SUPPLEMENTARY INFORMATION Articles https://doi.org/10.1038/s41559-017-0400-7 In the format provided by the authors and unedited. Regulated hunting re-shapes the life history of brown bears Richard Bischof 1, Christophe Bonenfant 2, Inger Maren Rivrud 3, Andreas Zedrosser 4,5, Andrea Friebe 1, Tim Coulson 6, Atle Mysterud 3 and Jon E. Swenson 1,7 1 Faculty of Environmental Sciences and Natural Resource Management, Norwegian University of Life Sciences, Ås, Norway. 2 Université de Lyon, F-69000, CNRS, UMR, 5558, Laboratoire de Biométrie et Biologie Évolutive, Villeurbanne, France. 3 Centre for Ecological and Evolutionary Synthesis, Department of Biosciences, University of Oslo, Oslo, Norway. 4 Department of Natural Sciences and Environmental Health, University College of Southeast Norway, Bø, Norway. 5 Institute of Wildlife Biology and Game Management, University of Natural Resources and Life Sciences, Vienna, Austria. 6 Department of Zoology, University of Oxford, Oxford OX1 3PS, UK. 7 Norwegian Institute for Nature Research, Trondheim, Norway. e-mail: richard.bischof@nmbu.no Nature Ecology & Evolution www.nature.com/natecolevol 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Supplementary Figure 1. Simplified state and transition structure of the hierarchical multi-state models for female (A) and male (B) brown bears. White circles indicate states that focal individuals can be in and black arrows transitions between these states. Grey circles and grey arrows refer to processes associated with recruitment (COY: cub-of-the-year; C1Y: yearling cub) and modelled peripherally to state transitions. Processes are modelled across multiple years and between seasons (mating, berry, and denning) within years. 1
Supplementary Figure 2. Tallies in the study population of brown bears for context. (a) Estimated brown bear population size in Sweden, reproduced from Swenson et al. 2017 1. (b) The number of bears killed annually within the study area was used as a proxy for hunting pressure in the analysis. 2
(c) The number of instrumented (with GPS or VHF telemetry collars) bears monitored each year; most individuals were monitored over multiple years. (d) The annual number of bears with which contact was lost during monitoring; 14 (56%) of these were later recovered dead (see also Methods section in the main text). (e) The annual number of monitored bears that were confirmed to have been killed illegally (only one of these had been lost from monitoring prior to its death). Supplementary Figure 3. Model-predicted female bear vital rates that exhibited dependence on temporal predictors. Where age-dependence (a and b) was present, predictions are based on primeaged animals (6 10 years old). For hunting mortality (a) and cub survival (b), points indicate mean predictions and vertical bars show their associated 95% credible intervals. Time period (10 years each) emerged as the only temporal predictor for the probability of weaning cubs during their second spring (c); estimates and 95% credible intervals are shown as horizontal lines and bands respectively. 3
Supplementary Table 1. Tally of cause-specific mortalities among brown bears monitored in the Swedish study area during 1985-2014. Cause Female Male Both Legal hunting 89 128 217 Management 6 31 37 Natural 24 9 33 Unknown 8 6 14 Confirmed illegal hunting 4 3 7 Traffic 1 4 5 Total 132 181 313 includes culling, monitoring-related deaths, and self-defense intraspecific killing, except 2: injuries sustained from a moose and one apparent starvation 4
Supplementary Table 2. Effects estimates (mean and the standard deviation of the posterior parameter distribution) on legal hunting mortality (h) of female brown bears during the fall hunting season (logit-link). Fall intercept: age=6-10, year= 2005-2014, yearling head circumference=0, state=1-2.220 0.292 age 1-0.734 0.386 age 2-0.455 0.358 age 3-0.628 0.395 age 4-0.181 0.367 age 5-0.592 0.444 age 11-15 0.221 0.318 age 16+ 0.872 0.353 year 1985-1994 -0.854 0.561 year 1995-2004 0.025 0.362 Yearling head circumference 0.321 0.132 Road density 0.336 0.112 Hunting pressure 0.364 0.182 state 2-3.837 1.238 Legal hunting confined to fall. No legal hunting mortality of individuals in state 3. 5
Supplementary Table 3. Effects estimates (mean and the standard deviation of the posterior parameter distribution) on mortality of female brown bears due to causes other than legal hunting (w, logit-link). Spring intercept: age=6-10, state=1 to 3-3.906 0.387 age 1 1.857 0.463 age 2 0.926 0.582 age 3-0.584 0.883 age 4-0.087 0.777 age 5 0.441 0.723 age 11-15 0.113 0.636 age 16+ 0.075 0.753 Fall intercept: all individuals in states 1, 2, and 3-5.029 0.401 Winter intercept: all individuals in states 1, 2, and 3-4.018 0.262 6
Supplementary Table 4. Effects estimates (mean and the standard deviation of the posterior parameter distribution) on survival of cubs-of-the-year (S, logit-link). Predictor mean sd Spring intercept: mother age=6-10, year=2005-2014, winter severity previous year=0, state=2 0.379 0.156 age 4-1.991 0.629 age 5-0.227 0.290 age 11-15 0.163 0.193 age 16+ 0.613 0.287 year 1985-1994 1.018 0.361 year 1995-2004 -0.014 0.178 Winter severity previous year -0.286 0.101 Fall intercept: all individuals in state 2 5.379 0.786 Winter intercept: all individuals in state 2 3.399 0.310 Covariates refer to mothers, not cubs-of-the-year Earliest age at first reproduction is 4 years. 7
Supplementary Table 5. Effects estimates (mean and the standard deviation of the posterior parameter distribution) on a female brown bear s probability of emerging with cubs-of-the-year from the winter den (f, logit-link). intercept: age =6-10, year=2005-2014, yearling head circumference=0, road density=0, state = 1 0.199 0.146 age 4-1.985 0.335 age 5-0.826 0.284 age 11-15 0.616 0.269 age 16+ 0.095 0.324 Yearling head circumference 0.416 0.118 state 3-3.609 0.837 Refers to the age of the mother at the time she emerges from the den, potentially with cubs (spring). Earliest age at first reproduction is 4 years. Females in state 3 (i.e. with dependent yearling cubs) reproduce with a lower probability than solitary females. Females that are in state 2 at the beginning of winter do not produce a new litter in the following spring. Supplementary Table 6. Effects estimates (mean and the standard deviation of the posterior parameter distribution) on a female brown bear s probability of weaning or losing her entire litter of yearling cubs (P, logit-link). Spring intercept: year=2005-2014, state = 3 0.646 0.242 year 1985-1994 1.421 0.770 year 1995-2004 1.163 0.440 For simplicity during modelling, we specify that all yearlings that are still with their mother after the spring are weaned during the transition from Fall to Winter. In reality, they are typically weaned during the following spring (as 2-year-olds). 8
Supplementary Table 7. Effects estimates (mean and the standard deviation of the posterior parameter distribution) on litter size (λ) conditional on emerging from the winter den with cubs-ofthe year. In the model litter size is the response of a linear regression (log-link, Poisson), followed by truncation ( 1; 4), to permit only observable litter sizes. Mean estimates (log-scale) and standard deviation shown below are based on posterior distributions prior to truncation. intercept: age=6-10 0.942 0.073 age 4-0.225 0.225 age 5-0.098 0.168 age 11-15 0.222 0.124 age 16+ 0.351 0.171 Earliest age at first reproduction is 4 years. Supplementary Table 8. Effects estimates (mean and the standard deviation of the posterior parameter distribution) on legal hunting mortality (h) of male brown bears during the fall hunting season (logit-link). Fall intercept: age=6-10 -1.914 0.173 age 1-1.269 0.377 age 2-0.206 0.295 age 3-0.226 0.314 age 4-0.050 0.319 age 5-0.613 0.384 age 11-15 0.166 0.297 age 16+ 0.247 0.356 Winter severity -0.284 0.095 Hunting pressure 0.651 0.098 Legal hunting is confined to Fall. 9
Supplementary Table 9. Effects estimates (mean and the standard deviation of the posterior parameter distribution) on male brown bear mortality due to causes other than legal hunting (w, logit-link). Spring intercept: age=6-10 -3.387 0.391 age 1 1.869 0.442 age 2 0.756 0.591 age 3 1.746 0.491 age 4 0.736 0.583 age 5-0.247 0.744 age 11-15 -0.891 0.854 age 16+ 0.738 0.623 Fall intercept: all individuals -3.935 0.255 Winter intercept: all individuals -5.623 0.639 References 1 Swenson, J. E. et al. Challenges of managing a European brown bear population; lessons from Sweden, 1943 2013. Wildlife Biology, wlb.00251, doi:10.2981/wlb.00251 (2017). 10
Additional Supplementary Data 1. Data for Figure 1.txt Individual states and fates as shown in Figure 1. Rows represent individual histories and columns indicate ages, starting at year 1 (3 seasons in each year: mating, berry/hunting, and denning). States are defined as follows: 1= alive without dependent cubs, 2=alive with dependent cubs-of-the-year, 3=alive with yearling cubs, 4=newly dead from legal hunting, 5=newly dead from other mortalities, 6=dead (absorbent state). Histories are composed of both observed and model-predicted states. Text file; male and female state histories provided as separate comma delimited matrices. 2. Data for Figure 3.txt Age-class-specific posteriors used to draw the violins in Figure 3 for the following parameters: female and male bear hunting mortality (logit), mortality due to causes other than legal hunting (logit), mortality due to causes other than legal hunting (logit) for cubs-of-the-year, litter size (log), and probability of weaning (logit). Text file; data are arranged in consecutive comma-delimited tables with associated figure panel letters indicated above each table. 3. Data for Figure 4.txt Data used to draw age-class-specific hunting mortality (logit) associated with different levels of hunting pressure, as well as associated life expectancies and reproductive values. Text file; different metrics are arranged in a series of commadelimited tables, with rows representing age classes and columns different levels of hunting pressure. 4. Multistate model for JAGS.txt Model definition of the hierarchical multistate model for female bears, in the JAGS programming language. Text file. 5. Data for Supplementary Figure 2.txt Data used for drawing the bars in Supplementary Figure 2b-c: number bears shot annually, number of bears monitored, number of bears lost to monitoring, and number of bears confirmed to have been killed illegally. See the reference listed in the caption for Supplementary Information Figure 2 for data sources for panel a (estimated annual population size). Comma-delimited text file. 6. Data for Supplementary Figure 3.txt Data used for drawing the time-dependent predictions and associated 95% credible intervals in Supplementary Figure 3: annual mortality due to hunting (logit), estimated survival (logit) of cubs-of-the-year, probability (logit) of weaning cubs after they have reached 1 year of age. Text file; data are arranged in a series of commadelimited tables with time interval in rows and the prediction and 95% credible interval boundaries in columns. 11