GxE

Estimating genotype-by-THI interactions in laying hens using factor analytic models and egg production data

---

Abstract

Climate change results in more frequent and prolonged periods of high temperatures. Laying hens may be particularly sensitive to heat stress due to their high metabolic rates and inability to sweat. Furthermore, some families may be more sensitive to temperature than others, which would suggest genotype by temperature (GxT) interaction. The extent of GxT interactions can be assessed using random regression models, which describe genetic (co)variances as continuous functions of temperature and enable estimation of correlations between any pair of temperatures. However, when data are sparse at temperature extremes, these models may yield unreliable estimates due to extrapolation of regression trends. In contrast, factor analytic models represent the genetic (co)variance structure across temperatures through a small number of latent factors, capturing GxT patterns more directly and robustly with fewer parameters. The objective of this study was to estimate genetic correlations of laying performance across different temperatures in laying hens using factor analytic models. The dataset included cage-level laying rate (number of eggs / number of hens in the cage) of Brown laying hens from three commercial farms in the Netherlands and one in Belgium, each contributing two to three successive production batches. Each cage consisted of offspring from the same sire. Although data were collected during the full production cycle, we used only the period where production levels are relatively stable (day 99 to 286 since start of lay). We obtained weather data from the nearest stations and created 8 classes based on the temperature-humidity index (THI), which ranged from 25.1 to 73.6. The total number of records was 3,585,019, collected from 21,370 cages with offspring from 3,590 sires. Data were analyzed using a factor analytic model with 6 latent factors to estimate permanent environmental and genetic correlations across all THI classes. The model included a fixed effect for farm, dam line, number of days since start of lay, and THI class. Estimated genetic correlations ranged from 0.53 to 0.99 (Table 1), with standard errors all below 0.05. Genetic correlations were generally lower between more distinct THI classes. Given the limited range of THI values, these results suggest strong GxT interaction for laying rate in laying hens. In our extended conference paper, we will 1) present a more detailed methodology, including a sequential modelling approach where latent factors were added one at a time, with estimates from simpler models serving as starting values for more complex ones, 2) report heritabilities and variance explained by permanent environmental effects, and 3) report the permanent environmental correlations across THI classes.

Keywords: 2026

How to Cite:

Duenk, P., Büttgen, L., Visscher, J., van der Sluis, M., Ellen, E. & Veerkamp, R., (2026) “Estimating genotype-by-THI interactions in laying hens using factor analytic models and egg production data”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2284545. doi: https://doi.org/10.31274/wcgalp.23586

Rights: 1