R.G. White and S. Murphy
Institute of Arctic Biology and Alaska Biological Research

A.B. Kugler
BBN Systems and Technologies

Supporting Agencies: USAF NSBIT Program and U.S. Fish and Wildlife Service (USFWS Award 14-16-0009; AKCFWRU RWO#36)

Use of low-flying jet aircraft for military training has caused concern for the effects of these overflights on wildlife. Northern residents and resource-management agencies have concerns for the effects these training exercises may have on productivity of caribou (Rangifer tarandus granti). Response to these concerns, coupled with National Environmental Policy Act (NEPA) requirements for military exercise, prompted the U.S. Air Force (USAF) to convene a "Research Needs Workshop" in April 1988 on the "Effects of Aircraft Noise and Sonic Booms on Fish and Wildlife." This research program was designed in response to a workshop task and was subsequently selected for funding by the USAF Noise and Sonic Boom Impact Technology (NSBIT) program. Caribou (R. t. granti) from the Delta Herd in interior Alaska were selected for study primarily because detailed energetics models were available for this species and because this herd occurs near Eielson Air Force Base in Interior Alaska.

The effects of disturbance by military aircraft on the behavior of caribou are reviewed in a companion report ("Behavioral Responses of Caribou to Low-Altitude Jet Aircraft;" Murphy et al. 1993, Technical Report #20). Results addressed field objectives of this study:

1) to measure the noise exposure experienced by caribou overflownby jet aircraft;

2) to record behavioral reactions of caribou to overflights by low-altitudejet aircraft by direct observation;

3) to record activity cycles and movements of caribou exposed tooverflights by low-altitude jet aircraft using telemetry; and

4) to evaluate the behavioral responses of caribou to overflights as afunction of noise exposure.

Data derived from activity monitoring radiocollars (Wildlink,© Inc.) (objective 3) then were evaluated for their energetic and population productivity implications using computer simulations and prediction equations based on tame caribou and reindeer (R. t. tarandus) at LARS. This report (White et al. 1993, Technical Report #21) details the rationale, objectives, results and implications of this analysis.

The rationale for linking empirical behavioral and energetic responses of caribou from field and laboratory studies to computer simulation of population responses, is based on the assumption that the effects of a stressor, in this case a dose of sound from a low-altitude jet aircraft, on a given wildlife population could be expressed at the population level if the stressor is significantly disruptive to the time/energy budget of animals. However, such responses can be empirically measured only in long-term studies, often placing a costly chronic impact on the exposed (treatment) population. A predicted measure of population response can be derived through modeling. Such models are driven by factors that affect energy intake and energy expenditure of individuals, while energy flow through populations, and the effect of energy limitation on population productivity, are intermediate variables and outputs of the model.

Energy expenditure and activity energy costs are well documented for caribou and the effect of energy and nutrition on fecundity is known quantitatively (see K. Gerhart this report). In addition, seasonal energy budgets of caribou have been assessed from an evolutionary and ecological perspective and this knowledge provides a strong framework with which to assess disturbance impacts. The effects of disturbance on caribou varies depending on type of disturbance, time of year and animal group (nursery bands, bull groups, mixed aggregations). This information has been used to develop a caribou energetics model (CARIBOU) that links animal activity and habitat-forage data to the annual body condition cycle, the milk production and calf growth rate for caribou. In the present study, annual energy budgets of female caribou were simulated and the effects of jet overflights were assessed in terms of reproductive output (Luick and White 1994, Technical Report #22). Computer simulations also incorporated a wide array of influencing variables, such as weather and parasitic insects.

The first step in the modeling process was to develop an independent measure of energy expenditure for use in the field. We evaluated, and eventually rejected the use of heart-rate monitoring to determine average daily metabolic rate (ADMR) and the incremental energy cost (IEC) of activity associated with field existence. IEC is a measure of the energy expenditure of an activity (e.g., walking) relative to a minimal standard activity (lying IEC = 1.00). IEC of walking for caribou is 1.81. We also rejected the use of visual observations to determine daily activity as under natural conditions, animals are frequently out of sight, and it is impossible, therefore, to determine the daily activity budgets of free-ranging caribou by direct observation. We used remote sensing of activity by a mercury tilt-switch, fitted to a data acquisition and transmission system, built into a caribou neck-collar (Wildlink activity system) to determine the daily activity budget (see J. Kitchens, this report). We converted Wildlink activity counts to the activity budget IECs and ADMR. Conversion, or calibration, equations were constructed using tame animals at LARS (White et al. 1993, Technical Report #21).

The following hypothesis was evaluated to test whether caribou altered their energy budgets in response to jet aircraft overflights.

Ho: Daily IEC and ADMR did not differ between caribou that had been exposed recently to jet aircraft overflights and animals that had not been overflown.

IEC24 for animals in the control groups ranged from 1.138 to 1.286 during the three sampling periods (Table 1). Although IEC24 was higher in 4 of 6 cases for the treatment animals by about 1-3%, only the post-calving increase was significant (P=0.039). In control animals IEC24 was similar for each season with highest value for the insect season.

Estimates of ADMR for control and treatment days were not significantly different in late-winter and insect season, but ADMR on treatment days exceeded controls (P=0.040, Table 1) post-calving. In late winter, ADMR of control animals was lower than the following seasons (P = 0.0001) while estimates for post-calving and the insect season were almost identical (P=0.6465).

Influence of the "dose" of sound, as given by the variables, number of overflights >85 dBA (NF), loudest overflight each day (LOUDSEL) and average sound exposure for the treatment day was minimal for late-winter and the insect season. LOUDSEL was consistently the sound variable with highest predictive power and the full data set for this variable is shown in Table 2. For post-calving females both IEC24 and ADMR increased significantly (P<0.10) in response to LOUDSEL (Table 2).

Table 1. Daily incremental energy cost (IEC24 + SEM) and average daily metabolic rate (ADMR ± SEM, kJ/kg^0.75) of control and treatment (i.e. exposed to overflights by low-altitude jet aircraft) Delta Herd caribou in Alaska, 1991.

r, correlation coefficient; P, significance of r.

Table 2. Linear regression models evaluating relationships between IEC24 and ADMR (kJ/kg^0.75) of caribou and the loudest sound exposure of the day (loudsel) resulting from overflights by low-altitude jet aircraft in Alaska, 1991.

r, correlation coefficient; P, significance of r.

Even under ideal circumstances, when energy expenditure could be predicted from Wildlink collar output in response to overflights, these results cannot be interpreted directly in relation to measures of animal performance such as conception rate, calf survival and calf growth rate. The approach can, however, be used to assess an immediate energy cost, that may be evaluated in-turn for its effect on these population parameters through computer simulation modeling (White et al. 1993, Technical Report #21; Luick and White 1994, Technical Report #22).

In this modeling exercise, we assumed that the effects of sound exposure could be translated to animal productivity through a decrease in time spent lying and an increase in time spent active. Because we could not allocate time spent in each component of the active period, foraging (eating and searching), walking or running, we made the explicit assumption that disturbance by jet aircraft did not change the time distribution within the active period. Therefore, it is implicit that caribou could, and did, make a compensatory increase in time spent foraging, and feeding, which is a finding for caribou disturbed by transportation corridors. This compensatory feeding could offset energetic costs associated with an increase in activity. We then used the computer simulations to explore the hypothesis that increased time spent active was attributable to walking and that no compensatory feeding occurred, i.e., time was spent "searching" for, rather than "eating," forage.

Figure 2. Representative intermediate variables (y-axis) plotted in real time during a one-year (x-axis) simulation commencing January 1. Dependent variables are plotted with relative scale. All energy variables, including costs, are in kJ × d^-1 , weights in kg and PROP. is proportion of day spent in the main activities of the daily activity budget.

A description of CARIBOU is given in a manual, Computer Simulation Model for Jet Aircraft Disturbance of Free Ranging Caribou (Luick and White 1994, Technical Report #22), which accompanies an executable version of the CARIBOU model that allows simulation of effects of low flying jet aircraft. Over 50 variables relative to energetics, foraging and activity are stored within the model. An example of 16 outputs is shown in Figure 2 and at the completion of each run the model calculates the probability of conception at mid-rut (Oct.15) from the body fat reserves of the cow (see K. Gerhart, this report) (Fig. 3). Also shown in Figure 3 is a plot of the seasonal trends in body weight and body fat reserves.

Computer simulation of the daily responses of caribou to jet overflights allowed us to test the predictability of the model CARIBOU by comparing its projected estimates of IEC24 and ADMR with our empirical estimates. Based on activity counts of undisturbed (control) caribou, it was clear that caribou were less active and had a lower ADMR in late winter than in either post-calving or the insect season (Table 1) CARIBOU predicted seasonal differences in ADMR, but the values for ADMR were higher in late winter than was estimated by our study. For post-calving caribou, and those in the insect

season, ADMR values predicted by the CARIBOU model were slightly lower but essentially agreed with our estimates derived from activity monitoring (Table 1). Seasonal changes in activity patterns and, thus, ADMR result from seasonal differences in environmental conditions (e.g., snow depth, insect activity) and from physiological adaptations in energy metabolism and nutrition (e.g., appetite and heat increment of feeding, lowered resting metabolic rate). The model CARIBOU takes into account these seasonal differences, and differences between our empirical and model derived estimates of ADMR highlighted the effects of a heavy snow year (spring 1991) and the importance of insects. These are potentially sensitive times of the year that should get special consideration for further research.

Figure 3. Probability of pregnancy of the model caribou based on body fat weight on October 15 calculated at the completion of each simulation run. Displayed are seasonal body weight (top curve), fat weight (bottom curve), and predicted pregnancy (top-most line). Body weight and fat weight are scaled along the y₁ axis, which has a range of 0 - 150kg.

In this modeling effort we were working with the hypothesis that cows and calves were not habituated to jet-overflights. The Delta Caribou Herd has been described as "the most highly disturbed herd in Alaska." Low level flight training of USAF A-10, F-15, and F-16 jets of Eielson AFB, Alaska occurs within the range of the herd and the "bombing range" borders on the calving grounds. Although research on Canadian caribou suggests that calf survival is compromised by the frequency of low-altitude jet aircraft overflights during and immediately after calving in barren-ground caribou, Alaskan researchers conclude disturbance of the Delta Caribou Herd by military aircraft and other factors do not adversely affect productivity. Our study showed a significant increase in energy expenditure in response to disturbance by military aircraft in post-calving caribou and supports concerns expressed for Canadian caribou. Whether this disturbance can lead to an overall decrease in herd productivity is not answered by this study. The model CARIBOU shows that the general outcome of a set of overflight scenarios depends on habitat quality, and the possibility of an interaction between habitat quality and disturbance on fecundity should be considered in developing management plans. Since fecundity is lower in "bad" than "good" years according to CARIBOU, our simulations suggest that management of overflight schedules should be reconsidered in years during, and following, heavy snow-fall and in hot-dry summers, i.e., "bad" year habitat characteristics.

Ultimately the value of CARIBOU is to predict the consequences of noise disturbance on fecundity. This process was initiated in the current analysis by modeling a hypothetical scenario of consecutive jet aircraft overflights. Post-calving caribou are emerging from a winter diet, they have limited fat reserves and are undergoing peak lactation. In general, total body weight and fat weight cannot increase, and when this is coupled with mid-summer insect harassment, recovery of body condition may not be initiated before late summer. For these reasons, the effects of disturbance may be biologically significant during the post-calving period.

Jet overflights in late-winter were associated with no change in the IECs of activities and the ADMR. Post-calving caribou exhibited a significant, linear increase in both IEC and ADMR with both maximum daily sound exposure and the mean daily sound exposure. In the insect season, caribou appear to expend no extra energy in response to low-flying jet aircraft. Extrapolation of these energetic responses to likely effects on fecundity using computer simulation showed no significant lowering of the probability of conception. If female caribou of these simulations are representative of all caribou overflown in this study, then we suggest that the energetic responses associated with changes in the daily activity budget, were likely not to affect herd productivity.

Left unstudied are the indirect effects of disturbance by low flying jet aircraft. These effects are due to using different habitats and can be expressed through population regulatory factors operating through nutrition and stochastic effects, as well as differences in predation rate. Indirect effects of overflights can be modeled by CARIBOU, but the full analysis also warrants further field study.

We made no attempt to "gain" with the model to determine when and under what condition caribou would be most vulnerable to overflights. Nor did we attempt to determine seasonal "thresholds" of noise exposure that would result in a significant decline in individual fecundity. These effects warrant analysis and the simulation program produced for this study can be used to make such assessments. The confidence placed on such a gaming exercise can be judged by the simulated output relation of variables compared to those reported for this herd. Our validation of CARIBOU was based on Wildlink activity data sets for non-overflown animals. CARIBOU simulations agreed closely with these outputs. The true value of the model in its present form is to test scenarios, challenge intuition, educate those associated with management and explain likely effects to air crews associated with making jet overflights of caribou.

A recommendation from our work is that early post-calving caribou should not be subjected to repeated low-altitude jet overflights until further research shows the significant behavioral and energetic effects shown in this study do not carry with them a serious indirect effect on fecundity and calf growth and survival.