Supplementary Information Model Parameters Inbreeding depression The median reported number of lethal equivalents (i.e. number of single alleles or combination of partially deleterious alleles per diploid genome which cause death when homozygous) in a survey of 40 populations of 38 mammalian species is 3.14, with 50% of the effects of inbreeding depression being due to recessive lethal alleles [1], and we use this default value because specific data on inbreeding depression in brown bears are limited. The isolated Kodiak Island brown bear population in Alaska is relatively large (>2500 individuals) and shows high reproductive performance despite very low levels of genetic diversity measured at neutral markers, indicating that under some circumstances (e.g. favourable, high-productivity habitat), brown bear populations may be able to survive through many generations of reduced fitness and purge strongly deleterious alleles [2]. However, low reproductive success associated with inbreeding depression has been reported in both captive [3] and small fragmented wild [4] European brown bear populations that are phylogenetically close to the extinct Irish bear population, and intensive management of threatened brown bear populations to minimise the effects of inbreeding through processes such as translocation is a key component of current conservation activities for the species [4-6]. Single-parameter sensitivity analysis on the lethal equivalents parameter showed that even at half the 3.14 value, median extinction time is not likely to exceed 3000 years (Figure S4). 1 Age at first reproduction The most common estimate of age at first reproduction for female brown bears is between 4 and 6 years old [7-8], and a comprehensive study of over 4,700 radio-collared female brown bears [9] found the modal age of primiparity to be 4.3 years. VORTEX only accepts integer values for this parameter, and so a value of 4 years was used. For males, age at first reproduction ranges from 3 years [10] to 9 years [11]; the mean value of 6 years was used for this parameter. The mean age at first reproduction for North American barren-ground brown bear populations was calculated from eight estimates as 7.48 years, ranging between 6 and 9.6 years [12-13]; a value of 7 years was used. Age at senescence In many extant bear populations, human-caused mortality is a major cause of adult mortality [8], which can result in a younger age structure in many populations than would occur in the absence of humans [9]. As Ireland was one of the last countries in Europe to be colonised by human settlers, around 9000 years BP [14], age at senescence in our Late Pleistocene brown bear population must be estimated from upper survival bounds that may be rare today. We took our estimate of age at senescence as 28 years, the age at which there is a sudden drop in reproductive output [9]. Percentage females reproducing annually Estimates for interbirth interval vary between 2-5 years, but 3 years is the most commonly quoted estimate [8, 15-16]. A 3-year interbirth interval also fits with the per capita litter production of 0.32 litters/female per year reported by ref. [9]. However, while 3 years is the most common estimate for interbirth interval, this parameter has been found to vary between populations occupying different 2 environments [12]; interbirth interval of Arctic barren-ground bears is 3.9 years [12, 17]. Age-specific mortality Data were taken from grizzly bears in Yellowstone National Park, because this population is well-monitored and experiences a degree of protection from poaching, and large, rigorously obtained life-history datasets are available [18]. For cubs and yearlings, only data from animals inside the park were used. However, for adults and subadults (3-4 years old), mortality rates were not disaggregated by location, so overall data were used [18]; adult mortality may therefore be overestimated due to human-caused mortality being included. The vast majority of human-caused mortality of adult bears is inflicted on males, not females, and female mortality is accepted as far more important for bear survival [18]. Environmental variation in mortality rate This is estimated for cubs in Yellowstone [18] at around a quarter of the mortality rate; the same proportion was applied to each of our mortality rates. Population density The lowest recorded brown bear population density is 3.8 bears/1000 km², and the highest recorded density is 29.5 bears/1000 km² [11-12, 17, 19]. We consider that the median of these carrying capacities (575 individuals) is most likely to reflect real conditions in the refugium, as the maximum population would be applicable only if the entire area constituted optimal bear habitat, an unrealistic assumption given LGM environmental conditions, whereas the lowest estimate may conversely be too pessimistic. 3 Scenario settings Species description Reproductive system Scenario name Number of iterations Number of years Lethal equivalents * % due to recessive alleles EV concordance of reproduction and survival Number of catastrophe types Reproductive system Female age at first reproduction * Male age at first reproduction Reproductive rates Mortality rates Mate monopolization Initial population size Carrying capacity Maximum age of reproduction Max number of broods / year Max progeny per brood % sex ratio at birth (M) % adult females breeding * Average Bear Model 1000 10,000 3.14 50 Barren-Ground Bear Model 1000 10,000 3.14 50 Yes 0 polygynous [refs 11, 20] 4 Yes 0 polygynous [refs 11, 20] 7 6 28 1 4 [ref 15] 50 [refs 21-22] 6 28 1 4 [ref 15] 50 [refs 21-22] 33 100 7 [ref 15] 55 [ref 15] 32 [ref 15] 6 [ref 15] 50.5 26.8 18.6 18.6 6 0.2546*mortality 50.5 26.8 26.8 26.8 21.2 21.2 21.2 0.2546*mortality 25.5 100 7 [ref 15] 55 [ref 15] 32 [ref 15] 6 [ref 15] 50.5 26.8 18.6 18.6 6 6 6 6 0.2546*mortality 50.5 26.8 26.8 26.8 21.2 21.2 21.2 0.2546*mortality 49 [ref 11] 49 [ref 11] 575 575 58 575 575 58 Distribution: 1 brood Probability: 1 offspring (%) P: 2 offspring (%) P: 3 offspring (%) P: 4 offspring (%) Female mortality: 0-1 F mortality: 1-2 F mortality: 2-3 F mortality: 3-4 F mortality: adult * EV in female mortality Male mortality rate (%): 0-1 M mortality: 1-2 M mortality: 2-3 M mortality: 3-4 M mortality: 4-5 M mortality: 5-6 M mortality: 6+ EV in male mortality (%) % males successfully siring offspring SA Range, or Justification for Exclusion from SA Low importance Adequate Varied as low as 1 Low importance Low importance Not necessary Good data available 3-7 Not as important; only important for genetic diversity Low importance Good data available Good data available Good data available 20-50% Good data available Good data available Good data available Good data available Good data available Low importance Low importance Low importance Low importance 2.5-19% Good data available Low importance Low importance Low importance Low importance Low importance Low importance Low importance Good data available Low relative importance Initial population size * 111-865 Carrying capacity * EV in carrying capacity 111-865 Low importance Table S1. Values used for each parameter in the two historical population viability models and sensitivity analyses. The main differences between the models (highlighted in blue) are female age at first reproduction and interbirth interval. F = female; M = male; P = probability; EV = environmental variation. Parameters on which the sensitivity of model outcomes was tested are indicated with asterisks. 4 Comparison between PVA Models Mills et al. [23] investigated the outputs of four different PVA models, and demonstrated how the conclusions of a PVA can vary substantially between different models. In a declining population such as that of the bears we are modeling, and in a situation such as ours that models environmental stochasticity but no density dependence, VORTEX is the most pessimistic model, in that in “average N at year 48” and “proportion of the replicates extinct from 500 runs”, VORTEX gives the lowest N and the highest proportion extinct. In an approximately linearly declining population, both N and proportion extinct at a particular time are linearly related to extinction time. Therefore we can estimate that other models with methods comparable to our VORTEX scenarios (GAPPS and INMAT) would have estimated extinction time for the Irish bear population at approximately 1.5 and 2.1 times longer respectively than our extinction time estimates. Given that neither of these alternative extinction times comes close to the duration of the LGM (approximately 10,000 years), we remain convinced of our conclusion that bears are unlikely to have persisted through the LGM in Ireland. 5 Supplementary Procedures Sensitivity Analysis This was conducted on key parameters that were either data-deficient or had a wide range of values in the literature, in order to assess the effect of parameter value choice on our conclusions about brown bear survival. Parameters investigated were adult female mortality rate, initial population size (K), interbirth interval, and female age at first reproduction. These parameters were explored both as single-parameter variations on the general model, and together using a Latin hypercube (LHC) sampling design. The single-parameter variation gives information on the shape of the relationship between parameter value and time to extinction. The LHC sampling brings out the most important parameter(s) and shows whether model relationships hold when combinations of parameter values within the range of uncertainty are tested. If there are strong interactions between parameters, or if a parameter is considerably less influential than others, a weak or non-existent relationship will be apparent. However the most important use of LHC is by testing how often (i.e. what proportion of) parameter combinations within plausible limits would cause the conclusions of modelling work to change. Parameters were varied in randomlychosen integer steps between high and low values as follows: adult female mortality rate between 3-19%; interbirth interval (i.e. 1 / % of females breeding annually 20-50%) between 2-5 years; female age at first reproduction between 3-7 years; initial population size between 111-865. Fifty LHC sample scenarios were made; each scenario was run 1000 times, for 10,000 years. 6 a) female mortality rate b) % females breeding each year %B Female mortality 7000 6000 6000 Time to extinction 5000 5000 4000 4000 3000 3000 2000 2000 1000 1000 0 0 2 7 12 17 10 22 c) female age at first reproduction Female min reproductive age 20 30 40 50 60 d) initial population size K 3500 2500 3000 2000 2500 1500 2000 1500 1000 1000 500 500 0 0 0 2 4 6 8 0 200 400 600 800 1000 Figure S1. Sensitivity analyses based on single-parameter variations, showing median and 95% confidence intervals. 7 a) female mortality rate b) % females breeding each year 6000 6000 5000 5000 y = 1.1927x2 - 26.882x + 4.0282 R² = 0.2545 4000 y = 17.245x2 - 479.25x + 3480.6 R² = 0.2871 4000 3000 3000 Median time to extinction 2000 2000 1000 1000 0 0 0 0 5 10 15 20 40 60 -1000 c) female age at first reproduction d) initial population size 6000 6000 5000 5000 4000 4000 y = -94.92x2 + 1038.2x - 2184.2 R² = 0.0365 3000 20 y = 0.0021x2 - 1.9978x + 866.25 R² = 0.0126 3000 2000 2000 1000 1000 0 0 0 2 4 6 8 0 200 400 600 800 1000 Figure S2. Sensitivity analyses based on 50 Latin hypercube samples of the same parameter space as the single-parameter variations shown in Figure S1. Circles show median extinction time of 1000 model runs; lines of best fit were generated by fitting quadratic models to these median values. 8 Frequency of LHC SA results 25 20 15 10 5 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200 3400 3600 3800 4000 4200 4400 4600 4800 5000 0 Median time to extinction Figure S3. Distribution of median extinction times based on Latin hypercube sampling of parameter space. LE 7000 6000 5000 4000 3000 2000 1000 0 0 1 2 3 4 Figure S4. Sensitivity analyses on variation of lethal equivalents alone, showing median and 95% confidence intervals. 9 In the absence of variation in other parameters (Figure S1), it is clear that small increases in % mortality of female bears will dramatically reduce the time that bears can persist during the LGM. By contrast, the time to extinction varies more linearly with respect to the other three parameters. When the likely effect of parameter variations is taken into account together in the LHC sampling sensitivity analysis, female mortality is seen to have a large impact, as does the % of females breeding each year, which also showed a large range in the singleparameter tests. Only 1/50 LHC samples returned an expected extinction time of >3000 years. 10 Supplementary References 1. Ralls, K., Ballou, J. D. & Templeton, A. R. 1988 Estimates of lethal equivalents and the cost of inbreeding in mammals. Cons. Biol. 2, 185-193. 2. Skrbinsek, T., Jelencic, M., Waits, L. P., Potocnik, H., Kos, I. & Trontelj, P. 2012 Using a reference population yardstick to calibrate and compare genetic diversity reported in different studies: an example from the brown bear. Heredity 109, 299-305. 3. Laikre, L., Andrén, R., Larsson, H-O. & Ryman, N. 1996 Inbreeding depression in brown bear Ursus arctos. Biol. Cons. 76, 69-72. 4. Chapron, G., Wielgus, R., Quenette, P.-Y. & Camarra, J.-J. 2009 Diagnosing mechanisms of decline and planning for recovery of an endangered brown bear (Ursus arctos) population. PLoS ONE 4(10), e7568. 5. Miller, C. R. & Waits, L. P. 2003 The history of effective population size and genetic diversity in the Yellowstone grizzly (Ursus arctos): implications for conservation. Proc. Natl Acad. Sci. USA 100, 4334-4339. 6. Swenson, J. E., Taberlet, P. & Bellemain, E. 2011 Genetics and conservation of European brown bears Ursus arctos. Mammal Rev. 41, 87-98. 7. Ward, P. & Kynaston, S. 1995 Bears of the world. London: Blandford. 8. Swenson, J. E., Gerstl, N., Dahle, B. & Zedrosser, A. 2000 Action plan for the conservation of the brown bear in Europe (Ursus arctos). Strasbourg: Council of Europe. 9. Schwartz, C. C., Keating, K. A., Reynolds, H. V., Barnes, V. G., Sellers, R. A., Swenson, J. E., Miller, S. D., McLellan, B. N., Keay, J., McCann, R. et al. 2003 11 Reproductive maturation and senescence in the female brown bear. Ursus 14, 109-119. 10. Zedrosser, A., Bellemain, E., Taberlet, P. & Swenson, J. E. 2007 Genetic estimates of annual reproductive success in male brown bears: the effects of body size, age, internal relatedness and population density. J. Anim. Ecol. 76, 368-375. 11. Craighead, L., Paetkau, D., Reynolds, H. V., Vyse, E. R. & Strobeck, C. 1995 Microsatellite analysis of paternity and reproduction in Arctic grizzly bears. J. Hered. 86, 255-261. 12. Ferguson S. H. & McLoughlin, P. D. 2000 Effect of energy availability, seasonality, and geographic range on brown bear life history. Ecography 23, 193-200. 13. McLoughlin, P. D., Taylor, M. K., Cluff, H. D., Gau, R. J., Mulders, R., Case, R. L., Boutin, S. & Messier, F. 2003 Demography of barren-ground grizzly bears. Can. J. Zool. 81, 294-301. 14. Wickham-Jones, C. R. & Woodman, P. C. 1998 Studies on the early settlement of Scotland and Ireland. Quatern. Int. 49-50, 13-20. 15. Wiegand, T., Naves, J., Stephan, T. & Fernandez, F. 1998 Assessing the risk of extinction for the brown bear (Ursus arctos) in the Cordillera Cantabrica, Spain. Ecol. Appl. 68, 539-570. 16. Chapron, G., Quenette, P. Y., Legendre, S. & Clobert, J. 2003 Which future for the French Pyrenean brown bear (Ursus arctos) population? An approach using stage-structured deterministic and stochastic models. C. R. Biol. 326, S174-S182. 12 17. McLoughlin, P. D., Ferguson, S. H. & Messier, F. 2000 Intraspecific variation in home range overlap with habitat quality: a comparison among brown bear populations. Evol. Ecol. 14, 39-60. 18. Schwartz, C. C., Haroldson, M. A., White, G. C., Harris, R. B., Cherry, S., Keating, K. A., Moody, D. & Servheen, C. (Eds.) 2006. Temporal, spatial and environmental influences on the demographics of grizzly bears in the greater Yellowstone ecosystem. Wildlife Monographs 161. 19. Clarkson, P.L. & Lipeins, I. S. 1994 Grizzly bear population estimate and characteristics in the Anderson and Horton Rivers area, Northwest Territories, 1987-89. Int. Conf. Bear Res. Manag. 9, 213-221. 20. Bellemain, E., Zedrosser, A., Manel, S., Waits, L. P., Taberlet, P. & Swenson, J. E. 2006 The dilemma of female mate selection in the brown bear, a species with sexually selected infanticide. Proc. Roy. Soc. B 273, 283-291. 21. Adamic, M. 1997 The analysis of key sources of mortality of the brown bear in Slovenia in the last 6 years period: 1.4.1991–31.3.1997. Zbornik Gozdarstva in Lesarstva 53, 5-28. 22. Jerina, K., Debeljak, M., Dzeroski, S., Kobler, A. & Adamic, M. 2003 Modelling the brown bear population in Slovenia: a tool in the conservation management of a threatened species. Ecol. Model. 170, 453-469. 23. Mills, L. S., Hayes, S. G., Baldwin, C., Wisdom, M. J., Citta, J., Mattson, D. J. & Murphy, K. 1996 Factors leading to different viability predictions for a grizzly bear data set. Cons. Biol. 10, 863-873. 13