F:\ALCES\Vol_38\Pagemaker\3816. ALCES VOL. 38, 2002 MERCER AND McLAREN – CARRYING CAPACITY IN NEWFOUNDLAND 123 EVIDENCE OF CARRYING CAPACITY EFFECTS IN NEWFOUNDLAND MOOSE W. E. Mercer 1 and B. E. McLaren 2 18 Virginia Place, St. John’s, NL, Canada A1A 3G6; 2Government of Newfoundland & Labrador, Department of Forest Resources & Agrifoods, P.O Box 2222, Gander, NL, Canada A1V 2N9 ABSTRACT: Newfoundland moose (Alces alces americana) increased following 1904, the year of successful introduction, to peak numbers in 1958. The population subsequently decreased to record low numbers by 1973, when an area-quota management system was instituted throughout the island (112,000 km2) in 38 moose management areas, in part, to respond to issues related to habitat and accessibility for hunting. Subsequent quota-management manipulations permitted the island- wide population to increase in accessible areas to record high numbers by 1986, after which populations again decreased, to a 1999 estimate of 125,000 animals (post-hunt). We hypothesise that, unlike most studied irruptions of cervid populations, moose populations in Newfoundland, and subsequently habitat carrying capacity (K), decreased on inaccessible range following 1958 to very low density, from which both have never recovered. Decreases in relative numbers of young moose seen while hunting and during winter classifications are consistent with increases in the number of moose seen during increase phases during 1966–99. These observations are less obvious for less accessible management areas. We explore other recruitment and density relationships as they have been developed in association with our estimate of K in moose for Newfoundland. We illustrate that, although some decrease in moose numbers following 1958 and 1986 was the result of management, changes to population size and to K also resulted in reduction in productivity, such that density dependence explains > 10% and up to 76% of hunter-observed recruitment. ALCES VOL. 38 : 123-141 (2002) Key words: accessibility, Alces alces, carrying capacity, hunter reports, moose, Newfoundland, population dynamics, quota management, recruitment Moose (Alces alces) populations are often difficult to compare because of geo- graphic differences in scale, habitat conti- nuity, and the relative importance of various mortality factors, including hunting, which introduces issues of hunter accessibility and its effect on population dynamics (Van Ballenberghe and Ballard 1998). Funda- mental questions remain about what limits or regulates moose populations as there is much geographic variation in the relative effects of predators, human hunting, and primary production (Gasaway et al. 1992, Crête and Courtois 1997, Saether 1997, Crête and Daigle 1999). Habitat carrying capacity (K) is a concept that assists man- agers in area comparisons and in resolving when density-dependent effects are pre- dicted to occur (McCullough 1999, Person et al. 2001). Approximation of K for cervids has been discussed in theory by Caughley (1976), and as an experiment by McCullough (1979). Crête (1989: 378) described K as a bounded rather than a constant value, vary- ing with effects of winter snowfall, annual primary productivity, and forest succession. The usefulness of McCullough’s approach to calculate K in areas where habitat changes is a problem also discussed generally in the original text (McCullough 1979: 156). In this paper, we attempt to estimate K using hunter reports of several local populations of moose. We compare areas where moose management has had more CARRYING CAPACITY IN NEWFOUNDLAND – MERCER AND McLAREN ALCES VOL. 38, 2002 124 Fig. 1. Selected management areas in Newfound- land in which trends observed by either-sex, resident moose hunters are compared. Shaded region is referred to as central Newfoundland in this paper. Gros Morne National Park (GMNP), Terra Nova National Park (TNNP), and other areas referred to in the text are indicated. and less success in Newfoundland over the past 4 decades. We hypothesise that, unlike most studied cervid populations (McCullough 1997), moose populations in Newfoundland, and subsequently K, decreased on inacces- sible range to very low density following initial irruptions, from which both have never recovered. We also predict that with an approach of more accessible, hunted populations to our estimates of K for these areas, lower numbers of young moose would be seen during hunting trips. These predic- tions allow us to illustrate density depend- ence in Newfoundland moose based on hunter reports during 1966–99. The relative importance of density-dependent versus density-independent factors for fluctuations in ungulate populations in the absence of predation can be tested by the following specific questions: (1) are indices related to moose abundance (i.e., moose observed by hunters), or to hunting (i.e., total days of hunting and total licences issued) more im- portant in explaining recruitment observed by hunters in autumn; (2) do all manage- ment areas experience declines following observed peaks in moose density; and (3) are observations of abundance related to observations of recruitment by hunters? Our contention that variation in density- dependent reproduction depends on varia- tion in K in time and space (Crête 1989) is contrasted to claims that regulation of moose density is dependent on hunter functional response, and that habitat or food supply may influence only the synchronicity of p o p u l a t i o n c y c l e s i n N e w f o u n d l a n d (Ferguson and Messier 1996). We show limitations in hunter functional response to moose density in Newfoundland, and we offer guidelines that may assist moose man- agement in the future. POPULATION AND MANAGEMENT HISTORY The island of Newfoundland (Fig. 1), which encompasses 112,000 km2, forms a test case of the question of population regu- lation, because of the absence of predators of adult moose other than hunters. Four adult moose (A. a. americana), 2 females and 2 males, were successfully introduced to Newfoundland from New Brunswick, Canada, in 1904 (Pimlott 1953, Broders et al. 1999). Rapid dispersal and low densities characterized the first 25 years of popula- tion increase (Pimlott 1953) and wolves (Canis lupus) were extirpated during that period (Mercer 1995). During 1953–56, an increase rate was estimated for insular Newfoundland at 0.33, based on observa- tions of young moose in mid-winter surveys (Pimlott 1959a). Keith (1983) later calcu- lated an average intrinsic rate of increase of r = 0.23 for North America in situations of CARRYING CAPACITY IN NEWFOUNDLAND – MERCER AND McLAREN ALCES VOL. 38, 2002 126 change may have caused the decrease in licence sales from 8,660 that year to 6,523 in 1953 (Pimlott 1959b). Later, minor changes to season length did not appear to affect licence sales (Mercer 1974). Gradu- ally, hunting seasons have increased in length, beginning in September or October and ending between December and Febru- ary; in 2002 the moose season was 15 weeks long, including a two-week, bow- hunting season. In 2002, 27,820 moose licences were available in insular New- foundland (Mercer 1995). METHODS We define K as the maximum density of moose that can be supported at equilibrium, in a stable environment and in the absence of time lags, as McCullough (1979) defined KCC. More generally, our definition agrees with Odum (1953), who defined K as the upper asymptote of the logistic or sigmoid curve describing unimpaired population growth. The inflection point of the S- shaped curve describing such growth has been used to determine maximum sustained yield (MSY), also termed inflection point carrying capacity (ICC) by McCullough (1979). MSY has been shown to occur near 0.6 K for ungulates (McCullough 1984, Per- son et al. 2001). Much of the information for determin- ing changes to the issue of moose licences (“quotas”) in Newfoundland remains the same as that reported in Mercer and Manuel (1974). Annual response to a questionnaire (“return”) attached to every moose licence is usually > 75%, but only following 1–2 reminders mailed to nonrespondents in Feb- ruary or March after the season closes. Initial returns, including a completed ques- tionnaire and the lower mandible of any moose taken on a licence (individual li- cences have always been for 1 moose), are about 50% of licence sales to resident hunt- ers without a reminder. The mailed remind- ers are the extent of enforcement, although wildlife regulations stipulate returns are mandatory within 7 days of a kill or by the end of the hunting season. Data from completed questionnaires are coded digit- ally and archived by June of each year, and records in this form date back to the 1966 hunting season. Hunter trends are calcu- lated annually from the resulting time series up to the hunting season of the previous year. On this timetable, the most recent hunter information can help set quotas only for the following hunting season, because draw notices (indicating successful licence applicants for the next season) are mailed by June each year. Among those questions answered by moose hunters (including co-licence hold- ers in a party hunt), the most reliable infor- mation has been considered for the calcula- tion of trends by moose management area, as follows: (1) average number of animals reported killed by licence type (either-sex or male-only hunters) as a percentage of all licences sold by type (“hunter success”); (2) average number of days spent hunting by all licence holders spending at least 1 and at most 24 days hunting by licence type (“days of hunting”); and (3) average number of moose seen by the licence holder divided by the average number of days of hunting reported (“moose seen / day of hunting”). In instances where management areas were subdivided only for a portion of our study period, 1966–99, we pooled data from subareas for our trend calculations. We adjusted hunter success from the calcula- tion using initial questionnaire respondents by assuming success of all nonrespondent hunters was represented by reports from reminder respondents. Hunter respondents given first reminders only reported no dif- ference in success from second-, third-, or fourth-reminder respondents, although all nonrespondents reported lower success than initial respondents without reminders (Wild- ALCES VOL. 38, 2002 MERCER AND McLAREN – CARRYING CAPACITY IN NEWFOUNDLAND 127 life Division, Newfoundland and Labrador, unpublished data). We did not exclude any either-sex, resident licence holders in the calculation of days of hunting and moose seen / day of hunting. We reported, in addition to hunter success, a fourth trend; kill rate; expressed as total estimated moose killed / 10 days of hunting, a factor of the division of hunter success by days of hunt- ing. We tracked licence issue and kill estimates from 1945, the year of the first general season for all Newfoundland ex- cluding the Avalon and Burin Peninsulas, to 1999, the last year of records available to us (Pimlott 1953, Mercer and Manuel 1974, Mercer 1995). From these figures, we calculated hunter success for either-sex licence holders. Additional information on questionnaire returns has been collected and archived but not regularly used in management, including information regarding co-licence holders, area access, age, and sex of moose seen during days of hunting, number of moose seen by calendar day, and date and location of moose kill. Herein, we calculate a fifth trend, using age and sex classification of moose reported seen by hunters as an index of recruitment in autumn. For all hunters reporting “calves seen,” we calculate the number of young as a ratio of the number of “cows seen” (young seen / 100 adult fe- males). We excluded, in this instance, hunters who did not report seeing “calves,” because there exists considerable bias in the sighting and identification of young moose, as discussed by Pimlott (1959a) and Mercer (1974). We show trends for the either-sex, resident moose hunter in insular Newfoundland (averages for the island por- tion of the province of Newfoundland and Labrador), in central Newfoundland (aver- ages for management areas 16, 17, 22, and 24), and in management area 26 on the south coast (Fig. 1). A network of forest access roads makes areas 16, 17, 24, and 22 the more accessible hunting areas studied, with mean geographic distances to the near- est road in each of these areas 2.4, 1.1, 1.0, and 0.7 km, respectively (Mercer 1995). Area 26 is among the least accessible hunt- ing areas in Newfoundland, with a mean distance to the nearest road of 9.6 km. Thus, area 26 offers a likely example of a moose population at K (Mercer and Manuel 1974). To correspond our estimates of moose seen / day of hunting from hunter reports to moose density, and to calibrate hunter esti- mates of recruitment in autumn with mid- winter, aerial observations, we considered area 24, in which sufficient aerial surveys had been conducted to form a time series similar to hunter trends. We used 8 esti- mates of population size, between 1973 and 1997, obtained from helicopter counts in winter with stratified random surveys, in which we adjusted all counts by a factor of 2.7 to correct visibility bias, an average correction factor for forested areas in New- foundland (Oosenbrug and Ferguson 1992, Gosse et al. 2002). Counts were divided over the entire survey area (including unforested regions) of area 24 to obtain average density in animals per square kilo- metre. We reported all linear relationships be- tween indices with adjusted r2. Stepwise multiple regressions, from which we re- ported Mallow’s statistic (Cp), were used to determine significant predictors of re- cruitment observed by groups of hunters, from insular and central Newfoundland, and from management areas 16, 17, 22, 24, and 26. In each stepwise procedure, we allowed variables to enter if they were significant at P < 0.10, and to stay in the model if they were significant at P < 0.05. We included among those variables, for the corresponding hunting area, moose seen / day of hunting for the same year and for the previous year, young / 100 adult females ALCES VOL. 38, 2002 MERCER AND McLAREN – CARRYING CAPACITY IN NEWFOUNDLAND 131 Table 1. Mean values for moose seen / day of hunting (M), hunter reports of autumn recruitment in young seen / 100 adult females (R), and hunter success (S, %), during 1984–89 and 1966–99 (excluding 1984–89). Theoretical habitat carrying capacity, K, is derived from the longer period, by extrapolating hunter reports, where possible, to an autumn recruitment rate of 20 young seen / 100 adult females using linear regression (see Fig. 7). K is reported in moose seen / day of hunting, ± 95% confidence intervals (from linear regression), and as density ( / km2) using the correspond- ence from Fig. 8. We also report slope and adjusted r2 for the regressions, where P < 0.05. 1984–89 1966–99 K Hunting area M R S M R S slope r2 P Moose seen ( / km2) Insular NF 1.10 49.4 85.2 0.69 56.4 68.0 -30.4 0.80 0.000 1.88 ± 0.51 7.5 Central NF 1.23 45.7 83.0 0.60 57.3 61.7 -25.1 0.42 0.000 2.00 ± 0.61 8.0 Area 16 1.47 46.1 95.7 0.48 57.8 72.6 -30.9 0.17 0.016 1.60 ± 0.48 6.4 Area 17 1.85 42.6 95.2 0.76 58.1 72.1 -26.0 0.63 0.000 2.24 ± 0.64 9.0 Area 22 1.59 47.2 93.0 0.66 59.4 79.1 -39.1 0.26 0.007 1.66 ± 0.44 6.6 Area 24 0.83 49.1 91.7 0.56 57.9 73.9 -28.1 0.31 0.001 1.94 ± 0.59 7.8 Area 26 1.32 35.9 89.5 0.98 28.7 75.9 — — 0.131 — — reports were 14–36% higher, ranging from 56–59 seen / 100 adult females. During 1984–89, the relationship between young seen / 100 adult females and moose seen / day of hunting formed a consistently shal- lower slope than the rest of the series, making the graphed relationship for the entire study period appear slightly curvilin- ear (Fig. 7). Hunter reports from area 26 suggest that there was a lower threshold in young moose reported in autumn of about 20 seen / 100 adult females. When we extrapolated the relationship excluding the peak period in other areas to estimate po- tential moose seen / day of hunting at 20 young seen / 100 adult females (i.e., at a theoretical zero population increase, or K), based on area 26, all areas produced similar estimates of 1.6–2.2 moose seen / day of hunting (Table 1). Consistently in all areas, according to stepwise regression, moose seen / day of hunting and the autocorrelated series lagged 1 year (young seen / 100 adult females in previous years), explained the most variance in young seen / 100 adult females. The best model fit occurred in areas 17 and 22, and in insular Newfound- land as an average (Table 2). In no areas were indices related to hunting (i.e., total days of hunting and total licences issued) significant in explaining recruitment in au- tumn according to stepwise regression. The principal component regressions also showed our models to be significant in all cases except in area 22, and the first principal component, explaining nearly 100% of the covariance in the 2 series, indicated that moose seen / day of hunting was negatively associated with the autocorrelated series, as in Fig. 7, and explained a similar amount of variance as the linear regressions, as in Table 1, r2 = 0.11 to 0.76. The match resulting from comparing hunter trends to aerial surveys in area 24 appeared to be approximately 4x the number of moose seen / day of hunting to arrive at CARRYING CAPACITY IN NEWFOUNDLAND – MERCER AND McLAREN ALCES VOL. 38, 2002 132 Fig. 7. Young seen / 100 adult females plotted against moose seen / day of hunting by either-sex, resident moose hunters in insular Newfoundland, central Newfoundland, and area 26, 1966–99 (solid circles). A linear re- gression is fit through the period, excluding observations from 1984–89 (open circles), and is extended toward the x-axis to make predic- tions about habitat carrying capacity, K (see Table 1). Fig. 8. Correspondence between moose seen / day of hunting by either-sex, resident moose hunters (left axis, solid circles) and moose density ( / km2 ) estimated from mid-winter, aerial surveys (right axis, open circles) in area 24, (A) as a time series and (B) as a regression for corresponding years. The 90% confidence interval for moose density is a vertical line over the 1997 aerial survey estimate in (A), used to calibrate the series. survey result in 1997 weighed heavier than the estimates from the 1980s in our com- parison. Expressed in both instances as young / 100 adult females, the hunter obser- vations of recruitment in autumn and aerial surveys of mid-winter recruitment in area 24 produced a near match, although the relationship is also weak statistically, r2 = 0.03 (Fig. 9). Matching our predictions for moose seen / day of hunting at 20 young seen / 100 adult females with the corre- spondence to aerial survey data in area 24, we suggest that densities of 6–9 moose / density estimates in moose / km2 in this area, although the relationship between the 2 indices was weak, r2 = 0.08 (Fig. 8). The 4 survey results from the 1970s and the ALCES VOL. 38, 2002 MERCER AND McLAREN – CARRYING CAPACITY IN NEWFOUNDLAND 133 Table 2. Results of principal component regressions of autumn recruitment in young seen / 100 adult females in Newfoundland moose populations by hunting area, using either-sex, resident hunter reports of “calves” and “cows” seen during their trips, 1966–99. We report eigenvalues and factor loadings for 2 principal components, PC 1 and PC 2 , of the covariance matrix between moose seen / day of hunting (M) and the autocorrelated series of young seen / 100 adult females in the previous year (R t–1 ). We report from the stepwise regression procedures producing M and R t–1 as the only significant predictive variables: Mallow’s statistic (Cp); we report from the regressions using PC 1 to predict autumn recruitment: F, P, adjusted r2, sum of squares (SS), and error sum of squares (SSE). PC 1 PC 2 Factor Factor Hunting Eigen- loadings Eigen- loadings Statistics from regression models area values M R t–1 values M R t–1 Cp F P r2 SS SSE Insular NF 67.59 -0.03 1.00 0.02 1.00 0.03 2.9 102.5 0.000 0.76 1,600.7 484.0 Central NF 95.21 -0.02 1.00 0.05 1.00 0.02 2.1 39.6 0.000 0.55 1,587.6 40.1 Area 16 130.41 -0.02 1.00 0.13 1.00 0.02 1.1 4.8 0.036 0.11 557.5 3,612.8 Area 17 167.57 -0.02 1.00 0.10 1.00 0.02 3.5 17.3 0.000 0.34 1,815.2 3,244.9 Area 22 233.40 -0.02 1.00 0.14 1.00 0.02 2.2 3.7 0.065 0.07 785.2 6,626.7 Area 24 108.79 -0.01 1.00 0.04 1.00 0.01 1.8 16.9 0.000 0.33 1,123.9 2,058.7 Area 26 280.21 -0.02 1.00 0.19 1.00 0.02 1.0 4.2 0.052 0.11 1,027.1 6,149.2 km2 approach K for moose in Newfound- land (Table 1). DISCUSSION Habitat Carrying Capacity for Moose McCullough (1979) extended the linear regression of rate of recruitment on post- hunt population size to a theoretical zero population increase to obtain an estimate of K for white-tailed deer ( Odocoileus virginianus) in the George Reserve, Michi- gan, USA. Although McCullough’s (1979) regression suggested linearity, our data in- dicate curvilinearity in the relationship be- tween recruitment observed in autumn and population size (Fig. 7). Linearity appears to be a reasonable interpretation of recruit- ment in autumn for Newfoundland moose during the increase phase (1973–83), until peak densities were achieved. This differ- ence from McCullough’s (1979) interpreta- tion may be explained by the possibility that the George Reserve deer herd, which peaked at 34 deer / km2 during 1952–71, before a hunting experiment, was not allowed to reach densities high enough to show a de- cline to K. Post-hunt peak density was 19 deer / km2 and mean density was much lower. In 1935, estimated density for the George Reserve deer herd reached 48 deer / km2. Deer population densities in other areas with no hunting have been higher yet. For example, in Saratoga National Park, New York, USA, without hunting, densities ranged between 37–74 deer / km2 and aver- aged 53 deer / km2 during 1985–94 (Underwood and Porter 1997). This den- sity may be similar to densities observed in other parks where hunting is prohibited or restricted. McCullough (1997) reported very high densities in a population of black- tailed deer (O. hemionus columbianus) on Angel Island (2.2 km2), California, USA. There, during a period of 20 years (1965– 84), 5 peak populations were recorded to average more than 100 deer / km2. These densities did not appear to be declining over time. CARRYING CAPACITY IN NEWFOUNDLAND – MERCER AND McLAREN ALCES VOL. 38, 2002 134 Fig. 9. Correspondence between young seen / 100 adult females by either-sex, resident hunt- ers (open circles) and by observers during mid-winter, aerial moose surveys (solid cir- cles) in area 24, (A) as a time series and (B) as a regression for corresponding years. We estimate that K in Newfoundland boreal forests approximates 6–9 moose / km2, calculated from theoretical hunter re- ports of 1.6–2.2 moose seen / day of hunting at an extrapolated hunter-observed autumn recruitment of 20 young / 100 adult females (Table 1). This estimate depends on the assumption that changes to young seen / 100 adult females did not occur with changes to moose density for area 26 and that moose seen / day of hunting varied presumably only with annual weather or other random effects (i.e., area 26 was at K throughout our period of observation, when autumn recruitment was 20–40 young seen / 100 adults; Fig. 7). The related assumption of zero population growth means that mean mortality between autumn and the following spring is about 30 young moose / 100 adult females. Our hunter trend involving moose seen / day of hunting produces a conserva- tive estimate of real moose density at high values, because: (1) we used a more con- servative correction factor for visibility bias than indicated specifically for area 24 (Oosenbrug and Ferguson 1992); (2) there are many kills that occur with < 1 day of hunting, but our index of hunting effort does not measure less than that period; and (3) on average hunters see more moose when days of hunting are fewer, and hunting trips are very short when hunter success is > 80%. Thus, our investigation of K from extrapolating moose seen / day of hunting is likely to be an underestimate particularly at h i g h d e n s i t i e s ; a g a i n , t h i s c o n t r a s t s McCullough’s (1979) conclusion that linear regression overestimates K. Combining observations along several hunting routes, we also generalize the effect of changing habitat on our estimate of K. However, our estimate of K for moose is higher than estimates elsewhere, especially in the pres- ence of wolves (Gasaway et al. 1992, Van Ballenberghe and Ballard 1998). Our estimate of K in Newfoundland should be compared to past density esti- mates for this unique situation. As an example of estimates during the first peak of moose densities in the boreal forest, densities in area 17 in 1960 were observed at 4.6 moose / km2 (Bergerud and Manuel 1969). This figure was likely an underesti- mate of the real population size, because the equivalent of 5.0 moose / km2 were shot along roads in 1962 (Bergerud et al. 1968). Multiplying the 1960 estimate by our visibil- ity correction factor for forest of 2.7 would result in 12 moose / km2 as a minimum density for area 17, such that this earlier peak exceeds our upper estimate of K. Estimates of moose density calculated only ALCES VOL. 38, 2002 MERCER AND McLAREN – CARRYING CAPACITY IN NEWFOUNDLAND 135 for areas of forest cover, in inaccessible parts of Newfoundland (cf. area 26), were stable but highly variable since 1960 (as in Fig. 4), and were often > 12 moose / km2 (Mercer 1995). To further support our estimates of K, which we suggest may have been exceeded by the peak densities achieved by all dispersing and expanding populations in Newfoundland, new densi- ties recorded in lowland forests in Gros Morne National Park, without legal hunting, were 3.4 moose / km2 (uncorrected) and 7.4–10.6 moose / km2 depending on the visibility correction factor (McLaren et al. 2000). In nearby area 40, where moose have recently reached their highest num- bers, the post-hunt density ranged from 3– 4 moose / km2 (uncorrected) or 8–10 moose / km2 (using the average visibility correction factor of 2.7) during 1989–99 (Mercer 1995). Comparisons may also be made among the estimates of deer and moose densities above if food production in different veg- etation types and biomass production dif- ferences of the ungulates are considered. Crête and Manseau (1996) and Crête and Daigle (1999) have performed reviews of this type, in which they suggested that vari- ation in moose biomass depends on the presence of other deer species and on the existence of predators. Moose on the south shore of the St. Lawrence River in the absence of wolves reach 740 kg / km2, while even in the presence of wolves, they reach a biomass exceeding 1000 kg / km2 on Isle Royale (Crête and Daigle 1999). In the forage-limited area these authors studied, on the Québec-Labrador Peninsula, where wolves are present, production estimates are 78 young / 100 adult females, and au- tumn recruitment is relatively high (90% survival to autumn) compared with our es- timates for Newfoundland. Our calculation of K in moose in the forested areas of Newfoundland may be as high as the equiva- lent of 3 young, 8 adult females, and 4 adult males (Table 1), which, according to Crête and Daigle’s (1999) estimates, approaches 5,000 kg / km2. To compare, 100 white- tailed deer / km2, with the same sex and age ratio, approaches 5,200 kg / km2. These estimates can apparently only be achieved in the absence of additive predation (Gasaway et al. 1992, Van Ballenberghe and Ballard 1998, Person et al. 2001). Population Dynamics in Newfoundland Moose 1904–99 Differences in K and resulting differ- ences in population dynamics reported in this paper can be explained by differences in hunting pressure / unit area relative to local density of moose. For example, if we compare central Newfoundland areas, then area 24, with the smallest amplitudes in moose seen / day of hunting (not shown), maintains a higher kill rate and a conse- quently lower density at peak, as well as a higher hunter-observed young to adult fe- male ratio (Table 1). In area 24, licence issue also explains more of the variance in the number of young relative to adult fe- males reported seen than in any of the other areas, suggesting a stronger influence of hunting. In contrast, an inaccessible area, such as area 26, has a much lower kill / unit area relative to its density (Fig. 4), and, as a result, a consistently low young to adult female ratio (Table 1). In general, we conclude that barren areas of Newfound- land (40% of insular Newfoundland and represented here by area 26), or other areas less accessible to hunting, following a popu- lation peak approaching K in the 1950s, moose populations decreased and remained very low thereafter (Fig. 10). In Terra Nova National Park (forested and accessi- ble but with negligible, illegal hunting of moose), the moose population now behaves similar to inaccessible areas in that it also experienced no recovery after a decline CARRYING CAPACITY IN NEWFOUNDLAND – MERCER AND McLAREN ALCES VOL. 38, 2002 136 Fig. 10. Summary of observed and theoretical population dynamics for moose in insular New- foundland: (A) young seen / 100 adult females by either-sex, resident hunters (solid line) as a function of estimated moose density ( / km2), with our range estimate of maximum sustained yield (MSY) for forested habitat as vertical dotted lines; (B) annual recruitment (solid line) as a function of moose density (MSY, range as in (A); and (C) a representation of population changes during 1940–2000 in forested, accessible areas (longer dashed line), which included 2 peaks above MSY (range as in A, now represented by horizontal dotted lines), in 1958 and in 1986, and in forested portions of less accessible areas (shorter dashed line), in which density-dependent ef- fects near or beyond our estimate of K re- sulted in a permanent decline in moose after 1 9 5 8 . T h i s f i g u r e w a s a d a p t e d f r o m McCullough (1984). following a 1958 peak in density. We pre- dict that in this area, the population will remain low for many years; current esti- mates of mid-winter recruitment from the last aerial survey are 20 young / 100 adult females in the park (Gosse et al. 2002), consistent with our estimates of zero popu- lation increase. We also predict that the moose population in Gros Morne National Park under present management will de- crease from its present high density and follow a similar trend. This population dy- namic is also different from that of other parks (e.g., Isle Royale National Park, Michigan, USA; McLaren and Peterson 1994) and other areas with wolves. In areas of Newfoundland more acces- sible to hunters, the population dynamic is much different. We describe 2 major peaks in moose density (1958 and 1986) both exceeding MSY but not necessarily K (Fig. 10). We maintain that the decline following the 1958 island-wide peak in moose resulted in part from hunting in some accessible areas, but also from a natural die-off caused by habitat destruction when populations grew > K in less accessible areas. Mercer and Manuel (1974) recorded low autumn recruitment in all areas they reviewed, where young observed as a percentage of winter- surveyed moose were 20–40% in accessi- ble areas and 10% in inaccessible areas. Those authors hypothesized that the differ- ence in winter recruitment was a result of destruction of winter food resources. Ac- cording to a study of areas of different forest productivity in Norway, differences in spring recruitment in moose begin with a measurable change in fecundity (Saether et al. 1996). In Newfoundland, a similar change in fecundity was observed in the 1980s in area 24 – consistent with the decline in young / 100 adult females observed in mid- winter, aerial surveys (Fig. 9), there were 44% young moose observed as twins in 1983, 21% in 1984, 18% in 1985, and gener- ALCES VOL. 38, 2002 MERCER AND McLAREN – CARRYING CAPACITY IN NEWFOUNDLAND 137 ally < 5% in current classifications (Mercer 1995). Nutrition-induced changes in vul- nerability to predation likely result in later changes to summer recruitment, owing to the importance of mostly compensatory pre- dation by black bear (Ursus americanus), estimated at 22% (area 24) to 38% (else- where in central Newfoundland) of young moose during summer (Mercer 1995). For central Newfoundland moose populations, we further estimate that density-dependent effects were not evident until a density of about 2 animals / km2 was reached (Fig. 10), corresponding to about 0.5 moose seen / day of hunting (Fig. 7). This density is below our estimate of MSY, taken at 0.6 K, or about 3.5–5.5 moose / km2. Using 2 methods, the moose population in insular Newfoundland was estimated in 1988 at 167,000 (post-hunt) and 217,000 (pre-hunt), and a population decline was predicted (Mercer 1995). If we use the post-hunt estimate, along with our assess- ment of the recent population decline by 25% of the peak density in 1986 (Fig. 4), then the current (1999) population of moose in Newfoundland is about 125,000 (post- hunt). Local moose densities range from < 0.1 to > 8 animals / km2 (Mercer 1995); thus, some populations are clearly < MSY, whereas others are probably > K (Fig. 10). Our estimate of the highest peak in moose densities from the first peak in hunter suc- cess (Fig. 3) indicates that, consistent with Caughley’s (1976) interpretation for ungu- lates, the first irruption occurred in 1958 and has never been exceeded. Management of Moose without Wolves Managers were late in responding with quota increases to the increase in moose populations in accessible areas throughout Newfoundland during the 1980s and 1990s, and were thus indirectly to blame for the latest observed declines in autumn recruit- ment. In the 1970s, increases in licence issue were primarily from the promotion of the male-only licence and the opening of new hunting areas; only by the late 1980s did total licence sales increase substantially as a response to increasing moose densities (Fig. 2). Since 1974, males continue to represent 65–75% of the legal (reported) kill. McCullough (1979) illustrated that a male-only harvest cannot move a popula- tion away from K, and that MSY cannot be achieved without harvesting females. Com- bined with relatively little change in density of moose (a 25% decline) during the 1990s (Fig. 4), harvest of most moose populations in Newfoundland has led to declines in sex ratio (Mercer 1995). Licence issue has not substantially changed throughout the 1990s, resulting in some new density-dependent declines in autumn recruitment in areas where moose have arrived more recently, such as the Northern Peninsula (McLaren et al. 2000). Although this generalized example of passive or precautionary man- agement in licence issue was a response to presumed stable or declining moose populations and declining hunter success (Fig. 3), this management was inconsistent with the decline in autumn recruitment also reported by hunters, especially in central Newfoundland (Fig. 6). Moose manage- ment in areas without wolves and a lack of rigid control of hunting is a difficult enter- prise. We illustrate the use of hunter statistics to show a biological phenomenon. For management of moose, it is useful to have measures of abundance and recruitment that are more cheaply obtained than by aerial survey, and hunter indices have often been suggested as an approach (Courtois and Crête 1993, Timmermann 1993). Other measures of abundance and recruitment based on cohort analysis, through more detailed investigation of the age structure of the hunter-killed population, rely on esti- mates of both kill and hunter effort, result- CARRYING CAPACITY IN NEWFOUNDLAND – MERCER AND McLAREN ALCES VOL. 38, 2002 138 ing in high and sometimes misleading corre- lations between reconstructions and the hunter indices, on which they are based (Fryxell et al. 1988, Ferguson 1993). At- tempts to calibrate these reconstructions with hunter indices result in circular argu- ments that have nonetheless been suggested for use in management (Fryxell et al. 1988, Ferguson and Messier 1996). For a discus- sion of biases in such methodology, see Caughley (1976). We apply hunter reports in an uncorrected fashion, taking advantage of their annual recurrence, to measure rela- tive moose abundance over time. We cali- brate these reports to real population dy- namics by including hunter observations of young moose as a measure of autumn re- cruitment. Trends from areas with frequent misreporting by hunters (not shown), such as coastal areas 23 and 29 (Fig. 1), do not include correlations between moose obser- vations as we have shown here, but we suggest that correlations between young and adult moose would not be directly mis- reported from areas that do show consistent trends. Our main conclusion is that, while some of the decrease in Newfoundland moose numbers following 1958 and 1986 is the result of management, changes to popu- lation size relative to K, as well as changes to K in some areas, resulted in density- dependent reproduction effects explaining > 10% and up to 76% of the decrease (Table 2). Our conclusions are consistent with what Saether (1997) terms a “general hypothesis” regarding the relative impor- tance of density-dependent versus density- independent factors for fluctuations in un- gulate populations in the absence of preda- tion. In contrast to our interpretation of moose population dynamics in Newfoundland (Fig. 10), the assumption in an argument for- warded by Ferguson and Messier (1996) is that Newfoundland hunters (and their man- agers) have always kept moose < K. Un- fortunately, we note that the basis of their argument, their measures of functional re- sponse in hunters, whether as effort (number of days spent hunting) or as kill rate (number of kills / licence / day), are not independent of their measure of moose density, which itself is based on hunter effort, measured by the number of days spent hunting (Ferguson 1993). Ferguson and Messier’s (1996) ar- gument for cycling in moose based on co- hort reconstruction of population size, prompted by the analysis conducted by Fryxell et al. (1988), is subject to non- independent validation of population esti- mates, against kills / day of hunting and moose seen / day of hunting. As we show (Fig. 4), these indices are highly correlated. Moreover, biases in behaviour of hunters affect kills / day of hunting, as discussed by Hatter (2001). The 2-year delay in man- agement response to information obtained from hunters (see Methods) is the most obvious of the “time lags” referred to by Ferguson and Messier (1996) in their argu- ment for delayed density dependence in Newfoundland moose hunting. We do not agree that this delay is responsible for cy- cling in moose. MSY densities have not been maintained in Newfoundland; moreo- ver, moose populations reached either 1 or 2 peaks, during which reproduction was observably, affected (Fig. 10). Such missed opportunities or mistakes have more to do with the past actions of moose managers or with the absence of predation and conse- quent delayed regulatory mechanisms in moose populations (Saether 1997) than with “socio-political changes [or] political events” (Ferguson and Messier 1996: 156). Careful interpretation is required to un- derstand population changes from indices of moose abundance, because these indices invariably represent a wide variety of populations with different dynamics in dif- ferent habitats. In our most precise esti- mate of K, from combined data for all of ALCES VOL. 38, 2002 MERCER AND McLAREN – CARRYING CAPACITY IN NEWFOUNDLAND 139 insular Newfoundland, we average the ob- servations of hunters for moose in different habitats and in different stages of popula- tion increase. Managers should note that our estimate of K is a range, and K varies naturally both in time and space (Crête 1989). Moreover, our estimate, particularly for forested regions, is affected by expan- sion of forestry operations and thereby con- tinuous supply of new habitat areas during the period of data collection; the same rea- son Pimlott (1953) accounted for the in- crease in moose during the early part of the 20th century. Also, extrapolation by linear regression does not explicitly take into ac- count density-dependent effects in moose that may become apparent only at higher densities than those observed over our man- agement period (we report a decline in density dependence at a threshold of 20 young seen / 100 adult females). Further, we have no indication from hunter returns whether both winter and summer habitat affect K, but we assume both are important. The structure, succession and composi- tion of natural forest communities have continued to be altered in inaccessible areas like area 26, so that their ability to produce moose has been marginalized (Fig. 6). Assuming as we have that area 26 repre- sents a population at K throughout our study period, we show how weather, as well as other random effects, can create variation in moose seen, young seen, and hunter success between years (Figs. 4 and 6). Our hunting indices reflect only areas that are hunted, mostly space that is < 2 km from a road. Even in accessible areas, there is considerable moose range that is > 2 km from road, which forms refugia, in which habitat quality is reduced when density is > K from lack of hunting. Throughout such areas, regenerating fir (Abies balsamea), birch (Betula papyrifera, B. cordifolia, a n d B . a l l e g h a n i e n s i s ) , a n d o t h e r hardwoods (e.g., Cornus, Prunus, Sorbus) have been eliminated except for some heav- ily browsed, sparsely distributed saplings. On the south coast, moose depend on atypi- cal foods, such as branches of blown down trees, lichens, and low shrub and herb com- munities (Albright and Keith 1987). Even- tually, population condition is affected un- der these circumstances (Ferguson et al. 1989), as for Gros Morne National Park (McLaren et al. 2000). During periods of increasing population size observed in the 1980s, yearling harvest as a proportion of the moose hunt increases, and following these increase phases, jaw size declines (Mercer 1995). An example of a population introduction and subsequent crash follow- ing poor condition occurred when moose were experimentally transported to Bru- nette Island, between the Burin Peninsula and the south coast of Newfoundland (Mer- cer 1995). In contrast, as we considered for central Newfoundland, accessible areas of moose in Newfoundland support a popula- tion fluctuating around a “long-term equilib- rium” density, unlike their erupting phase from 1904–58 (Fig. 10). We speculate, consistent with predictions by Saether (1997), that for moose in Newfoundland in the absence of hunting, a stable equilibrium between the population and food resources is not possible. ACKNOWLEDGEMENTS We are grateful to M. Strapp, former statistician for the Newfoundland Wildlife Division, for his careful attention to the hunter statistics over many years. Produc- tion of this paper was supported by the Department of Forest Resources and Agrifoods, Government of Newfoundland and Labrador. REFERENCES ALBRIGHT, C. A., and L. B. KEITH. 1987. Population dynamics of moose, Alces alces, on the south-coast barrens of CARRYING CAPACITY IN NEWFOUNDLAND – MERCER AND McLAREN ALCES VOL. 38, 2002 140 Newfoundland. Canadian Field-Natu- ralist 10:373–387. BERGERUD, A. T., and F. MANUEL. 1969. Aerial census of moose in central New- foundland. Journal of Wildlife Manage- ment 33:910–916. , , and H. WHALEN. 1968. The harvest reduction of a moose popu- lation in Newfoundland. Journal of Wildlife Management 32:722–728. BRODERS, H. G., S. P. MAHONEY, W. A. MONTEVECCHI, and W. S. DAVIDSON. 1999. Population genetic structure and the effect of founder events on the genetic variability of moose, Alces alces, in Canada. Molecular Ecology 8:1309–1315. CAUGHLEY, G. 1976. Wildlife management a n d t h e d y n a m i c s o f u n g u l a t e populations. Pages 183–246 in T. H. Coaker, editor. Applied biology. Vol- ume 1. Academic Press, London, UK. COURTOIS, R., and M. CRÊTE. 1993. Pre- dicting moose population parameters from hunting statistics. Alces 29:75– 90. CRÊTE, M. 1989. Approximation of K carrying capacity for moose in eastern Quebec. Canadian Journal of Zoology 67:373–380. , and R. COURTOIS. 1997. Limiting factors might obscure population regu- lation of moose (Cervidae: Alces alces) in unproductive boreal forests. Journal of Zoology, London 242:765–781. , and C. DAIGLE. 1999. Manage- ment of indigenous North American deer at the end of the XXth Century in relation to large predators and primary p r o d u c t i v i t y . A c t a V e t e r i n a r i a Hungarica 47:1–16. , and M. MANSEAU. 1996. Natural regulation of cervidae along a 1000 km latitudinal gradient: change in trophic dominance. Evolution and Ecology 10:51–62. FERGUSON, S. H. 1993. Use of cohort analysis to estimate abundance, recruit- ment and survivorship for Newfound- land moose. Alces 29:99–114. , W . E . M E R C E R , a n d S . M . OOSENBRUG. 1989. The relationship between hunter accessibility and moose condition in Newfoundland. Alces 25:36–47. , and F. MESSIER. 1996. Can human predation of moose cause population cycles? Alces 32:149–161. FRYXELL, J. M., W. E. MERCER, and R. B. GELLATELY. 1988. Population dynamics of Newfoundland moose using cohort analysis. Journal of Wildlife Manage- ment 52:14–21. GASAWAY, W. C., R. D. BOERTJE, D. V. GRANDGARD, K. G. KELLYHOUSE, R. O. STEPHENSON, and D. G. LARSEN. 1992. The role of predation in limiting moose at low densities in Alaska and Yukon and implications for conservation. Wild- life Monographs 120. GOSSE, J., B. MCLAREN, and E. EBERHARDT. 2002. Comparison of fixed-wing and helicopter searches for moose in a mid- winter habitat-based survey. Alces 38: 47-53. HATTER, I. W. 2001. An assessment of catch per unit effort to estimate rate of change in deer and moose populations. Alces 37:71–77. KEITH, L. B. 1983. Population dynamics of wolves. Pages 66–77 in L. N. Carbyn, editor. Wolves in Canada and Alaska: their status, biology and management. Report Series 45, Canadian Wildlife Service, Ottawa, Ontario, Canada. MCCULLOUGH, D. R. 1979. The George Reserve deer herd: population selec- tion of a K-selected species. University of Michigan Press, Ann Arbor, Michi- gan, USA. . 1984. Lessons from the George Reserve, Michigan. Pages 211–242 in ALCES VOL. 38, 2002 MERCER AND McLAREN – CARRYING CAPACITY IN NEWFOUNDLAND 141 L. K. Hall, editor. White-tailed deer: ecology and management. Stackpole Books, Harrisburg, Pennsylvania, USA. . 1997. Irruptive behaviour in ungu- lates. Pages 69–98 in W. J. McShea, H. B. Underwood, and J. H. Rappole, editors. The science of overabundance: deer ecology and population manage- ment. Smithsonian Institution Press, Washington, D.C., USA. . 1999. Life-history strategies of ungulates. Journal of Mammalogy 79:1130–1146. MCLAREN, B. E, C. MCCARTHY, and S. P. MAHONEY. 2000. Extreme moose demographics in Gros Morne National Park, Newfoundland. Alces 36:217– 232. , and R. O. PETERSON. 1994. Wolves, moose, and tree rings on Isle Royale. Science 256:1555–1558. MERCER, W. E. 1974. Recruitment rates of Newfoundland moose. Proceedings of the North American Moose Confer- ence and Workshop 10:37–45. . 1995. Moose management plan for Newfoundland. Report on file with Wildlife Division, Newfoundland and Labrador, St. John’s, Newfoundland, Canada. , and F. MANUEL. 1974. Some aspects of moose management in New- foundland. Naturaliste Canadien 101:657–671. ODUM, E. P. 1953. Fundamentals of Ecol- ogy. Saunders, Philadelphia, Pennsyl- vania, USA. OOSENBRUG, S. M., and S. H. FERGUSON. 1992. Moose mark-recapture survey in Newfoundland. Alces 28:21–29. PERSON, D. K., R. T. BOWYER, and V. VAN BALLENBERGHE. 2001. Density depend- ence of ungulates and functional re- sponses of wolves: effects on predator- prey ratios. Alces 37:253–274. PIMLOTT, D. H. 1953. Newfoundland moose. Transactions of the North American Wildlife Conference 18:563–581. . 1959a. Reproduction and produc- tivity of Newfoundland moose. Journal of Wildlife Management 23:381–401. . 1959b. Moose harvests in New- foundland and Fennoscandian countries. Transactions of the North American Wildlife Conference 24:424–448. SAETHER, B.-E. 1997. Environmental stochasticity and population dynamics of large herbivores: a search for mecha- nisms. Trends in Ecology and Evolution 12:143–149. , R. ANDERSEN, O. HJELJORD, and M. HEIM. 1996. Ecological correlates of regional variation in life history of the moose, Alces alces. Ecology 77:1493– 1500. SOKAL, R. R., and F. J. ROHLF. 1995. Biometry: the principles and practice of statistics in biological research. Third edition. W.H. Freeman and Co., San Francisco, California, USA. TIMMERMANN, H. R. 1993. Use of aerial surveys for estimating and monitoring moose populations – a review. Alces 29:35–46. UNDERWOOD, H. B., and W. F. PORTER. 1997. Reconsidering paradigms of over- population in ungulates: white-tailed deer at Saratoga National Historic Park. Pages 185–200 in W. J. McShea, H. B. Underwood, and J. H. Rappole, editors. The science of overabundance: deer ecology and population management. Smithsonian Institution Press, Wash- ington, D.C., USA. VAN BALLENBERGHE, V., and W. B. BALLARD. 1998. Population dynamics. Pages 247– 274 in A. W. Franzmann and C. C. Schwartz, editors. Ecology and man- agement of the North American moose. Smithsonian Institution Press, Wash- ington, D.C., USA.