Using a Thermal Camera to Measure Heat Loss Through Bird Feather Coats

Summary

This protocol describes the quantification of heat transmission through a flat skinned avian specimen using a thermal camera and hot water bath. The method allows obtaining of quantitative, comparative data about the thermal performance of feather coats across species using dried flat skin specimens.

Abstract

Feathers are essential to insulation, and therefore to the cost of thermoregulation, in birds. There is robust literature on the energetic cost of thermoregulation in birds across a variety of ecological circumstances. However, few studies characterize the contribution of the feathers alone to thermoregulation. Several previous studies have established methods for measuring the insulation value of animal pelts, but they require destructive sampling methods that are problematic for birds, whose feathers are not distributed evenly across the skin. More information is needed about 1) how the contribution of feathers to thermoregulation varies both across and within species and 2) how feather coats may change over space and time. Reported here is a method for rapidly and directly measuring the thermal performance of feather coats and the skin using dried whole skin specimens, without the need to destroy the skin specimen. This method isolates and measures the thermal gradient across a feather coat in a way that measurements of heat loss and metabolic cost in live birds, which use behavioral and physiological strategies to thermoregulate, cannot. The method employs a thermal camera, which allows the rapid collection of quantitative thermal data to measure heat loss from a stable source through the skin. This protocol can easily be applied to various research questions, is applicable to any avian taxa, and does not require destruction of the skin specimen. Finally, it will further the understanding of the importance of passive thermoregulation in birds by simplifying and accelerating the collection of quantitative data.

Introduction

Feathers are the defining characteristic of birds and serve many functions, among the most crucial being insulation¹. Birds have the highest average core temperatures of any vertebrate group, and feathers insulating them from environmental temperature changes are a vital part of energy balance, especially in cold environments². Despite the importance of feathers, the majority of literature on changes in thermal condition in birds has focused on metabolic responses to temperature variation rather than the function of feathers as insulation^{3,4,5,6,7,8,9,10} (for further details, refer to Ward et al.)^11,12,13. However, feathers themselves may vary across time, individuals, and species.

The method presented here is useful for quantifying the overall thermal value of the feather coat alone. It removes confounding factors in live birds, such as behavioral thermoregulation¹ and varying amounts of insulating fat. More widespread measurement of the thermal performance of feather coats is necessary to improve the understanding of how feathers contribute to insulation and how this varies among and within species throughout a bird’s life history and annual cycle.

Feathers insulate birds by trapping air both between the skin and feathers as well as within the feathers, and they create a physical barrier to heat loss¹⁴. Feathers consist of a central feather shaft, called a rachis, with projections called barbs¹⁴. Barbules are small, secondary projections on barbs that interlock with adjacent barbs together to “zip” up the feather and give it structure. Furthermore, down feathers lack a central rachis and have few barbules, therefore forming a loose, insulative mass of barbs over the skin¹⁴. Feather coats vary across species^15,16, within species^17,18, and within comparable individuals^{2,19,20,21,22,23,24}. However, there is little quantitative information about how variations in number of feathers, the relative abundance of different types of feathers on a bird, or changes in the numbers of barbs/barbules affect the overall thermal performance of a feather coat. Previous studies have focused on determining a single mean value of insulation and thermal conductivity for a given species^11,12,13.

The feather coat is known to vary among species. For example, most birds have distinct areas of skin from which feathers do, or do not, grow called the pterylae and apteria, respectively¹⁴. The placement of the pterylae (sometimes called “feather tracts”) varies across species and has some value as a taxonomic character¹⁴. However, some birds (i.e., ratites and penguins) have lost this pterylosis and have a uniform distribution of feathers across the body¹⁴. Additionally, different species, especially those inhabiting different environments, have different proportions of feather types¹⁵. For example, birds inhabiting colder climates have more down feathers¹⁵ and contour feathers with a larger plumulaceous portion¹⁶ than species inhabiting warmer environments.

The microstructure of certain types of feathers may also have an effect on insulation across species^25,26. Lei et al. compared the microstructure of contour feathers of many Chinese Passerine sparrows and found that species inhabiting colder environments have a higher proportion of plumulaceous barbs in each contour feather, longer barbules, higher node density, and larger nodes than species inhabiting warmer environments²⁵. D’alba et al. compared the microstructure of down feathers of common eiders (Somateria mollissima) and graylag geese (Anser anser) and described how these differences affect both the cohesive ability of the feathers and ability of the feathers to trap air²⁶. Quantitative comparative data about how these variations in feathering affect the overall thermal performance of the feather coat across species is limited (for more details, refer to Taylor and Ward et al.)^11,13.

Within a species, the feather coat’s thermal performance may vary. Some species, such as the monk parakeet (Myiopsitta monachus)¹⁷, inhabit very large and diverse geographical ranges. The different thermal stresses posed by these different environments may affect the feather coats of birds within a species regionally, but there are currently no data available on this topic. Additionally, Broggi et al. compared two populations of great tits (Parus major L.) in the northern hemisphere. They demonstrated that contour feathers of the more northern population were denser but shorter and less proportionally plumulaceous than those of the more southern population. However, these differences disappeared when birds from both populations were raised in the same place¹⁸.

Furthermore, Broggi et al. explained these findings as a plastic response to differing thermal conditions, but they did not measure the insulation values of these different feather coats¹⁸. The results also suggest that contour feather density is more important to insulation than the proportion of plumulaceous barbs in contour feathers, but Broggi et al. suggested that northern populations may be unable to produce optimal feathers due to a lack of adequate nutrients¹⁸. Quantitative measurements of the overall thermal performance of these feather coats would further the understanding of the significance of plumage differences.

Over time, the feather coats of individual birds vary. At least once a year, all birds molt (replace all of their feathers)¹⁹. As the year goes on, feathers become worn^2,20 and less numerous^18,21,22,23. Some birds molt more than once a year, giving them multiple distinct feather coats each year¹⁹. Middleton showed that American goldfinches (Spinus tristis), which molt twice a year, have a higher number of feathers and higher proportion of downy feathers in their basic plumage in winter months than they do in their alternate plumage during summer months²⁴. These annual differences in the feather coat may allow birds to conserve more heat during colder periods passively or shed more heat passively during warm seasons, but no studies have tested this conclusively.

Although birds thermoregulate behaviorally^1,27 and can acclimate metabolically to different thermal conditions^{3,4,5,6,7,8,9,10,26}, feathers play an important role in thermoregulation by providing a constant layer of insulation. The method described here is designed to answer questions about the feather coat alone and its role in passive thermoregulation (i.e., how much heat does a living bird retain without modifying its behavior or metabolism?) by isolating the feathers. While active and physiological thermoregulation are ecologically important, it is also important to understand how the feathers alone aid in insulation and how they influence the need for active behavioral and physiological thermoregulation.

Previous studies have established methods for quantifying thermal conductivity and thermal insulation of animal pelts^11,12,13,28. The method presented here is an extension of the “guarded hot plate” method^11,12,13,28. However, the method described here measures the temperature at the outer boundary of the feather coat using a thermal camera, rather than thermocouples. The guarded hot plate method gives very precise estimates of energy flow through a pelt, but it requires construction of a multi-material hot plate, some familiarity with the use of thermocouples and thermopiles, and destructive use of a pelt that must be cut into small pieces. These pieces are then greased to eliminate air between the sample and hot plate apparatus. With the exception of the few birds that lack apteria (e.g., penguins), cutting small squares from bird skins is problematic for comparative purposes, since the location of the cut has large effects on the number of feathers actually attached to (and overlying) the skin. This problem is exacerbated by the variation among taxa in the presence, size, and placement of ptyerlae¹⁴.

Furthermore, while museum specimens can be a potentially rich resource for assessing the variation in insulation among birds, in general, permission to cut and grease skin specimens in scientific collections is unattainable. Additionally, specimens taken from the wild for guarded hot plate measurements cannot be subsequently used as museum specimens. The method presented here differs from the guarded hot plate method in that it can be used with whole dried bird skins, without 1) requiring the destruction of the specimen and 2) greasing the underside of the skin. It uses thermal cameras, which are increasingly affordable (though still relatively expensive), precise, and used for live bird measurements of thermal relations.

This method does not measure energy flow (and therefore thermal conductivity or insulation value) through the skin and feathers directly as the guarded hot plate method does. Instead, it measures the temperature at the outer boundary of a feather coat using a thermal camera. The resulting values represent an integrated measure of the heat lost passively through the skin, feathers, and air trapped between them (compared to a heat source underneath). Specimens prepared as flat skins and measured using the described technique can be stored in collections, and indefinitely provide value for future research. This method provides a standardized, comparable, and relatively simple way to measure feather coat thermal performance in any flat skinned specimen, which is especially useful in inter- and intra-specific comparisons.

Protocol

This work did not involve any work with live animals and was therefore exempt from animal care review.

1. Set-up and materials (Figure 1)

1. If flat skins of the species of interest are not available, use Spaw’s protocol²⁹ to create skins from fresh or frozen specimens. Preen feathers into a neat, natural position, and dry to obtain a constant weight before proceeding with measurements.
2. Set up a constant temperature hot water bath.
   NOTE: This set-up is quite tall, so it is easiest to place the hot water bath on the floor.
3. Place a sheet of clear acrylic glass (Table of Materials) over the surface of the constant temperature hot water bath. The glass allows transmission of the heat to the underside of the skin without wetting the specimen.
   NOTE: This pilot study used a sheet of acrylic glass (0.125 in thick). The thickness of the glass will not affect the emissivity³⁰ of the material (it will always be 0.86), but it will affect the absolute temperature at the surface of the glass (i.e., thicker glass will result in a lower temperature). Thus, all measurements should be taken using an acrylic glass sheet of the same thickness.
4. Place a piece of foam core board (1 in thick) with a circular hole (0.5 in diameter) over the acrylic glass.
   NOTE: The size of the bird should guide the size of the hole and thus the size of uninterrupted feathering receiving heat from the source. Here, a 0.5 in diameter hole is used, because this size is large enough to achieve sufficient heat transmission to the specimen while still being small enough to center the heat under the breast feather tracts (of all birds except for the smallest ones). Regardless of the size of the thermal opening, to obtain a comparable and replicable value for every bird, make sure to perform measurements with holes of the same size.
5. Attach a thermal camera to a tripod directly above the set-up at the camera’s minimum focusing distance.
   NOTE: Here, a FLIR SC655 thermal camera is used (680 px x 480 px resolution, ±2 °C or ±2% accuracy of reading, 40 cm minimum focusing distance). Other cameras may differ in degree resolution.
6. Calibrate the camera by entering the following into the thermal camera software:
   1. Find the reflected temperature by placing aluminum foil (shiny side facing up) over the foam with the emissivity value set to 1.0 in the camera calibration software. Take a thermal image. The temperature at the surface of the aluminum foil is the reflected temperature, which should be similar to the ambient temperature of the room.
   2. Set the emissivity value³¹ to 0.95.
      NOTE: 1) Emissivity is the relative amount of heat that an object emits³² and ranges from 0 to 1. An object with a high emissivity value emits a large amount of heat, while an object with a low emissivity value radiates little heat³². This value represents the emissivity of feathers. 2) This value (0.95) is disputed. Cossins and Bowler claimed that feathers have an emissivity value between 0.90–0.95 but included no evidence³¹. Hammel reported a value of 0.98, but he obtained this value from a frozen specimen, so it may not be accurate³³. Despite the lack of evidence, 0.95 emissivity is the value most often used in the thermal camera literature (as evidenced by Cossins and Bowler³¹).
   3. Make sure the ambient temperature and humidity in the room are constant. These values should be measured before every measurement and updated in the camera calibration software. The temperature and humidity of all indoor rooms will fluctuate somewhat, so recording these values and updating them in the software reduces measurement error.
      NOTE: Here, FLIR ResearchIR Max software is used. This software does not save data for all images, so it is crucial to record all of these values for each image.

2. Performance of measurements

1. Set the constant temperature hot water bath to a target temperature (40 °C is a proxy for the mean internal core temperature of most passerine birds)³⁴.
   NOTE: If working with a species whose resting core temperature is higher (e.g., hummingbirds³⁴) or lower (e.g., penguins³⁴ or ratites^34,35), it may be appropriate to adjust the hot water bath accordingly. Figure 3 shows the relationship between the temperature of the hot water bath and temperature at the surface of the acrylic glass (e.g., the actual temperature of the heat source the flat skin is exposed to).
   1. Results obtained from this protocol (see Figure 5) suggest that obtaining measurements across a range of temperatures is also informative of thermal performance differences. To accomplish this, follow the protocol using 5 °C increments from 30–55 °C.
2. In the thermal camera software, draw a circle/ellipse over the hole in the foam where the heat is escaping. This makes it possible to visualize this area when placing the skin on the foam to ensure that the correct area on the flat skin is measured.
3. Place the flat skin specimen on the foam with the area of interest over the hole.
   NOTE: Here, the belly region of each bird is measured, because it is not obstructed by other parts of the body such as the wing and is centrally located enough not to be subjected to edge effects. The placement of the skin over the heat hole will vary according to the experimental question. In general, placement directly over a feather tract, and as far away from the skin edge as possible, is recommended. Make sure not to flatten or disorder the feathers when placing the skin. If necessary, preen them into a natural position once the skin is placed.
4. Wait 15 min to allow the skin to acclimatize to the heat source. If measurements are made too early, the temperature value at the surface of the feather coat will be too low. Here, the temperature transmission through the skin and feathers stabilized at 15 min, so waiting longer than 15 min will not yield an artificially high result.
5. Taking a thermal image of the flat skin.
   1. Set the emissivity value³¹ to 0.95 before taking the thermal image.
6. Remove the skin from the foam and immediately take a thermal image of the set-up without the flat skin on the foam. This quantifies the temperature at the surface of the acrylic glass and calibrates the area of the heat source with the measurement area on the flat skin.
   1. Here, the emissivity of the acrylic glass used³⁰ is 0.86. Make sure to record this in the thermal camera software before taking the picture without the skin.
      NOTE: The temperature displayed by the hot water bath is not necessarily the temperature at the surface of the acrylic glass (Figure 3), since its thermal conductance is not perfect. Using the temperature of the glass reduces errors in estimating how warm the underside of the skin is, and is therefore a combined estimation of how much heat is lost through the skin and feathers.
7. Place the skin back on the foam in the same position. Repeat steps 2.5–2.6 for a total of five trials.
   1. To place the specimen skin correctly, gently touch the feathers in the target measuring area with one fingertip, then remove the finger and view the thermal image. The residual heat from the finger will remain visible on the thermal image briefly. Check to ensure that the sampling area is within the visible circle drawn in the software, which represents the area of heat radiating through the skin from the hot water bath. If it is not, move the skin until it is. This process is illustrated in Figure 2.
      NOTE: While fresh skins (when available) may represent the natural thermal performance of the skin in a living bird more closely, using dry skins for these measurements allow comparable, repeatable results on a much larger pool of specimens. Therefore, all measurements should either be taken using skins dried to constant weight, or on both the fresh and dried skin of specimens.

3. Data collection from thermal images

1. Each measurement consists of two thermal images: one of the flat skin and one of the acrylic glass. First, open the image of the acrylic glass. Align the circle drawn in the software with the hole in the foam visible in the image. Record the temperature value at the center of the circle.
   NOTE: For more details on extracting data from thermal images, see Senior et al.³⁶.
   1. Make sure to calibrate the camera with the proper values. Set the emissivity³⁰ to 0.86 and set the ambient temperature and humidity to match the current conditions in the lab, before recording the temperature value.
2. Open the thermal image of the flat skin. Without moving the circle, record the temperature value at the center of the circle.
   NOTE: Because the circle is not recorded in the image, it is important to calibrate the placement of the circle with the image of the acrylic glass taken in section 2.6.
   1. Make sure to calibrate the camera with the proper values. Set the emissivity³¹ to 0.95, and make sure to set the ambient temperature and humidity to the current conditions in the lab, before recording the temperature value.
3. Repeat steps 3.1–3.2 for all measurements of all specimens.

4. Calculation of thermal performance

1. Subtract the temperature of the surface of the feather coat from the temperature of the acrylic glass. This value represents the heat retained by the feather coat.

Results

Representative results from a series of one individual of each of five species, measured at six temperatures, are presented in Figure 4 and Figure 5. These show that small variations in the placement of the skin can result in variations in the readings of up to 1.7 °C. Figure 4 shows how training of an investigator increases repeatability of the measurements. For example, the same individual house sparrow (Passer domesticus...

Discussion

This paper provides a protocol for repeatable, standardized thermal imaging measurements of avian flat skin specimens. This method makes it possible to compare thermal performance of the feather coat among species, within species, between comparable individuals, and at different locations on the bodies of individuals, all without destruction of the specimen.

The availability of necessary materials and equipment may be a limitation of this method. Although thermal cameras are rapidly becoming m...

Materials

Name	Company	Catalog Number	Comments
Aluminum Foil	Reynolds Wrap	109000831	30 square ft.; this exact model need not be used.
Foam Core Board	Foamular	20WE	1 in. x 4 ft. x 8 ft; this exact model need no be used.
General Purpose Water Bath	PolyScience	WB02	Ambiet +5 °C to 100 °C; ±.01 °C
PDF Data logger	Elitech	RC-51H	Built in temperature and humidity sensor
Plexiglass	AdirOffice	1212-3-C	Acrylic glass; 12 in. x 12 in. x 1/8 in.; this exact model need not be used.
Thermal Image Analysis Software	FLIR	ResearchIR Max v4.40.7.26 (64-bit)	Allows collection of precise, quantitative thermal data
Thermal Imaging Camera	FLIR	SC655	680x480-pixel resolution, ±2 °C or ±2% accuracy, 40 cm minimum focusing distance
Tripod	The Audubon Shop	The Birder Tripod with Manfrotto 700RC2 Rapid Release Head	65" maximum height; this exact model need not be used.