Delayed numerical response of goshawks to population fluctuations of forest grouse Risto Tornberg, Erkki Korpima ¨ki, Sven Jungell and Vitali Reif Tornberg, R., Korpima ¨ki, E., Jungell, S. and Reif, V. 2005.Delayed numerical response of goshawks to population fluctuations of forest grouse.  / Oikos 111: 408 /415. Delayed density-dependent mortality induced by delayed numerical response of predators can drive prey populations to fluctuate in high-amplitude cycles. We studied numerical response of goshawks Accipiter gentilis to varying densities of their main prey (forest grouse) in western Finland during 1979 /1996. Occupancy rate of goshawk territories tracked grouse numberswith a two year lag. Occupancy rate of goshawk territories and pooled number of adult and young goshawks correlated negatively with a 1  /2 year lag to the chick production of grouse. Goshawk to grouse ratio was negatively related to grouse density. This suggests that goshawk predation on grouse is inverselydependent on grouse density. We conclude that in northern Europe with few alternative preys, goshawk predation might contribute to the generation of multiannual cycles of forest grouse. This should be tested experimentally. R. Tornberg and V. Reif, Dept of Biology, Univ. of Oulu, PL 3000, FIN-90014 Oulu, Finland (risto.tornberg@oulu.fi).  / E. Korpima ¨ki, Dept of Biology, Univ. of Turku, FIN- 20014 Turku, Finland.  / S. Jungell, Jungarva ¨gen 527, FIN-66850 Jeppo, Finland. Predation is one possible mechanism maintaining popu- lation regulation provided that the predator acts in a density-dependent way (Hanski et al. 1991, Murdoch 1994, Sinclair and Pech 1996, Turchin 1999). Outcome of predation in prey population is largely dependent on timing and amplitude effect of predator’s response (Andersson and Erlinge 1977, Korpima ¨ki and Norrdahl 1989, 1991, Hanski et al. 1991, Korpima ¨ki 1994, Begon et al. 1996, Valkama et al. 2005). A time-lag, the necessity for coupled predator  /prey oscillations, in numerical response may be caused by a lower reproduction rate of the predator than its prey (Tanner 1975, Hanski et al. 1991, Valkama et al. 2005), which may not necessarily be due to lower intrinsic reproductive rate of the predator, but also due to some limiting factors, like territoriality (Rohner 1995). A time lag in the numerical response has most often been found in resident mammalian predators (Korpima ¨ ki et al. 1991, Krebs 1996, Korpima ¨ki and Krebs 1996, O’Donoghue et al. 1997), but in few cases it has been demonstrated also for raptor  /prey interaction: great-horned owls Bubo virginianus and goshawks Accipiter gentilis preying on snowshoe hares Lepus americanus in North America (Keith et al. 1977, Doyle and Smith 1994, 2001, Rohner 1995, 1996, Rohner et al. 2001), and gyrfalcons Falco rusticolus preying on ptarmigans Lagopus mutus in Iceland (Nielsen 1999). In addition, the brood size of goshawks living in boreal forests of northern Finland seems to lag one year after grouse (Huhtala and Sulkava 1981, Sulkava et al. 1994, Tornberg 2001). Common attributes of these predators, at least of the adult birds, is their year-round residency and territoriality as well as specialisation on one or few prey species. Four forest grouse species (the capercaillie Tetrao urogallus , the black grouse Tetrao tetrix , the hazel grouse Bonasa bonasia and the willow grouse Lagopus lagopus ) in Finland typically fluctuated synchronously in fairly regular 6  /7 year cycles till the late 1980s (Linde ´n 1988, Ranta et al. 1995), where after the cyclicity became less obvious while the synchrony Accepted 20 April 2005 Copyright # OIKOS 2005 ISSN 0030-1299 OIKOS 111: 408 /415, 2005 408 OIKOS 111:2 (2005)