3.5. Goshawks and grouse. Are they coupled?

Grouse constitute the main prey for goshawks throughout the year (II, IV). Although their proportion clearly decreases in winter, at least for females, it probably accounts for 10-50% by biomass (IV). During the breeding season goshawks accounted for the loss of between 2 and 22% of adult grouse, depending on species. In earlier investigations similar estimates have been stated with regard to raptor predation on grouse (Linden & Wikman 1983, Widen 1987, Redpath & Thirgood 1999, Nielsen 1999, Thirgood et al. 2000). Studies on grouse also show that mortality, caused mainly by goshawks, has been on the same magnitude (Angelstam 1984, Willebrand 1988, Valkeajärvi & Ijäs 1994, Wegge et al. 1990). Smaller species, like willow grouse and hazel grouse, were depredated more intensively than larger black grouse and capercaillie, which suffered relatively little from goshawk predation (V). In Scandinavia, goshawks killed more black grouse and capercaillie females, probably based on the lower availability of smaller grouse species being highly preferred in my study area (I, II, Widen 1985, Selås 1989). Predation on grouse chicks was fairly low, only 7% of hatched chicks. Estimating predation on juvenile birds is problematic because very little remains are left, rarely allowing quantitative estimates (Sulkava 1964, Höglund 1964, Huhtala 1976, II). Therefore, the grouse chick proportion is underestimated in nest samples. It is possible to obtain more quantitative estimates in fledging time (Huhtala 1976, II). On the other hand, large prey is probably respectively overestimated, (Sulkava 1964). In Scotland, hen harriers Circus cyaneus took between 30 and 40% of red grouse chicks. However, hen harriers are probably much more efficient chick predators than larger goshawks. Predation on grouse is very much age related in autumn but very little is known about it. Taking non-territorial birds, floaters, into account much higher predation losses are found. This part of raptor and owl populations is the least well known (Rohner 1995, 1996, Nielsen 1999, Korpimäki & Krebs 1996). I calculated that 1/3 of wintering hawks may be non-territorials, which may even be a cautious estimate (IV). In addition, a good food supply will attract hawks from other regions (Kenward 1977, Kenward et al. 1981), which will yield higher raptor densities than the assumed balanced population model (see Kenward et al. 1991, 1999).

I found no functional response of the goshawk for varying grouse numbers. However, density changes of grouse were rather mild and goshawks presumably were able to kill grouse at maximum intensity and also at the lowest stated density level of grouse, but not essentially to raise the killing rate at higher densities. This suggests the response type that is typical for a specialist predator, not being able to switch to other prey (Korpimäki et al. 1991, Nielsen 1999). In fact, goshawks have potentially alternative prey during the breeding season in my study area (I, II), but in early spring, hawks nesting in remote sites do not have many real alternatives. Although goshawk is considered to be a generalist predator, the lack of suitable sized prey compels it to be a specialist in certain circumstances, as also found in other raptors (Galushin 1974, Korpimäki & Norrdahl 1989, Huhtala et al. 1996, Nielsen 1999). The shape of the functional response curve in goshawks is probably concave (type II) (Wikman & Tarsa 1980, I, but see Linden & Wikman 1983). Similar response types have also been found in peregrines Falco pereginus and grey falcon Falco rusticolus (Redpath & Thirgood 1999, Nielsen 1999).

I could state a weak numerical response, expressed as number of goshawk nestlings/territory with a time-lag of one year for the density all grouse species pooled (r = 0.601, p = 0.066). Sulkava et al. (1994) comparing grouse densities of the previous year and the breeding performance of the goshawk contained significant positive correlations, which proves a time lag of one year (see also Huhtala & Sulkava 1981). In winter, study capture indices tracked the productivity of the local population (own obs.). In Alaska, sightings of goshawks peaked one year after a peak year for the snow-shoe hare (Smith & Doyle 1994). Thus, the total population (breeders and non-breeders) lag at least 14-15 months behind the grouse. Traditionally, time-lags are connected to the response pattern of mammalian predators, while avian predation would rather track prey density changes without time-lags (Galushin 1974, Korpimäki & Norrdahl 1989, 1991, Korpimäki 1994). This concerns, in particular, nomadic specialists that are not bound to stationary territories (Korpimäki 1993). The goshawk is a species that more or less stays all year round in its territory and neither are juveniles particularly migratory (Sulkava 1964, Höglund 1964, Saurola 1976, Marcström & Kenward 1981, Halley 1996). Because of this relatively high residency and synchronous fluctuations of grouse over large areas, time-lags are to be expected. Nielsen (1999) found gyr falcons (breeding adults + nestlings) to lag two years behind ptarmigans in Iceland. In Canada, the breeding population of great horned owls Bubo virginianus also lagged two years behind the snow-shoe hare peak (Rohner 1996). Rohner (1995) suggested that time-lag arises when the defence of territories decreases during prey decline creating scope for new territories. It may also rise from the lack of recruits to fill empty territories after high mortality rate of breeding birds in years of poor prey populations (Nielsen 1999).

The total response of the goshawk on grouse was inversely density dependent. Thus predation rate was higher in low grouse abundance than in peak densities. Similar total responses have been found in earlier goshawk studies (Wikman & Linden 1981, Wikman & Tarsa 1980) and in large falcons hunting on Lagopus species (Redpath & Thirgood 1999, Nielsen 1999). Predation patterns of this kind will prove a delayed density-dependence and destabilising effect of predator on prey population (Sinclair & Pech 1996). This relationship is based on predation of breeding birds, which corresponded with a time-lag in changes of prey. However, not much is known about the response of non-breeding birds, which are free to move to any site of good prey supply. Studies on goshawks in more southern regions and on great horned owls suggest that non-breeders are less residential and probably do not show extended delays for changes of prey (Kenward et al. 1981, Doyle & Smith 1994, Rohner 1996). This part of the population probably behaves similarly to nomadic raptors and owls, yet, not over such vast areas like true nomads. (Galushin 1974). By this dataset it is, however, not possible to say much about this issue because very little is known about the total response of the whole goshawk population, including breeders and non-territorial parts of a population. It is in any way clear that ‘floaters’ constitute a remarkable part of a goshawk population like they do in great horned owls (Rohner 1996, IV). They probably show aggregative responses for prey accumulations (Kenward 1977), which have a stabilising effect on prey fluctuations (Korpimäki & Krebs 1996).

In Southern Finland and also in nearby settlements in northern Finland alternative prey is available (I, S. Sulkava, pers.comm.). This enables relatively stable breeding for goshawks and probably a more rapid response for grouse density variations, which might dampen the grouse cycle. The goshawk operates in line with other general predators, like foxes and martens, which have the highest densities in the south (Linden et al. 1996). Hence, we might expect lower grouse densities on average and weaker cyclicity in southern Finland. According to Linden (1989) grouse populations in Southern Finland are partly non-cyclic, while strongest cyclicity and highest densities are found in Central Finland. This is what is stated for vole cycles: large number of general predators switching between prey types and dampening and shortening vole cycles in the south. In the north, where fewer generalist predators are present, resident specialists, mainly small mustelids, (Mustela erimea, M. nivalis) drive the vole cycle (Erlinge et al. 1984, Hansson 1984, Henttonen 1987, Korpimäki et al.1991, Hanski et al. 1991, 1993). This might also suit the grouse cycle, with the exception being that the resident specialist is an avian predator in the north, but the same species acts as a generalist in the south, in connection with other generalists. The vole cycle clearly affects the grouse cycle via an APH type effect (Angelstam et al. 1984). Irregularities in the grouse cycle may, in fact, result from such interference, which increases as significance of the small mammal community increases as a result of changes in landscape structure. Grouse declined three times during the study period, in 1990, 1994 and 1997, which were also the chrash years of voles. The disappearance of the regular 6-7 year pattern in the grouse cycle may partly result from the strong impact of small mammal predators on grouse after vole chrashes sensu APH. The fragmentation of forests have created suitable habitats for Microtus voles, which have increased average vole and predator densities, respectively (Henttonen 1989, Kurki et al. 1997). It has also been hypothesised that disturbances reducing breeding success of a population every now and then may sustain periodic fluctuation in a population that is otherwise regulated by a delayed density dependent manner (Kaitala et al. 1996). Korpimäki & Norrdahl (1997) suggested vole chrashes to be one such a source of disturbance in grouse dynamics.

The general predation theory makes several predictions on the outcome of predator-prey interaction (Rosenzweig & MacArthur 1963, May 1973, Begon et al. 1990). Stability of the system is reached when the predator is relatively inefficient but unstable in the opposite situation. At some point in between the system tends to show limit cycles, which are stable in their own way (Taylor 1984). Theoretical and field studies imply that resident specialist predators acting in a delay are able to cause limit cycles in the predator-prey system, while nomadic specialists and generalists tend to dampen cycles (Korpimäki & Norrdahl 1989, 1991, Korpimäki 1993, Hanski et al. 1991,1993, Korpimäki & Krebs 1996). Goshawk fulfilled the main criteria of the theory. However, to get a better understanding of it, a more detailed analysis is needed of the goshawk’s impact on grouse populations, especially during the non-breeding season. We need more data on the dynamics of non-territorial hawks, floaters, and also more data on grouse chick predation in late summer. To reach these goals intensive radio-telemetry studies on grouse and goshawks are needed in the same area simultaneously.