff, which represents an index of radiation. We used again GAMMs to test if Tdiff would affect Tbip, Tbsc and HR after adjusting for effects of Ta, hour of day, and the random factor animal ID.

For a comparison of species we also computed monthly means and SEMs of HR in wild boars, and created a graph of seasonal time courses in other ungulates as published in Arnold2 that were kindly provided by the author. If not stated otherwise we provide means ± SEM.

Source: Ecology -

The use of multi-criteria method in the process of threat assessment to the environment

3 Questions: Daniel Cohn on the benefits of high-efficiency, flexible-fuel engines for heavy-duty trucking