NATIONAL PARK SERVICE Wolf Ecology and Prey Relationships on Isle Royale

An accurate portrayal of interactions between wolves and long-lived prey such as moose requires extended study. Sixteen years of accumulated information now allow some conclusions about fluctuations in moose numbers, productivity, and survival patterns.

Recent changes in vulnerability of moose to wolf predation, lowered calf production, and some evidence of malnutrition have emphasized the role of food supply in regulating moose numbers, contrary to the initial conclusions of Mech (1966). These changes reflect complex ecological relationships. Changes in habitat and other environmental variables—primarily snow conditions—have modified food availability and have underscored food supply as the ultimate factor determining numbers of moose on Isle Royale.

Although wolf predation does not control moose numbers in the usual sense (i.e., hold the population below the level where food supply had an effect), it is the primary mortality factor operating on this herd. Predation is the key factor shaping the adult survival pattern.

An Historical Perspective

The moose is a rather recent colonizer of Isle Royale (Fig. 77), probably swimming from nearby Ontario early in the present century, about the same time that moose populations on the mainland increased (Mech 1966). With no effective predators and an abundant food supply, the population grew to very high levels by the late 1920s. From ground observations, Murie (1934) estimated 1000-3000 moose present in 1929 and 1930. Browse depletion was evident at that time, and significant mortality from malnutrition apparently reduced the population to several hundred animals by the mid-1930s (Hickie 1936).

Fig. 77. Young bull moose in summer.

During the dry year of 1936, two large fires burned over 100km² of Isle Royale. Aspen and birch regeneration in the burn provided a renewed food supply, and the population again grew until direct mortality from malnutrition was observed in the late 1940s (Aldous and Krefting 1946; Krefting 1951). In 1947, a population of 600 moose was estimated by aerial strip count (Krefting 1951).

Wolves became established on Isle Royale in the late 1940s (Mech 1966). Though wolves undoubtedly were killing moose soon after their arrival, the initial effect of their predation is unknown. By flying overlapping strips and buzzing moose to flush nearby animals, Cole (1957) counted 242 moose. He estimated the 1957 population at 300. Three years later, however, Mech (1966) saw 529 moose on a similar census. He estimated 600 moose, conservatively, and concluded that wolf predation was keeping the moose population within the limits of its food supply.

Recent Moose Numbers

In the years following Mech's study, other researchers refined moose inventory techniques; however, inherent biases and wide sample variances among moose population estimates remained a problem. Even so, the moose population apparently increased during the 1960s (Wolfe and Jordan unpubl. data) and leveled off or perhaps even declined from 1970 to 1974. Population trends from 1964 to 1970 are reviewed briefly and data from 1970 to 1974 are presented in detail.

Population Size, 1964-70

Estimates of the moose population made by aerial censuses (Table 17) suggest that moose increased from 1960 to 1969. Improved counting procedures and accumulated experience probably explain some of the increase. Mech's (1966) estimate of 600 moose in 1960 allowed for a 12% adjustment for moose that were missed, but probably no more than two-thirds of the moose could be seen in strip counts. Even if Mech's estimate for the 1960 population were adjusted accordingly, there is still enough disparity between the 1960 and 1969 counts to indicate an increase in the population in the intervening period.

TABLE 17. Population estimates for the Isle Royale moose population, 1960-74.

Wolfe and Jordan (unpubl. data) consider moose pellet counts, 1964-70, to be a more accurate population index than the aerial censuses. They used the 1969 aerial census to "calibrate" the pellet counts, producing their "best estimate" (Table 17). They concluded that the population rose considerably between 1964 and 1967 and stabilized somewhat at around 1300-1600 moose from 1968 to 1970. Moose pellet counts made by Krefting (1974) also suggest an increasing moose population during the 1960s.

Use of pellet counts as population indicators assumes that defecation rates are constant throughout the study. Defecation rates may vary with quality and digestibility of food. Krefting's pellet counts rose steadily from 1950 to 1970; during the same period, the available browse supply declined. Increased use of less preferred, less digestible, browse species may have contributed substantially to the increased pellet counts.

Population Size, 1970-74

Midwinter aerial censuses in February 1972 and 1974, produced estimates of 818 ± 234 and 875 ± 260 moose respectively (Table 18). In 1972, zone boundaries were based mainly on the distribution pattern shown in Jordan et al. (1967). In 1974, the zones were redesigned with reference to current signs of moose, vegetation types, and elevation to reflect moose distribution more accurately. Four zones initially were assigned both years, but the results from the two lowest-density zones were similar and were combined as one in the table.

TABLE 18. Results of aerial moose censuses on Isle Royale, 1972 and 1974.

The irregular distribution of moose in midwinter is shown in Figs. 78 and 79. In 1974, the low density zone included most of the 1936 burn and much of the predominantly deciduous forests in the interior of the island, especially at higher elevations. Almost half of the island but only 15% of the population was included in this zone in 1974. Conversely, only 14% of the island but over half (57%) of the estimated population was included in the high density zone of shoreline and small islands, primarily areas of coniferous vegetation.

If an arbitrary 20% is added to the 1972 and 1974 estimates to account for moose that were missed, the resulting estimates would be 1023 and 1094 moose, respectively. These figures are lower than the 1969 estimate of 1438 (assuming an adjustment of 20%), but the differences are not statistically significant.

Fig. 78. Relative distribution of moose in February 1972. (click on image for a PDF version)

Fig. 79. Relative distribution of moose in February 1974. (click on image for a PDF version)

Relative mortality and reproduction, when compared to earlier conditions, indicate a probable decline since 1970. Judging from the number of kills found during annual winter studies since 1970, predation by wolves, especially on calves, has been consistently high (Appendix L). In addition, summer observations and fall aerial counts indicate low calf production and survival.

In spring 1974, it was still obvious that moose densities were high, relative to food supply. During 3 months of field work, six moose were found which appeared to have died of malnutrition. Even if the population has declined since 1970, heavy browsing in critical wintering areas during previous years could have long-term effects on forage production. Recent studies of browse production and utilization at the southwest end of the island indicate that moose densities in that area have remained high (Belovsky et al. 1973).

Moose population fluctuations in the near future will probably reflect a slowly contracting food supply unless fire intervenes. Winter conditions, which greatly modify food availability and energy requirements, will contribute to short-term variations.

Sex Ratio and Productivity

The proportion of adult males to females has a bearing on potential calf production (productivity), which is basic to an understanding of population dynamics. While the sex ratio of the Isle Royale moose herd is currently even and has apparently fluctuated little over the past two decades, calf production has shown a downward trend from 1959 through 1974. Net productivity (calves surviving their first year) seems to be relatively low, especially in those generations born after severe winters.

Sex Ratio

Aerial classification of 423 moose (365 adults) in October 1972-74 showed that the sex ratio of the Isle Royale moose herd was essentially even (Table 19). Although such counts often were flown during the 1960s, changes in methods preclude use of these counts for sex ratio variations. 1959-74.

TABLE 19. Fall aerial classification counts of Isle Royale moose, 1972-74.

Date	Total observed	Bullsa		Cows	Calves	Males/ 100 females	% calvesb	% yearlingsc
		Ad.	Yrl.
17-19 Oct. 1972	114	47	2	53	12	92.5	10.2	4.1
23-25 Oct. 1973	192	73	8	81	30	100.0	15.6	9.9
22-25 Oct. 1974	117	43	7	51	16	98.0	13.6	14.0
Total			180		185

The collection of skeletal remains of moose was used to check for significant long-term trends in the adult sex ratio. There was no apparent shift from an even sex ratio among adult moose dying from about 1950 to 1974 (Table 28). The slight majority of males in the total collection (54.5%) is not statistically significant (95% confidence interval is ±4.5%) and can be explained by the fact that male skulls with antlers are more easily found than skulls of females.

Other studies of naturally regulated moose populations have found that adult sex ratios are about even (Peek 1971a; Shubin and Yazan 1959; Knorre 1959). After reviewing North American and European studies, Bubenik (1972:276-295) concluded that in naturally regulated populations males occurred in equal proportion to females, with perhaps a slight excess of males in some areas.

Productivity

The percentage of calves in the moose population has been determined from aerial counts in fall and winter; summer field observations provide an index to the twinning rate (Fig. 80). These data can be used to estimate net productivity. The age at which cows first produce calves is unknown for Isle Royale moose, although other studies indicate that most females older than yearlings become pregnant each year (Pimlott 1959; Simkin 1965; Schladweiler and Stevens 1973). Changes in twinning rates best reveal long-term trends in productivity of Isle Royale moose.

Fig. 80. One-month-old moose calf.

CALF RATIOS, 1970-74

Moose herd composition, summer-fall in 1970-74, is shown in Table 20. Numbers of calves classified during fall aerial counts are probably more accurate than those made on the ground in summer. However, the fall aerial counts show a similar relationship to summer ground observations 1972-74, suggesting that the latter may provide at least a relative index to the calf segment.

TABLE 20. Moose herd composition, summer and fall, 1970-74.

Low calf numbers in summer 1971 and 1972, correspond to unusually deep snow the previous winters. Severe winter conditions probably were responsible for the low calf production. The relatively high calf crop in 1973 after a particularly mild winter further indicates that winter conditions may be important to productivity the following season. Ling (1970) correlated moose calf numbers with the severity of the preceding winter. He indicated that severe winters could affect calf production two summers hence by retarding the development of calves in their first winter and lowering their productivity as yearlings. Pimlott (1959) and Markgren (1969) believed that fertility of yearlings was quite dependent on nutrition during their first winter. Since yearlings comprise a relatively large segment of the adult population, a decline in fertility of female yearlings would greatly decrease potential calf production. Variations in yearling fertility and twinning rates (see next section) in response to winter conditions may be responsible for annual variations in relative calf numbers on Isle Royale.

NET PRODUCTIVITY AND CALF MORTALITY

Ideally, composition counts from successive fall and winter seasons should indicate the magnitude of calf losses in the intervening period. For Isle Royale moose such records are available for 10 years since 1959 (Table 21). Because of relatively small sample sizes, these percentages are approximate. Year-to-year comparisons have little meaning, but all years can be lumped to provide an average fall and winter calf composition of 17% and 14%, respectively. Since the winter counts were made when calves were 9-10 months old, the average yearling recruitment rate would be somewhat less than 14%—more likely 11-12%. Use of a hypothetical age distribution derived from a life table gives 13% yearlings.

TABLE 21. Proportion of moose calves in successive fall and winter aerial composition counts on Isle Royale (Mech 1966; Shelton 1966; Jordan and Wolfe [unpubl. data[; present study).

Jordan et al. (1971) calculated annual calf mortality on Isle Royale at 72% using a hypothetical calf crop combined with a 12.5% yearling recruitment rate. High calf mortality is characteristic of moose populations in general. LeResche (1968) estimated moose calf losses in the first 5 months of life in south-central Alaska at 56%. Rausch and Bratlie (1965), working elsewhere in Alaska, found calf mortality ranged from 30% to 80%. Peek (1971a) determined loss between the ages of 5 and 17 months at 24% in mild winter conditions, and 61% mortality in a severe winter in northeastern Minnesota. Heptner and Nasimovich (1967; cited in Peek 1971a) stated that moose calf mortality in the Soviet Union normally ranged between 30% and 50%, but was as high as 80% in some areas. Shubin and Yazan (1959) documented calf mortality of 43% and 66% from the embryo state to 7 months of age in 2 successive years in a Russian moose population.

None of the above authors attributed high calf losses primarily to predation. While wolf predation is implicated at least partially in high calf mortality on Isle Royale, it does not follow that, in the long run, calf survival would increase greatly in the absence of wolves.

LONG-TERM TRENDS IN PRODUCTIVITY

Pimlott (1959) and Markgren (1969) indicated that the twinning rate was correlated with the ovulation rate, and thus was a reliable index to fertility. While Mech (1966) and Shelton (1966) found twinning rates of Isle Royale moose from 1958 to 1963 to be higher than most reported elsewhere in North America, it is clear that the productivity of Isle Royale moose dropped off considerably in the last half of the 1960s and the early 1970s (Fig. 81; Tables 22 and 23). Of 213 cows with young observed during summer in 1959-65, 72 (34%) had twins. The corresponding figure for the period 1966-73 was 33/232 (14%)—a significant decline. Rates obtained in this manner should be considered minimal, since ground observations are biased against calves (Pimlott 1959), and there is some loss concurrent with the observations which took place over a period of many weeks.

TABLE 22. Twinning rates for Isle Royale moose from summer ground observations.

TABLE 23. Average occurrence of twins in summer and fall, 1959-73.

Fig. 81. Calf production declined by the early 1970s.

When Murie (1934) recorded a low twinning rate (6%) on Isle Royale, the moose population was at an all-time high of 2000-3000. The high productivity indicated by the twinning rates in the early 1960s was attributed to intensive wolf predation, which presumably was stimulating reproduction (Mech 1966; Shelton 1966). However, this situation clearly has not been maintained.

Both twins rarely survive their first winter; even by October their numbers have decreased substantially. The percent of twins in fall aerial counts has remained essentially the same during the entire period 1959-74 (Table 23). On Isle Royale a high twinning rate probably contributes little to yearling recruitment, although twins may serve as a buffer against predation on single calves.

The observed drop in twinning rates could reflect lower fertility (ovulation rates), or higher neo-natal mortality. Increasingly severe winter conditions resulting from greater snowfall and an increased moose population may have raised mortality rates of newborn calves by lowering the winter nutritional plane of pregnant cows. Such effects have been reported for most North American ungulates (Swenson 1973). Recent generations of undersized calves (see Residual Effects of Severe Winters) further support this idea. Low twinning rates elsewhere have been attributed to severe winter-spring weather (Knorre 1959) or to low nutritional levels of pregnant cows in winter. (Schladweiler and Stevens 1973; Houston 1968).

Moose Distribution and Habitat Relationships

Early successional stage forests typically support higher densities of moose in North America (Aldous and Krefting 1946; Krefting 1951; Spencer and Chatelain 1953; Peterson 1955). In primeval times and in modern-day wilderness areas fires are the primary factor creating large areas of moose habitat. Social behavior and reproductive characteristics enable moose to exist at low densities in mature forest types and yet increase rapidly to fill expanses of new habitat created by fire (Geist 1974). As forests mature, moose densities shrink in accordance with a reduced food supply. Current high densities of moose on Isle Royale are supported by older, more mature forests, but over time the moose population will decrease unless new burns occur.

Winter Distribution of Moose

Conifer cover becomes more important as snow accumulates in winter, especially if snow interferes with moose mobility. Not only is the snow softer and less deep, but snow-covered conifers are effective windbreaks and serve as thermal insulators during cold periods. They may also be cooler in warm periods. Moose often seek shade under conifers in midday, even in winter. With a relatively low surface area:volume ratio and a dark coat, it seems reasonable that moose exposed to full sunlight on calm days in winter could become overheated (Fig. 82).

Fig. 82. Even in winter moose may seek shade.

With average midwinter snow depths of about 60 cm, moose on Isle Royale occur in highest densities in shoreline areas where conifer cover, especially balsam fir, is most common (Figs. 78, 79). Although snow is rarely a serious impediment unless greater than 70 cm deep (Coady 1974a), Peek (1971b:39-49) found moose moving into conifer cover when snow was only 46 cm deep. Thus snow appears to affect habitat selection even when movements are not seriously restricted.

Recent burns may attract moose in spite of deep snow. The 1936 burn was prime winter habitat in the 1940s (Krefting 1951) and as late as 1957 (Cole 1957). Apparently abundant browse in recent burns compensates for the increased energy expenditure required to inhabit such areas.

Midwinter use of old burns by moose on Isle Royale is affected greatly by snow conditions. (Fig. 83). During his census in 1960, Mech (1966) found many moose in burns dating from 1936 and 1948. Snow depths during that census ranged between 30 and 41 cm. Moose were probably more active in open habitats because of low snow levels. During the early 1970s, utilization of these burns was very low, particularly in midwinter. Significant use was last recorded in 1968, when moose were able to move freely about because of little snow (Wolfe and Allen 1973).

Fig. 83. Moose utilize old burns on Isle Royale only with low snow depths.

The locations of wolf-killed moose also indicate that during periods of deep snow moose leave open burns and congregate in heavy cover near shorelines. With more than 50 cm of snow, there were more kills near shorelines and fewer kills in burns than with less than 50 cm of snow (Table 24).

TABLE 24. Effect of snow depth on location of wolf-killed moose.

Cows with calves may instinctively seek denser cover than other moose, presumably because it provides greater protection from wolves. During aerial surveys in Minnesota, Peek (1971a) found that cows with calves. were more likely to be found in heavier cover. On Isle Royale, kills of calves are usually near shorelines—in heavy cover—while kills of adults are more randomly distributed (Appendices I-K).

Moose Distribution During Other Seasons

In spring and early summer, moose are often found in aquatic habitat, although little is known of their island-wide distribution pattern (Figs. 84, 85). Botkin et al. (1973) concluded that aquatic plants are the most important source of sodium for moose on the island they believe that heavy utilization of aquatics is linked to a critical need for this element. Excessive heat and insect pests also may prompt use of aquatic areas (Kelsall and Telfer 1974).

Fig. 84. Cow in beaver pond in late June.

Fig. 85. Cow lying in Lake Superior on hot summer day.

Belovsky et al. (1973:101-122) estimated summer densities of moose on two study areas at the southwestern end of the island indirectly from estimates of browse production and intake by moose. Their calculated density was 3.8 "adult equivalents"/km². The same general area commonly has even more moose in midwinter and probably supports one of the highest year-round densities of moose on the island. Forests in the area are old and relatively undisturbed.

Minimum densities were estimated for some areas of the island from aerial composition counts in October 1974. An area of 18.4 km² along the north shore of Rock Harbor, extending from Starvation Point to Lake Richie, contained at least 52 moose, implying a minimum density of 2.8/km². This area supports a mature forest dominated by paper birch and aspen, with some balsam fir and white spruce. Most of it was burned shortly before 1846 (Janke 1973). Likewise, densities of moose were quite high in the mature forest of similar composition near Malone Bay. The isthmus between Malone Bay and Siskiwit Lake had a minimum density of 1.7/km². Since we made no attempt at a total count, actual densities were undoubtedly higher. By contrast, very few moose were observed in the 1936 burn north of Siskiwit Lake.

Moose-Habitat Relationships

Moose browsing has an obvious effect on the composition of the forest of Isle Royale, reducing or eliminating browse species in some areas. American yew (Taxus canadensis), once very abundant, now occurs only where inaccessible to moose. The effect of browsing on forest composition is most striking in the 1936 burn east and south of Lake Richie. Here white spruce—which is not browsed—is rapidly becoming a dominant tree species (Fig. 86). Aspen and paper birch regeneration was greatly retarded and in some cases killed by moose browsing.

Fig. 86. White spruce unbrowsed by moose, grows unimpeded in old burns.

Growth form has also been altered by moose. Krefting (1974) found that moose browsing depressed the growth of all favored species—aspen, balsam fir, Canadian honeysuckle (Lonicera canadensis), mountain ash (Sorbus americana), mountain maple (Acer spicatum), paper birch (Betula papyrifera), and redosier dogwood (Cornus stolonifera) (Fig. 87).

Fig. 87. Variations in balsam fir growth form are attributable to moose browsing.

Intensive moose browsing actually may prolong browse production in some species. Hazel (Corylus cornuta) and redosier dogwood show extensive lateral sprout growth. Both can tolerate heavy browsing and survive. The degree of browsing that a plant species can withstand is quite variable and differs greatly from site to site depending on soil conditions (Bergerud and Manuel 1968). Some aspen- and birch-stands on ridgetops where soils are poor have been killed by browsing, but just off the ridge there are stands of the same species that have grown beyond the reach of moose.

All important browse species except balsam fir show an island-wide downward trend (Krefting 1974). The more rapid, recent changes in the 1936 burn essentially have taken this area out of production of moose browse. Accordingly, moose have shifted from the burn to older forest stands.

Isle Royale continues to support high densities of moose in spite of its old forests. Its insular nature prevents normal dispersal and probably contributes to higher densities than would otherwise occur. The Isle Royale moose population is approximately twice as dense as populations in optimum moose habitat in north eastern Minnesota (Peek 1971a).

The reproductive potential of moose is adapted toward maintenance of the species at low densities in mature forests, rapid expansion into large areas of good habitat following fire, and slow contraction following forest maturation (Geist 1974). The Isle Royale moose population appears to be entering the phase of slow contraction. In the absence of fire, their food supply will continue to slowly shrink, leading eventually to a decreased moose population. National Park Service policies are directed toward restoration of a natural fire regime on Isle Royale, with moose densities adjusted accordingly.

Direct Mortality Factors

The Isle Royale moose population has been regulated naturally for its roughly 70-year history, since hunting has never been allowed on the island. From the inception of wolf-moose studies on Isle Royale, emphasis has been placed on the collection of carcasses and skeletal remains in order to determine patterns of natural mortality when moose are preyed upon intensively by wolves.

Calves on Isle Royale exhibit high mortality from year-round wolf predation and to a lesser extent malnutrition and accidents. While significant mortality of newborn calves may occur from uncertain causes, direct predation probably accounts for most mortality after about the first month of life.

As an adult, a moose generally lives for several years with a reduced threat of mortality from wolves, but predation losses increase steadily after 7 years of age. Adult losses occur mainly during winter and early spring. There is evidence suggesting that in late winter, bulls become especially vulnerable. Calves are quite vulnerable when snow is deep. Winter severity has a marked influence on the degree of malnutrition in the population—the record of recent years indicates that malnutrition early in life may account for increased vulnerability of maturing, young adults to wolf predation.

While wolf predation is the primary direct cause of mortality for the Isle Royale herd, accidents and malnutrition also claim some moose. Wolf predation and malnutrition mainly affect calves and old moose, but accidents (including drowning) occur more randomly among adults (Table 25). Since carcass collections are subject to several biases, the frequency of recorded deaths may not reflect mortality in accurate proportions. Calves are undoubtedly underrepresented in all categories because their bones are small and disintegrate rapidly, especially when a calf is very young. Drowning is probably overrepresented because moose that drown and remain floating usually are reported by boaters in summer, so proportionately more of these are examined. Much of the mortality from "unknown" causes probably is attributable to wolf predation.

TABLE 25. Cause of death for 836 moose, dying of natural causes, examined on Isle Royale, 1958-74.

Wolf Predation on Adults

Wolf kills and probable wolf kills account for 58% (307/531) of the known-age adults in the collection. Most of the known wolf kills listed in Table 25 were first located from the air in winter; probable kills mostly were skeletal remains found during summer field work. The age distributions of adults in the wolf-kill and probable wolf-kill groups are not significantly different (G_H = 18.18, 17 d.f., P <0.50). Certainly some of the dead moose in the probable wolf kill group were fed on by wolves subsequent to death from other causes, but the resulting error probably is small (Fig. 88).

Fig. 88. Most adult moose on Isle Royale are eventually killed by wolves.

Most predation on adults occurs during a relatively short period of time in winter and early spring. While predation in midwinter is often heavier on cows than bulls the bulls apparently suffer greater losses in later winter and early spring.

Usually, predation is most intensive on older moose, with the rate of loss increasing steadily after 7 years of age; however, the generalization that wolves prey most intensively on older moose has not applied in recent years, when young adults (1+ to 5+ years) accounted for over half of the wolf kills. These moose were apparently vulnerable to wolf predation primarily because of malnutrition early in life, even prior to birth.

SEASONAL ASPECTS

Season of death can be roughly determined for adult males using characteristics of antler development (Table 26). Season of death for females cannot be determined from skeletal remains.

TABLE 26. Incidence of antlers and antler growth in skeletal remains of male moose on Isle Royale.

Most bull moose are antlerless for about 4 months, from late December through April. The period of antler development is also 4 months in length (May-August), as is the period with polished antlers (September-late December) (Peterson 1974). Minor fluctuations in this pattern are unimportant. Moose with the largest antlers have begun antler development by the beginning of May, shed velvet early in September, and probably drop their antlers in late December. Moose with smaller antlers, such as yearlings, probably lag behind the larger bulls by at least 2-3 weeks. Losses of adult males are heaviest in winter and early spring. Using the percentages in Table 26, relative rates of adult male mortality are: May-August, 1.0; September-late December, 2.0; late December-April, 12.4. Mortality may be higher among males than females in late winter and early spring due to depleted energy reserves. Therefore, the weighting factor for winter mortality for the total population may be slightly less than 12.4. These figures represent a considerable revision of estimates of Jordan et al. (1971), which resulted in the following relative rates of adult losses: May-August, 1.0; September-December, 1.6; January-April, 1.8.

A rough picture of seasonal patterns of adult mortality can be derived in a different manner, using the occurrence of adult remains in the winter-kill collections versus the summer collections of wolf scats. Adult moose made up 67% (206/307) of the total carcasses from winter studies, 1959-73. In scat collections from the nonwinter period 1973, adults made up only 14% of the total remains of moose; the rest were calves. Adult remains comprised 25% of the classifiable moose remains in fresh summer scats collected from 1958 to 1960 (Mech 1966). While prey occurrence in scats may be affected by biases (Peterson 1974), these data further demonstrate the low vulnerability of adults in the nonwinter months.

AGE DISTRIBUTION OF KILLS

While wolves prey most heavily on old adults, significant variations in predation patterns over a 16-year period have occurred (Appendix L). Variations in the age distribution of the kill may reflect relative abundance of animals in the older age group and, more recently, increased vulnerability among young adults.

The general pattern. The known wolf kills and probable wolf kills were combined to provide a generalized picture of vulnerability according to age. The relative frequency of kills in each age group was compared to the hypothetical occurrence of each age group in the population (calculated from the life table) to illustrate the differential vulnerability of young and old adults (Table 27).

TABLE 27. Age-distribution of 307 wolf-killed adult moose compared to hypothetical occurrence in the Isle Royale moose herd.

The marked selectivity of wolf predation is readily apparent (Fig. 89). Mortality, relatively low from age 1 to 5 years, increases after 6 years. After 8 years, the percent occurrence in the kill is greater than the calculated percent occurrence in the population.

Fig. 89. Age-distribution of wolf-killed adult moose from Isle Royale compared to the hypothetical age-distribution of the population. (click on image for a PDF version)

Variations in predation patterns. Predation on the young adult group increased dramatically in 1971-73, compared to the early years of the project (Fig. 90). In the mild winter of 1973, moose aged 1+ to 5+ years accounted for 65% (13/20) of the adult wolf kills collected, indicating that the increased vulnerability did not arise from the effects of deep snow on moose mobility. Rather, as records were gathered in successive winters, it appeared that cohorts of calves born after a winter of nutritional distress, or perhaps experiencing such a winter as calves, were permanently more vulnerable to wolves.

Fig. 90. Temporal variations in age-distribution of 200 adult moose killed by wolves in late January, February, and March on Isle Royale, 1959-74 (see Appendix L for data sources).

Deep snow conditions greatly increased winter severity in 1969, 1971, and 1972, and nutritional distress in the intervening winter, 1970, was evidenced by the small size of calves born the following summer. It is significant that of 13 wolf-killed moose aged 1+ to 5+ years that were killed in 1973, 12 were born after one of these four winters. The other moose was a calf during the deep-snow winter of 1968-69.

In retrospect, the increase in frequency of kills in the group aged 1+ to 5+ years (1959-64:6% to 1965-69:22%) may reflect a slightly increased level of nutritional stress during the 1960s. Increasing trends in population size and snowfall (winter severity) plus a slowly declining food supply because of forest succession combined to cause a progressively lower winter nutritional plane.

While changes in vulnerability are responsible for changes in predation pressure on young moose, relative abundance may have influenced variations in predation on old moose. The majority of adults killed by wolves from 1959 to 1964 were aged 6+ to 10+ years, while from 1965 to 1969 the 11+ to 17+-year-olds were killed most often. In both cases the animals in the age groups most heavily preyed upon were born in a 10-year period, 1947-57. Possibly a series of years with exceptionally good yearling recruitment (calves surviving to age 1) produced a "bulge" in the age structure of the moose population. When these animals became vulnerable, predation may have increased simply because of higher availability.

SELECTION RELATED TO SEX

Mech (1966), Shelton (1966), and Wolfe (in press) found more females than males in their samples of adult moose killed by wolves during the winter—primarily February and March. Since the sex ratio of the herd was probably even, a proportionate number of bulls died at another time, perhaps in late winter and early spring as suggested earlier. While females predominated in winter wolf kills, males predominated among skeletal remains found randomly in summer (Table 28).

TABLE 28. Percentage of males (±95% confidence interval) in total sexed remains of carcasses and skeletons of adult moose on Isle Royale.^a

PREDATION COMPARED TO HUNTING

Wolf predation represents a natural mortality factor that has shaped survival patterns in moose populations for thousands of years. Predation is usually selective for calves and old moose, and takes equally from each sex in the long run. By contrast, modern hunting of moose in North America takes from all age groups and usually takes more males than females (Cumming 1972; Karns 1972:115-123).

Hunting may lower the percentage of males in a moose population to the point where calf crops suffer. A bull probably fertilizes only one female during the primary breeding period in early October, especially in low-density populations (Fig. 91). Cows that are not bred at this time will again be fertile in 25-30 days, but calves produced from late breeding will be smaller at the onset of winter, which may be critical in harsh environments (Markgren 1974). Additionally, calves born after the peak calving period might be highly vulnerable to wolves in summer.

Fig. 91. Bull moose following cow during rut.

Hunting is a more random mortality factor—humans tend to remove animals in proportion to their occurrence in the population. The age distribution of hunter kills in Ontario is compared to that of wolf kills on Isle Royale in Figure 92.

Fig. 92. Comparison between age distribution of wolf-killed adult moose from Isle Royale and hunter-killed moose from Ontario (Addision and Timmerman 1974).

Over time, unless harvests are restricted by age and sex, intensive hunting will result in a younger overall age structure, especially among males. Restrictions in Norway and Sweden allow a hunting kill that more closely approximates natural mortality (Bubenik 1972:276-295). Large numbers of prime-breeding-age animals, both males and females, are spared to maximize productivity (Lykke 1974). Similar intensive management in North America is a likely development as human populations and hunting pressure increase throughout moose range.

We should recognize that hunting never could fully duplicate the selectivity that is characteristic of wolf predation. Elimination of wolves probably removes the most important selection force operating on moose populations throughout their evolutionary history, and we should attempt to maintain these natural predators in their ecological role wherever feasible.

Wolf Predation on Calves

Calves comprise the age group most often preyed upon by wolves. The evolutionary significance of predation on calves is readily seen in the development of a strong social bond between cows and their calves during the first year of life. Calves have little defense against wolves, even though they may weigh 3-4 times as much as wolves in winter, so their survival depends on adequate defense provided by their mother (Fig. 93).

Fig. 93. A calf depends on its mother for protection in the first year of life.

Moose calves are preyed upon at all times of the year on Isle Royale. After the first few weeks of life predation is the most important cause of calf mortality. In winter certain snow conditions may increase the vulnerability of calves to wolves.

CALF LOSSES IN SUMMER

Calf mortality is heaviest during the first 5 months of life, when the proportion of calves in the population drops from a hypothetical 31% (Jordan et al. 1971) to 17% (Table 21). The causes of this mortality are difficult to document. Twenty-four percent (77/305) of the calves represented in the bone collection were less than 6 months of age. Of these 77 remains, cause of death was recorded as follows: known and probable wolf kills 53%, accidents 12%, unknown 35%.

Calf remains were found in 56% of the wolf scats from nonwinter months in 1973—accounting for 86% of the total identifiable remains of moose. Since young calves seem quite susceptible to malnutrition and accidents, mortality is high whether wolves are present or not. Some of the feeding may simply be scavenging. Occurrence of calves in wolf scats does not decrease during the summer, however, soil seems reasonable to assume that predation causes most of the loss after calves are 1-2-months-old.

CALF LOSSES IN WINTER

Predation causes most calf losses in winter, with known and probable wolf kills totaling 82% (175-213) of the remains of 6- to 12-month-old calves in the bone collection. Winter observations indicate that scavenging is minor relative to direct killing of moose. Calves account for 34% (111/328) of the kills located during 16 winters, but the average proportion of calves in the population during this time of year was estimated at only 14% (Table 21), indicating a marked selectivity.

Snow conditions are extremely influential on calf vulnerability (Fig. 94). During a severe winter in southcentral Alaska, 56% of 57 wolf-killed moose examined were calves (Stephenson and Johnson 1973). On Isle Royale, deep snow made calves especially vulnerable during the winter studies of 1969, 1971, and 1972. Peterson and Allen (1974) commented on the effect of reduced mobility of calves:

Fig. 94. Calves are more vulnerable to wolves when snow is belly-deep.

In deep snow a calf finds it easiest to follow in the tracks of its mother. But, when defending their young, cows invariably move to the rear of their calves, providing protection for the area most vulnerable to wolf attack. While this is highly adaptive behavior under normal circumstances, when snow is deep the calf must break trail under a great handicap. This slows the movements of the pair when wolves approach. In 1972, wolves supported by a crust were observed trotting easily beside a calf that was struggling to move through deep snow, protected from behind by its mother. In another case, wolves surprised a cow and calf who were browsing a short distance from one another. They wounded the calf before the mother could come to its aid. In 1971, we recorded two cases of calves probably abandoned on lake edges by their mothers, who may have gone inland to feed.

On Isle Royale snow depths greater than 75 cm, which is about chest height for a 6-month-old calf (82 cm, Kelsall 1969), cause a significant increase in the percentage of calves in the kill (Table 29; Fig. 95).

TABLE 29. Effect of snow depth on occurrence of calves in wolf kills.^a

Fig. 95. A cow moves to the rear end of her calf when confronted by wolves.

Snow conditions may adversely affect yearling recruitment. On Isle Royale in October 1972, yearlings totaled only 4.1% of the moose population (Table 19). This low yearling percentage probably resulted initially from low 1971 calf production followed by intensive predation the following deep-snow winter, when calves accounted for 47% of the wolf-killed moose examined (n = 37).

There is no conclusive evidence that calf survival is higher without wolves, primarily because comparisons of moose populations with and without predation are invariably complicated by differences in other environmental factors. Calf losses to wolves are not always compensatory, for in some winters wolf predation may cause more mortality than probably would have otherwise occurred (Bishop and Rausch 1974). However, calf survival may be very low even without wolf predation (Houston 1968; LeResche and Davis 1973), demonstrating the potency of other mortality factors.

Malnutrition

Although most important as an indirect cause of mortality, malnutrition may act directly at times. Probably the most significant direct loss of individuals from malnutrition occurs immediately after birth (Fig. 96). Although it is very difficult to determine its actual occurrence, five calves less than 1-week-old were found dead in 1975, probably from fetal malnutrition. Calves may be born weak and underweight, unable to stand and nurse.

Fig. 96. Calves may die of nonpredatory causes soon after birth.

Twin fetuses have a higher nutritional requirement than a single fetus, so prenatal malnutrition will affect twins the most. One of the five Isle Royale calves found dead in 1975 was known to be a twin. Neonatal mortality was implicated in low twinning rates among Shiras moose (Schladweiler and Stevens 1973). Knorre (1959) mentions several instances of one calf from a set of twins dying soon after birth.

Direct loss to malnutrition in late winter and spring is more easily documented. During the late spring of 1972, which followed a particularly severe winter, I found carcasses of several moose that appeared to have died of malnutrition. Most were near conifer cover, close to lakeshores. The immediate area around the carcass was often depleted of food, indicating that the moose had been there for some time. Most died resting on their sternum, with legs beneath them. Usually their bone marrow was entirely depleted of fat.

Older records list some moose which seem to fit the above description, and for which there was no other obvious cause of death. These are included in the malnutrition group in Table 25. Of 27 such deaths, 19 came from the severe winters of 1970-71 and 1971-72. Calves comprised 26% of the deaths from malnutrition in these 2 years, suggesting a somewhat higher vulnerability in this age class.

The age distribution of adults suspected of dying of malnutrition is skewed toward older animals, as in the wolf-killed sample, and males outnumber females almost 2 to 1. For moose 9 years and older, nearly 3 times as many males die (although the sex ratio derived from Table 35 is 42:58). Several studies have shown that bulls, especially those over 2.5-years-old, lose weight during the rut, while females lose little or no weight (Skuncke 1949; Knorre 1959; Yazan 1959). McGillis (1972) found in Alberta that males older than yearlings lost much fat during the rut and were unable to recover this substantial loss before the onset of winter. Because of lower fat reserves, males may be expected to show the effects of negative energy balance before females do.

Most of the known deaths from malnutrition on Isle Royale occurred in late winter. In the spring of 1972, when deep snow lasted well into April, it was obvious that there were more vulnerable moose than the wolves could "cull." Actually, considering the severe conditions, it is somewhat surprising that there was not more direct mortality from malnutrition that year—we found only 12 carcasses in more than 1600 km of hiking.

Direct mortality from malnutrition in late winter is a good indication that the relationship between the moose population and their food supply has changed since the early 1960s. Recent winters of above-average severity or duration, in effect, reduced the carrying capacity of the island to the point where available forage was insufficient to sustain moose in some areas.

Drowning and Other Accidents

Accidents appear to be an insignificant cause of mortality. Except for calves, such mortality appears to be nonselective.

Drowning is mentioned frequently as a cause of mortality among moose. Calves on Isle Royale are more susceptible to drowning than adults, making up 40% of the total recorded. Six of the eight calves were less than 3 months of age. Probably most of these young calves drowned while following their mothers across bays of Lake Superior. High waves may be a factor, since no drowned calves were reported from inland lakes. LeResche (1968), however, reported calves swimming well "in the face of 2-ft waves and high winds." He believed that drowning probably was insignificant after the age of 1 month.

Most of the adults that drowned probably fell through thin ice in early winter or late spring. Two were found in inland lakes, the other 10 were floating in bays of Lake Superior. In females, the approximate date of drowning can sometimes be determined by the relative size of fetuses. A cow that drowned in Rock Harbor contained two well-developed fetuses, indicating death in April (Mech 1966). During a late spring in 1972, the National Park Service boat Ranger III reported a cow moose frozen in the ice of Rock Harbor. From fetal development we judged it to have died in February 1972, when ice was just forming. Most drowned adults on Isle Royale were young, corresponding to the higher occurrence of these animals in the population. There is no evident selectivity for older animals, as is the case with some mortality factors.

Most accidental mortality other than drowning is caused by falls (Fig. 97). Three of the 14 carcasses were found in old copper-mine shafts. Four others. (including three calves) died of broken legs; one calf had a broken neck. Falls down steep slopes or over precipices and banks accounted for four others. (Five known deaths from falls which occurred during encounters with wolves are listed as wolf kills.)

Fig. 97. Injuries, such as this broken leg, may predispose moose to predation.

Ironically, the oldest moose recorded from Isle Royale died accidentally in February 1974. This 19.5-year-old cow fell 5 m from a steep shoreline coated with ice.

There is only one record of a bull moose on Isle Royale dying of puncture wounds incurred during the rut (Hakala 1953). Two others died after locking antlers during the rut in 1968 (Wolfe, in press) (Fig. 98).

Fig. 98. These two bulls drowned in a beaver pond after locking antlers, a rare occurrence during the rut.

Mortality from Unknown Causes

A substantial portion (30%) of the autopsy records list cause of death as "unknown." In most cases these were skeletal remains found months or even years after the actual death of the animal. There are significant differences in the age distribution of wolf-killed adult moose and those dying of unknown causes (G_H = 37.7, 17 d.f., P<0.01). Adults in the "unknown" mortality group have a younger overall age structure than wolf kills (32% vs. 16% less than 6 years old, respectively). Wolfe (in press) concluded from these disparities in age distribution that wolf predation was an unlikely cause of "unknown" deaths. However, the only other mortality factors known to be operating on this moose population are malnutrition and accidents. As previously discussed, malnutrition does not seem to be an important cause of death for the relatively young animals in question; accidents appear to be insignificant.

Thus wolf predation appears to be the only cause of death that could be responsible for losses of such magnitude. Significantly, males predominate in the "unknown" sample (67:33); most of them died between December and April, since 72% had no antlers. This suggests late winter or spring predation, since at that time of year critical malnutrition is more common among males than females. The younger age distribution of "unknown" deaths might result from a late winter-spring increase in moose vulnerability.

Indirect Mortality Factors

Residual Effects of Severe Winters

For moose, food availability is closely related to winter severity. Deep snow restricts movements, reducing energy intake and increasing energy expenditure. Snow conditions are particularly important on Isle Royale, where densities of moose relative to food supply are high. Generally, with deeper snow, moose become more limited to areas of conifer cover near shorelines. This may restrict food consumption or reduce the quality of forage consumed and lead to heavy browsing which may adversely affect future browse production. With deep snow, the kill rates of Isle Royale wolves generally increase, primarily because of greater vulnerability of calves.

Although deep snow could have a direct effect on survival of young adults (Peterson and Allen 1974), the long-term effect of severe winters and associated malnutrition on young moose is more important. Predation increased on moose born after a winter when malnutrition was evident—usually a winter of deep snow. Early growth and development in these moose may have been retarded to the point where survival in subsequent years was lowered (Fig. 99).

Fig. 99. Abnormal antlers of 2.5-year-old moose killed by wolves in 1972.

In moose, body size at birth is largely dependent on the nutritional status of the mother during the latter part of gestation (Knorre and Knorre 1956; Knorre 1959). Metatarsal length was used in this study as an index of both calf size and the mother's nutritional state during pregnancy. Malnutrition most retards the bones having the highest growth rates during the period of nutritional distress. The major leg bones of young animals show different growth rates—the metatarsus has a higher in utero rate of growth, making it the most developed hind limb bone at birth (Peterson 1974). Metatarsus length of domestic lambs up to 10-months-old was correlated with prenatal nutritional plane and depended very little on postnatal nutrition (Palsson and Verges 1952); this also was assumed true for moose. Extensive collections were possible only from 9- to 10-month-old calves. Fortunately, wolves commonly left the lower hind leg (the metatarsus) intact.

The collection of calf metatarsi began in the winter of 1971 and continued through the study. These were used to compare calves born in different years. (Table 30). Analysis of variance (model II) indicates significant differences between cohorts (P < 0.025). Mean metatarsal length increased in each succeeding year, with the greatest jump occurring in 1973. A multiple comparison test (Newman-Keuls) indicates that the metatarsi of the 1973 calves were significantly longer than those of the 1970 and 1971 calves (P < 0.05).

TABLE 30. Mean metatarsal length (in mm) for 6-to 12-month-old moose calves according to year of birth.

In addition to fetal malnutrition, a delayed birth (because of delayed estrus) might cause shorter metatarsal length in a moose calf killed in its first winter. Late breeding would be expected most often among yearling females. (Coady 1974b). However, because of their low winter nutritional plane we would expect few Isle Royale yearlings to be breeding. Additionally, calves born late would be highly vulnerable to wolves in summer and few would survive until winter.

In 1972, we extended the collection of metatarsal to provide baseline data for all ages of moose (Table 31). Little increase in metatarsi length occurred after 4 years of age, so these animals were placed in one group. Metatarsal length in females generally averaged slightly less than in males, but the difference was statistically significant (P < 0.03, t-test) only in the relatively large sample from moose more than 4-years-old.

TABLE 31. Mean metatarsal length (mm) of Isle Royale moose (primarily wolf kills).^a

The age at which epiphyseal union occurs provides another indicator of growth. Between the epiphysis (the bony cap at the end) and the metaphysis (main shaft) of long bones is a layer of cartilage—the center for longitudinal bone growth. After growth has been complete, the cartilage is replaced by bone, which fuses the epiphysis to the main shaft.

Studies of black-tailed deer have shown that deficient diet could delay epiphyseal fusion for at least 12 months, and that fusion of metatarsal epiphyses in this species occurs by 27-29 months of age (Lewall and Cowan 1963).

Epiphyseal closure in moose metatarsi can also occur as early as 29 months of age (Table 31). Thus, 9 of the 17 wolf kills between 2 and 4 years-old (seven, 2-3 years old and two, 3-4 years old) suffered obvious growth retardation. These young adults were born in years when calves were undersized at birth, indicating a long-term vulnerability resulting from early malnutrition.

Factors Affecting Adult Vulnerability

Increased age is an obvious liability to moose on Isle Royale. Of special interest, then, are pathological conditions associated with advanced age. Usually, examination of wolf-killed moose is limited to well-chewed skeletal remains, making it impossible to check for abnormalities in soft tissues. Even so, bone abnormalities. (arthritis, jaw necrosis, severely fat-depleted bone marrow, or broken bones) were found in a high percentage (38% of 205) of wolf-killed adults examined, 1958-74. Arthritis and jaw necrosis are rare among moose less than six years old. Arthritis and malnutrition affects males much more than females, probably contributing to some differential mortality between the sexes. In addition to bone pathology, a high load of parasites may increase vulnerability in some moose.

ARTHRITIS

Arthritis, through restricted mobility, probably plays an important role in increasing the vulnerability of moose to wolf predation (Fig. 100). The arthritic lesions observed in Isle Royale moose closely resemble osteoarthritis, or degenerative joint disease (Neher and Tietz 1959; Stecher 1963). Sokoloff (1969:2) defines this condition as a "noninflammatory disorder of movable joints characterized by deterioration and abrasion of articular cartilage, and also by formation of new bone at the articular surface." This type of arthritis occurs primarily in old individuals and is most common in weight-bearing joints. Among Isle Royale moose, arthritis was found most frequently in the hip (Table 32).

TABLE 32. Site of arthritic lesions in Isle Royale moose.

Fig. 100. Normal vs. arthritic pelvic joint of moose.

Arthritis is usually not evident before 7 years of age in either sex, but after this age is attained the overall incidence in males is significantly greater than in females (Fig. 101): 40% of 102 males compared to 13% of 88 females. (t_s = 4.47, P <0.001). While arthritis is often a normal consequence of aging, the differences between males and females pose several interesting questions about arthritis which research has not yet answered.

Fig. 101. Incidence of arthritis in the coxofemoral (hip) joints of moose dying on Isle Royale.

JAW NECROSIS

Necrotic lesions of the bone surrounding the cheek teeth of older moose were often found (cf. Murie 1934; Mech 1966; Shelton 1966). This seems to be a chronic infection caused by microorganisms that gain entrance to the soft tissues surrounding the teeth.

It is probable that microorganisms responsible for jaw necrosis in moose are universally present in the oral cavity. Entrance into the soft tissues of the jaw can be gained through any opening. The molariform teeth of a moose wear down considerably with age, and necrotic lesions commonly develop when the tooth surface is worn to or below the gum line (Passmore et al. 1955). Food that becomes impacted between teeth or around their base may also provide a focus for infection (Fig. 102). The infection may spread to the entire toothrow or remain localized. In severe cases, teeth may loosen and fall out, and, in rare cases, the jaw may break.

Fig. 102. Jaw necrosis in Isle Royale moose, apparently resulting from (a) excessive wear, and (b) food impaction.

Jaw necrosis is limited to older individuals, and the incidence of necrotic lesions is directly related to age (Fig. 103). Of 344 adults over 7 years of age, 25% had some jaw pathology at death. Ten percent showed severe necrosis (loss of teeth or massive bone necrosis.

Fig. 103. Incidence of jaw necrosis among moose dying on Isle Royale.

MALNUTRITION

Moose suffering from malnutrition are undoubtedly more vulnerable to wolf predation. Moose with class 3 bone marrow probably had seriously depleted their energy stores. This may be a conservative estimate, since some marrow placed in the class 2 category had very low actual fat content.

Bone-marrow condition was noted for 269 wolf-killed moose (including probable kills) during 1958-74 (Table 33). All calves and almost all of the adults were killed in winter, usually during January, February or March.

TABLE 33. Incidence of fat-depleted bone marrow in Isle Royale moose killed by wolves, 1958-74.

The percentage of calves with serious fat depletion (45%), 1970-74, is significantly higher than for 1958-69 (21%) (t_s = 2.21, P < 0.03). During the recent period, marrow fat of calves was quite variable but correlated to a certain extent with winter severity. In 1969-70, (a winter of apparent nutritional distress) all four calves examined had severely depleted bone marrow. During the difficult winters of 1970-71 and 1971-72, of the calves examined, severely depleted marrow was present in 59% (13/22). By contrast, none of the 10 calves examined during the mild winter of 1972-73 had severe marrow fat depletion. Of 11 calves examined, 1973-74, 27% had severely depleted marrow. These samples are small, but they provide some indication of the effect of winter conditions on calf welfare.

The low incidence of fat-depleted marrow in wolf-killed adults in 1970-74 resulted from the large number of young adults in the sample. All these animals had normal bone marrow.

Males are more susceptible to death from malnutrition than females, as discussed previously. The incidence of severe marrow-fat depletion in adult males (21% of 78 kills) was significantly higher than in adult females. (9% of 103 kills) (t_s = 2.27, P <0.03). Like arthritis, malnutrition apparently contributes to a differential mortality between males and females.

PARASITES

The incidence of parasites in Isle Royale moose is difficult to document, since soft tissues are rarely available for examination. Nevertheless, since wolves are sensitive to slight changes in the health of a moose, heavy parasite loads might contribute to increased vulnerability. Parasites may be quite innocuous in healthy hosts, but they constitute a greater liability with increased age and reduced vigor.

In addition to two common ectoparasites—moose fly (Lyperosiops alcis) and winter tick (Dermacentor albipictus)—three internal parasites were noted during early Isle Royale investigations: a lungworm (Dictycaulus hadweni) and cysts of two tapeworms (Echinococcus granutosus and Taenia hydatigena) (Murie 1934; Mech 1966). Jordan (unpubl. data) found two additional organisms: Sacrocystis (a fungus) and Nematodirus (an intestinal nematode).

According to Mech (1966), heavy infestations of ticks are often found on animals in poor condition and in some populations are an important, direct cause of death (Fig. 104). Ticks cause irritation sufficient to make moose scratch and rub their hides, even with their teeth, often resulting in hair loss in areas of heavy tick infestation. Mech (1966) suggested that irritation from ticks accounted for bare areas seen on moose in spring. However, since the bare areas on the upper back and neck correspond to those where winter hair is first shed, this loss of hair in spring is more likely a molt pattern. Such bare areas are rarely seen in Alaskan moose (A. Franzmann, pers. comm.), indicating an unusual condition on Isle Royale.

Fig. 104. Male tick (second from left) and three female ticks in varying stages of engorgement with moose blood (scale in centimeters).

While it is often possible to state whether ticks were present on an animal killed by wolves, the importance of such infestations cannot be determined in most cases. There is little evidence that ticks are a primary factor predisposing Isle Royale moose to wolf predation. However, in some cases ticks may act with other influences to increase vulnerability.

Hydatid cysts. The internal parasite of Isle Royale moose which seems the most important is Echinococcus granulosus. Cysts of this parasite contain the larval stage, hundreds of tiny protoscolices, each of which is capable of developing into an adult tapeworm if eaten by a definitive host (Smyth 1964). Cysts are usually found in the lungs, occasionally the liver, and were found once in a kidney (Shelton 1966) (Fig. 105). They commonly reach an inch in diameter.

Fig. 105. Hydatid cysts in moose lung.

The only definitive host for this parasite on Isle Royale appears to be the wolf (Peterson 1974). The mature tapeworm develops in the intestine of wolves; eggs are shed with the feces. The eggs are somehow ingested by moose, and the cycle is completed as the larval stage migrates to the lungs and liver via the bloodstream, finally forming a hydatid cyst (Smyth 1964).

In an individual moose, hydatid cysts apparently accumulate with age, as shown by other studies (reviewed by Mech 1966) and by data gathered by the many investigators involved in Isle Royale wolf-moose studies since 1958. In a sample of 32 moose, those older than 3 years of age had at least a few hydatid cysts (Table 34). Older moose on Isle Royale probably all harbor E. granulosus.

TABLE 34. Incidence of hydatid cysts in lungs of Isle Royale moose.^a

After reviewing several studies of the effects of hydatid cysts on the health and vigor of ungulates Mech (1966) concluded that, in a stressful situation such as a wolf attack, a heavy load of hydatid cysts could increase the vulnerability of a moose. Under persistent attack by wolves, even a slight impairment in the ability of the moose to escape probably is significant. (At the same time, the parasite increases the possibility of transmission to the definitive host, thus completing its own life cycle.)

Life Table for Isle Royale Moose

Life tables are a valuable means of summarizing mortality patterns in relation to age. Mech (1966) and Shelton (1966) used a life-table approach, based on age according to toothwear, to analyze mortality patterns in the Isle Royale moose herd. Wolfe (in press) constructed a life table for the same population based on collections through 1969 using the more accurate cementum annulations method. Because major shifts have occurred recently in the mortality of certain age groups, and sampling of year-round mortality has greatly increased since 1970, a new life table has been constructed. This life table, based on carcass collection spanning the period 1958-74, should result in a more realistic representation of long-term mortality patterns in the Isle Royale moose herd.

Techniques

The life table for Isle Royale moose was constructed according to methods described by Deevey (1947) and Caughley (1966). Included are estimates of the following parameters:

(1) Mortality (d_x): the number of deaths within the age interval x to x + l. This is determined from the proportion of recorded deaths in an age interval to the total deaths recorded for all ages, applied to the initial cohort size (usually 1000).

(2) Mortality rate (q_x): the proportion of a cohort dying between age x and x + l. This is determined by dividing d_x by l_x.

(3) Survivorship (l_x): the probability of an individual surviving from time of entry into the table to any age x. Usually l_x is multiplied by a constant K = 1000. Then the l_x series indicates the number alive at any subsequent age.

Since natality rates for Isle Royale moose are unknown, I did not attempt to combine mortality rates with estimates of fecundity to arrive at an estimate of age distribution. Instead, I calculated the "L_x" statistic (Deevey 1947), which is simply an average of successive "l_x" figures. This is the hypothetical age distribution at a point midway in each age interval, assuming equal recruitment each year and constant mortality rates.

I have assumed that the aggregation of records from all sources adequately represents year-round mortality patterns in the Isle Royale moose population over roughly the last two decades. Data on sex, age, and approximate time of death for the specimens used for the analysis are provided in Appendix M. In addition, 54 skeletons that could not be assigned to a specific 5-year period were included in the life-table calculations.

Possible Biases

While simple in principle, the construction of life tables for wild populations may be complicated by various difficulties (Caughley 1966; Eberhardt 1969). If a life table is calculated from mortality data spanning several years, it is necessary to make the assumption, often untenable, that population size and age structure have remained stable during those years. Sampling must adequately represent mortality during the entire year. Small sample sizes in each age interval may be a problem with long-lived animals. An accurate method of age determination is necessary. Survivorship curves made by plotting successive "l_x" values may not be representative of a population when derived from "time-specific" records, since there are variations resulting from unusually high survival or mortality in particular cohorts (Bubenik 1972:276-295; Collier et al. 1973).

I must assume that population size and age structure have remained relatively constant during the period of collection. Caughley (1966) stated that the error introduced by population fluctuation probably is insignificant if the period of collection exceeds the average wave length of the fluctuations; under such conditions variations in mortality would be compensatory. Because there have been some population changes, the life table will be biased to the extent that fluctuations in population levels and mortality patterns. are not compensatory—thus the current life table for Isle Royale moose is only an approximation of population parameters.

The total skeletal collection was used as the data base for the life table. Initially, the data derived from moose found during random summer ground searches were used. However, the 60 :40 sex ratio in the random collection indicates inadequate sampling for all seasons, since the actual sex ratio has probably varied little from equality. It is true that in the total collection the January-March period is more heavily sampled than April-December, but exclusion of the winter-spotted kills would result in more bias than when they are included. Since many of the remains of moose found during winter aerial searches would have been found during later random searches, the entire collection appears to be the least biased representation of year-round mortality.

Since the entire collection spans 25 years there is little reason to believe the data are biased by a year, or even a series of years, of unusual calf production or survival. Possible sex bias stemming from differential ease in finding remains appears insignificant (Peterson 1974). Calves, especially those 1-6 months old, are certainly underrepresented in the skeletal collection, so the life table has been constructed for the adult segment of the population only. The estimate of calf mortality by Jordan et al. (1971) is incorporated into the survivorship curve.

The Life Table

Life table series were calculated for adult males, females, and combined sexes (including moose of unknown sex) (Table 35). The resulting survivorship curve illustrates the age-selectivity of wolf predation—the major mortality factor for adult moose (Fig. 106). This curve would be a straight line if mortality rates were constant for moose of all ages. Using this curve we can follow the fate of a hypothetical cohort of 1000 yearlings. Moose that have survived their first year of life have a mean life expectancy of 7.3 years. Between the ages of 1 and 7 years, annual mortality rates are about 10% or less; rates increase steadily thereafter. Males have essentially died out by age 15.5 years, females by age 19.5.

TABLE 35. Life table for Isle Royale.

Fig. 106. Survivorship curves for male and female adult moose on Isle Royale.

The mortality rate for Isle Royale moose (Fig. 107) shows the same "U"-shaped pattern that Caughley (1966) found typical of a wide variety of mammals, ranging from laboratory populations of rats and voles to wild and domestic ungulates. Similar mortality trends in African ungulate species have also been reported (Spinage 1970, 1972).

Fig. 107. Age-specific mortality rates for Isle Royale moose.

Calves averaged 14% of the population in mid-winter on Isle Royale; yearling recruitment was estimated at 11-13%. Average annual adult mortality was calculated at 13% (Table 35). The similarity between these figures suggests long-term stability in the moose population. This does not imply that substantial fluctuations could not have occurred. The calculated average mortality rate is rather insensitive to small differences between recruitment and mortality which could, over a period of several years, lead to considerable population change. A large difference between recruitment and mortality in a single year could also result in a significant population change but would not be reflected in these statistics.

Sex Difference in Survival

Isle Royale bulls have not been found to live longer than 15.5 years, while cow moose have lived as long as 19.5 years. Males of a given age tend to have a slightly higher mortality rate than females of the same age (Table 35). The greater incidence of malnutrition and arthritis in bull moose has already been discussed. Sex differences in survival, while apparently real, are not sufficient to significantly alter the calculated age distribution of males and females, even when only the oldest age groups. (11+ years and up) and compared (G_H = 12.3, 8 d.f., P <0.15). Larger sample sizes are needed to confirm a small difference in survival.

From the age distribution (L_H, Table 35) of the moose population the proportion of males in each age interval was calculated (Fig. 108). The sex ratio remains essentially even from ages 1-8. After age 8, the proportion of males drops gradually for several years, with the rate of decline accelerating in the oldest age classes. The proportion of males does not fall below 45% until after 10 years of age. Since only about 12% of the population is calculated to be older than 10 years, differential mortality in males and females has little effect on the sex ratio of the entire population.

Fig. 108. Calculated proportion of males alive in each age interval, based on a hypothetical age distribution for adult moose on Isle Royale.

Higher mortality rates are common in males of various ungulate species. The reasons suggested for disproportionate male mortality include greater growth rates in males and stress during the rut, when male activity is greatest and food consumption drops markedly (Robinette et al. 1957; Borg 1970; Flook 1970). These factors may explain the greater incidence of malnutrition among bull moose on Isle Royale. Since wolves are capable of detecting slight incapacities, arthritis is probably an important additional reason for a shorter average lifespan among males.