2006 Poultry Science Association, Inc. Midday and Nighttime Cooling of Broiler Chickens J. C. Segura,* J. J. R. Feddes,* and M. J. Zuidhof 1 *Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton, Alberta, Canada, T6G 2P5; and Agriculture Research Division, Alberta Agriculture, Food and Rural Development, Edmonton, Alberta, Canada, T6H 5T6 Primary Audience: Flock Supervisors, Researchers, Technical Service Personnel SUMMARY Hot weather along with high stocking densities can lead to high mortality and decreased performance of broilers, especially during the last week of rearing. Two trials were conducted to test the hypotheses that reduced nighttime and midday temperatures improve broiler live performance and reduce mortality under warm cyclic temperature conditions. In each trial, groups of 306 male broilers were placed in each of 6 environmentally controlled chambers. The warm temperature treatments were control, nighttime cooling, and midday cooling (4 replicates). In the control treatment, diurnal temperatures ranged between 15 and 25 C on d 29 and progressively increased to between 20 and 35 C on d 42. In trial 2, 2 additional chambers housed broilers under thermoneutral conditions. No differences in feed consumption were found within or between trials due to the temperature treatments. The cooling treatments did not improve BW, weight gain, feed conversion, or livability. The birds housed under thermoneutral conditions did not have improved BW. These results suggest that broilers subjected to regular warm cyclic temperature fluctuations for 2 wk prior to shipping are able to acclimatize with no negative impact on BW. Effective environmental temperatures predicted from broiler surface thermal resistance and thermal mass were within 5 C among the treatments. Key words: broiler chicken, cyclic temperature, cooling, heat stress 2006 J. Appl. Poult. Res. 15:28 39 DESCRIPTION OF PROBLEM High temperatures during the summer months are a major concern for broiler producers. Hot days associated with high stocking densities can create a poor environment for the broiler, especially during the last week of rearing. This increases the risk of poor live performance and significant economic loss to producers. There has been considerable research on the effects of environmental temperatures on broiler performance, body temperature, and water con- sumption. Much of the emphasis has been on heat stress treatments with constant ambient temperature or minimum temperature variations. Few ventilation systems are able to maintain desired constant temperatures. The environmental conditions can be very difficult to control during warm weather. At times, the indoor temperatures remain 3 to 5 C above the outside temperatures when the ventilation rate is at maximum. Sudden temperature increases can affect the bird performance if they are not acclimatized to high temperatures. The current project included a wide range of tem- 1 Corresponding author: martin.zuidhof@gov.ab.ca
SEGURA ET AL.: COOLING OF BROILER CHICKENS 29 peratures from 15 C at night to 35 C during the day, thus simulating the conditions in many poultry barns during hot weather. When a bird has difficulty dissipating heat, it will lower its heat production by reducing feed intake. May et al. [1] confirmed that production efficiency is reduced because energy is expended for panting, and growth is slowed due to lower feed consumption. Furthermore, because birds develop (and reach market weight) at increasing rates due to improvements in genetics, feed, and management, they also produce heat at a higher rate. Several other studies have looked at live performance of broilers at high environmental temperatures. Lott [2] found that water consumption is a main component of acclimation of broilers to hot temperatures. Water must be available to meet this peak demand. Donkoh [3] found that water consumption increased with rising ambient temperatures from 20 to 35 C. At the same time it was found that the higher temperature decreased broiler growth rate, feed intake, and increased feed conversion ratio (FCR). May and Lott [4] suggested that mortality is the best indicator of the effect of rearing temperature on broiler performance. Little information was found concerning broiler chicken mortality related to high temperature fluctuations in commercial farms. Producers sometimes incur high losses during summer because of high temperatures, especially during the last 2 wk of the growth cycle [5]. In the Canadian Provinces of Alberta, Ontario, and Quebec, more than 600,000 chickens died in June of 2002 because of high temperatures [6]. Poultry mortality caused by high temperatures is an important concern that needs to be investigated through barn temperature control. The focus of this project was to investigate the effect of environmental temperature fluctuations on broiler production parameters (weight gain, feed consumption, and FCR) and mortality. The main objective was to investigate the effect of high diurnal fluctuating environmental temperatures on feed and water consumption, live performance, and bird mortality. Both midday and night cooling strategies were considered in reducing bird thermal loads. A secondary objective was to consider the thermal mass of a broiler in simulating its response to temperature, using temperature data from a filled water bottle hanging in the bird space. MATERIALS AND METHODS The Faculty Animal Policy and Welfare Committee of the Faculty of Agriculture, Forestry and Home Economics of the University of Alberta approved the experimental protocols [7]. In both trials, 306 male Hubbard Hubbard broiler chicks [8] were placed in each of 6 environmental chambers after being weighed in groups. In trial 2, two additional chambers housed birds at thermoneutral temperatures recommended by the industry. All chicks originated from the same parent flock. The dimensions of each environmental chamber were 4.45 m (length) by 3.85 m (width) by 3 m (height) giving an area of 17.13 m 2 and a stocking density of 0.05 m 2 /bird. Unusable space was 0.5 m 2, the area used by the 4 feeders, 0.4 m in diameter (Figure 1). A personal computer was used to operate the virtual instrument program written in LabView Version 5.1.1 [9]. This computer facilitated continuous monitoring of temperature and consumption of feed and water as well as monitoring and control of ventilation rate, heating, and cooling. Birds were provided with continuous lighting throughout the 42-d trial. Each chamber was equipped with a polyvinyl chloride nipple drinker system with 34 nipples (9 birds per nipple; 0.25 m between nipples; Figure 1). To continuously monitor water intake, the water line was connected to a 20-L water container set on a digital scale [10] with a personal computer interface. The software was designed to control the filling of the water container when the container mass was less than 7 kg (7 L) and to stop filling at 14.1 kg (14.1 L). The weight of the water container was read every 10 min. The difference between the 10-min readings represented the water consumption for that period of time. Four pan feeders with a capacity of 22.6 kg each were located in the middle of each chamber (Figure 1). The feeders were suspended from a beam that was supported at the ends by 2 scales similar to the one used for the water weighing system. The weight of the feeders (sum of the weight readings from the 2 scales) was read every 10 min. Feed consumption for each 10-min period was calculated by subtraction. The BW gain was determined weekly at d 7, 14, 21, 29, and 36 for
30 JAPR: Research Report Figure 1. Individual chamber layout, top, and lateral views. a group of 30 birds selected at random from each environmental chamber. In both trials, all the birds were fed a standard starter wheat-soybean based diet (3,000 kcal of ME/kg of diet, 21.3% crude protein) from 1 to 3 wk of age and a standard grower diet (3,000 kcal of ME/kg; 18.8%
SEGURA ET AL.: COOLING OF BROILER CHICKENS 31 crude protein) from 4 to 6 wk. In all treatments, feed and water were supplied ad libitum. The ambient temperature during the preexperimental period (d 1 to 28) in each treatment followed the same temperature curve: 35 C at d 1 and decreased linearly to 25.4 C by d 28. In each chamber, ventilation was provided by an exhaust fan (9,440 m 3 /h at 50 Pa; Figure 1) [11]. The computer software controlled the ventilation rate and the heating system according to the desired temperature in each chamber. Each chamber had one overhead recirculation fan that provided adequate air movement at bird level (Figure 1). Two HOBO data logger temperature sensors [12] were located in the center of each room. One sensor measured the water temperature in a 1-L black water bottle 15 cm above the floor surface. The response of the birds to the warm temperatures was thought to be similar to the water temperature rather than the dry-bulb temperature sensor at 30 cm above the water bottle. The set point temperature of each chamber was maintained by a finned pipe that could provide heating or cooling and through controlling the airflow from the exhaust fan (Figure 1). Control (Trials 1 and 2) The control treatment consisted of diurnal temperature fluctuations typical of those occurring during the summer months in the Prairie Provinces. An increasing cyclic temperature curve started at 20 C at midnight on d 29 and decreased to 15 C at 0500 h, after which the temperature rose linearly to 25 C at 1300 h.temperature progressively increased daily by approximately 0.8 C until the final temperature points on d 42 were 20 C at 0500 h and 35 C at 1300 h (Figure 2). Midday Cooling (Trials 1 and 2) Midday cooling treatment followed the same temperature as the control treatment except that the maximum temperature was 28 C (Figure 2). This cooling strategy is commonly used in industry by operating pressurized misting systems during the warmest part of the day when the outdoor air temperature is too high to use increased ventilation rates to maintain an indoor temperature of 28 C. Nighttime Cooling (Trials 1 and 2) The nighttime cooling treatment was the same as the control treatment, except that the minimum daily temperature was 15 C from 2300 to 0300 h (Figure 2). This temperature regimen took advantage of cooler outside air during the night. Thermoneutral Temperature (Trial 2) A thermoneutral temperature treatment was applied only during the second trial. The birds were assumed to be confined in thermoneutral temperature condition over their entire growth cycle. The ambient temperature on d 29 decreased approximately by 0.4 C each day until 20 C was reached on d 42 (Figure 2). To facilitate the statistical analysis [13], the 4 treatments from the 2 trials were grouped into the following 2 statistical models (see Table 1). Model A: control, midday cooling, and nighttime cooling treatment results were compared across trials 1 and 2 (4 replicates); Model B: control, midday cooling, nighttime cooling, and thermoneutral temperatures, trial 2 only (2 replicates). RESULTS AND DISCUSSION Many of the results approached significance levels. One of the difficulties with this type of experiment is the ability to replicate due to facility constraints. It is important to note that results approached significance consistently in Models A and B, and the specific treatment effects were repeated across trials. Larger sample sizes would improve the confidence of the inferences made in the current study. Mortality Mortality is summarized across Models A and B in Table 2. No significant differences were found for overall mortality or any specific cause of death, including sudden death syndrome, ascites, omphalitis, or culling. For both models, most of the culled birds exhibited leg problems. The results for bird live performance and feed and water consumption for Models A and B were divided as follows. Live Performance Model A. Lower BW at d 29 in the control treatment compared with midday and nighttime
32 JAPR: Research Report Figure 2. Temperature treatments (d 29 to 42): A) thermoneutral, B) control treatment C) nighttime cooling, and D) midday cooling treatments.
SEGURA ET AL.: COOLING OF BROILER CHICKENS 33 Table 1. Treatment-trial separation in Models A and B Trial 1 Trial 2 Model B Model A Midday cooling 1 Midday cooling 1,2 Nighttime cooling 1 Nighttime cooling 1,2 Control 1 Control 1,2 Thermoneutral 2 1 Treatment is part of Model A. 2 Treatment is part of Model B. cooling treatments approached significance (P = 0.065; Table 3). Pairwise testing demonstrated significant differences between these treatments. The reason for the weight difference was not apparent, as temperature treatments were not applied until d 29. By d 42, BW in the control treatment was higher than in the nighttime cooling treatment (P = 0.058). Birds in the control treatment gained more than those in the nighttime cooling treatment (P = 0.056). The cooling treatments were assumed to decrease the heat stress and result in higher gains. When considering bird strain and their ability to physiologically adapt or compensate, these results demonstrate that bird performance can be remarkably similar when a wide range of temperature regimens are applied during the later stages (last 2 wk) of bird growth. Model B. Although not significant overall, pairwise differences were significant between 29- d BW in the control treatment and the nighttime cooling treatment (Table 3). The reason for the weight difference was not apparent considering that the temperature treatments were applied after d 28. As in Model A, the final bird weights at d 42 in the control treatment were greater than those in the nighttime cooling treatment (P = 0.090). Overall differences in gain approached significance (P = 0.090). A pairwise comparison of means showed that birds in the control treatment gained more than those in the nighttime cooling treatment. This finding was unexpected, as the cooling treatments were developed to decrease heat stress and result in higher gains. Surprisingly, the birds in the thermoneutral treatment did not show an increased gain over those in the warm temperature treatments. These results again demonstrate that bird performance can be remarkably similar under a wide range of temperature regimens in the later stages of the growth cycle. Feed and Water Consumption In Model A, no significant treatment differences were observed in overall feed consumption, water intake, water to feed ratio, and feed conversion rate (Table 3). In Model B, the water to feed ratio in the daytime cooling treatment was significantly higher than those values in the other treatments. The water to feed ratio in the control treatment was the lowest (P > 0.05). This finding suggests the birds with the highest gain have the lowest water to feed ratio. Overall feed and water intake and feed to gain values were not different between treatments (Table 3). There were minor differences in diurnal feed and water intake patterns that corresponded to temperature fluctuations (Figure 3). Although birds may have experienced heat stress to varying degrees, they appear to have adapted, resulting in similar daily feed and water consumption rates. In all treatments except the thermoneutral control, higher levels of Table 2. Mortality (%) during all treatments from 0 to 42 d of age and the type of disease causing the mortalities Mortality cause Marketable Birds Mortality SDS 3 Ascites 4 Omphalitis Culled 5 Other 4,6 Treatment (%) (%) (%) (%) (%) (%) (%) Control 1 93.4 6.62 2.37 0.90 0.65 0.90 1.80 Midday cooling 1 94.1 5.88 1.72 1.23 0.65 0.82 1.47 Nighttime cooling 1 94.0 6.05 2.04 0.98 0.65 0.65 1.72 Thermoneutral 2 94.9 5.07 2.12 0.33 0.16 0.65 1.80 1 From 4 replicates, Model A. 2 From 2 replicates, Model B. 3 SDS = sudden death syndrome. 4 Included condemnation from processing plant. 5 Included birds culled because of leg deformities. 6 Other diseases are cellulitis, emaciation, hepatitis, and pendulous crop.
34 JAPR: Research Report Table 3. Live performance data for the period from 29 to 42 d of age for birds exposed to a range of temperature conditions. Initial and final body weight, gain, feed and water consumption, water:feed ratio, and feed:gain ratio for statistical Models A and B Model and Feed Water Water: Feed: treatment Gain BW 29 d BW 42 d intake intake feed gain (TRT) (g/d) SEM (g) SEM (g) SEM (g) SEM (g) SEM (g:g) SEM (g:g) SEM Control 2 77.4 a 1.2 1,008 b 17.7 2,086 a 16.3 2,182 45.3 3,752 70.9 1.72 0.02 2.116 0.095 DC 3 74.1 ab 1.2 1,059 a 17.2 2,042 ab 16.4 2,189 38.7 3,874 60.6 1.77 0.02 2.184 0.081 NC 4 73.5 b 1.2 1,058 a 17.2 2,035 b 15.8 2,206 36.7 3,809 57.4 1.73 0.02 2.221 0.077 Control 80.9 a 1.7 1,012 b 24.1 2,102 a 21.9 2,240 82.3 3,763 138.3 1.68 c 0.01 2.096 0.074 DC 77.8 ab 1.8 1,055 ab 23.8 2,061 ab 23.1 2,200 68.9 4,066 115.7 1.85 a 0.01 2.193 0.062 NC 74.8 b 1.7 1,079 a 23.8 2,022 b 21.8 2,193 74.1 3,830 124.5 1.75 b 0.01 2.280 0.067 TN 6 77.6 ab 1.7 1,044 ab 23.5 2,060 ab 22.4 2,268 64.5 4,007 108.4 1.77 b 0.01 2.238 0.058 Tests of significance P > F A B BW 29 d 7 <0.0001 NA <0.0001 0.5563 0.7226 0.2188 0.8410 Trial 0.0001 0.4927 0.9654 0.4307 0.0725 0.1390 0.8140 Trial TRT 0.4054 0.5978 0.4147 0.9200 0.1213 0.0927 0.9053 TRT 0.0564 0.0646 0.0577 0.9052 0.5213 0.3100 0.7435 BW 29 d <0.0001 NA <0.0001 0.3223 0.8067 0.0022 0.8611 Tmt 0.0899 0.2596 0.0899 0.8813 0.3453 0.0031 0.4353 a c Means within model with no common superscript differ significantly (P < 0.05). 1 Combined analysis over trials 1 and 2; 4 replicates per treatment. 2 Control treatment: diurnal minimum and maximum temperatures increasing from 15 and 25 C (d 29) to 20 and 35 C (d 42), respectively. 3 Midday cooling treatment: same as control except that the maximum temperature was 28 C. 4 Nighttime cooling treatment: same as control except that the daily minimum temperature was 15 C from 2300 to 0300 h. 5 Trial 2, 2 replicates per treatment. 6 Thermoneutral control: initial (d 1) temperature 35 C, decreasing 0.4 C/d to 20 C (d 42). 7 Initial (29 d) BW was used as a covariate in the statistical model.
SEGURA ET AL.: COOLING OF BROILER CHICKENS 35 Figure 3. Overall comparison of 2-h feed and water intake (d 29 to 42) per bird and the temperature for each treatment (Model A) and thermoneutral treatment (Model B).
36 JAPR: Research Report Figure 4. Cyclic ambient and water bottle temperatures during d 37 and 38 in the control treatment. feed and water intake occurred between 0900 and 1100 h (Figure 3), before the elevated temperatures occurred. This response prior to an increase in temperature agrees with that reported by May et al. [1]. The lack of differences among final BW, feed consumption, and water consumption in Models A and B indicated the birds adapted to their environments. Although there were diurnal temperature differences between treatments (Figure 3), the birds appeared to respond more to the daily average temperature than to ambient temperature. The BW and FCR for the thermoneutral treatment, in which the average temperature was coolest over the experimental time frame, were not different from the other treatments, suggesting that the birds were able to adapt to the increased heat, perhaps because they were able to cool down during the night (Figure 3). Bird Response to Temperature Similarities in live weight among the different temperature treatments may be explained by behavioral, physiological, and physical responses of the birds to their environment. Because broilers have thermal mass properties, the effective environmental temperature that a broiler perceives or responds to may be quite different from the ambient temperature. In this project, a 1-kg water-filled bottle was suspended in the bird space. The rate of change in the temperature of the water bottle contents may provide insights into bird response to its thermal environment. The rate of heat exchange between the broiler and its environment is dependent on the surface thermal resistance and the simulated effective environmental temperature of the chicken. The simulated effective environmental temperature is dependent on the rate of change in temperature of 2.2 kg of water. This approximated the mass of a 40-d-old broiler chicken. The film of stagnant air that adheres to the surface and acts as insulation [15]; insulating feathers, floor thermal resistance, contact area during resting, and the velocity of the air in the vicinity
SEGURA ET AL.: COOLING OF BROILER CHICKENS 37 of the bird determine the effective surface thermal resistance to broiler body heat loss [16, 17]. As thermal resistance increases, broilers have more difficulty losing heat and tend to reduce their feed intake. As shown in Figure 4, the temperature of the water in the bottle was cyclic (sinusoidal), similar to that of the ambient temperature. However, the bottle temperature fluctuations had reduced amplitude and lagged behind the ambient temperature measurements. To simulate the effect of the thermal mass of the bird and the surface thermal resistance of the bird to heat loss or heat gain from the environment, the thermal properties of the bird were applied to a 2.2 kg container filled with water. The following equations describe the 2 effects: Q = mc p (T container, t+1 T container, t ) Equation [1] where Q is the heat (kj) available to heat or cool the water, m is the mass (2.2 kg) of the container, C p is the specific heat of water (kj/ kg per C), and T container is the container temperature ( C) at time t or t + 1 (h). Table 4. Hourly heat stress indices (HSI) over the last 14 d (Model B) 1 HSI (C h) Treatment Ambient Bottle Control 2 1,708 2,102 Midday cooling 3 1,371 2,148 Nighttime cooling 4 1,534 2,122 Thermoneutral 5 277 1,525 CV % 53.1 23.4 1 Trial 2 2 replicates per treatment. 2 Control treatment: diurnal minimum and maximum temperatures increasing from 15 and 25 C (d 29) to 20 and 35 C (d 42), respectively. 3 Midday cooling treatment: same as control treatment except that the maximum temperature was 28 C. 4 Nighttime cooling treatment: same as control treatment except that the daily minimum temperature was 15 C from 2300 to 0300 h. 5 Thermoneutral: initial (d 1) temperature 35 C, decreasing 0.4 C/d to 20 C (d 42). substantial fluctuations in ambient temperatures [20]. Heat stress indices (HSI), as proposed by Emmanuel [21], were calculated for each treatment to investigate the similarity in the bird response. The formula is as follows: Q = A(T ambient T container )/(R airfilm + R feathers ) Equation [2] HSI = Σ ( T t) Equation [3] where Q is the heat entering or leaving the container over a 1-h period (kj/ h), A is the assumed body surface area of a 2.2-kg bird based on Walsberg and King [18] equation (0.137 m 2 ). R feathers is the assumed thermal surface resistance of a 42-d broiler fed ad libitum (0.11 m 2 C h Kj 1 ) found by Ozkan et al. [17]. The surface thermal resistance also included the airfilm (R airfilm ) value of 0.028 m 2 C h kj 1 [19]. Figure 5 shows that the effective environmental temperature pattern that birds responded to was reduced in amplitude. The 4 simulated effective environmental temperature patterns were quite similar. The simulated diurnal temperatures were within 5 C of the temperatures recommended by industry. Also, the birds were acclimated to this diurnal temperature pattern, which might explain why the growth response did not vary among the 4 treatments in spite of where T is the difference between ambient and recommended temperature, and t is bird age (d). This index weights differences in ambient and recommended temperatures more heavily as birds become older. Using the hourly ambient and water bottle temperature readings, an hourly HSI was calculated for each treatment (Table 4). The birds in the control treatment had a higher HSI than the other 3 treatments by design, as periods of lower temperatures were provided in the midday and nighttime cooling treatments. Higher HSI values were calculated from the temperatures collected from the water bottles compared with ambient temperatures; however, the coefficient of variation for HSI (23.4%) was substantially lower compared with that calculated using ambient temperatures (53.1%). This reduced variability demonstrates how thermal mass of birds tempers heat stress.
38 JAPR: Research Report Figure 5. Hourly and simulated temperature perceived by broiler chickens among the treatments. CONCLUSIONS AND APPLICATIONS 1. The midday, nighttime and the thermoneutral treatments during the last 2 wk of rearing had no significant effect on feed and water consumption. However, diurnal differences were observed. Birds adjusted their feed and water intake patterns to compensate for diurnal temperature fluctuations. 2. The period with the greatest feed and water intake was from 0900 to 1200 h for Model A. This preceded the temperature peaks, which were also a factor in increased feed and water consumption because birds are more likely to eat when not under heat stress. For the thermoneutral treatment, the feed consumption did not peak as in the other treatments, most likely because of the constant temperature during the day. 3. Cooling treatments did not increase market BW. 4. The birds in the thermoneutral treatment did not have the highest BW, suggesting that the birds in the warmer cyclic temperature treatments were able to acclimatize. 5. The temperatures recorded with the black water bottle provided insight as to how birds responded to the warm temperatures treatments. The effective temperature difference among treatments was only 5 C. Without a sudden increase or decrease in temperatures, high mortality was unlikely. 6. The HSI values based on water-bottle temperatures were more consistent with the live performance data than HSI values calculated with ambient dry-bulb temperatures. REFERENCES AND NOTES 1. May, J. D., B. D. Lott, and J. D. Simmons. 2000. The effect of air velocity on broiler performance and feed and water consumption. Poult. Sci. 79:1396 1400. 2. Lott, B. D. 1991. The effect of feed intake on body temperature and water consumption of male broilers during heat exposure. Poult. Sci. 70:756 759.
SEGURA ET AL.: COOLING OF BROILER CHICKENS 39 3. Donkoh, A. 1989. Ambient temperature: a factor affecting performance and physiological response of broiler chickens. Int. J. Biometeorolol. 33:259 265. 4. May, J. D., and B. D. Lott. 2000. The effect of environmental temperature on growth and feed conversion of broilers to 21 days of age. Poult. Sci. 79:669 671. 5. Hyink, D. 2001. Alberta Chicken Producers, Edmonton, AB, Canada. Personal communication. 6. Gerein, K. 2002. Chickens roast to death on Camrose-area farm. Edmonton Journal, 13 July, sec. A1 and A20. 7. Canadian Council on Animal Care. 1984. Guide to the Care and Use of Experimental Animals. Vol. 2. Canadian Council on Animal Care, Ottawa, ON, Canada. 8. Hubbard L.L.C., Walpole, NH. 9. Virtual instrument program LabView, Version 5.1.1, National Instruments Inc., Austin, TX. 10. Pelouze, Model 4010CN, Signature Brands Inc., Bridgeview, IL. 11. 558 mm Exavent exhaust fan, Fancom, Panningen, the Netherlands. 12. HOBO data logger temperature sensor, Onset Computer Corporation, Bourne, MA. 13. The data were analyzed by analysis of variance using the mixed procedure of SAS [14]. Daily diurnal feed and water consumption data were analyzed as 12 periods of 2 h each. Live performance data were analyzed using individual BW data with treatment, trial (Model A), and their interactions as sources of variation (fixed effects) and 29-d BW as a covariate. Differences between means were determined using pairwise tests and are reported as significant where P < 0.05, unless otherwise indicated. Mortality was analyzed with chi-squared analysis using the CATMOD procedure [14]. Mortality data were not transformed, because transformations did not satisfy normality assumptions. 14. SAS. 2001. SAS system for elementary statistical analysis. Release 8.1. SAS Institute Inc., Cary, NC. 15. Simmons, J. D., and B. D. Lott. 1997. Heat loss from broiler chickens subject to various air speed and ambient temperatures. Appl. Eng. Agric. ASAE 13:665 669. 16. Ward, J. M., D. C. Houston, G. D. Ruxton, D. J. McCafferty, and P. Cook. 2001. Thermal resistance of chicken (Gallus domesticus) plumage: a comparison between broiler and free-range birds. Br. Poult. Sci. 42:558 563. 17. Ozkan, S., W. K. Smith, and H. M. Bath. 2002. The development of thermal resistance of the feather coat in broilers with different feathering genotypes and feeding regimes. Br. Poult. Sci. 43:472 481. 18. Walsberg, E. G., and J. R. King. 1978. The relationship of the external surface area of birds to skin surface area and body mass. J. Exp. Biol. 76:185 189. 19. Van Beek, G., and F. F. E. Beeking. 1995. A simple steady state model of the distribution of vertical temperature in broiler houses without internal air circulation. Br. Poult. Sci. 36:341 356. 20. Van Kampen, M., B. W. Mitchell, and H. S. Siegel. 1979. Thermoneutral zone of chickens as determined by measuring heat production, respiration rate, and electromyographic and electroencephalographic activity in light and dark environments and changing ambient temperatures. J. Agric. Sci. 92:219 226. 21. Emmanuel, E. J. 2001. Effect of ventilation rate, air speed and bird disturbance on the incidence of cellulitis and broiler performance. M.S. Thesis, Univ. Alberta, Edmonton, Alberta, Canada. Acknowledgments The authors acknowledge financial support for this project from the Alberta Agricultural Research Institute, Alberta Chicken Producers, and Lilydale Cooperatives Ltd. The assistance of the staff at the Poultry Research Center was invaluable to the study. The advice and technical assistance of C. Ouellette and G. Van Veen are gratefully acknowledged. The assistance of the management and staff at Lilydale Cooperatives Ltd. in Edmonton to process the birds in groups was very much appreciated.