Effects of Capture on the Reproductive Performance of Female Moose

Understanding ecological and anthropogenic factors that drive wildlife population dynamics is essential to developing and maintaining population sizes consistent with management goals. Reproduction and recruitment of individuals influence population growth, and therefore understanding factors that affect these population characteristics is critical to elucidate population changes. In particular, it is important to understand whether study methods affect research subjects. If individuals are adversely affected by research methods, it may produce erroneous results leading to poor management decisions. For example, one study found using flipper bands to mark king penguins Aptenodytes patagonicus to track survival and reproduction of individuals significantly reduced survival rates and breeding success (Wilson 2011). Another example of reduced survival of marked individuals was reported by Swenson et al. (1999) where moose Alces alces calves marked with ear transmitters experienced a significantly higher mortality rate than control calves or those marked only with ear tags.

Immobilized animals may experience direct and indirect adverse effects as a result of capture and immobilization procedures and vary depending on the species, age of the animal, drug and dosages used, capture procedures utilized, and timing of capture during the pregnancy cycle for females (Ballard and Tobey 1981; Larsen and Gauthier 1989; and Roffe et al. 2001). One potential negative effect includes aspiration pneumonia if the head does not remain elevated during handling procedures (Roffe et al. 2001). Immobilized animals also may encounter renarcotization postcapture, wherein they may become more vulnerable to predation or injury because of terrain or habitat conditions in the vicinity of the capture site. In addition, pregnant females may experience negative impacts to the fetus, including increased abortion rates, lower birth weights, offspring abandonment, or decreased survival and recruitment of neonates (Ballard and Tobey 1981; Côté et al. 1998; Larsen and Gauthier 1989). Effects from immobilization may be long term or short term. For example, one commonly used immobilization drug, xylazine HCl, may cause short-term physiological effects such as hypertension, hyperglycemia, or anorexia (Côté et al. 1998; Larsen and Gauthier 1989). Another commonly used drug for immobilizing large ungulates, carfentanil citrate, may cause increased body temperature and decreased heart and respiratory rates (Larsen and Gauthier 1989).

Moose distribution, population size, and female reproductive and survival rates were studied on Togiak National Wildlife Refuge in southwestern Alaska since 1998 (Aderman 2008; Reference 1, Supplemental Material). These data were obtained by following female moose that were captured and fitted with Very High Frequency radiocollars and were tracked during monthly aerial surveys. A previous study conducted in the Southwest Yukon, Canada used similar capture and immobilization methods for female moose and detected a significant reduction in calf recruitment during the years females were captured and fitted with radiocollars (Larsen and Gauthier 1989). Although several studies have evaluated the effects of capture on female production and postnatal survival of young in several species (king penguins: Wilson 2011; mountain goats Oreamnos americanus: Côté et al. 1998; moose: Ballard and Tobey 1981; Larsen and Gauthier 1989), empirical evidence obtained from long-term studies following the same marked individuals is limited.

We examined the reproductive performance of female moose captured as part of a population demographics and distribution study, and we sought to determine whether production and recruitment of calves differed during capture years compared with noncapture years for radiocollared females. We predicted female moose would produce and recruit fewer calves during a year when they were captured because of the potentially negative effects associated with capture and immobilization of animals.

The study area comprised approximately 21,000 km² and included Togiak National Wildlife Refuge (including private inholdings) and nonrefuge lands in the vicinity of Goodnews Bay (Figure 1). The Wood River and Ahklun Mountains begin at the southern boundary (coastline) and rise to >1,500 m in the northern portion of the study area. Numerous rivers and creeks, bordered by willows Salix spp. and cottonwood Populus balsamifera, begin in alpine tundra and alder-dominated Alnus spp. slopes and drain through wet and dry tundra uplands. Caribou Rangifer tarandus from the Mulchatna and Nushagak Peninsula herds frequented the area. Large carnivores included brown bears Ursus arctos, black bears U. americanus, wolves Canis lupus, coyotes C. latrans, Canada lynx Lynx canadensis, and wolverines Gulo gulo. Almost one-half of the study area is federally designated wilderness. Petersen et al. (1991) and U.S. Department of Interior, Fish and Wildlife Service (2009; Reference 2, Supplemental Material) provides further detail of the study area. Legal harvest of female moose was not permitted within the study area.

We captured and fitted radiocollars on female moose in early April over several years (1998, 2000, 2002, 2003, 2004, 2006, 2007, and 2008). We captured female moose as young adults or at 10 mo of age (short-yearlings) in late winter prior to the calving period, which typically occurs from the second week in May through the second week in June (Aderman 2008). Capture procedures utilized an observer to locate moose from a fixed-wing aircraft. In late winter, moose in this study area often congregate in large groups in the major river drainages at lower elevations forming spatially separated groups. We selected individual moose for capture opportunistically based on age (short-yearling females were preferred in most capture years) and river drainage. Moose captures were stratified among major drainages containing moose to obtain samples across the entire study area.

Once located, we pursued moose from a helicopter and darted them using a powder-fired Cap-Chur rifle (Palmer Cap-Chur Inc., Powder Springs, GA) or a CO₂ powered Cap-Chur pistol. We immobilized adults with 3.0–4.5 mg (1.0–1.5 cc) carfentanil citrate (Wildlife Pharmaceuticals, Fort Collins, CO) and 150–200 mg (1.5–2.0 cc) xylazine HCl (Cervizine®; Wildlife Pharmaceuticals) via a 3-cc dart with a 3.8-cm barbed needle. Short-yearlings were immobilized with a mixture of 1.2 mg (0.4 cc) carfentanil citrate and 60 mg (0.6 cc) xylazine HCl via a 1-cc dart with a 1.9-cm barbed needle. We did not consistently record the amount of time a moose was pursued (beginning from when the moose first reacted to the helicopter to when it was darted); however, most chase times were <1 min and none >4 min. Handling times (from time moose were immobile to when they were standing) averaged 30.1 min for adult females (n  =  27) captured in 1998. Handling times averaged 37.5 (range  =  27–64) min for moose (n  =  15) that we performed ultrasound on (to determine rump fat thickness) as opposed to an average handling time of 20.8 (range  =  5–29) min for moose (n  =  12) on which we did not perform ultrasound.

We fitted adult moose with a Very High Frequency radiocollar (Telonics Inc., Mesa, AZ), while we deployed radios with expandable collars on short-yearling moose. Immobilized adults were reversed with 300–450 mg (6.0–9.0 cc) naltrexone HCl (Trexonil®; Wildlife Pharmaceuticals) given two-thirds intramuscularly and one-third subcutaneously and 400 mg (4.0 cc) tolazoline HCl (Tolazine®; Lloyd Laboratories, Shenandoah, IA) intravenously. Short-yearlings were reversed with 125 mg (2.5 cc) naltrexone HCl (two-thirds intramuscularly and one-third subcutaneously) and 300–400 mg (3.0–4.0 cc) tolazoline HCl (either intravenously or intramuscularly). We recaptured several moose (n  =  19) and fitted them with new radiocollars approximately 4–6 y after initial capture as radio batteries neared their life span and to replace expandable collars deployed on short-yearlings (Aderman 2008). We determined pregnancy status for moose captured in 1998 and 2006 (n  =  38) from blood assays using the pregnancy-specific protein B (PSP-B) test (Sasser et al. 1986; Stephenson et al. 1995). We considered multiple capture options and selected this method and combination of immobilization drugs in our study because of 1) accessibility on the refuge, 2) cost efficiency, and 3) prior success by others using these methods (Keech et al. 2000 and Boertje et al. 2007). Togiak National Wildlife Refuge extends over >18,000 km² of tundra and mountainous regions, which makes ground-based capture methods ineffective and costly. Research methods were approved and conducted in accordance with International Animal Care and Use Committee protocol (Alaska Department of Fish and Game protocol no. 2012-022).

We monitored female moose (radiotracked and visually located) monthly via fixed-wing aircraft. We monitored radiocollared females ≥2 y old weekly from mid-May to mid-June to determine annual calf production. Females that were not observed with a calf by mid-June were classified as barren for that year of production. We determined annual autumn recruitment of calves in November, and based it on whether or not the radiocollared females were still accompanied by their calves (Aderman 2008). We assumed that detection rates for calf production and recruitment were consistent among individuals and years. We acknowledge that detection rates for calf production may have been <100% because of lack of visibility in some areas of dense vegetation, and we conducted only 1 flight/wk during the highest predation period on newborn calves. However, we made every effort possible to determine calf status for captured and noncaptured moose when radiocollared females were located and search effort remained consistent every year. Female moose captured as short-yearlings (n  =  32) were censored from the calf production and recruitment analysis for the initial year of capture because they were not of reproductive age in that year. If females died or went missing after producing a calf, we included them in the production analysis but censored them from the recruitment analysis because we were unable to track the status of the calf in the November aerial survey without the presence of the radiocollared female (n  =  20). In addition, for years when female moose were barren (n  =  94), we censored them from the recruitment analysis. Barren moose were still included in the production analysis; however, retaining them in the recruitment analysis would bias the results because no calf was produced. The goal of the recruitment analysis was to assess success of recruitment for those females that produced a calf(s) during that year. We classified as capture-related mortalities any moose that died ≤30 d after capture (Beringer et al. 1996).

We measured moose population growth using minimum counts conducted on 75% of the study area and calculated the exponential rate of increase (Caughley 1977). This excluded moose located in the southeastern portion of the study area where data were unavailable.

The equation for the exponential rate of increase is N_t  =  N₀e^rt , where N₀ is the population size at some initial year; N_t is the population size at year t; e is the base of natural logarithms; and r is termed the instantaneous rate of increase, which is converted to

We analyzed the reproductive performance data with a generalized linear mixed model fit by the Laplace approximation method (Bolker et al. 2009) to assess the effects of capture on calf production and recruitment by radio-collared female moose (lme4 package; Program R; v. 2.15.2; R Development Core Team 2012). Generalized linear mixed models are a powerful tool used to analyze nonnormal data with repeated measures on individuals. Generalized linear mixed models combine properties of linear mixed models to incorporate random effects and generalized linear models to allow for response variables from different distributions (i.e., binary, Poisson, negative binomial; Bolker et al. 2009 and Pan and Lin 2005). We fitted two types of models (logistic and Poisson) for both calf production and recruitment, treating each as a binomial and count-type response. We used the logistic model to analyze the odds of presence or absence of a neonate(s) produced and for the odds of calves successfully recruited based on whether or not the female was captured and fitted with a radiocollar in that year (Table S1 and S2, Supplemental Material). We used the Poisson model to analyze the likelihood for the number of neonates produced depending on capture status, and to analyze the likelihood for the number of calves recruited (limited from one to three calves) by each radiocollared female to vary based on capture status of the female in that year.

The binary response variable for calf production was yes (calve(s) was produced) or no (no calve(s) was produced), and the Poisson response variable was the number of calves produced (one: singleton, two: twins, or three: triplets) by the female in a year. The binary response variable for calf recruitment was yes (calve(s) was successfully recruited) or no (no calves were recruited). The Poisson response variable for calf recruitment was the number of calve(s) successfully recruited by the female (one: singleton or two: twins).

We included the capture status for each year as a binary (yes or no) fixed effect. Because of the potential for cumulative capture effects that may increase with the age of the animal (Cattet et al. 2008), we also included the number of times the moose was captured during our study. Although our primary interest was to evaluate whether or not capture status had an effect on calf production and recruitment, we considered two additional covariates for which we had information available to improve the model: age of the moose and year of the study. Other studies have reported decreased calf production with increased age for female moose (Boertje et al. 2009 and Ericsson et al. 2001). We calculated 95% binomial confidence intervals for age-specific calf production and recruitment of radiocollared females (Boertje et al. 2009) to examine the relationship in our data, and then included age (adjusted each subsequent year after initial capture) as a fixed effect in our model. Year of the study was included as a random effect to explain variation that may have occurred as a result of variable environmental conditions or fluxes in predator abundance not directly measured in our study. We included year as a random effect rather than a fixed effect because we sought to account for differences between years; however, we were not directly interested in analyzing the differences between years of the study. Individual moose were modeled as random effects to account for multiple observations on the same individuals. When repeated measures are collected on individuals and the main interest is the variation that exists among the units, it is appropriate to model the individual as a random effect (Bolker et al. 2009). We used the Wald z-statistic (P-values) to evaluate the fixed effects in each model to assess whether capture status, age, or number of captures had a significant influence for each fitness parameter of radiocollared female moose (calf production and recruitment) and response type (binomial and count; Bolker et al. 2009).

We captured and radiocollared 83 female moose, of which 78 were included in the analysis of reproductive performance from 1998 to 2011 (Table 1). We censored moose that died (n  =  4) or went missing (n  =  1) prior to their first calving season. Two of the mortalities were considered capture-related because they occurred ≤30 d postcapture. The other two censored moose were captured as short-yearlings and died 325 and 428 d postcapture, both from unknown causes. Individual females were radiotracked from 1 to 14 consecutive years and averaged 5.3 y (±3.4 SD) radiotracked. Of the moose tested for pregnancy status in 1998 and 2006 (n  =  38), we found a combined 95% pregnancy rate. We observed radiocollared female moose with 128 singletons, 191 sets of twin calves, and 3 sets of triplets (519 total calves) during this study. The total number of recruited calves was 245 during this study and included 54 sets of twins. We did not observe more than two calves successfully recruited from any of the three sets of triplets. In 1992, 6 (N₀) moose were counted within the surveyed portion of the study area. By 2011, this population had grown to 1,367 (N₁₉) moose. Thus, the exponential rate of increase was 0.286.

The results of the four generalized linear mixed models we tested indicated that capture status did not significantly influence reproductive performance of female moose during years when they were captured throughout the study (Table 2). Female moose had very similar odds of producing calves during years when they were captured compared with noncapture years (−0.33, P > 0.290; Table 2). In total, 76% of radiocollared females (n  =  63 moose-years) produced calves in the same year as being captured, whereas 78% of radiocollared females (n  =  235 moose-years) produced calves during noncapture years (Figure 2). Similarly, the number of calves (from one to three calves) produced by radiocollared females did not vary substantially between capture and noncapture years in the study (−0.176, P > 0.134; Table 2). Likewise, we did not detect a substantial influence of female capture status on calf recruitment rates (−0.072, P > 0.789; Table 2). Radiocollared females that were captured had the same likelihood of recruitment for neonate(s) as radiocollared females that were not captured (Figure 3). Additionally, the number of calves recruited in a year by a radiocollared female moose did not vary considerably according to capture status during the year (−0.051, P > 0.755; Table 2).

Although our results did not reveal a strong influence of capture status or number of captures throughout the study on reproductive performance of radiocollared moose, it is apparent that age of the female was one important variable to consider when evaluating reproductive success of moose (Table 2). The proportion of radiocollared females producing and recruiting calves declined as age increased, although this trend was not significant because of small sample sizes of older females (Figures 4 and 5). We also examined differences in the proportion of radiocollared female moose producing and recruiting calves from 1998 to 2011. Calf production and recruitment varied throughout the study (Figures 6 and 7); therefore, including year of the study in the models was warranted.

Our results indicate that capturing female moose utilizing aerial darting techniques and immobilization drugs to affix radiocollars on the animals did not result in lowered reproductive performance. The number of radiocollared females producing calves (and the number of calves produced by each female) was similar during capture and noncapture years. Similarly, autumn recruitment of calves was not significantly lowered during capture years compared with noncapture years of radiocollared females. Our results indicate the variation we observed in calf production (Figure 6) and recruitment (Figure 7) rates during this study are more likely attributed to other factors known to influence reproductive performance than the capture procedures employed during this study (i.e., age of female, predation on calves, etc.). We cautiously interpreted the results of our analysis because the appropriate interpretation of results from generalized linear mixed model remains controversial (Bolker et al. 2009). In conjunction with the results of the models, we evaluated the overlap of binomial confidence intervals around the mean proportions for calf production and recruitment (Figures 2 and 3) for captured and noncaptured females. Based on consistent outcomes, we are confident our conclusions are not misinterpreting the results of this study.

The lack of a significant effect of capture methods on female moose reproductive performance in our study is partially consistent with a similar study that analyzed capture effects on fitness of female moose in the Yukon in early 1980s (Larsen and Gauthier 1989). Similar to our study, they did not detect a difference in calf production by females that were captured and fitted with radiocollars. However, they detected lower calf survival for females that were captured and marked with radiocollars. We measured calf survival as “calf recruitment” in our study, but the measures are comparable between the two studies (calf survival until November in the year the calf is born). We did not detect an influence of capture on calf recruitment, whereas the study in the Yukon did find a significant negative influence on calf survival for captured females. We speculate the difference in results is an artifact of following multiple females across consecutive years in our study versus the other study where all moose were followed within 2 y. By following the same individuals over multiple years and obtaining data from females that were captured multiple times, our study offers an improved design to assess whether the effect is directly related to capture or whether the population coincidently experienced a low recruitment year because of other factors. We examined variability of recruitment from year to year in our study (Figure 7), and we would not have observed a representative range of values if our study was only conducted over a 2-y period. The study conducted in the Yukon may have occurred during a period of low recruitment and may have been unrelated to capture events.

There was a clear trend in diminishing proportion of females producing calves with increased age of the female moose (Figure 4), and this should be considered when designing future studies assessing factors influencing female reproductive performance. However, further research is needed to determine more conclusively that these capture methods do not have other negative impacts that were not measured in this study. For example, there may be more subtle effects that we did not measure that could indicate negative impacts of capture such as weight loss of the female after capture procedures, lowered calf birth weight, and increased abortion rates. These negative aspects of capture were not specifically addressed by covariates in our analysis and may improve explanation of variability we observed in these reproductive performance measures. The applicability of our results is limited to moose populations that are experiencing similar population trends (i.e., population growth, geographic expansion, high rates of individual survival, and high twinning rates). Different methods of capture and immobilization of female moose may be more suitable in other areas. In addition, future studies may attempt a more rigorous monitoring schedule where female moose are monitored more frequently during the calving period. We acknowledge the 7-d monitoring schedule may not have been sufficient to detect 100% of the calves produced given the increased vulnerability to predation newborn calves experience. This was the most frequent monitoring schedule feasible given constraints associated with accessing the study area, such as weather and the expense associated with conducting multiple aerial surveys over a large geographic area. Our monitoring schedule was consistent across study years and among captured and noncaptured radiocollared females; therefore, we believe any bias due to nondetection would not affect the outcome of our results in this analysis.

A potential explanation for the lack of effects observed from capture activities in this study is the high level of nutrition this moose population has access to. Preliminary analyses of major browse species in our study area in 2007 suggests they provide more digestible protein and are available for a longer period than in the same species measured by McArt et al. (2009) in Denali National Park and the Nelchina Basin, Alaska (B. Collins, Alaska Department of Fish and Game, personal communication). This high plane of nutrition manifests itself in the rapid growth and sexual development of moose within the study area. Compared with five other populations in Alaska, Boertje et al. (2007) found moose in our study area (GMU 17A) ranked highest in female short-yearling average mass (213 kg), age first parturient (24 mo; 74% of 19), age of first twins (24 mo; 29% of 14), and average multiyear twinning rates of parturient females ≥36 mo of age (0.67).

We recommend the continued use of current capture and immobilization methods utilized for moose in this region of southwestern Alaska. Production and recruitment of calves by radiocollared females should continue to be monitored for the duration of the population dynamics and distribution study to determine whether effects of capture methods result in altered reproductive performance once the habitat quality or moose population numbers stabilize or begin to decline. Further, we suggest conducting a comprehensive analysis to determine natural factors contributing to moose calf production and recruitment in this population to better elucidate population changes in the future.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Table S1. Moose Alces alces calf production histories and covariates for female radiocollared moose on Togiak National Wildlife Refuge, southwestern Alaska, from 1998 through 2011. Includes information on moose identification number, year of study, age of female moose, capture status, calf production status, and number of calves produced.

Found at DOI: 10.3996/032013-JFWM-028.S1 (63 KB XLS).

Table S2. Moose Alces alces calf recruitment histories and covariates for female radiocollared moose on Togiak National Wildlife Refuge, southwestern Alaska, from 1998 through 2011. Includes information on moose identification number, year of study, age of female moose, capture status, calf recruitment status, and number of calves recruited.

Found at DOI: 10.3996/032013-JFWM-028.S2 (52 KB XLS).

Reference S1. Aderman AR. 2008. Demographics and home ranges of moose at Togiak National Wildlife Refuge, Southwest Alaska, 1998–2007. Progress report. Dillingham, Alaska: Togiak National Wildlife Refuge.

Found at DOI: 10.3996/032013-JFWM-028.S3 (1006 KB PDF).

Reference S2. U.S. Department of Interior, Fish and Wildlife Service. 2009. Togiak National Wildlife Refuge comprehensive conservation plan. Anchorage, Alaska: U.S. Fish and Wildlife Service.

Found at DOI: 10.3996/032013-JFWM-028.S4; also available at http://digitalmedia.fws.gov/cdm/ref/collection/document/id/564 (52 KB XLS).

The success of this project is due to the cooperation of many people. We thank fixed-winged pilots M. Hinkes, M. Hink, P. Liedberg, R. MacDonald, G. Howell, T. Tucker, J. Wittkop, D. Cox, R. Grant, T. Schlagel, and G. Dobson for providing safe flying during capture operations and radiotracking. R. Swisher, B. Merkley, J. Woolington, L. Van Daele, P. Valkenburg, M. Keech, B. Dale, L. Butler, and J. Crouse were instrumental during capture operations. P. Walsh, M. Lisac, G. Collins, and P. Abraham assisted in various aspects of this project. We also thank N. Roberts and A. Benson for helpful comments and suggestions on the analysis for this study. We thank several anonymous reviewers and the Subject Editor whose comments, criticisms, and suggestions greatly improved this manuscript. Financial support was provided by Togiak National Wildlife Refuge, Alaska Department of Fish and Game, and the U.S. Fish and Wildlife Service Office of Subsistence Management.

Any use of trade, product or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.