Non-Invasive Methods Developed for Studies of Captive Wild Birds
J.R. Millam & M.J. Delwiche
This article is reprinted from a past issue of the Exotic Bird Report, the Project's print publication.
Certain key questions in avian biology can only be answered by studying birds in captivity. But keeping stress-sensitive species in captivity may confound experimental treatments, as well as the interpretation of results. Non-invasive methods provide one means of combating such problems by minimizing exposure of captive animals to humans. Automated, non-invasive methods may also improve animal welfare, provide more detailed observations and reduce experimental bias. Recognition of these advantages is illustrated by the common use of non-invasive methods in several research and commercial applications: monitoring body weight of broiler chickens by automatic, remote balances; identifying dairy cows by microchip radio frequency ID ear tags; and in operant psychology (e.g., rats pressing levers for food rewards). Few examples exist, however, in which these techniques have been used in conjunction with one another to study multiple variables simultaneously, automatically and remotely. Clearly, such methods would be advantageous for studying wild birds in captivity.
To this end we tested the use of three non-invasive techniques for characterizing the baseline behavior of domestic pigeons. Pigeons were chosen for developing the techniques in our laboratory because of the commercial availability of automated key-pecking and feeding devices and because it made sense to test the equipment on a non-wild species first.
A pair of King pigeons were housed in a 3 ft x 3 ft x 6 ft welded wire cage. A nest box was attached to one end of the cage. A solenoid-actuated pigeon feeder and a key-pecking switch (i.e., "operant response key," about the size of a US 25c coin) were mounted on the end opposite the nest box. A pigeon's peck on the switch energized the solenoid, thereby moving the food hopper up so that the pigeons could eat for a computer-determined amount of time. Near the feeder was a 15" diameter antenna (AVID, Norco, CA). Because the welded wire of the cage absorbs the electromagnetic field of the antenna, thereby reducing its range and thus its ability to detect radiofrequency 1D tags, the antenna was mounted onto a plywood panel along one side of the cage. Within the circumference of the antenna, a hole in the plywood gave the birds access to a nipple-type plastic drinking font with cup. Slight pressure by a bird's beak released water, which would then accumulate in the cup. Thus, a bird in the act of drinking would have its head within the center of the field of the antenna. A microchip radiofrequency 10 tag, about the size of a large grain of rice, was implanted subcutaneously on the top of each bird's head.
To estimate body weight, an electronic balance equipped with an RS-232 serial port for computer control was placed on the cage floor in front of the drinking font and antenna. Birds usually stood on the balance while drinking. Operation of the feeder, response key, balance and antenna was controlled by a computer equipped with an analog/digital input-output card (Keithley, Cleveland, OH) and associated software which enabled programming in BASIC. ID tag detection thus permitted identification as to which bird was on the balance. Number of key pecks resulting in food presentations, i.e., "reinforcements," (if any), number of seconds spent drinking (if any), and mean body weight (if weighed) were automatically summarized by the computer program and stored in a data file every 15 min.
The pigeons were observed remotely via video camera to facilitate operant key training. They were first habituated to the action of the pellet feeder by repeated remote human activations of the feeder until they ate readily when the feeder was operated. They were then "shaped" to key peck by reinforcing (with feeder activation) successively closer approximations of pecking on the key, until pecking only the key itself earned a reward of food. After key pecking for food was established, the BASIC computer program provided a 20 sec food presentation following each key peck. Food was available 24 hr per day under this regimen, although birds tended to consume food only during their "day" time.
EATING
During preliminary testing, pigeons were exposed to long day lengths (15L:9D) to stimulate egg production, and number of food reinforcements was summarized hourly. A representative day's feeding pattern shows a crepuscular ("dawn" and "dusk" peaks) pattern of food reinforcement. Most data, however, were collected after day length was changed to short days (1OL: 14D) to prevent the confounding feature of egg laying and associated changes in body weight.
Short day lengths profoundly altered the pattern of feeding, changing the overall daylight profile of intake from crepuscular to unimodal, with most food consumed during midday.
Because we couldn't identify individual birds at the feeder, the data on feed use is a composite of both animals' behavior. Video monitoring showed that the presence of one bird often appeared to interrupt feeding attempts made by the other, i. e., food guarding behavior. For example, it was often observed that the female would press the food reward key but then be inhibited from consuming the feed by the presence of the male· or even be chased away by the male; the male was clearly dominant. We even observed instances when the female "earned" a food reward, which the male then consumed.
Observation of birds by video also showed that feeding bouts tended to last only a few minutes. With our data summarization interval of 15 min, we could not discriminate whether the increased feed intake during midday was due to increased frequency of eating bouts or whether the bouts of eating just lasted longer.
DRINKING
Video observation showed that birds often engaged in a bout of eating which was then followed by a bout of drinking. Time spent drinking was inferred from the amount of time the ill tags were in the field of the antenna. By this measure, the male spent about three times as much time drinking as the female, with much of his consumption occurring at and just after midday. The difference in time spent drinking by the male and female could be genuine. The male bird may have drunk more slowly, as no differences were apparent in wetness of droppings. Alternatively, it is possible that the male may have spent non-drinking time within the field of the antenna; this was observed casually on at least two occasions. To clarify this, drinking behavior measured via ID tags could be validated by time-lapse videography. A more precisely targeted antenna field would help solve this (potential) problem.
BODY WEIGHT
The electronic balance recorded large differences in daily body mass for both male and female birds (Fig. 2). In a representative 9-day period, body weight changed daily by about 7 % for the male and 6% for the female. For the most part, birds were at their body weight minimum at the beginning of their day (9 a.m.), then gained mass progressively until a couple of hours before darkness (7 p.m.). Their average weights were about 685 g and 620 g, respectively, and changed 40 to 50 g over the course of a single day. Measurements of body weight in many experimental situations are often made at midday, with the assumption that birds are behaviorally quiescent then because of a crepuscular feeding pattern, and therefore variability in body mass measurements may be minimized. In fact, for pigeons held under short day lengths, the opposite is true: body mass determinations made at midday can vary profoundly in less than one hour. This variability could decrease the statistical sensitivity of experiments relying on body mass determinations as a dependent variable.
CONCLUSIONS
In sum, these data demonstrate that operant feeding, remote body mass sensing and radio frequency ID tagging can provide a finely detailed characterization of the eating and drinking patterns of group-living birds in a typical captive environment. They also suggest that such measures could be used to monitor the effects of experimental treatments in a minimally invasive manner, permitting studies of captive wild species that might not otherwise tolerate the human intrusion required to gather these types of data.
Mike Delwiche is a professor in the Department of Biological and Agricultural Engineering His research and teaching interests include bioinstrumentation and control.
