| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ANIMAL PRODUCTION |
Department of Animal and Range Sciences, Montana State University, Bozeman 59717
 |
Abstract |
|---|
 Top Abstract INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
Key Words: beef cattle • metabolic requirement • orientation • simulation model • solar radiation • winter
|
INTRODUCTION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
). To reduce costs, cattle may graze semiarid foothill range, although this exposes them to cold temperatures and high winds while grazing areas with low nutritional value.
Cattle energetic requirements and metabolism under cold conditions are often studied in controlled environments or metabolic chambers, or by using heat sources covered with a pelt. Field and pen studies often include abundant feed (Blaxter, 1967; Christopherson et al., 1979). Cattle may respond differently while grazing during winter (Yousef, 1989). Cattle and wildlife conserve energy by lowering metabolic rate or resting heat production (Cuyler and Oritsland, 1993; Bergen et al., 2001; Han et al., 2003), seeking shelter (Redbo et al., 2001), altering activity patterns (Schaefer and Messier, 1996; Olson and Wallander, 2002), and orienting to the sun and wind (Gonyou and Stricklin, 1981).
Extrapolating energetic requirements from controlled to natural environments may lead producers to feed more than needed. Resting heat production of cattle increased by only one-third of predicted NRC (1981) maintenance energy requirements at –6°C and –15°C (Bergen et al., 2001). The NRC (1996) model overestimated NEm for feedlot steers in winter, underpredicting ADG when temperatures averaged –18°C (Block et al., 2001).
Thermal balance of cattle can be determined using a simple biophysical equation that incorporates metabolic heat production and all pathways of heat gains and losses (Campbell and Norman, 1998). This equation has been used to predict metabolic heat production in hot environments (Brosh et al., 1998; da Silva, 2000; Berman, 2004), but it has not been used to predict metabolic requirements of cattle during winter until recently (Keren and Olson, 2006). The objective was to apply this model to other situations.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
) and procedures were approved by Montana State University’s Animal Care and Use committee. The mechanistic model we evaluated and applied was developed (Keren and Olson, 2006) based on data from 12 pregnant Angus x Hereford cows, 3 or 8 yr old (6 in each group) grazing native rangeland, and experienced with winter at the Montana Agricultural Experiment Station’s Red Bluff Research Ranch near Norris, Montana (45°35'N; 111°39'W; elevation 1,600 m). The study site is sandy and silty range common to the foothills of the Northern Rockies. The dominant plant species are bluebunch wheatgrass (Agropyron spicatum) and Idaho fescue (Festuca idahoensis). Nutrient composition of the plants (DM basis) was 4.6% CP, 56% ADF, and 70% NDF. The gently sloping west-facing bench is exposed to prevailing south winds that keep the pasture snow-free most of the time. The 2 age groups grazed together as 1 group, were supplemented 3 times each week (1.8 kg/animal, 21% CP supplement), and had free access to mineral salt and water in two 570-L troughs on the east side of the pasture, which were kept ice-free with propane stock tank heaters. Supplements were provided on days when the cattle were not observed.
Cattle were observed every day between 0800 and 1700 for 3 d/wk from 28 November 2003 to 21 January 2004. Cows were weighed, body condition was scored (1 = thin, 9 = fat; Vizcarra and Wettemann, 1996), and backfat was measured by ultrasound (Aloka Model 200E, Wallingford, CT) at the beginning and end of the field study. To create realistic dimensions for the computer model, body dimensions (shoulder to tail length, shoulder height, chest girth) and hair length (undercoat and guard hair) were measured on all cows at the beginning of the study.
A weather station (Campbell Scientific, Inc., Logan, UT) was located near the south end of the pasture at the height of a cow’s spine (1.3 m). The station recorded net radiation, short-wave radiation, ambient temperature, humidity, and wind velocity and direction every 10 min throughout the study (Keren and Olson, 2006).
Model Development
From the thermal balance equation (Campbell and Norman, 1998; Eq. 12.1), latent heat loss, conductance, and storage were assumed to be negligible. Cattle do not sweat during winter, and heat loss via respired water vapor is also negligible (Diesel et al., 1990; Giesbrecht, 1995). Conductance of heat of cattle standing or walking on hard ground is only through the hoofs, and assumed to be near zero. Despite greater energetic costs, cattle stood rather than lay when the ground temperature was very cold (Keren and Olson, unpublished data), avoiding heat loss conducted to the ground. Further, cattle in good body condition are well insulated; thus, we never noticed melting associated with bed areas, indicating low, if any, conductance of heat from the animal to the snow.
By removing the components of thermal balance described in the preceding paragraph from the model (Campbell and Norman, 1998), we modified the remaining components to:
 | [1] |
in which R is all forms of short- and long-wave radiation absorbed by a body; hb is the coefficient for heat transfer through tissue, skin, and coat; body core temperature (Tb) is assumed to be relatively constant at 39°C (Kennedy et al., 2005; Lefcourt and Adams, 1998); Ts is surface temperature; 
s 
represents long-wave radiation emitted by a body that consists of surface emissivity (s), the Stephen-Boltzman constant (), and ambient temperature (Ta); and he is the environmental heat transfer coefficient of the surrounding boundary layer.
Equation 1 describes the gradient between a constant body core temperature and the environment, which drives heat flow through a series of layers of insulation and produces
 | [2] |
in which Mr (in W/m2) is the added heat required to maintain the balance described by Equation 1; and Tes is the standard operative temperature defined as the temperature of an enclosure with irradiative and free-convection conditions, in which heat loss for an animal of Tb is the same as in its natural environment (Bakken, 1992).
Campbell and Norman (1998) refer to Mr as the rate of metabolic heat production. The actual value of metabolic rate may be influenced by levels of activity and plane of nutrition but is not predicted by the heat transfer mechanistic model, in which a cow is treated as a 3-layer (tissue, coat, boundary layer) shell with no internal heat source. At high Tes, predicted values of Mr may be lower than the minimum basal rate physiologically possible (i.e., cattle are above their lower critical temperature). We did not observe adverse effects of cold exposure in our cattle during winter 2003 to 2004; the model predictions for metabolic requirements were never more than 180 W/m2 that winter. This does not approach the hypothetical maximum metabolic rate of 500 W/m2 (Campbell and Norman, 1998), and therefore an upper limit on metabolic responses to acute cold at the point of hypothermia could not be set in the model.
Model Analysis
The model was created and evaluated in R (Ihaka and Gentleman, 1996). We evaluated results from 2 model simulations. First, we predicted metabolic requirements based on the winter 2003 to 2004 data and compared subsets of this simulation’s predictions to independent data sets (simulation 1). Second, we simulated all possible permutations of 5 model variables: short-wave irradiance on a horizontal surface (Ih), ambient temperature (Ta), wind velocity (µ), orientation of the body’s primary axis to the sun’s azimuth (Øpa), and agreement or wind direction relative to the sun’s azimuth (Øws; simulation 2).
Simulation 1
To describe and clarify the effect of behavior and environmental variables on thermal balance, we analyzed model predictions using a mixed effects model (Pinheiro et al., 2004) on the group of 12 cows selected from the larger Red Bluff herd, with cow as the random variable.
During winter 2003 to 2004, we measured midbody surface temperatures (Ts, °C), with surface emissivity (s) = 0.97 (Campbell and Norman, 1998), on the left and right sides of each cow using a hand-held infrared thermometer (Model 08407-20, Cole Parmer, Vernon Hills, IL). These measures integrate a cow’s thermal environment into 1 variable (Ts), analogous to the calculated operative temperature (Keren and Olson, 2006). Replacing Tes in Eq. 2 with Ts removed a potential source of error from the Tes equation and increased the accuracy of model predictions.
We selected subsets of data from the winter 2003 to 2004 data to match the environmental conditions reported in 3 empirical studies (Young, 1975; Rutley and Hudson, 2000; Han et al., 2003). The studies were selected based on subjects, environment, and level of feeding (grazing or low quality hay, not feedlot) to match conditions from the database.
The model was designed to predict metabolic requirements based on complex scenarios of free-ranging cattle and therefore used variables such as time of day, long-wave radiation emitted from the ground, cloud cover, wind direction, and cattle activity and orientation. Other research has not described these variables in enough detail to use with the model.
Environmental variables are often highly correlated, and cattle respond differently to different combinations of these variables (Keren and Olson, 2006). Predicting metabolic requirements, holding many of the model variables constant at a mean level, would have been unrealistic because of inherent animal-to-animal variation. By using subsets of the winter 2003 to 2004 data to compare with the empirical studies, we accounted for correlations among variables and for potential responses of cattle under the conditions of their studies. Thus, we believe that we accurately simulated their environments despite the limited weather and animal behavior information.
Simulation 2
In the study, cattle were conditioned to different combinations of environmental variables that can be considered treatment cells (e.g., warm/windy/cloudy, cold/calm/clear skies). However, we had several missing cells of weather combinations and varying sample sizes because weather variables are correlated and do not adhere to a complete multifactorial design.
On overcast days, long-wave radiation that was emitted reradiates from the atmosphere back toward the ground, resulting in warmer ambient temperatures than on clear days. In the study area, strong winds are associated with an atmospheric pressure gradient formed between warm air in the intermountain area and arctic airflow from Canada. Cold days reduce this pressure gradient and are associated with light north winds (from the data, r = 0.47) compared with strong south winds on warm (>–10°C) days.
The large variation in the number of observations in different weather cells partly reflects the systematic sample design; cows were supplemented 3 times each week and responded to human activity near the water troughs, so that natural behavior in response to environmental conditions could not be observed on those days. Therefore, we observed animal behavior and measured surface temperature on the other days, regardless of the weather conditions that day. The observation days in winter 2003 to 2004 were relatively warm (Weatherbase, 2004), so most cells in the data represented mild weather conditions and only a few reflected extreme cold conditions.
We constructed a matrix of sky and ground conditions on a clear day at 1238 and simulated all possible permutations of 5 model variables at 5 levels. Generating a balanced matrix of weather variables for the second simulation was a 2-step process. First, we ran multiple simulations with different combinations of extreme weather variables (simulation 1). Second, we generated environmental variables in equal increments within the limits established from the first step (simulation 2).
Simulation 2 defined the interactions among the model variables, tested the boundaries of the model, and accounted for some of the missing cells in simulation 1. Simulation 2 allowed us to transform a complex model to a simplified polynomial regression model (Venables and Ripley, 2002) with only 5 variables. The best model was selected based on the results of simulation 1, on Akaike’s information criterion (lower is better), and by removing biologically irrelevant interactions among variables after reviewing the plotted data. This allowed us to achieve maximum information with minimum complexity.
|
RESULTS AND DISCUSSION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
Behavioral and environmental factors had varying effects on predicted requirements for heat production of the cattle under various climatic conditions (Table 1). We predicted requirements for different activities: lying, walking, grazing, or other, relative to standing (the intercept). Lying reduced metabolic requirements by 10% (Table 1). Based on a review of several studies, standing increased metabolic requirements of cattle 8 to 25% compared with when they are lying down (Susenbeth et al., 2004).
|
). The general shape of a cow is similar whether it is standing or lying; both were represented well by the ellipsoid (E. Keren, unpublished data). Thus, we used the same calculations for surface area exposed to wind or sun for a standing cow as for a cow lying down.
Metabolic requirements associated with grazing are not reported in Table 1
because they were similar to requirements for standing. Also, we did not model heat production as it fluctuates between different activities. Metabolic requirements from standing to the various activities (lying or walking) can be adjusted by using the appropriate estimates in Table 1. They are the result of applying simple regression to model predictions based on surface temperature of cattle during those times.
Short-wave radiation reduced (P < 0.001) metabolic requirements by 25 to 36% during the day. Average short wave radiation measured by the pyranometer was 184.2 W/m2 from 0800 to 1700, and 265 W/m2 between 1000 and 1500 when the pasture was exposed to direct solar radiation (Table 1). The positive coefficient for the temperature x short-wave interaction may reflect warm, sunny days when cattle were above their upper critical temperature, increasing metabolic requirements, e.g., panting (Renecker and Hudson, 1986).
Metabolic requirements were not directly affected by orientation to the sun but varied depending on short wave radiation (orientation x short wave radiation interaction, P < 0.001). Orienting perpendicular to the sun increases a cow’s surface area exposed to the direct beam component of short-wave radiation. Therefore, the thermal advantage of a perpendicular orientation was most pronounced on clear days when the direct beam component made up a larger percentage of the total incoming solar radiation, but this advantage varied some depending on an animal’s dimensions and time of day (Keren and Olson, 2006). In simulation 1, the relatively small effect that orientation had on metabolic requirements might reflect the few clear days when cows could experience high levels of direct short-wave radiation.
Orienting to the sun must also be evaluated in the context of a potential tradeoff when orienting to the wind. Wind velocity and radiation increase and decrease metabolic requirements, respectively. Often a particular orientation does not allow a cow to maximize heat gain and minimize heat loss simultaneously. The angle between wind direction and the azimuth can be used as a reference to evaluate this tradeoff between these 2 opposing factors (Keren and Olson, 2006). Wind blowing from the same direction as the sun’s azimuth is defined as low agreement because at any orientation a cow must trade off minimizing convective heat loss or maximizing irradiative heat gain. A 90° angle between the 2 variables is defined as high agreement because cows orienting in this way minimize convective heat loss while maximizing heat gain.
The effect of orientation may have been lower because of low agreement. Wind direction was fixed at 180° at all times; therefore agreement ranged only from 0° to 60° and was low (less than 45°) during midday when the effect of radiation was most pronounced.
Wind velocity influenced the cows less than short-wave radiation (Table 1). Average wind velocity at the study site was 6 m/s, with a maximum of 16 m/s. Relative humidity between the first and fourth quartiles ranged from 43 to 66%. The lower impact of wind may have been the result of the dry winter air in Montana. Wind velocity may contribute to greater heat loss under humid conditions, i.e., high water vapor density in the air. Water droplets condense on an animal’s surface under such conditions. High wind velocity would enhance the evaporation of this water, thereby cooling the surface.
Besides altering orientation relative to general environmental conditions, 2 windbreaks on the pasture provided an opportunity for cattle to avoid high wind velocities. However, cattle used the windbreaks during the day only twice but were lying or standing behind the windbreak almost every morning as the observer arrived at dawn (0800).
Cattle faced away from strong winds on cold days. Resistance to heat transfer of fur pelts mounted on a flat heated plate parallel to the direction of low-velocity airflow was 5% greater than those mounted perpendicular to airflow (Cena and Monteith, 1975; Campbell et al., 1980). In a 3-dimensional simulation, heat loss was greater for a cow orienting parallel to 1 to 4 m/s wind than when orienting perpendicular (Wu and Gebremedhin, 2001). During the 2003 to 2004 winter, wind did not seem to be an important factor influencing orientation of the cattle, except when loose snow on the ground was lifted by strong gusts of wind (E. Keren, personal observation).
The model settles this apparent disagreement. At low wind speeds, a perpendicular orientation reduces overall coat conductance because the leeward side is completely protected. With strong perpendicular winds, coat insulation is disturbed (Ames and Insley, 1975); whole-body conductivity is lower with a parallel orientation. In the model, the inflexion point was at approximately 5.2 m/s wind (Figure 1). This value will vary depending on hair characteristics, animal shape, ambient temperature, and the irradiative environment.
|
). Resting metabolic rates of these 422- to 491-kg heifers were 4 kcal·h–1·kg–0.75 at mild temperatures (–1 to 3°C), and 6 kcal·h–1·kg–0.75 at cold temperatures (–24 to –33°C). Adjusted to their average body mass (456 kg), the model predicted metabolic requirements of 3.3 kcal·h–1·kg0.75 under mild conditions and 4.6 kcal·h–1·kg0.75 under cold conditions
The model predicted that metabolic requirements of the lightest cow (480 kg) were consistently 20% greater than the average metabolic requirements for all 12 cows, on a body mass basis. Predicted requirements were 10% lower for the heaviest cow (757 kg) than the average. Smaller animals, with greater surface area to mass ratios, have greater energetic requirements (Willmer et al., 2000). After adjusting the model predictions to account for Young’s (1975) lower body mass, the predicted requirements were 4.8 and 5.7 kcal·h–1·kg0.75, which were somewhat greater and slightly lower, respectively, than Young’s measures in the environmental chamber.
Cattle in Young’s (1975) study were acclimated in an open pen, exposed to solar radiation and ambient temperatures of –24 to –30°C, and then brought into the metabolic chamber of –30°C for measurements. His reported monthly ambient temperatures do not account for solar radiation and the various ways that cows could conserve heat in the yard such as orientation or huddling with other animals. Even for an individual acclimated to cold, his measured values presumably represent a response to an acute cold incident, not basal metabolic rate, because certain behavioral responses are not possible in an environmental chamber. This may explain his slightly greater values.
Second, estimates of the metabolic requirements of free-ranging plains bison (Bison bison) were determined over a 2-yr period with 2 methods (Rutley and Hudson, 2000). Estimates of DM intake were 4.5 to 5 times greater based on the product of bite rate, bite size, foraging intensity, and feeding time compared with intake based on a chromium marker. This difference caused large variation when estimating MEm. Requirements determined by the marker method were 2.7 ± 0.7 kcal·h–1·kg–0.75 in December 1994, and 2.6 ± 0.2 kcal·h–1·kg–0.75 in December 1995; requirements based on the product of bite rate, bite size, foraging intensity, and feeding time were 8.1 ± 0.62 kcal·h–1·kg–0.75 in December 1994. They did not report estimates of intake based on bite rate, bite size, foraging intensity, and feeding time in December 1995.
Rutley and Hudson (2000) thought they underestimated MEm with the chromium marker and that the true value was somewhere between the 2 methods. Differences between methods were consistent across seasons. Estimates based on the marker were also lower than apparent MEm, determined from actual feed consumption in pen trials. Using their reported body mass, the model predicted 2.6 kcal·h–1·kg–0.75 for their 1994 study, and 2.9 kcal·h–1·kg–0.75 for their 1995 study, similar to their marker estimates. The predictions indicate that the chromium marker method did not underestimate requirements during winter and that the true value was close to the chromium marker value.
Third, cattle (Bos taurus) grazing on the Tibetan Plateau (ambient temperature in January ranges from 0 to –30°C), supplemented with a maintenance ration, were fasted for 4 to 7 d before fasting heat production was measured with a respiratory mask (Han et al., 2003). At an ambient temperature of –5°C and a wind velocity of 2.5 m/s, fasting heat production was 2.5 kcal·h–1·kg–0.75. Under these conditions, the model predicted similar values for cows weighing 300 kg (the average mass of their mature cattle). Predicted heat production was 2.6 kcal·h–1·kg–0.75 at a mean ambient temperature of –5°C and a mean wind velocity of 2.5 m/s.
The predicted metabolic requirements for the fasted cattle measured by Han et al. (2003) were closer than the predicted values for those measured in the 2 Canadian studies. This may be attributed to the more detailed description of weather conditions provided by Han et al. (2003) that allowed us to match a relatively accurate subset of weather conditions.
At an ambient temperature of –15°C and a wind velocity of 5.2 m/s, Han et al. (2003) measured fasting heat production at 3 kcal·h–1·kg–0.75. Because of site conditions, we could not exactly match their conditions. The predicted heat production was 3 kcal·h–1·kg–0.75 at mean ambient temperature of –15°C and wind velocity of 3.3 m/s, and 3.3 kcal·h–1·kg–0.75 at an ambient temperature of –13.6°C and a wind velocity of 5 m/s. At an ambient temperature of –12.9°C and a wind velocity of 5.5 m/s, predicted heat production was 3.7 kcal·h–1·kg–0.75.
The cows were affected by wind more than cattle described by Han et al. (2003), which have heavier coats than other breeds of cattle. Therefore, their cattle may be more insulated than our cattle and thus not as affected by wind velocity.
On cold clear days, cattle used by Han et al. (2003) may have absorbed more solar radiation (Han et al., 2002) because the ratio of direct to diffuse components of short-wave radiation is greater at lower latitudes (36° N at their site compared with 45° N at the site). Further, the higher elevation of those study sites may have contributed to clearer skies.
Simulation 2
The first run of simulation 2 was used to detect possible limitations of the model. As a reference point for the 3 environmental variables, we used 1 SD lower than the mean, the mean, and 3 SD greater than the mean weather variables from the winter 2003–2004 data. All permutations of temperature, wind, radiation, and orientation were repeated with wind direction at 0° and 90° relative to the sun, which correspond to disagreement and agreement of wind and radiation, respectively.
The model evaluated the amount of radiation absorbed by the surface of a cow according to its orientation relative to the sun’s elevation and azimuth (Keren and Olson, 2006). If sin–1 of the total short-wave irradiance on a horizontal surface (Ih) exceeds the solar constant at the zenith, the model returns a negative surface area value. At the date and time (21 January, 1238) chosen for this simulation, 569 W/m2 was the maximum value possible for Ih.
In the model, metabolic requirements increased linearly as ambient temperatures decreased to –60°C, which is the lower temperature limit of the model. At this extremely cold temperature and below, average metabolic requirements exceeded 400 W/m2, reaching values of 1,200 W/m2. More importantly, the relationship between ambient temperature, wind velocity, and radiation did not make sense at these extremes; requirements increased due to greater radiation loads and lower wind velocities.
The effect of wind velocity on metabolic requirements and the way that wind interacted with other variables was constant to velocities of 50 m/s. Thus, the model apparently predicted response to extreme wind velocities.
Once these limits were established, a 4 (variables) x 5 (levels) matrix of equal increments was expanded to include all possible permutations of environmental factors and orientation. Levels of the different variables were: 0, 142, 285, 427, 570 W/m2 for short-wave radiation; –5, –10, –15, –20, –25°C for ambient temperature; 0, 4, 8, 12, 16 m/s for wind velocity; and 0°, 22.5°, 45°, 67.5°, 90° for orientation relative to the sun.
Metabolic requirements were best described as a function of all 5 variables in which short-wave radiation, temperature, and wind were third order polynomials (Table 2). The effect that environmental variables had on metabolic requirements in simulation 2 was rather intuitive (Figures 2 and 3), and similar to those discussed in simulation 1, except for orientation and agreement.
|
|
|
). In the completely balanced design of these data, agreement reduced metabolic requirements somewhat (P = 0.10). However, agreement was important biologically because it represented surface area exposed to the wind in the context of tradeoffs with radiation (Figure 4). Orientation to the wind is not an independent variable in the model but is reflected in the relationship between orientation relative to the azimuth, wind velocity, and agreement.
|
). This range of values brackets estimates of lower critical temperature for beef cattle (–23°C; Webster, 1974). The –30°C low before metabolic requirements are predicted to increase may reflect that culling cows that fail to rebreed after winter selects cattle that tolerate the winters in Montana.
|
indicates metabolic requirements when there is agreement, whereas Figure 6b indicates requirements when there is no agreement. A parallel orientation to the wind lowers metabolic requirements compared with other orientations at wind speeds greater than 4 m/s (Figure 6a). Solar radiation contributes to this difference (Figure 6a–c) because a parallel orientation to the wind is a perpendicular orientation to the sun when in agreement. The advantage of a parallel orientation to the wind over a perpendicular one is reduced (Figure 6b–d) with low agreement.
|
). At –15°C, orienting 0° (parallel) to the wind direction does not have a thermal advantage over orientations that increase surface area exposed to 285 W/m2 of short-wave radiation. But at wind velocities greater than 11 m/s (the inflection point – e), orienting to minimize convective heat losses becomes thermally advantageous. |
IMPLICATIONS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
|
Footnotes |
|---|
2 Corresponding author: bolson{at}montana.edu
Received for publication June 15, 2005. Accepted for publication December 22, 2005.
|
LITERATURE CITED |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONIMPLICATIONSLITERATURE CITED |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
