CHAPTER 14
Population Fluctuations

ANIMALS OF DIFFERENT SPECIES, including birds of the grouse family, often fluctuate in abundance together, despite living in different habitats and having different life histories. An obvious explanation, that each species is influenced by the same weather, proved difficult to demonstrate until the discovery of the North Atlantic Oscillation (NAO).¹ Paradoxically, it is often easier to show a link between animal population fluctuations and this crude index of prevailing weather than with local meteorological measurements such as rainfall or temperature. We illustrate the explanatory value of the NAO with bag records from the four species of British grouse.

Within this broad climatic framework, detailed studies give insights into the biological processes that determine the dynamics of thriving populations on the scale of a scientific study area, typically a few square kilometres. It happens that such processes vary with the structure of the landscape. Thus, the dynamics of grouse populations in large tracts of homogeneous habitat differ from those in more fragmented habitats.

The scientific study of population fluctuations is known as ‘population dynamics’. This chapter is mostly about the fluctuations in grouse numbers that occur from year to year. Other chapters give more on how grouse abundance varies from place to place.

Landscape and weather set the stage upon which short-term fluctuations in grouse numbers occur. Such fluctuations are often cyclical. There is a long-standing debate about causes of population cycles in boreal herbivores such as voles, lemmings and hares, and in birds of the grouse family. Studies of grouse, especially red grouse, are now resolving this problem.

WEATHER AND GROUSE NUMBERS

Grouse abundance and the North Atlantic Oscillation

Bag records from all four species of British grouse (Fig. 170) tend to fluctuate together, reflecting periods of general tetraonid² abundance and scarcity.³ Moran showed that bags of all four species on ‘a number of Scottish estates’ in east-central Scotland fluctuated together between 1866 and 1938 (Table 29), and suggested that bags of all four species were influenced by weather.⁵

TABLE 29. Cross-correlation coefficients between bags of four grouse species in east-central Scotland in 1866-1943. Bottom left, raw bags; top right, filtered bags.⁴

Notes

^**P < 0.01, ^***P < 0.001, ^****P < 0.0001, based on filtered bags.

These results resemble those of Moran (1952), but are not identical because the data that we used differed somewhat from his.

To illustrate this, we compare a measure of general tetraonid abundance in east-central Scotland with the NAO (Fig. 171). The fit is imperfect, but peaks in the NAO during the early 1900s and 1920s were evidently associated with peaks in tetraonid abundance. More formally, the equation that best explains tetraonid abundance included a running mean of four NAO values, including the current value and the previous three years’ values.⁶

The 1866-1943 bag records did not distinguish between young and old birds, but recent records of red grouse shot show the NAO to be related to the proportion of young in the bag.⁷ Hence, for red grouse at least, some aspect of weather that is related to the NAO seems to affect breeding success, and breeding success is known to influence subsequent population densities.

This broad relationship raises several questions. For example, the NAO reflects the strength and persistence of the Azores area of high pressure, which is associated with mild weather and westerly winds in western Europe. But what are

FIG 170. Annual bags of red grouse, ptarmigan, black grouse and capercaillie from east-central Scotland, 1866-1943.

FIG 171. Five-year means of general tetraonid abundance (PC1 score) in east-central Scotland and the North Atlantic Oscillation.

the relevant local meteorological factors affecting grouse?⁸ The next section tackles this question for rock ptarmigan and red grouse on the Cairngorms massif and the Cairnwell hills.

Population cycles and weather – entrainment

The idea of entrainment elucidates the relationship between population fluctuations and weather. Thus, grouse numbers fluctuate irrespective of weather, but weather influences the fluctuations. If weather shows a regular pattern, grouse numbers reflect this pattern, though not exactly. Also, different populations of the same species might be entrained by different aspects of weather.

Many but not all species of the grouse family are known to show regular population cycles. In species that do show cycles, not all populations are cyclic, and not all cyclic populations show the same period. The quoted period of a cyclic population fluctuation is an average value, because successive fluctuations often differ in length. For grouse species, commonly reported cycle periods include three to four years, six to seven, and about ten (also called eight to eleven) years. Longer periods may also occur but evidence for these is equivocal (Fig. 172).⁹ Cycle period is not species-specific – red grouse, for example, have shown cycles of four to five years, six to seven years, seven to eight years, and ten years. Furthermore, populations of different grouse species that coexist in the same area tend to fluctuate together. Hence fluctuation pattern and cycle period seem to depend more upon local circumstances than upon species. One reason for this may be that local population fluctuations adopt the period of local weather patterns. The fluctuations do not mirror weather patterns exactly, but they do seem to be under their influence.

Red grouse and rock ptarmigan on the Cairngorms massif in Scotland have fluctuated together in ten-year cycles (see Fig. 173a). But in the nearby (20km distant) Cairnwell hills, the same species show unstable but erratic fluctuations (see Fig. 173b), similar in amplitude to the cycles in the Cairngorms but with no fixed period.¹⁵ Within each range of hills, the two species fluctuate more or less together, but fluctuations in the two ranges are out of phase with each other (Figs 173a and 175). In this case, the role of weather can be understood only when we take soils into account. The study sites in the Cairngorms lie over granite that gives rise to generally infertile soils, supporting ptarmigan food plants of lower nutritive value, while on the more fertile Cairnwell hills with mostly base-rich bedrock the same plant species are of higher nutritive value. Interactions between weather and diet quality can explain the different dynamics on the two hill ranges.

The ten-year bird cycles on the Cairngorms massif relate closely to a ten-year cycle in June temperature. Records from the nearby village of Braemar show peak

FIG 172. Equivocal evidence for a 20-year cycle in bags of red grouse at French Park, Ireland.¹⁰ Top: bags (log_e basis) and trend (dashed line). Gaps represent missing data. Bottom: autocorrelations.¹¹ The significant negative autocorrelation at a lag often years suggests a 20-year cycle, but the data are equivocal because they represent only three putative cycles and the 20-year autocorrelation coefficient is insignificant.¹²

FIG 173. (a) Numbers of rock ptarmigan and of red grouse in the Cairngorms (recorded as adults seen per 10km of transect walks in May-September), a measure closely related to spring density (birds per km²). The numbers of both species peaked more or less together once a decade, hence the two fluctuated in partial synchrony and showed ten-year cycles.¹³ (b) Numbers of rock ptarmigan in the Cairngorms and the Cairnwell hills (adults per 10km of transect walks). The birds in the Cairngorms showed clear ten-year cycles, and their raised baseline in the 1960s was associated with an exceptional run of warm Junes. The Cairnwell birds fluctuated erratically and out of phase with those in the Cairngorms.¹⁴

FIG 174. Numbers of rock ptarmigan (adults per 10km of transect walks) in the Cairngorms, as observed and as postdicted from the June mean temperature at Braemar village by an ‘omnibus’ model with no input from observed ptarmigan numbers after 1944.¹⁶

June temperatures occurring at ten-year intervals, intriguingly as isolated single-year spikes and not as high points of a smooth curve. Peaks in ptarmigan and red grouse come one to two years after the cyclic peaks in June temperature, but the bird cycles do not mimic the temperature fluctuations exactly.

A simple model, of the same type as the equation relating tetraonid abundance to the NAO (see endnote 6), gives a good account of ptarmigan cycles on the central Cairngorms sub-massif (Fig. 174).¹⁷ It is essentially a running mean of June temperatures dating back four years, together with a delayed density-dependent term representing a damped cycle.¹⁸ A plausible explanation depends on the observation that, in the infertile Cairngorms, the breeding success of ptarmigan is related to June temperature, as follows.

June temperatures spike every ten years or so, and this gives a decennial boost to reproduction, probably through better food for the chicks – more insects, or improved plant growth or plant quality. Better reproduction results in more youngsters recruited to the breeding population in the following year, and hence an increase in spring density. As in red grouse (see below) and irrespective of weather, high recruitment in one year begets high recruitment in the next, a form of positive feedback in population growth. Spring density peaks one to two years after the initial increase in reproduction. At peak density, the population becomes crowded and birds are more aggressive. This causes a continued decline in density due to continued aggression and reduced recruitment, associated with lower breeding success and probably increased emigration of youngsters to other sub-massifs.¹⁹

Whatever the biological mechanism, the ten-year cycle of ptarmigan and red grouse in the Cairngorms was clearly linked to a ten-year cycle in June temperature. This raises the question of why the nearby Caimwell fluctuations (see Fig. 173b) were not entrained to June temperatures. On the more fertile Caimwell, ptarmigan occurred at higher densities, and had better nutrition and also better reproduction that, crucially, was not related to June temperature.²⁰ Instead, fluctuations were related negatively to spring snow-lie and positively to temperature (Fig. 175). Neither of these shows regular cycles, nor do Caimwell ptarmigan.

The equation describing the Caimwell fluctuations incorporates the same damped cycle as the Cairngorms equation, but modified by different weather (snow-lie and temperature), and so results in an erratic fluctuation.²¹ Although this fluctuation has a cyclic component, it would be confusing to call it cyclic, because it does not show a single dominant period. A term that covers both kinds of fluctuation is ‘unstable’.

Weather cycles themselves need not be permanent features of a climate, for they seem to come and go in an unpredictable and chaotic fashion.²² This might explain how grouse cycles come and go, and how they change their period over the years (see below).

POPULATION DYNAMICS AND BEHAVIOUR

The story so far is that grouse populations show unstable dynamics, and that interaction with weather and soil fertility can result in cyclic or erratic fluctuations. However, this explanation does not include the biological mechanism that causes

FIG 175. Numbers of rock ptarmigan (top) and red grouse (bottom) in the Cairnwell hills: number of spring cocks observed, and number postdicted from spring snow-lie and mean air temperature in early May. Postdictions are from ‘omnibus’ models, with no input from observed numbers after 1966.

the unstable dynamics, a topic that has long fascinated ecologists. The equations that were used successfully to ‘postdict’ grouse numbers (see Figs 174 & 175) do not necessarily correspond to biological mechanisms. Thus, annual variations in numbers are caused initially by something in the birds themselves. This ‘something’ changes in relation to weather. The equations do not specify the something, nor how weather affects it, and so they are empirical, not mechanistic.²³

Some of the first mathematical models in ecology showed that model population cycles can result from predator-prey interactions.²⁴ Models also substantiate the yet earlier insight that ‘overcrowding creates disease’, such that population cycles might result from successive episodes of overcrowding, disease-induced decline, and subsequent recovery.²⁵ Another possibility is that animals at high densities deplete their food, starve into decline, and then recover from low densities when their food regains abundance.

These mechanisms, conceptually similar, each fall into the category of ‘trophic interactions’ or ‘extrinsic processes’.²⁶ Fieldworkers, however, are often impressed by the time and energy that individuals of the same species spend competing with each other. Such observations led to the view that animals are able to regulate their own density below any limit set by resources or natural enemies,²⁷ which implies that spacing behaviour can regulate density and hence is an ‘intrinsic process’.²⁸

Critics assert that classical theories of population dynamics do not explain how unstable dynamics can arise from intrinsic processes. So entrenched is this view that cycles are often cited as evidence of trophic interactions. Biologists and sociologists interested in the fundamental question of how cooperation between organisms evolved, however, have refuted this criticism. Cooperation occurs when individuals benefit more from helping each other than from behaving selfishly. Models with individuals either cooperating or acting selfishly often result in unstable population fluctuations.²⁹ Increasing populations go along with more cooperation, whereas declines are associated with increased selfishness. This result provides a broad theoretical basis for the idea that intrinsic processes involving the behaviour of individuals can cause unstable population dynamics.³⁰

Population cycles are fascinating in themselves, but are also useful in population biology, because each successive cyclic fluctuation is seemingly a repetition of the same event. This allows scientists to develop hypotheses about population regulation during one fluctuation and to test them during the next. In particular, work on red grouse has developed, refined and tested the hypothesis that spacing behaviour can cause unstable population fluctuations, of which cycles are one obvious manifestation.³¹

Basic concepts in population dynamics

To explore population fluctuations further, we need some basic ideas. Obviously, animal numbers are limited by resources such as food and cover, other species that compete for the same resources, and natural enemies such as predators and parasites. For gamebirds we include shooters as natural enemies. Weather, in turn, influences the availability of resources and the impacts of natural enemies. These limiting factors determine the size of animal populations through effects on reproduction, death and movement. One approach to explaining fluctuations in population size is therefore to measure reproduction and mortality, the vital rates of a studied population, and to find whether variations in vital rates are explained by fluctuations in limiting factors.

Movement is also an important determinant of population size, but in practice is harder to study than reproduction or death. Population ecologists often delimit an area of ground and define the animals within it as their study population. Some animals move out of the study area and become difficult or impossible to follow, while others of unknown provenance move in. This complicates the calculation of reproduction and mortality, and so early modellers, pretending that inward and outward movement (immigration and emigration, respectively) cancel each other out, ignored movement and emphasised ‘closed’ populations, in which changes in population size are determined solely by the balance between reproduction and death.

This sweeping simplification led to the compelling concept of density dependence. Clearly, an animal population could not increase without limit. Presumably, some factor in the environment, such as disease or lack of food, imposed a limit through its effect on a vital rate, thereby regulating the balance between reproduction and mortality, causing the model population to decline from high densities, and allowing it to increase from low densities. Therefore, to detect the factor regulating a real population, one could take the mortality rate due to each candidate factor and plot it against population size.³² The regulating factor should be revealed by a plot that shows the candidate mortality rate increasing with rising density.³³ A mortality factor that passes such tests is said to ‘regulate’ population size.

This school of thought, greatly elaborated, remains popular. Some models now go further by incorporating movement. Also, regulating factors can act with a time lag, thereby delaying their effect. Models incorporating such delayed density dependence can produce unstable population fluctuations, including cycles. Despite its present numerical complexity, this view remains essentially one of passive population dynamics, in which the size of an animal population is determined by its vital rates and perhaps movement, these being regulated solely by the population’s resources and natural enemies. This nurtures a conservative view that studies of behaviour are unnecessary for an understanding of population dynamics.

If you watch birds in the field, however, a very different impression can emerge. Most red grouse, for example, live in a sea of heather, which provides them with abundant food and cover. They spend relatively little time eating or taking cover, and usually rear far more young than are needed to replace adult losses. This is, of course, why they are gamebirds – in most years the autumn population can be heavily cropped without depressing the breeding population the following spring. The size of breeding populations can certainly be limited by damage to the heather food or by the ravages of natural enemies, but these do not explain all the observed fluctuations. If the birds are marked so that they can be recognised individually and are then watched carefully, a pattern of social behaviour emerges.

Red grouse are territorial birds, each spring territory being held by a cock, which is usually paired with a hen. In summer, the territorial system relaxes, and families – pairs with their broods of young – roam unhindered by territorial boundaries. In the half-light of dawn, the father of a wandering brood wings back to his territory and briefly crows title before returning to his family.

By autumn the young are fully grown, each family returns to its home territory and the father aggressively reasserts his old territorial boundaries. The young cocks also begin to display aggression, and, using the parental territory as a base, attempt to establish their own territories nearby. Soon, families break up as each young cock tries to get his own territory, perhaps to some extent with the help of his father. Some young birds succeed, forcing old cocks to cede ground or ousting them. Some fail, and join with ousted old cocks, often in wandering bands of ‘floaters’.³⁴

By the end of winter, most floating cocks have died because they did not have a territory. On grouse-moors where gamekeepers kill most predators, almost all territory owners typically survive the winter to form the following year’s breeding population. In more natural circumstances, some territory owners are killed by predators, allowing an equivalent number of floaters to occupy the vacated territories and so survive. Hence the number of territories in spring is more or less determined by the competition for territories in autumn.³⁵

A hen gains her place in the breeding population by consorting with a territorial cock. Some hens choose a cock during the autumn contest and stay with him until spring, some move from one cock to another, and a few wander from cock to cock until they finally settle, just before laying eggs in spring. A territory is typically occupied by a pair, although occasional cocks have two hens and bachelor territorial cocks are common. Even so, cocks largely determine hen numbers. If a territorial cock dies and is not replaced, the hen soon leaves and looks for another cock. Hens that fail to pair with a cock become floaters, sharing the fate of cock floaters.

Some territorial birds emigrate to less crowded ground just before laying. Such pre-laying emigration is density dependent and, in our experience, comprises mostly young birds.³⁶ It is observed most frequently at very high spring densities, exceeding about 100 territories per km².

Such observations suggest that social behaviour determines population density, in this case largely the cocks’ territorial behaviour during the autumn territorial contest. Some, unconvinced by such evidence, have argued that social behaviour serves merely to distribute animals after their numbers have been limited by resources or natural enemies. Only in the last few years has unequivocal evidence been provided that animals can regulate their own densities.

Role of territorial behaviour in population dynamics of red grouse

Red grouse territories typically occupy most of the available moorland,³⁷ and so spring population density is essentially the inverse of mean territory size, as determined largely during the territorial contest of the previous autumn. Aggressiveness influences territory size, and territory size probably influences aggressiveness. Thus, on a given tract in a given year, the more aggressive cocks have the larger territories. Then again, comparing different years on the same moor, cocks at higher average density, with more numerous smaller territories crowded together, show more aggressiveness.³⁸ This leads to complex interactions between aggressiveness and density, which are now being unravelled, partly by the experimental use of testosterone implants to increase the cocks’ aggressiveness, and partly by mathematical deduction.

Aggressiveness in cock red grouse is related to the amount of testosterone in their blood, and a few milligrams of testosterone can have profound effects on population dynamics. This was first illustrated by an experiment at Glen Tanar in which testosterone was implanted into half the cocks on a study area in spring (see Fig. 176). The implanted cocks took larger territories and ousted many unimplanted males, thereby causing an increase in average territory size and a decline in breeding density.³⁹

This experiment showed that an increase in aggressiveness can reduce population density. Other than that, however, implications for population dynamics were not clear. A year after the implants, spring territory size and density on the experimental area had reverted to that of the control, and so the

FIG 176. Effects of testosterone implants on spring territory size and numbers of red grouse at Glen Tanar in spring 1993. Shaded territories are of implanted cocks before (left) and after (right) implants. Similar results occurred in 1992, except that cocks in the opposite half of the study area were implanted.

increase in aggressiveness had no long-lasting effect on population density. Also, the implants were done in spring, not in autumn when the territorial contest that fixes population density occurs. In addition, implants of testosterone are obviously unnatural.

The next experiment rectified these drawbacks. To appreciate its ingenious design, we must understand the likely role of kinship in population cycles of red grouse. Population density is determined largely by the cocks’ mean territory size in a given autumn. The general idea that population fluctuations are caused by changes in mean territory size, in turn due to changes in aggressive behaviour, is the ‘territorial-behaviour hypothesis’.⁴⁰ This is now well established, but it does not explain what causes the changes in aggressiveness. The ‘kinship hypothesis’ is a specific version of the territorial-behaviour hypothesis that does explain the changes in aggressiveness.

Changes in population density from one spring to the next are determined partly by the mortality of established old territorial cocks, and partly by the recruitment of young cocks to the territorial population. Although the mortality of old cocks varies from year to year, it bears no consistent relation to the population cycle’s ‘phase’ (its increase or decline). Recruitment varies more than adult mortality and is the main determinant of population change, being greater during the increase phase of cycles and less during declines. And kinship seems to influence the proportion of young cocks competing for territories that gets recruited into the territorial population.

Kinship

That related animals behave differently towards each other than towards unrelated individuals is well documented. So, too, is ‘helping’, whereby young animals help their parents to rear further offspring. This can be understood in terms of kin selection: doubtless it is selectively advantageous to help with the rearing of kin. It should also be advantageous to help kin get recruited into the breeding population.

Kin are less aggressive towards each other than towards non-kin.⁴¹ Young red grouse cocks tend to get territories close to their father’s territory, thereby forming kin clusters (extended family units) of related territory owners. Such clusters build up during the increase phase of a cycle and decay during the decline.⁴²

Each kin cluster is thought to influence future densities through its effects on recruitment. Comparisons within years show that a young cock that is hatched into a larger cluster has a greater chance of getting a territory.⁴³ Extrapolate this result to differences between years, and it follows that young cocks should recruit more successfully in contests that start on average with more or larger kin clusters, provided that the population is not already so crowded as to make further recruitment impracticable. Similarly, at equal densities, recruitment should be low in contests that start with fewer or smaller kin clusters, unless density is low enough to make recruitment easy. Observations during a recent population cycle confirmed that the proportion of cocks within kin clusters rose along with recruitment during the increase phase of the cycle, that both began to fall just before peak density, and that they continued to fall during the decline.⁴⁴

Mathematical modelling brings observations together to show how the growth and decay of kin clusters might lead to population cycles (see Fig. 177).⁴⁵ Kin clusters are dynamic entities that grow through recruitment and that decay from the death of established members. Some recruitment is necessary for a cluster to maintain itself In an increasing population, recruitment exceeds mortality by a sufficient amount to allow kin clusters to grow in size. Eventually, however, the population becomes so crowded that it cannot increase further and recruitment is inhibited.⁴⁶ Unable to sustain themselves, kin clusters break up, and so aggressiveness among neighbours increases, triggering a population decline. There are now no kin clusters to facilitate recruitment, and the population decline continues for a few years until it has reached a low density. This makes recruitment easy, whereupon kin clusters reform as density begins to increase again.⁴⁷

The experiment that we next describe was designed with the kinship hypothesis in mind. Old cocks (those that had held territories in the previous spring) were implanted with testosterone in September-November to make them more aggressive. It was predicted from the kinship hypothesis that increased

FIG 177. Model fluctuations in population density, kin-cluster size and aggressiveness of red grouse during a population cycle, according to mathematical formulations of the kinship hypothesis.⁴⁸

aggression would reduce the recruitment of young cocks and hence the size of kin clusters.

It was not known whether the enlarged autumn territories would revert to their original size when the testosterone in the implants had been exhausted. If so, this might allow young cocks to be recruited, albeit later than usual, and any effect of the experiment might be as transient as in the previous experiment carried out in spring at Glen Tanar (see Fig. 176). Alternatively, if territories were fixed in the autumn, the enlarged territories should remain, resulting in reduced densities and smaller kin clusters the following spring. According to the kinship hypothesis, these smaller kin clusters should cause reduced recruitment in the autumn a year after the implants. In consequence, the size of kin clusters and density should remain lower on the experimental areas for at least two springs after the autumn of implantation. The boldness of these predictions meant that this was a risky experiment, in the best sense of the word.⁴⁹

Manipulating recruitment

The experiment was carried out twice,⁵⁰ once in northern England and once in northeast Scotland, where the kinship hypothesis originated. In each country, the experiment used two separate moors simultaneously.

Implants of testosterone changed the outcome of autumn territorial contests. On 1km² experimental areas of moorland, most of the old cocks were implanted with testosterone. During the territorial contests, implanted old cocks gained bigger territories, which entailed lower densities and prevented most of the young cocks that had been reared that summer from establishing territories on the study area. The testosterone in the implants was exhausted after three months, but the density of the experimental populations remained depressed throughout the winter, and onwards into the following spring and summer (Fig. 178). Meanwhile, on nearby control areas, the normal course of events developed: many young cocks gained territories and densities remained higher than on the experimental areas.

FIG 178. Effects of testosterone implants (T, arrow) on the density of red grouse (cocks and hens) on four 1km² study areas in England (Catterick and Moorhouse) and in Scotland (Edinglassie and Glen Dye) during autumn and spring. Control areas, black solid lines; implants, red dashed lines. The increases between spring t+1 (April) and autumn t+1 (late July-early August) are due to young being reared. In each case, the implanted population declined faster than the control (overwinter losses were greater, as shown by steeper slopes between autumn and spring), not only in the first winter after implants, but also in the following winter (between autumn t+1 and spring t+2, highlighted grey), when no birds were implanted. Hence, well after they were exhausted, the testosterone implants continued to affect all four populations.⁵¹

This confirmed that breeding density was reduced by more aggressive territorial behaviour in the previous autumn. Furthermore, it showed that the reduced density of territory owners, established during the autumn contest and held long after the cause of the reduction (a transitory, artificial pulse of testosterone) had vanished.

Indeed, the effects of testosterone upon density did not stop at the breeding season immediately after the implantation, but carried through until the following breeding season a year later. Despite no further implants, grouse density and the recruitment of young cocks remained significantly lower on the experimental than on the control areas for at least two consecutive breeding seasons. Furthermore, grouse density again declined more on the treated areas than on the controls.

As expected from the kinship hypothesis, the declines in density caused by testosterone implants were associated with a reduction in kin clusters of territorial cocks. Parasites could not explain these results, for the declines were not associated with increased burdens of the parasitic caecal threadworm Trichostrongylus tenuis.

Implications of the experiment

This experiment provided clear evidence that red grouse have the ability to regulate their own density below any limits set by resources or natural enemies, in line with the territorial-behaviour hypothesis.⁵² The outcome of a territorial contest can be influenced by the previous contest. In effect, the recruitment of young cocks in one contest is influenced by the birds’ ‘memory’ of the previous contest. The kinship hypothesis suggests that this memory is held in the form of kin clusters, but one can deduce the effect of the memory without knowing in what form it is held.

Based solely upon the results of this experiment and other known facts about the birds’ life history, mathematical deduction shows how territorial behaviour, interacting with present density and a memory of past density, can moderate recruitment and so lead to unstable dynamics.⁵³ Together with vital rates, the mathematical form of the two-way interaction between recruitment and density determines whether dynamics are stable or unstable.⁵⁴

The main result of the modelling is that unstable dynamics should occur if changes in aggressiveness at least match changes in density. In the real world, two forms of aggressiveness were measured during a 14-year (1964-77) study of two population cycles on Kerloch moor, during which population density fluctuated fivefold.⁵⁵ Meanwhile, the rate at which cocks gave territorial songs fluctuated sixfold among years, and the rate at which they had territorial boundary disputes with each other fluctuated more than 20-fold. Hence proportional changes in density were more than matched by proportional changes in two measures of aggressiveness. It seems that the theoretical requirements for unstable dynamics to be caused by territorial aggressiveness are realistic.

In general, resources and enemies affect animal numbers by influencing vital rates and hence population dynamics. But the experimental and theoretical evidence from red grouse shows that we should not expect all fluctuations in density to relate directly to factors such as food, cover, disease or predation. According to the kinship hypothesis, for example, unstable dynamics can be the result of competition for land amongst extended families. But vital rates should affect the size of these families. Hence resources and enemies might well influence animal numbers indirectly, through their effects upon competition among families. The idea that resources and enemies can influence population dynamics indirectly, through their effects upon social structure, brings fruitful insight to the dry, deficient mantra that animal populations are regulated by density-dependent vital rates.

Breeding success during cycles

If breeding success is so poor that insufficient young are reared to replace adult losses, then, without immigrants, a population must decline. During population cycles in birds of the grouse family, breeding success is usually lower during the decline phase than during the increase. Such observations tempt some to conclude that reduced breeding success causes declines. An alternative view is that cyclic declines engender poorer breeding.

The main documented demographic cause of population cycles in seven studies – rock ptarmigan in Alaska, Iceland and Scotland, willow ptarmigan in Norway and Newfoundland, and two of red grouse in Britain – was variation in the recruitment of young birds into the breeding population.⁵⁶ Recruitment occurred in two stages. First, during cyclic declines the birds’ breeding success was lower -in other words, fewer young were reared to independence per adult. Second, a smaller proportion of these independent young survived the winter and recruited into the breeding population. However, mean breeding success differed greatly among populations within species, such that cyclic increases in some populations occurred with lower breeding success than during cyclic declines in other populations (see Table 30). It follows that variations in breeding density were not driven by the same variations in breeding success in all populations of the same species. Further, there was no inverse relation between mean breeding success and average adult survival. Hence populations with better mean breeding success had greater overwinter losses of independent young. An equivalent statement is that, in populations with better breeding success, smaller proportions of the reared young recruited into the breeding population.

TABLE 30. Mean breeding success (young reared per adult) during the increase and decline phases of population cycles, with average annual survival rates of adult hens.⁵⁷

The above evidence shows that variations in winter losses of independent young rock ptarmigan and Lagopus lagopus were the key determinant of recruitment, and, in studies of red grouse, many such losses were due ultimately to territorial behaviour. In the other cases just mentioned, and perhaps in other grouse species, some form of behaviour, not necessarily territorial, may be the ultimate cause also (see the last paragraph of ‘Causes of Grouse Population Cycles’, below).

The idea that poor breeding success during cyclic declines is an effect and not a cause of decline is still largely untested. The hypothesis that population cycles and other manifestations of unstable dynamics are due to variations in recruitment, mediated by behaviour, does not require variations in breeding success to occur. From an evolutionary point of view, however, it seems reasonable that a hen should invest less in rearing young during decline years, when the stresses associated with increased aggressiveness are probably greater than in increase years and the ‘fitness benefits’ per egg laid are smaller.⁵⁸

THE ROLE OF MOVEMENT AND LANDSCAPE IN POPULATION FLUCTUATIONS

Movement

To understand how grouse populations fluctuate, we must take their spatial organisation into account. Kin clusters in red grouse are a specific example. More broadly, spatial organisation involves movement. A basic fact about movement is that, during natal dispersal, young hen red grouse move further than cocks and do not usually settle next to their natal territory. Another fact emerges from the spatial scale upon which several population experiments have been carried out. In the testosterone experiments above, for example, the 1km² experimental and control study areas were only 0.5km apart, and yet large differences in density were maintained for more than a year after the implants. This suggested a strong spatial structuring that prevented immigration and compensatory recruitment from neighbouring tracts that were much larger and had a higher density. This fits with other population studies, in which the demographic effects of experiments on areas as small as 16ha have been confined largely to these areas.⁵⁹ The evidence suggests that cocks are largely responsible for maintaining such spatial structuring.

Cocks and hens that fail to get territories during the autumn territorial contest become floaters that range widely, often in bands that provide some protection against attacks from territory owners. Old birds that have lost their territories sometimes get a new one elsewhere, but most die landless. Some floaters get territories by replacing territory owners that have been killed, for example, by predators. Others disperse and may establish territories in areas that are perhaps less crowded than their natal patch, where they are unlikely to form part of any kin cluster. There is evidence that immigrants increase aggressiveness on their new area.⁶⁰ Hence they might depress local densities or even precipitate a cyclic decline that would otherwise have been delayed until a later year. Dispersing floaters, in any case, are likely to suffer higher mortality than territorial residents. Most floaters are usually young birds. Hence the increased rate of dispersal associated with cyclic declines should be accompanied by higher mortality of young birds.

The root, or ultimate, cause of death in floating birds is their failure to join a breeding population.⁶¹ Chased by territory owners, prevented from feeding where they will and from using cover effectively, such birds are stressed and lose condition. Thus, floaters are more vulnerable to predators than are territorial birds, so that the immediate, or proximate, cause of their death is often predation.⁶² Stressed birds are also more vulnerable to parasitic or other disease. So, a study that failed to account for the birds’ social status might incorrectly conclude that predation or disease was the main cause of a cyclic decline, when it was, primarily, due to the workings of the social system.

Travelling waves

When cycling populations of red grouse or ptarmigan are counted over several square kilometres, it becomes apparent that the fluctuations across the study area do not occur together and are only partly synchronous. Peaks and troughs seem to follow each other over the ground like ripples across a pond; this pattern is called a travelling wave,⁶³ and has also been observed in cycling populations of voles and snowshoe hares. Such patterns might be due to dispersing floaters triggering successive declines in contiguous areas, but evidence is lacking.

Theoretical models suggest that population cycles and travelling waves could arise in a constant environment, through unstable dynamics alone. In reality, however, variations in natural enemies and resources certainly affect reproduction and survival. Hence, fluctuations in numbers are likely to be due to interactions between extrinsic factors that affect vital rates directly, and intrinsic factors that affect recruitment through variations in spacing behaviour and movement. We now consider one of these interactions.

Habitat fragmentation and population cycles

Much anecdotal evidence suggests that grouse cycles worldwide occur mostly in large tracts of fairly homogeneous habitat, and that fragmentation of such habitat by, for example, agriculture is associated with the disappearance of cycles. This has been attributed to increased dispersal out of fragmented habitat patches into population sinks, and to increased predation from generalist predators. For example, forest grouse and rodents show cycles in north Fennoscandia, but not in the more fragmented agricultural south where increased generalist predators are thought to suppress the cycles. Work on red grouse and ptarmigan in Scotland provides examples of generalist predators suppressing cycles.

The illegal killing of raptors is common practice on grouse-moors, because raptors are supposed by moorland managers to reduce stocks of red grouse available for shooting. An experiment at Langholm moor in southern Scotland aimed to test this hypothesis (see Chapter 12).⁶⁴ The killing of raptors, especially hen harriers, was stopped and their numbers increased. More harriers ate more grouse, grouse numbers declined, and the population cycle that had previously been apparent in shooting records stopped, while the cycle continued on nearby moors where the traditional, albeit illegal, killing of raptors continued.

A ski development on Cairn Gorm provides another example where generalist predators damped a population cycle (Fig. 179).⁶⁵ Though alien to this Arctic-Alpine habitat, carrion crows were attracted from the valley below by people’s discarded food scraps. On the most developed area near the main car park, rock ptarmigan at first occurred at high density, but then lost nests to the influx of crows and reared abnormally few broods. They also died flying into ski-lift wires, and declined until none bred for many summers. On a nearby higher area with fewer wires, ptarmigan lost nests to frequent crows and reared abnormally few broods, but seldom died on wires. Here, adult numbers declined and then

FIG 179. Non-cyclic fluctuations in rock ptarmigan (adults on 271ha) near the ski development at Cairn Gorm. Development began in 1961, after which ptarmigan numbers declined and then stabilised at a lower density, while numbers nearby (see Figs 173b & 174) continued to show cycles.⁶⁶

became unusually steady for over two decades, with no significant cycle (Fig. 179). On a third area further from the car park, ptarmigan lost fewer nests to the less frequent crows, but bred more poorly than in the massif’s centre and showed cycles of lower amplitude than there. On a fourth area yet further away, with few or no crows, ptarmigan bred equally as well as in the massifs centre and showed cycles of the same amplitude as there.

These examples help to show the role of large areas of fairly homogeneous habitat in generating population cycles. Fragmented habitat is likely to engender more losses of eggs or chicks from increased numbers of generalist predators, or more emigration of young followed by death in population ‘sinks’, which are typically areas of ground occupied by generalist predators and lying between habitat fragments. Because cyclic populations occur in homogeneous tracts, with lower densities of predators and longer distances to sinks, they should rear and recruit a higher proportion of young than non-cyclic populations in fragmented

FIG 180. Number of spring territories of red grouse during an unperturbed population cycle on a 318ha control area (top), and during a cycle perturbed by removing cocks in the increase phase of the population fluctuation on a 203ha experimental area (bottom) at Rickarton moor. The vertical red dashed lines (bottom) show numbers before and after removals in spring. The black dashed lines show predictions made in spring 1982, using a model developed previously from fluctuations observed on a different study area.⁶⁸ (The brief 1986-7 decline on the experimental area followed severe damage to heather by winter weather in 1985 and increased caecal threadworm burdens in 1986.)

habitat. This should facilitate the development of greater population densities and larger kin clusters.

The next step in the argument is illustrated by an experiment with red grouse,⁶⁷ which prevented a population cycle by removing cocks in the increase phase, thereby stopping them from attaining peak density (Fig. 180). The control population peaked and then showed the predicted cyclic decline, but the experimental population showed no cyclic decline. Hence peak numbers may be necessary to initiate the process – presumably the catastrophic breakdown of kin clusters – that marks the beginning of cyclic declines.

In short, it seems that large tracts of fairly homogeneous habitat, where generalist predators are scarce and opportunities for dispersal limited, may be necessary for peak numbers. Peak numbers seem to be necessary for cyclic declines, and, consequently, large tracts engender population cycles.

CAUSES OF GROUSE POPULATION CYCLES

Having explored the roles of weather, soil fertility, habitat continuity or fragmentation, and behaviour in population regulation, we can now assess some alternative explanations for the population cycles that occur in birds of the grouse family worldwide. The multiplicity of cycle periods within species among areas, and the similarity of cycle periods among species within areas, makes it clear that we are not dealing with a single process. Hypotheses about population cycles must explain three issues: what drives cycles (this is the ‘something’ described under ‘Population Dynamics and Behaviour’, above); intra-specific variations in cycle period; and inter-specific synchrony.

Entrainment by weather (see p.350) or other extrinsic processes can explain the second and third issues. Another explanation of inter-specific synchrony is the ‘Moran effect’. This postulates that random variations in weather can bring independently cycling populations into synchrony. The main difference between the Moran effect and entrainment is that entrainment involves a weather cycle (see Figs 174 & 175), whereas the Moran effect does not. The widespread low numbers of grouse and other animals in northwest Europe in the early 1940s (see Figs 170-174) might be seen either as an example of the Moran effect or of entrainment, as might the remarkable synchrony among the six moors in Figure 168. The distinction is slight. Neither, however, explains what drives cycles.

Grouse cycles seem to be an expression of intrinsically unstable population dynamics.⁶⁹ Even in a constant environment, models of unstable dynamics can produce population cycles, chaotic fluctuations, and erratic fluctuations that are part cyclic and part seemingly random. More realistically, however, intrinsically unstable dynamics are likely to interact with local conditions, which themselves might fluctuate, to produce the observed fluctuations in grouse numbers. In modelling terms, the outcome should depend partly on vital rates and partly on the form of the interaction between density and recruitment. In turn, vital rates should depend upon local conditions.

At least three types of erratic fluctuation have been recorded. First, like weather cycles,⁷⁰ grouse cycles may come and go. In Finland, for example, forest grouse (capercaillie, black grouse and hazel grouse) showed such clear cycles during 1964-84 that a paper based on these years was entitled ‘The clockwork of Finnish tetraonid population dynamics’.⁷¹ Between 1989 and 1999, however, grouse populations in Finland were quite steady.⁷² Second, cycles may change in period. Shooting bags of red grouse at Rickarton moor gave evidence of a six-year cycle in the late nineteenth century, but a ten-year one after the Second World War.⁷³ In Finland, Siivonen gave evidence of three- to four-year cycles in forest grouse in 1929-49, whereas the ‘clockwork’ cycles of 1964-84 had a six-year period.⁷⁴ Third, each fluctuation can be cyclic in the limited sense that it comprises some years of population increase followed by some of decline, but the periods of successive fluctuations can differ. Thus, the counts of ptarmigan in the Cairnwell hills show a mixture of six- and ten-year fluctuations (see Fig. 173b).

An obvious problem with generalising from British results to population cycles in grouse worldwide is that the results and models involving behaviour apply only to red grouse, in which recruitment is moderated by autumn territorial behaviour. Most species of grouse have a polygamous social system, unlike that of red grouse, in which monogamous pairs maintain exclusive breeding territories. In migratory populations in particular, it seems improbable that autumnal territorial behaviour plays a major role in moderating recruitment. Even so, one can restate the territorial behaviour hypothesis for grouse cycles in the more general form that intrinsically unstable dynamics in birds of the grouse family involve behaviour that moderates recruitment (and hence loss of non-recruits),⁷⁵ even if the social systems through which this mechanism operates, as well as the seasonal timing, differ within and among species. In this form, the hypothesis might also apply to mammals that show population cycles, such as voles, lemmings and hares, although there is little evidence.

Cycles and predation

The evidence that behaviour can cause unstable population dynamics in red grouse does not rule out the hypothesis, popular in the literature, that natural enemies can cause predator-prey cycles. Modelling suggests that simple two-part predator-prey cycles are most likely when the predator is a specialist. The best relevant example of a single main cyclic prey species and a largely specialist predator is rock ptarmigan in Iceland, where the birds are preyed upon by gyrfalcons. During a population cycle in Iceland,⁷⁶ increased numbers of falcons took the heaviest toll of ptarmigan during the decline and the low phase of the cycle, as required by the predator-prey hypothesis. The evidence did not show, however, that the number of ptarmigan killed was in itself sufficient to bring about the decline. Ptarmigan, already excluded from the potential breeding population by aggressive behaviour, might have been more vulnerable to falcons. Or, falcons might have amplified a behavioural decline in ptarmigan by reducing the size of groups of related cocks that would otherwise have formed kin clusters.

In Britain, there are no grouse-specialist predators, and so the predator-prey hypothesis for population cycles does not seem relevant. Also, most predators have been routinely killed on many grouse-moors for over a century, while grouse there have continued their cycles.

The alternative prey hypothesis for tetraonid population cycles suggests that generalist predators might switch from their primary, cyclic, mammalian prey to grouse when the former are scarce. This has been suggested for three- to four-year cycles in parts of Fennoscandia and ten-year cycles in North America, where small rodents and snowshoe hares are the putative primary prey, respectively. The usual version of the alternative prey hypothesis is that predators cause the cycles simply by killing grouse. If so, grouse cycles are a mere adjunct of mammal cycles, themselves unexplained. More interestingly, it is also possible that prey switching by predators might entrain unstable population fluctuations in grouse, so affecting their timing, but not causing their instability.

There are, however, differences in tetraonid cycle length between Sweden and north Finland, where the period is three to four years, and south Finland, where it was six years in 1964-84 and now seems erratic. As the period of the small-rodent cycle in both countries is typically three to four years, the southern Finnish cycles cannot be explained by the alternative prey hypothesis. Similarly, vole cycles in Britain are also three to four years in length, but cycles of red grouse and Scottish ptarmigan can be ten years. In Britain, in short, it seems that predators can suppress cycles, but it is unlikely that they are a widespread cause of them.

Parasites and cycles

Parasites, like predators, are natural enemies of grouse. The idea that regular fluctuations in the numbers of red grouse are a parasite-host cycle dates back well over a century. Outbreaks of fatal ‘grouse disease’ due to the caecal threadworm Trichostrongylus tenuis were, in the nineteenth century, of unknown aetiology, but James Dalziel Dougall⁷⁷ considered that ‘overcrowding creates disease’ and that the ‘whole mischief arose, primarily from over-stocking, and, secondarily from gamekeepers most improperly killing off the falcons’, which he termed ‘sanitary commissioners’ because they killed the most heavily infected birds. He ‘personally ascertained by enquiry amongst the more aged Highlanders that “disease” was in their early days quite unknown’, and quoted the bitterly eloquent words of an old lady, who had a sheep farm in Argyllshire, that the disease was ‘a curse of God Almighty on the grouse, because now-a-days the lairds let the moors to Sassenach gentlemen’. He explained that ‘after widespread havoc…a period of health sets in’ and that these alternations between health and disease were so regular that he had been able to predict in print the years when outbreaks would occur.

Most red grouse carry some caecal threadworms, and burdens are often in the thousands. The parasite seems to occur primarily in red grouse and in the conspecific willow ptarmigan, with rock ptarmigan as an accidental host, and to be confined largely to the more temperate parts of the birds’ range. The red grouse is especially adapted to temperate conditions, being the only Lagopus subspecies that does not turn white in winter. Worm burdens reported in willow ptarmigan are typically lower than in red grouse. Hence proponents of the parasite-host hypothesis for other tetraonid cycles must find other parasites to explain them.

The ‘overcrowding creates disease’ hypothesis is easy to understand. Red grouse should increase until they are overcrowded, when caecal threadworms should proliferate, heavily infested grouse should die in large numbers and the survivors should produce few young, thereby generating a crash in numbers. Such crashes have been recorded by Dougall and by many authors subsequently.

A modern version of this hypothesis is more subtle, and postulates that parasites can cause cycles without necessarily involving mass mortality of adult birds (see Chapter 13). Parasite burdens should increase with increasing grouse density, until the birds’ reproductive rate and perhaps the adults’ survival are adversely affected. This should lead to population decline, but not necessarily to a precipitous crash involving birds dying in large numbers.

An outbreak of grouse disease occurred at Glen Esk in the late 1950s.⁷⁸ During the springs of 1958 and 1959, some territorial adults died, but survivors reared more than enough young to replace adult deaths. Territorial behaviour took place as usual in the autumn, thus excluding some old and some young birds from the next spring’s breeding population. Mean territory size increased following each year of disease, and so density declined between 1958 and 1960. Disease and territorial behaviour hence acted together, the parasites reducing the young available for recruitment, and territorial behaviour regulating their recruitment. With hindsight, the kinship hypothesis suggests how parasites might have influenced territorial behaviour. Lower breeding success in one year presumably engendered smaller kin clusters in the next, thereby causing fewer young birds to be recruited to the territorial population.

Outbreaks of grouse disease range in severity from declines associated with poorer breeding but little extra mortality, to precipitous crashes involving mass mortality associated with huge worm burdens. High grouse density and wet summer weather seem to be predisposing conditions for increased worm burdens.⁷⁹ Weather plays an important role because it affects the parasites’ transmission rate. When conditions favour the transmission of parasites from bird to bird, outbreaks of disease can occur at fairly low grouse densities. These obviously involve fewer deaths, and are less often remarked than the spectacular mass mortality of artificially overcrowded grouse-moors.

Dougall’s perceptive account of the fluctuations engendered by grouse disease needs some amendment in the light of a century and a half of further study. We now understand that the agent of disease is the caecal threadworm, and that weather together with grouse density are major determinants of worm burdens. Moderate outbreaks of grouse disease are likely to engender changes in territorial behaviour, which amplify the effects of the disease, inducing longer declines than would be expected from deaths due to disease alone. Severe outbreaks of disease seem to be commoner in parts of northern England, these perhaps engendered by the wet climate and the unnaturally high peak population densities associated with muirburn and the killing of Dougall’s ‘sanitary commissioners’. Even in the drier conditions of northeast Scotland, however, population crashes associated with mass mortality and heavy worm burdens have been reported on grouse-moors where predators were rigorously reduced.

LANDSCAPE AND POPULATION DYNAMICS

The animal populations of entire landscapes are harder to comprehend than those of the usual smaller area used for scientific study. This more speculative section, therefore, is woven from a mixture of fact, experience and insight.

The observation that population cycles are associated with large tracts of fairly homogeneous habitat provides a link between population dynamics at the scale of a study area or a grouse-drive (a few square kilometres) and habitat on a larger scale. To understand population fluctuations on the scale of a grouse-moor or woodland estate, something must also be known about the surrounding landscape.

Lowland Britain and Ireland are now largely dominated by an agriculture that provides few toeholds for grouse. Over the centuries, man has drained wetlands, and has converted forest to heath vegetation and heath to arable crops or grass pasture. This broad succession has proceeded more slowly in the uplands, where most of our grouse now remain.

The main factors affecting past and present trends in numbers of British and Irish grouse have been fivefold: destroying forest and replacing it with scrub or moorland; destroying scrub and heather moorland, and replacing it with grassy vegetation; planting uplands with exotic conifers; exterminating some species of predators and altering the abundance of others; and weather and climate change. Many of these are covered separately in other chapters. Here we bring them together in a broad sketch of how past land management continues to affect British grouse.

Forest destruction and fragmentation

Below the timberline, those tracts of the British and Irish uplands that retain the dwarf shrubs so crucial for grouse are unnaturally devoid of trees, for in past centuries the uplands were deforested and kept without trees by grazing and muirburn. Black grouse, birds of forest edge, scrub and moorland, survived this historic deforestation. However, capercaillie, souls of the forest, depend upon open coniferous forest and reportedly became extinct in the eighteenth century, when tree cover reached its lowest ebb.

In Fennoscandia, fragmentation of native forest occurred more recently than in Britain, and its effects upon woodland grouse are therefore better documented. During the twentieth century, especially in southern parts of Norway, Sweden and Finland, the forest industry created patchworks of old-growth forest, broken up by clear-felled areas. One consequence of this, in Sweden for example, was that black grouse increased during the 1990s as the openings created by clear-felling provided good habitat, including open scrub and dwarf shrubs, whereas capercaillie declined as their mature forest habitat was reduced to fragments.

The effect of clear-felling upon woodland grouse, however, is much influenced by the type of ground vegetation that flourishes after logging, especially the balance between scrub, heath vegetation and grassy vegetation. Open scrub and heath vegetation support black grouse, but grassier vegetation supports more voles, which in turn sustain more predators such as foxes and crows. These hunt not only in the clear-fells but also within the old-growth fragments, so that the capercaillie remaining in woodland fragments amongst clear-fells suffer more predation than those in extensive tracts of old-growth forest.

Destruction of scrub and heather moorland

In Britain, the destruction of mature forest and its replacement by scrub and heather moorland would at first have favoured black and red grouse at the expense of capercaillie. Then, from the eighteenth century onwards in the uplands, sheep numbers increased, and in the nineteenth century red deer, previously scarce or absent, were encouraged on Scottish sporting estates. Browsing by cattle, sheep and deer eliminated much scrub such as willow, hazel and birch, thereby removing an important source of food and cover for black grouse. Any dwarf shrubs remaining on heavily browsed ground were reduced in height, so providing less cover for birds. A century ago, black grouse were still abundant in widespread open scrubland, but declined as sheep and deer eliminated this important habitat.

The effects of browsing on upland vegetation depend partly upon soils. On richer soils, browsing by sheep and the effects of other agricultural practices favour grassy vegetation by reducing or eliminating dwarf shrubs. As a result, red grouse decline and then largely disappear, as in much of the now grassy Welsh uplands and the border country between England and Scotland. Above the timberline, prolonged browsing of heath vegetation by many sheep has eliminated much ptarmigan habitat on the hilltops of southern Scotland and northern England. On poorer soils there are fewer sheep and deer, and hence dwarf shrubs can grow and red grouse and ptarmigan are more likely to survive, though scrub is widely eliminated by browsing.

Reforestation

Extensive reforestation with pine and larch began in the mid-eighteenth century and accelerated in the nineteenth century, so helping the reintroduced capercaillie to thrive. Twentieth-century plantings of exotic conifers such as Sitka spruce and lodgepole pine are, however, mostly so dense that insufficient light reaches the ground to sustain dwarf shrubs, making these forests of little value to woodland grouse after canopy closure. A few degraded remnants of the old Caledonian forest, now greatly treasured, support some of our densest populations of capercaillie, but the birds can also thrive in open planted pinewood with heather and blaeberry underneath.

Black and red grouse often increase where young conifers have been planted on heather moorland, but then decline as the canopy closes and shades out dwarf shrubs. It was once hoped that black grouse would also thrive in Sitka spruce clear-fells and restocked areas. Experience shows that this does not generally occur, and one reason seems obvious: the birds depend upon heather and blaeberry, which are scarce on most Sitka restocks.

Predators

Most of our terrestrial wildlife survives on agriculturally poor or marginal ground, where it depends largely upon the activities of humans. In particular, unless they are killed by gamekeepers on sporting estates, medium-sized predators such as the fox are unnaturally abundant and kill many ground-nesting birds such as grouse. This is partly because grassy moorland provides more vole, pipit and rabbit food for predators, but perhaps also because man has exterminated large predators that kill medium-sized predators.

Large predators, near the top of their food chain, typically occur at low densities. Medium-sized predators, however, breed more quickly, and if unchecked reach greater abundance. Hence the extermination of large predators should lead to a greater abundance of predators overall, and so more predation of ground-nesting birds.

Wolves and lynxes, for example, eat foxes, and in parts of Europe where wolves are abundant the foxes are scarce. Foxes eat pine martens. Eagles and goshawks eat many crows and also kill smaller raptors. When hen harriers became unusually abundant on Langholm moor, for example, they apparently suppressed the numbers of red grouse. The archetypal grouse-moor-owner’s answer to such a problem is to instruct his keepers to kill hen harriers. Part of a more balanced approach might be to allow eagles to kill harriers, but many landowners persist in killing eagles as well as the other raptors that comprise Dougall’s ‘sanitary commissioners’.

Weather and climate

Some fluctuations in the numbers of grouse and other wildlife are so widespread that local factors, such as changes in land use or management, cannot explain them. During the twentieth century, shooting-bag records of red grouse in northwest Scotland and willow ptarmigan in southeast Norway were remarkably similar (see Fig. 26).⁸⁰ Large bags in the early part of the twentieth century fell to a common trough about 1918, recovered to a peak about 1930, and then declined shortly before the Second World War. After the war, numbers remained much lower than they had been before. Scottish red grouse and Norwegian willow ptarmigan have different diets, live in different habitats, suffer in the main from different diseases, and are subject to different management practices by humans. It seems obvious that a widespread common factor such as weather or pollution affected numbers of grouse in both countries.

Consequences of land management practices on grouse in Britain

In the much altered British countryside, red grouse and rock ptarmigan continue to show population cycles where large tracts of unfragmented habitat remain to them. Cyclic declines in red grouse and ptarmigan are natural and should be followed by cyclic increases. This can be inconvenient for managers of grouse-moors, who in principle can prevent a cyclic decline by shooting hard in the increase phase, but cyclic declines are no cause for alarm about the survival of the local population.

The habitat remaining to black grouse and capercaillie in Britain is much more patchy and fragmented. This can explain why these two species now show no evidence of population cycles. Indeed, today their habitat is so depleted that both are in danger of extinction.

SUMMARY

All four species of British grouse tend to fluctuate in numbers together, under the influence of weather. On large tracts of continuous habitat, red grouse and rock ptarmigan show unstable dynamics, sometimes manifested as regular cycles. In other parts of their world range, capercaillie and black grouse also cycle, but not in modern Britain, where their habitat is greatly fragmented. Large tracts of homogeneous habitat engender unstable fluctuations in grouse species, but fragmented habitat does not.

Weather, food, predators, parasites and disease influence recruitment and behaviour. They therefore influence numbers directly via reproduction and mortality, and indirectly via social structure, recruitment and movement. Numbers of red grouse are regulated by territorial behaviour. This moderates recruitment, and recruitment in turn determines densities. Territorial behaviour and recruitment are influenced by social structure, particularly the size of clusters of related cocks. In turn, recruitment affects cluster size. Results from red grouse cannot be applied directly to other species that do not show autumnal territorial behaviour. Even so, we commend the insight that resources and enemies influence population dynamics indirectly through their effects upon social structure and movement, as well as acting directly upon vital rates. We suggest that these two generalisations may apply in other grouse species, and perhaps more widely in animals that show unstable population dynamics.