It is not the temperature that kills.It is how fast you got there.

Heat management in a broiler house is usually written as a set-point problem. Keep the index below a number and the birds are fine, cross it and they are not. There is a published broiler threshold and it is worth knowing. But the most operationally useful result in the heat-stress literature is not about the level at all. It is about how quickly the house got there, and it implies that a warning system watching only for a line being crossed will be late on exactly the days that matter.

The one commercial broiler threshold

Vale, Moura, Naas and Pereira published a study in 2010 based on 20 straight-run flocks in 14 commercial poultry houses in Sao Paulo state, Brazil, over the 2007 to 2008 summer, at a stocking density of 13 birds per square metre. Using a temperature-humidity index weighted 75 percent dry bulb and 25 percent wet bulb, they found that a maximum index above 30.6 degrees Celsius was the necessary condition for a high-mortality day, which they defined as 0.2 percent or more per day, twice the mortality of a normal day. Below that index, no high-mortality day occurred. At an index of 30.6, maximum air temperature is around 31 degrees.

The refinements are age-conditional. Birds between 29 and 31 days showed no effect. Between 31 and 40 days, high mortality required both the index above 30.6 and a maximum temperature above 34.4 degrees. Above 40 days, it required the index above 30.6 combined either with a minimum index at or below 15.5, or with a minimum index above 15.5 and the daily maximum temperature occurring later than three in the afternoon. The timing of the peak, in other words, is part of the rule.

The performance figures deserve the same prominence as the threshold. The classifier reached 89.34 percent overall accuracy, but precision on the high-mortality class was 0.73. Roughly a quarter of the warnings it issued would be wrong. This is also a single country, a single season, 20 flocks, and it should never be presented as globally validated. A competing published broiler threshold of an index of 28 degrees for 40-day-old birds appears in the layer literature, so even the number itself is not settled.

The rate of rise

Kang and colleagues, in Frontiers in Veterinary Science in 2020, ran the experiment that changes how this should be read. Working with 70-week-old Hy-Line Brown laying hens in a windowless commercial layer facility in South Korea, 24 birds per treatment against 12 controls, they raised the temperature-humidity index at three different speeds.

Fast: from 24.2 to 32.1 degrees in one hour, then held for four and a half hours. No mortality at one hour, more than 95 percent by five hours, 100 percent at five and a half. Intermediate: from 25.2 to 34.3 over three hours. Twenty-nine percent at four hours, 75 percent at five, 79 percent at eight. Slow: to 31.2 degrees over six hours, then held for three. No mortality at all.

An index of 31.2 reached over six hours killed nobody. An index of 32.1 reached in one hour killed more than 95 percent in five.

The slow condition ended up above the Brazilian broiler threshold and produced nothing. The fast condition was barely a degree higher and produced near-total loss. The authors state the conclusion directly: the faster and higher the increase in the index, the more serious the mortality. Panting tracked it, with no panting at an index of 25.3, 40 percent of hens panting at 29, all hens panting above 30 at more than 200 counts per minute, and more than 250 counts per minute alongside 60 percent mortality at 280 minutes.

Three caveats before this is carried across. These are laying hens, not broilers. They are 70 weeks old, which is a very different animal from a 40-day-old broiler in body mass, feathering and metabolic rate. And it is one facility with 24 birds per treatment. The direction of the finding is strong and the magnitude should not be transplanted.

How much heat stress costs when it does not kill

A peer-reviewed systematic review and meta-analysis by Liu and colleagues in Poultry Science in 2020 pooled 12 studies covering 470 broiler chickens, screened from 4,107 records. The mortality odds ratio under heat stress was 3.74, with a 95 percent confidence interval from 1.39 to 10.12, across eight trials with no detected heterogeneity. Feed intake fell by a weighted mean of 97.95 grams, interval 141.70 to 54.20 grams. Body weight gain fell by 151.40 grams, interval 198.59 to 104.21. Feed conversion worsened by 0.17 after sensitivity analysis, though the initial pooled estimate of 0.12 was not statistically significant.

The confidence interval on the mortality odds ratio spans an order of magnitude, and the whole meta-analysis is built on experimental studies rather than commercial houses. The performance figures are the more transferable part: roughly 100 grams of feed and 150 grams of gain, per bird, in flocks that survived. Multiply that across a house and the sub-lethal cost of a heat event is larger than the pile at the door.

The coefficient nobody has settled

If you are going to watch water during hot weather, you need to know how much water a degree is worth, and the published answer depends on who you ask. A University of Georgia extension bulletin states that broiler water consumption increases approximately 7 percent for each degree Fahrenheit. A doctoral dissertation attributes to the NRC an increase of 7 percent for every 1 degree Celsius above 21 degrees. An Australian industry report, citing Manning and colleagues in 2007, gives 6 percent per degree Celsius above 20 degrees, with feed intake falling 1.23 percent per degree. A fourth extension source restates the 7 percent figure without specifying the unit at all.

Seven percent per degree Fahrenheit is about 12.6 percent per degree Celsius. The spread between the highest and lowest published reading is therefore about a factor of two. We are not going to pick one and present it as settled, because the sources are extension bulletins, a dissertation and an industry report rather than a primary measurement that resolves the discrepancy. Anyone setting a water band from the wrong coefficient will set it wrong by a large margin, and that is worth saying out loud rather than quietly averaging.

Where the lead time comes from

There is no published day-scale lead time between an environmental crossing and a broiler mortality response. The Brazilian work says the effect of a heat wave lasted an average of 2.7 days across events of one to five days, and that a single heat-shock day is capable of producing high broiler mortality. The Korean work says that inside a house the response unfolds over hours. Neither gives you a day of warning from the bird.

The warning has to come from the forecast, and that approach is validated in another species. USDA ARS runs a heat-stress forecast driven by seven-day National Weather Service predictions of temperature, humidity, wind speed and cloud cover, feeding a breathing-rate equation with Normal, Alert, Danger and Emergency bands, and states the equation has been validated by subsequent studies. That system is for feedlot cattle, not poultry. The nearest poultry equivalent, a Wageningen project, describes itself as developing a proof-of-principle for a real-time early warning system and a prototype cooling system, and its page makes no validation claim.

So the honest position is this. A forecast plus a rate-of-rise rule is better founded than a static temperature threshold, and it is an extension of an approach validated in cattle rather than a validated poultry system. Losses on the scale trade press has reported this year, with French industry bodies estimating 500,000 to 700,000 laying hens and 2.5 to 3 million broilers lost in one late-June event, are estimates rather than official statistics, and the same reporting notes no national assessment has been published. That is all the more reason to build the warning on the day that is coming rather than the reading that has already happened.