Ecological Bulletins 47: 137-144. Copenhagen1999 Nigel G. Yoccozand Rolf A. Ims Yoccoz,N. G. and Ims, R. A. 1999.Demography of small mammalsin cold regions:rhe importanceof environmentalvariability. - Ecol. Bull. 47: 137-144. Environmental variability in arctic and alpine regionsis large,and consistsof predicta- ble (seasonality)as well asless predicrable components (e.g. between-years variability). We comparedemography of alpine and arctic microtines basedon two long-term stud- ies in the French Alps (snow vole, Chionomysnivalis) and at Svalbard(sibling vole, Microtus rossiaemeridionalis),as well as data from a short-term study on the common vole, M arvalis, in the Alps. While the length of the vegetationperiod is of the same order of magnitude (3-4 months in the Alps, 3 months at Svalbard),the population dynamics and demographyof the snow vole and of the sibling vole population are strikingly different.The alpine snowvole population is very stable,with little variability in survival and reproduction betweenyears, particularly so during the winter. The sib- ling vole Svalbardpopulation is highly fluctuating, with low variability in summerde- mographicrates and largevariability in winter population rate of change.These differ- ent patternsof variability in the dynamicsand demographyof small mammalsare relat- ed to the climatic patternsin both areas,particularly the pronouncedseasonal fIuctua- tion in climatic variability at Svalbard,and the somewhatconstant level of climatic variability in the Alps. We arguethat suchpatterns may be of generalrelevance to arctic environmentson Olle hand and alpine environmentsat middle latitudes on the other hand. We funher discussthe possibleconsequences of changesin variability patternson the demographyof small mammals. N G. Yoccoz([email protected]), Div. o/Arctic Ecology,Norwegian Insti- tute for Nature Research(NINA), Polar Environmental Centre, N-9296 Ti'omsø,Norway. -R.A. Ims, Div. o/Zoology, Depto/Biology, Univ. o/Oslo, EO.B.I050Blindern, N-O316 Oslo, Norway. "Environmental variability is a virtual Pandoras box of selec- short distances of very different habitats such as broad- rive forces which can injluence the evolution of life histories, leaved forests, coniferous forests and alpine meadows be- and there is still much we do not understand about the nature cause of steep altitudinal gradients (Gerrard 1990). High of selectionin jluctuating environments" (Boyce 1988: 16). altitude climates generally differ from high latitude cli- Cold regions are found either at high latitudes or at mates, but the variability of mountain climates prevents high altitudes. They share common characteristics, first of any simple comparison (Barry 1992). Climate of the west- all a marked seasonality: short summers favourable to veg- ern European Alps is, for example, wetter as altitude in- etation growth separated by lang winters with a permanent creases(mainly becauseof orographic effects), but is not as snow cover. They also differ by important aspects,such ås severe with respect to temperature as compared to arctic climate, photoperiod and spatial heterogeneity: alpine re- environments: extremely cold temperatures - < -30°C - gions will aften be characterized by a juxtaposition over are rare in the European Alps, while they are common in . ECOLOGICAL BULLETINS 47,1999 37 northem Fennoscandia or in the high Arctic. European Alps (snow yale, Chionomys niva/is, and com- The recent focus on clirnatic change have led to detailed mon yale, M arva/År;northern Alps, France). Two of these studies of lang-term trends both in the Alps and in the studies (sibling and snow voles) are based on intensive Arctic. There is now strong evidence for a significant lang-term (7 yr) capture-recapture methods, and allow for warming trend in the Alps, particularly so during the last a detailed analysis of the variability in demographic rates. two decades (e.g., Diaz and Bradley 1997, Sommaruga- aur main objective is to assessif environments with simi- W6grath et al. 1997). The arctic regions on the other hand lar average seasonality, as measured by the length of the show large spatial heterogeneity, with no clear trend (Przy- breeding season,may select for different life-history tactics bylak 1997). The Svalbard area, for example, has not as a consequence of differences in tempora! variability. warmed up since the 1920s (Førland et al. 1997). The cli- matic variability may also increase as a consequence of glo- bal change (Katt and Brown 1992, Mearns et al. 1997). Material and methods Studies of climatic variability have mainly focused on dec- adal scale variability (e.g., Beniston 1997, Hurrell and Van Svalbard Loon 1997), but, to aur knowledge, the interannual cli- matic variability has not been thoroughly studied in arctic The sibling vaie - a sibling speciesof the common vole- and alpine regions. was introduced to Svalbard from Russia between 1930 and Small mammals present good opportunities for study- 1960. It is a relatively small species (adult females: 30 g, ing the effects of environmental fluctuations because they adult males: 35 g). This speciesis the only small mammal have a short generation time (often less than ane year) and speciesin this high arctic archipelago, and there is only ane are amenable to detailed studies based on capture-mark- persistent population (Fredga et al. 1990, Yoccoz et al. recapture methods (e.g., Leslie et al. 1953, Krebs 1966, 1990; seeFig. 1). The Svalbard population probably origi- Yoccoz et al. 1993b, Ims andYoccoz 1997). Most popula- nates from the region around St. Petersburg, and was first tion studies of small mammals have focused on the issue of identified as M arvalis (Nyholm 1966, Alendal 1977). population cycles, with relatively litcle work being done on The difference in behaviour between the two species- M demography (see e.g. Hansson and Henttonen 1988, rossiameridionalis using human dwellings such as barns Stenseth and Ims 1993). Short-term environmental varia- much more than M arvalis - probablyexplains why the bility may have both lang-term effects through the evolu- sibling vaie was introduced to Svalbard. After an initial tion oflife-history traits as well as direct short-term effects survey in 1989, we started in 1990 a capture-mark-recap- through annual changes in demographic rates and conse- ture study of the population found around the ghost min- quencly population sizes. It is for example known that ing town ofGrumantbyen (78°10'N, 15°16'E). We moni- there may be a trade-off between the mean and the vari- tored the population by following selected areaswithin a 1 ance of a trait, an increasing variance having a negative ef- km strip along the shore inhabited by the voles. We cap- fect on the lang-term fitness, everything else being equal tured individual voles three times a year, at the end ofJune, (Orzack and Tuljapurkar 1989, Yoshimura and Clark just after the snow melt, at the beginning of August, and at 1991). Therefore, different patterns in environmental var- the beginning of September, befare the first snowfalls lead- iability and predictability may well select for different de- ing to permanent snow cover. We based aur trapp ing re- mographic tactics (Boyce 1979, 1988). Population dy- gime on the robust design (Pollock 1982) with 7 -10 trap- namics will be ultimately affected through the demograph- ping secondary sessionswithin each of the three primary ic parameters: populations with large potential turnover sessions.This ensured reliable estimates of population sizes rates (early age at first reproduction, high reproductive based on closed population models. Survival rates was esti- rates) might be more variable than populations with in- mated using generalizations ofCormack-Jolly-Seber mod- verse demographic characteristics. els (Yoccoz et al. 1993b, Yoccoz et al. unpubl.). There have been relatively few empirical studies ad- dressing the issue of environmental variability and its im- pact on demographic tactics, moscly on plants (Carlsson FrenchAlps and Callaghan 1994, Watson et al. 1997; but see, e.g., Benton and Grant 1996, Gaillard et al. 1997). This is is for Three microtine speciespermanently inhabit alpine (above the most due to the lack of detailed lang-term studies the timberline) areasin the western European Alps (Le where demographic parameters have been reliablyestimat- Louarn 1977,Janeau 1980). The snowvole, C niva/is,is a ed (Gaillard et al. 1998). In this paper, we relate demo- middle-sizedvaie (adult females:40 g, adult males:45 g), graphic and population dynamic patterns to patterns in and is restrictedto boulder and rocky areas(Claude 1995, environmental variability. This we do on the basis of field Janeauand Aulagnier 1997), including housesand alpine studies of three small mammal species performed in the huts (it may thereforebe fOWld up to 4000 m, far above high arctic (sibling vaie, Microtus rossiaemeridionalis= M. the snow line). It is also found at the sealevel closeto the epiroticus; Svalbard, Norway) and in the alpine zone of the Mediterraneansea (Jones and Carter 1980), and therefore 38 it is not a species restricted to the high altitude roDe. It is more common in mountain habitats because rocky areas are more frequent at high altitudes. The common vaie is on the other hand an inhabitant of alpine meadows up to the snow line (Spitz 1977, Meylan 1995). It is not found in the subalpine roDe, and the "alpine" common vaie has sometimes been distinguished from its lowland conspecific under the Dame M incertus (Le Louarn et al. 1970). The lowland M arva/is has a very wide geographic range in Europe (Fig. 1). The common vaie is the smallest rodent speciesin the alpine zone (adult females: 20 g, adult males: 25 g). The third species, the fossorial form of the water vaie, Arvicola terrestrissche1man, is for the most part found at lower altitudes (600-1800 m a.sl. in the areas studied here), and is usually cyclic with a 5-6 yr periodicity (Saucy 1994). It was uncommon or absent at the altitudes consid- ered hete (2200-2700 m), and is much larger (80-100 g) than the two other microtines.