- Email: [email protected]
Relation between tonic immobility and production estimated by factorial correspondence analysis in Japanese quail
Relation Between Tonic Immobility and Production Estimated by Factorial Correspondence Analysis in Japanese Quail S. Mignon-Grasteau* and F. Minvielle†,1 *Station de Recherches Avicoles, INRA, 37380 Nouzilly, France; and †UMR Ge´ne´tique et Diversite´ Animales, INRA-INAPG, 78352 Jouy-en-Josas, France ABSTRACT Duration of tonic immobility (TI) is a wellestablished criterion of fearfulness in birds, but its relation to production traits has been little studied. TI and 27 variables related to egg production and quality, feed consumption, and body condition were collected for 355 female Japanese quail. Experimental birds were the F2 generation from a reciprocal cross between a line selected for long TI and a line from a different origin and selected for early egg production. TI and several other variables had very asymetrical distributions so all variables were
categorized, and the association between traits was studied by factorial correspondence analysis (FCA). Attributes of good versus poor layers were well separated by the FCA, but no association was obtained between TI and other traits on the primary axes of the FCA. Fearfulness, as measured by TI, was not related to levels of performance of egg laying quail. Secondary negative associations between TI and body temperature and between TI and residual feed consumption were obtained; however, they were marginal and need to be confirmed.
(Key words: quail, fear, behavior, egg, feed intake) 2003 Poultry Science 82:1839–1844
INTRODUCTION Research on the welfare of domesticated birds raised for meat and egg production has identified a number of traits related in one way or another to the well-being of birds. Fearfulness is one of them, and it has been greatly studied, especially in experiments with fear-related traits, such as open-field activity (Jones and Faure, 1982; Jones et al., 1992) and tonic immobility (TI; Benoff and Siegel, 1976; Gallup, 1979; Jones, 1986). TI is a passive reaction of the birds that freeze after being exposed to severe stress. It is measured easily in chickens and quail as early as 1 wk after hatching and clearly has a genetic foundation, as divergent lines showing little or extreme fear (Mills and Faure, 1991) have been developed by artificial selection on the duration of TI in Japanese quail. Consequently, TI has some appeal as a possible representative behavioral trait to be included in the selection index of poultry. The relation between measures of fearfulness, and, more generally, between welfare-related traits has been widely studied (Mills and Faure, 1986; Mignon-Grasteau et al., 2003). On the contrary, experimental evidence for the existence of a relation between measures of well-being and production is still lacking even though both are thought to
2003 Poultry Science Association, Inc. Received June 2, 2003. Accepted for publication August 19, 2003. 1 To whom correspondence should be addressed: minvielle@dga2. jouy.inra.fr.
be adversely affected by extreme fear (Jones and Hocking, 1999). In a recent study on feed conversion and behavior, however, the 20 best and 20 worst broilers from a randombred line showed similar median durations of TI (Skinner-Noble et al., 2003). Also, a preliminary intrapopulation analysis in a quail line selected for early egg production (Minvielle et al., 1999b, 2000b) did not show any association between TI and egg production (Minvielle et al., 2002). However, the sample size was small, the population used showed little variation for TI, and measuring the association through the calculation of linear correlations was approximate at best because normalizing the distribution of TI by the Box-Cox transformation (Besbes et al., 1993) was imperfect. The objective of the present study was to evaluate the extent of the association between TI and varied productionrelated traits, using Japanese quail as a model animal. It could not be achieved by simply comparing lines differing for TI or for production traits but needed to be done within a single population to avoid finding nonsense correlations. In order to have a large experimental group with a wide genetic base, we used F2 birds from an initial cross between a line selected for long duration of TI (LTI; Mills and Faure, 1991) and a line selected for early egg production (DD; Minvielle et al., 2000a). Abbreviation Key: AFE = age at first egg; BT = body temperature; DD = Japanese quail line selected for early egg production; EN = egg number; EW = egg weight; FCA = factorial correspondence analysis; FI = feed intake; LTI = Japanese quail line selected for a long duration of tonic immobility; RFI = residual feed intake; TI = tonic immobility.
1839
1840
MIGNON-GRASTEAU AND MINVIELLE
MATERIAL AND METHODS F2 Experimental Population Reciprocal crosses were carried out between Japanese quail from different families of the LTI and DD lines that had been compared for fearfulness and performance in a previous study (Minvielle et al., 2002). All 12 LTI breeders chosen to produce the F1 had the largest possible value of TI (300 s), and TI of their 12 DD mates did not exceed 55 s. Next, 1 male and 3 F1 females were randomly drawn from each F1 family. TI recorded for these birds varied between 31 and 300 s. Finally, 438 F2 birds (30 full-sib F2 families) were produced in 3 consecutive hatches by 10 single-pair matings of a F1 male with 3 full-sib F1 females from another family. Complete information with data on all the traits described above was only available for 355 F2 females (17 s ≤ TI ≤ 300 s) from the 30 F2 families and constituted the data set used in the present experiment.
Husbandry The experimental work was carried out at the Unite´ de Ge´ne´tique Factorielle Avicole, Nouzilly, France. Obtained in 3 successive hatches, chicks were reared in group cages for 5 wk. They received a mash starter diet (2,901 kcal ME/kg, 11.5% moisture, 7% ash, 27% total protein, 8% fat, and 4% crude fiber). Initial temperature in the group cages was 37°C, and it was decreased to 30°C by the end of the rearing period. At 5 wk of age, females were assigned randomly to individual cages of a 4-tier egg laying battery maintained at 25°C, where they remained until the end of the experiment. The quail received a commercial layer diet (2,709 kcal ME/kg, 11.5% moisture, 12% ash, 20% total protein, 4% fat, and 4% crude fiber) and were under artificial lighting for 14 h/d. Throughout the experiment, feed and water were available ad libitum.
Traits TI. At 9 or 10 d of age, the duration of TI was measured on each quail. It was defined as the length of time during which a chick remained immobile after it had been held on its back for 10 s to induce the freezing reaction. For practical reasons, the measure of TI was terminated if the bird had remained immobile for 5 min (TI = 300) or if it failed to remain at all immobile after 5 attempts at inducing TI (TI = 0). BW and Body Temperature. Birds were weighed at 5, 27, and 31 wk of age. Rectal body temperature (BT) of fasted birds was measured at 5 and 31 wk of age. Egg Production and Egg Quality. Individual egg production was recorded daily for 64 wk between the ages of 5 and 69 wk. Age at first egg (AFE), total egg number (EN), average clutch length, and the mean proportion of defective eggs (soft shell, broken, and double yolk) were obtained for each quail. Three normal eggs laid consecutively were collected around the age of 29 wk to estimate average egg weight (EW) and egg composition (shell, yolk, and albumen weights) for each quail.
FIGURE 1. Distribution of the duration of tonic immobility (TI).
Feed Consumption. When females reached 27 wk of age, a 24-d feed trial was started. In this trial, feed intake (FI) was measured and used to estimate residual FI (RFI), egg mass, mean BW, and BW gain (BWG) during the feed trial.
Statistical Analysis Egg Laying Curve. The egg laying curve of each quail was estimated by using the Wood equation: ENt = A tB e−Ct (Timmermans, 1973). In this equation, ENt is the number of eggs laid until time t (t = 2, 4,. . ., 64 wk), and A, B, and C are the parameters of the curve. Parameter A relates to the level of egg production in the first period of test, C is an indicator of the decline of egg production, and B/C estimates the time at peak egg production. RFI. The multiple regression FI = 108.95 + 5.53 mean BW0.75 + 1.99 BW gain + 0.877 egg mass was adjusted to individual measures of FI. The residuals of the regression line, adjusted for the effect of the hatch, were used to estimate RFI. Factorial Correspondance Analysis. Correspondance analysis is an inductive exploration of a data set for finding patterns in the data when several traits have been recorded for each individual (Benzecri, 1992). By reducing the number of primary traits into a smaller number of factors that are combinations of the original variables and that account for the most variance, the redundancy in the original data set is eliminated, and true associations between traits may be found. Two applications of correspondance analysis are available: principal component analysis for normally distributed traits and factorial correspondence analysis (FCA) for categorical traits. In the present work, the distribution of TI (Figure 1) was not readily transformable, and the distributions of several production traits were very asymetrical (Table 1) so a FCA was carried out on the data set with the SPAD 4.0 software (Lebart et al., 1999). First, raw values were corrected for the effects of the hatch (production-related traits) and technician (TI) by using, respectively, the GLM and NPAR1WAY procedures from the SAS program library (SAS Institute, 1988). Then, Spearman rank correlations were estimated between all variables taken 2 × 2, and for each pair of redundant variables (rs >
1841
TONIC IMMOBILITY AND PRODUCTION IN JAPANESE QUAIL TABLE 1. Elementary statistics on the 27 variables measured for the 355 Japanese egg laying quail and repartition of the data into low and high categories for the second factorial correspondance analysis Category Trait Tonic immobility, s BW at 35 d, g Body temperature at 35 d, °C Body temperature at 31 wk, °C Feed intake in 24 d, g Residual feed intake in 24 d, g BW gain in feed test, g Mean egg weight, g Mean yolk weight, g Mean albumen weight, g Mean shell weight, g Ratio yolk:albumen Age at first egg, d Total egg number Soft-shell eggs, % Broken eggs, % Double yolk eggs, % Defective eggs, % Clutch length, d Number of pauses Months in test at first egg, mo Maximum 1 mo egg number Age at peak egg production, mo A1 B1 C1 tinf2
Mean ± SD
Skewness
Kurtosis
Minimum
Maximum
Low if trait ≤
High if trait >
± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
0.13 0.25 0.28 0.32 0.10 0.84 −0.34 0.32 0.14 0.26 0.50 0.60 2.70 −1.40 9.24 3.89 8.49 5.58 4.02 1.77 9.64 −2.21 1.72 −0.76 2.60 3.42 0.40
−1.33 0.19 0.52 −0.42 0.05 1.79 0.66 0.34 0.08 0.49 0.98 1.37 13.75 1.84 110.0 21.32 102.8 45.63 31.29 3.02 91.63 16.10 2.51 6.35 10.57 17.41 0.66
17 147.5 40.7 40.9 453.4 −96.4 −41.5 9.07 2.73 5.14 0.66 0.43 37 112 0 0 0 0 2.1 0 1 18 2 2.904 0 0 1
300 238.8 43.0 42.4 858.0 183.3 19.4 15.27 5.15 9.30 1.25 0.78 79 415 29.4 11.4 4.8 11.3 47.9 70.5 2 30 15 35.23 3.035 0.986 11.85
100 180.4 — 41.40 — −34.4 — 11.23 — — — — — 320 — — — 0.2 — — — — — — — 0.05 —
259 199.2 — 41.84 — 34.2 — 12.69 — — — — — 370 — — — 0.5 — — — — — — — 0.10 —
178.9 189.0 41.66 41.60 658.8 0.0 −3.79 11.93 3.92 7.13 0.88 0.55 44.1 333.0 0.89 0.80 0.12 0.61 8.1 12.2 1.0 27.1 4.25 20.765 0.493 0.112 5.23
87.7 16.6 0.38 0.32 68.3 473.3 9.19 1.04 0.43 0.68 0.09 0.06 4.9 58.4 2.2 1.3 0.36 1.0 4.2 14.1 0.10 1.1 2.81 3.206 0.382 0.120 1.58
1 A, B, and C are coefficients of the Wood egg-laying curve: ENt = A tB e−Ct, where EN is the egg number, and t is the age expressed in 2-wk periods. 2 tinf is the age at peak egg production estimated from the Wood curve.
0.7), only 1 trait was included in the analysis. Finally, for each trait, the data were assigned to 1 of 3 categories (high, low, and mean) defined (Table 1) in such a way that each category was equally frequent. For 2 variables, only 2 classes could be defined. A first FCA analysis was performed with 14 nonredundant variables and explained only 55% of the total variation in the first best 10 axes. Consequently, a second FCA with a subset of 8 variables representing the different types of traits (fear, egg production, feed consumption, and egg quality), but with only one or two variables per type, was performed to obtain more significant FCA axes.
RESULTS AND DISCUSSION Briefly, in the first FCA (data not shown), axis 1 (accounting for 8.3% of the total variation) distinguished between poor (low EN and high C) and good (high EN and low C) layers where C is a coefficient of the Wood egg-laying curve. Axis 2 (7.4% of the total variation) compared late (older AFE and low A) and early (younger AFE and high A) layers, and axis 3 (6.4% of the total variation) distinguished between small females laying small eggs (low BW and low EW) and large females laying larger eggs (high BW and mean EW) where A is a coefficient of the Wood egg-laying curve. Results for these 3 axes were not surprising, because in the Wood model (Timmermans, 1973), C is an indicator of the decline of egg production and A
relates to the level of early egg production, and positive correlations between BW and egg size have been reported extensively in quail (e.g., Sefton and Siegel, 1974; Strong et al., 1978; Minvielle et al., 1999a). TI appeared as the predominant variable for the first time in the eighth axis (4.4% of the total variation) and seemed to explain only a minor part of the total variability. In the second analysis, the first 5 axes accounted for 45% of the total variation, and for each axis, the variables explaining a major part of the variability are listed with their contribution to the axis (Table 2). In Figure 2, axis 1 distinguished between poor (low EN and high C) and good (high EN and low C) layers, as in the previous analysis, and an association (inside the ellipses) among RFI, BW, and EW could be observed 3 times (low BW with low EW and mean RFI, mean BW with mean EW and low RFI, and high BW with high EW and high RFI). It indicated that the average-sized, egg-laying F2 quail used feed most efficiently, which is in partial agreement with the small negative correlations between BW and RFI reported in chickens (Bordas and Me´rat, 1981; Luiting and Urff, 1991; TixierBoichard et al., 1995). TI contributed only 1.2% to the variation of FCA axis 1 and remained close to the origins of the axes in Figure 2. A better spread of the 3 classes of TI was obtained with secondary axes 3 and 4 in Figure 3, in which a 3-variable association (inside the ellipses) among TI, defective eggs, and BT could be visualized 3 times. A similar but indirect
1842
MIGNON-GRASTEAU AND MINVIELLE TABLE 2. Contributions of the first 5 axes of the factorial correspondence analysis (FCA), and of the main variables1 on each axis, to the total variation Axis Rank 1 2 3 4 5
Variation (%)
List of the main2 variables (% contribution to the axis)
12.5 9.2 8.3 7.7 7.2
C (36.7); EN (36.2); DEF (17.0) EN (22.8); BW (19.6); RFI (18.5); C (18.2); EW (16.8) BW (27.7); EW (25.3); EN (10.5%) TI (23.0); BW (19.0); DEF (15.7); RFI (12.9); BT (12.5); C (11.6) BT (36.3); TI (20.0); EW (19.3)
1 C = coefficient of the Wood egg-laying curve; EN = total egg number; DEF = percentage of defective eggs; RFI = residual feed intake; EW = mean egg weight; TI = duration of tonic immobility; and BT = body temperature at 31 wk. 2 Explaining more than 10% of the variance of FCA axis.
relation between stress and BT was reported previously by de Jong et al. (2002) who observed lower BT in restricted-fed, and therefore chronically-stressed, broiler chickens than in control birds, but their observation resulted likely from the added effects of stress and lower FI. There was also some indication of a negative association between RFI and extreme classes of TI, maybe in relation to the stress susceptibility, as evaluated by TI, of high emotivity quail. Indeed, in chickens, birds with low RFI are less active physically than high RFI birds (Katle et al., 1984; Gabarrou et al., 1998). Finally, the widest spread of F2 quail for TI was obtained with axes 4 and 5 (Figure 4). All other traits, however, were located away from TI in this figure, and no association was visible. Concomitant differences in TI and performances have been reported for several line comparisons (Mills et al., 1994; Jones et al., 1997; Minvielle et al., 2002), and they have
been sometimes interpreted as evidence for the existence of a relation between those traits (Jones and Hocking, 1999). Associations between traits, however, can only be tested on animals with similar genetic backgound. Overall, the present results obtained within an F2 population of Japanese quail showed that TI measured early in life was not strongly associated with any of the main poultry production traits, in agreement with the preliminary indications reported by Minvielle et al. (2002). Using egg shell quality as an example, it has been proposed that stress might be needed to trigger the association with TI (Mills et al., 1994). In the present experiment, no special stress was imposed, but laying quail were kept in an environment (lodging, feed, and contact with man) similar to that of commercial quail egg production in France. Consequently, including TI as an overall indicator of fearfulness in a selection programme should have little impact on the genetics of the
FIGURE 2. Spread of the variables according to the first 2 axes of the factorial correspondance analysis. Filled diamond = tonic immobility (TI); filled square = BW; open triangle = body temperature (BT); circle = residual feed intake (RFI); multiplication sign = egg weight (EW); addition sign = egg number (EN); double cross = coefficient of the Wood egg-laying curve (C); and open diamond = defective eggs (DEF). The size of the symbols increases with their mean contribution to the 2 axes. Consistent associations among traits across categories are circled.
TONIC IMMOBILITY AND PRODUCTION IN JAPANESE QUAIL
1843
FIGURE 3. Spread of the variables according to the axes 3 and 4 of the factorial correspondance analysis. Filled diamond = tonic immobility (TI); filled square = BW; open triangle = body temperature (BT); circle = residual feed intake (RFI); multiplication sign = egg weight (EW); addition sign = egg number (EN); double cross = parameter of the Wood egg-laying curve (C); and open diamond = defective eggs (DEF). The size of the symbols increases with their mean contribution to the 2 axes. Consistent associations between traits across categories are circled.
production traits, at least in a well-managed production unit because the level of fear of humans might be related negatively with egg production (Hemsworth and Barnett, 1989). Interestingly, a similar experiment with TI and fear-
related behavior traits (Mignon-Grasteau et al., 2003) has also shown that TI was not associated with the other traits. Consequently, TI might not be a good choice as a unique variable of stress response to be included among the selec-
FIGURE 4. Spread of the variables according to the axes 4 and 5 of the factorial correspondance analysis. Filled diamond = tonic immobility (TI); filled square = BW; open triangle = body temperature (BT); circle = residual feed intake (RFI); multiplication sign = egg weight (EW); addition sign = egg number (EN); double cross = parameter of the Wood egg-laying curve (C); and open diamond = defective eggs (DEF). The size of the symbols increases with their mean contribution to the 2 axes.
1844
MIGNON-GRASTEAU AND MINVIELLE
tion objectives, and the association between stress response and production will need to be studied further by using several other stress-related traits.
ACKNOWLEDGMENTS The technical assistance of Jean-Louis Monvoisin (INRA, Jouy-en-Josas), David Gourichon and Chantal Moussu (INRA, Nouzilly) is gratefully acknowledged. The authors thank Jean-Michel Faure and Andrew Mills (INRA, Nouzilly) for kindly providing quail from their LTI line to start this experiment.
REFERENCES Benoff, F. H., and P. B. Siegel. 1976. Genetic analyses of tonic immobility in young Japanese quail (Coturnix coturnix japonica). Anim. Learn. Behav. 4:160–162. Benzecri, J. P. 1992. Correspondance Analysis Handbook, Dekker, New-York. Besbes, B., V. Ducrocq, J. L. Foulley, M. Protais, A. Tavernier, M. Tixier-Boichard, and C. Beaumont. 1993. Box-Cox transformation of egg-production traits of laying hens to improve genetic parameter estimation and breeding evaluation. Livest. Prod. Sci. 33:313–326. Bordas, A., and P. Me´rat. 1981. Genetic variation and phenotypic correlations of food consumption of laying hens corrected for body weight and production. Br. Poult. Sci. 22:25–33. de Jong, I. C., S. van Voorst, D. A. Ehlhardt, and H. J. Blokhuis. 2002. Effects of restricted feeding on physiological stress parameters in growing broiler breeders. Br. Poult. Sci. 43:157– 168. Gabarrou, J. F., P. A. Geraert, N. Franc¸ois, S. Guillaumin, M. Picard, and A. Bordas. 1998. Energy balance of laying hens selected on residual food consumption. Br. Poult. Sci. 39:79–89. Gallup, G. G., Jr. 1979. Tonic immobility as a measure of fear in domestic fowl. Anim. Behav. 27:316–317. Hemsworth, P. H., and J. L. Barnett. 1989. Relationships between fear of humans, productivity and cage position of laying hens. Br. Poult. Sci. 30:505–518. Jones, R. B. 1986. The tonic immobility reaction of the domestic fowl: A review. World’s Poult. Sci. J. 42:82–96. Jones, R. B., and J. M. Faure. 1982. Open-field behaviour of male and female domestic chicks as a function of housing conditions, test situations, and novelty. Biol. Behav. 7:17–25. Jones, R. B., and P. M. Hocking. 1999. Genetic selection for poultry behaviour: big bad wolf or friend in need? Anim. Welfare 8:343–359. Jones, R. B., D. G. Satterlee, and H. L. Marks. 1997. Fear-related behaviour in Japanese quail divergently selected for body weight. Appl. Anim. Behav. Sci. 52:87–98. Jones, R. B., D. G. Satterlee, and F. H. Ryder. 1992. Open-field behavior of Japanese quail chicks genetically selected for low or high plasma corticosterone response to immobilization stress. Poult. Sci. 71:1403–1407. Katle, J., H. B. Bentsen, and B. O. Braastad. 1984. Correlated traits with residual feed consumption. Pages 136–138 in Proceedings of the 17th World’s Poultry Congress and Exhibition, Helsinki, Finland. Lebart, L., A. Morineau, T. Lambert, and P. Pleuvret. 1999. Syste`me SPAD. CISIA-CERESTA, Montreuil, France.
Luiting, P., and E. M. Urff. 1991. Residual feed consumption in laying hens: 2. Genetic variation and correlations. Poult. Sci. 70:1663–1672. Mignon-Grasteau, S., O. Roussot, C. Delaby, J. M. Faure, A. Mills, C. Leterrier, D. Gue´mene´, P. Constantin, M. Mills, G. Lepape, and C. Beaumont. 2003. Factorial correspondence analysis of fear-related behaviour in Japanese quail. Behav. Proc. 61:69–75. Mills, A. D., and J. M. Faure. 1986. The estimation of fear in domestic quail: correlations between various methods and measures. Biol. Behav. 11:235–243. Mills, A. D., and J. M. Faure. 1991. Divergent selection for duration of tonic immobility and social reinstatement behavior in Japanese quail (Coturnix coturnix japonica) chicks. J. Comp. Psychol. 105:25–38. Mills, A. D., K. M. Herron, M. M. Bain, S. E. Solomon, and J. Faure. 1994. Eggshell quality in Japanese quail Coturnix Japonica genetically selected for high or low levels of fearfulness. Pages 292–293 in Proceedings of the 9th European Poultry Conference, Glasgow, UK. Minvielle, F., J. L. Coville, A. Krupa, J. L. Monvoisin, Y. Maeda, and S. Okamoto. 2000a. Genetic similarity and relationships of DNA fingerprints with performance and with heterosis in Japanese quail lines from two origins and under reciprocal recurrent or within-line selection for early egg production. Genet. Sel. Evol. 32:289–302. Minvielle, F., E. Hirigoyen, and M. Boulay. 1999a. Associated effects of the roux plumage color mutation on growth, carcass traits, egg production, and reproduction of Japanese quail. Poult. Sci. 78:1479–1484. Minvielle, F., A. D. Mills, J. M. Faure, J. L. Monvoisin, and D. Gourichon. 2002. Fearfulness and performance related traits in selected lines of Japanese quail (Coturnix japonica). Poult. Sci. 81:321–326. Minvielle, F., J. L. Monvoisin, J. Costa, A. Frenot, and Y. Maeda. 1999b. Changes in heterosis under within-line selection or reciprocal recurrent selection: an experiment on early egg production in Japanese quail. J. Anim. Breed. Genet. 116:363–377. Minvielle, F., J. L. Monvoisin, J. Costa, and Y. Maeda. 2000b. Long-term egg production and heterosis in quail lines after within-line or reciprocal recurrent selection for high early egg production. Br. Poult. Sci. 41:150–157. SAS Institute. 1988. SAS User’s Guide: Statistics. SAS Institute Inc., Cary, NC. Sefton, A. E., and P. B. Siegel. 1974. Inheritance of body weight in Japanese quail. Poult. Sci. 53:1597–1603. Skinner-Noble, D. O., R. B. Jones, and R. G. Teeter. 2003. Components of feed efficiency in broiler breeding stocks: is improved feed conversion associated with increased docility and lethargy in broilers? Poult. Sci. 82:532–537. Strong, C. F., Jr., K. E. Nestor, and W. L. Bacon. 1978. Inheritance of egg production, egg weight, body weight and certain plasma constituents in Coturnix. Poult. Sci. 57:1–9. Timmermans, M. P. F. C. A. 1973. The statistical and genetical significance of the application of mathematical models to explain egg production curves in poultry. Arch. Geflu¨gelkd. 37:37–45. Tixier-Boichard, M., D. Boichard, E. Groenveld, and A. Bordas. 1995. Restricted maximum likelihood estimates of genetic parameters of adult male and female Rhode Island Red chickens divergently selected for residual feed consumption. Poult. Sci. 74:1245–1252.
categorized, and the association between traits was studied by factorial correspondence analysis (FCA). Attributes of good versus poor layers were well separated by the FCA, but no association was obtained between TI and other traits on the primary axes of the FCA. Fearfulness, as measured by TI, was not related to levels of performance of egg laying quail. Secondary negative associations between TI and body temperature and between TI and residual feed consumption were obtained; however, they were marginal and need to be confirmed.
(Key words: quail, fear, behavior, egg, feed intake) 2003 Poultry Science 82:1839–1844
INTRODUCTION Research on the welfare of domesticated birds raised for meat and egg production has identified a number of traits related in one way or another to the well-being of birds. Fearfulness is one of them, and it has been greatly studied, especially in experiments with fear-related traits, such as open-field activity (Jones and Faure, 1982; Jones et al., 1992) and tonic immobility (TI; Benoff and Siegel, 1976; Gallup, 1979; Jones, 1986). TI is a passive reaction of the birds that freeze after being exposed to severe stress. It is measured easily in chickens and quail as early as 1 wk after hatching and clearly has a genetic foundation, as divergent lines showing little or extreme fear (Mills and Faure, 1991) have been developed by artificial selection on the duration of TI in Japanese quail. Consequently, TI has some appeal as a possible representative behavioral trait to be included in the selection index of poultry. The relation between measures of fearfulness, and, more generally, between welfare-related traits has been widely studied (Mills and Faure, 1986; Mignon-Grasteau et al., 2003). On the contrary, experimental evidence for the existence of a relation between measures of well-being and production is still lacking even though both are thought to
2003 Poultry Science Association, Inc. Received June 2, 2003. Accepted for publication August 19, 2003. 1 To whom correspondence should be addressed: minvielle@dga2. jouy.inra.fr.
be adversely affected by extreme fear (Jones and Hocking, 1999). In a recent study on feed conversion and behavior, however, the 20 best and 20 worst broilers from a randombred line showed similar median durations of TI (Skinner-Noble et al., 2003). Also, a preliminary intrapopulation analysis in a quail line selected for early egg production (Minvielle et al., 1999b, 2000b) did not show any association between TI and egg production (Minvielle et al., 2002). However, the sample size was small, the population used showed little variation for TI, and measuring the association through the calculation of linear correlations was approximate at best because normalizing the distribution of TI by the Box-Cox transformation (Besbes et al., 1993) was imperfect. The objective of the present study was to evaluate the extent of the association between TI and varied productionrelated traits, using Japanese quail as a model animal. It could not be achieved by simply comparing lines differing for TI or for production traits but needed to be done within a single population to avoid finding nonsense correlations. In order to have a large experimental group with a wide genetic base, we used F2 birds from an initial cross between a line selected for long duration of TI (LTI; Mills and Faure, 1991) and a line selected for early egg production (DD; Minvielle et al., 2000a). Abbreviation Key: AFE = age at first egg; BT = body temperature; DD = Japanese quail line selected for early egg production; EN = egg number; EW = egg weight; FCA = factorial correspondence analysis; FI = feed intake; LTI = Japanese quail line selected for a long duration of tonic immobility; RFI = residual feed intake; TI = tonic immobility.
1839
1840
MIGNON-GRASTEAU AND MINVIELLE
MATERIAL AND METHODS F2 Experimental Population Reciprocal crosses were carried out between Japanese quail from different families of the LTI and DD lines that had been compared for fearfulness and performance in a previous study (Minvielle et al., 2002). All 12 LTI breeders chosen to produce the F1 had the largest possible value of TI (300 s), and TI of their 12 DD mates did not exceed 55 s. Next, 1 male and 3 F1 females were randomly drawn from each F1 family. TI recorded for these birds varied between 31 and 300 s. Finally, 438 F2 birds (30 full-sib F2 families) were produced in 3 consecutive hatches by 10 single-pair matings of a F1 male with 3 full-sib F1 females from another family. Complete information with data on all the traits described above was only available for 355 F2 females (17 s ≤ TI ≤ 300 s) from the 30 F2 families and constituted the data set used in the present experiment.
Husbandry The experimental work was carried out at the Unite´ de Ge´ne´tique Factorielle Avicole, Nouzilly, France. Obtained in 3 successive hatches, chicks were reared in group cages for 5 wk. They received a mash starter diet (2,901 kcal ME/kg, 11.5% moisture, 7% ash, 27% total protein, 8% fat, and 4% crude fiber). Initial temperature in the group cages was 37°C, and it was decreased to 30°C by the end of the rearing period. At 5 wk of age, females were assigned randomly to individual cages of a 4-tier egg laying battery maintained at 25°C, where they remained until the end of the experiment. The quail received a commercial layer diet (2,709 kcal ME/kg, 11.5% moisture, 12% ash, 20% total protein, 4% fat, and 4% crude fiber) and were under artificial lighting for 14 h/d. Throughout the experiment, feed and water were available ad libitum.
Traits TI. At 9 or 10 d of age, the duration of TI was measured on each quail. It was defined as the length of time during which a chick remained immobile after it had been held on its back for 10 s to induce the freezing reaction. For practical reasons, the measure of TI was terminated if the bird had remained immobile for 5 min (TI = 300) or if it failed to remain at all immobile after 5 attempts at inducing TI (TI = 0). BW and Body Temperature. Birds were weighed at 5, 27, and 31 wk of age. Rectal body temperature (BT) of fasted birds was measured at 5 and 31 wk of age. Egg Production and Egg Quality. Individual egg production was recorded daily for 64 wk between the ages of 5 and 69 wk. Age at first egg (AFE), total egg number (EN), average clutch length, and the mean proportion of defective eggs (soft shell, broken, and double yolk) were obtained for each quail. Three normal eggs laid consecutively were collected around the age of 29 wk to estimate average egg weight (EW) and egg composition (shell, yolk, and albumen weights) for each quail.
FIGURE 1. Distribution of the duration of tonic immobility (TI).
Feed Consumption. When females reached 27 wk of age, a 24-d feed trial was started. In this trial, feed intake (FI) was measured and used to estimate residual FI (RFI), egg mass, mean BW, and BW gain (BWG) during the feed trial.
Statistical Analysis Egg Laying Curve. The egg laying curve of each quail was estimated by using the Wood equation: ENt = A tB e−Ct (Timmermans, 1973). In this equation, ENt is the number of eggs laid until time t (t = 2, 4,. . ., 64 wk), and A, B, and C are the parameters of the curve. Parameter A relates to the level of egg production in the first period of test, C is an indicator of the decline of egg production, and B/C estimates the time at peak egg production. RFI. The multiple regression FI = 108.95 + 5.53 mean BW0.75 + 1.99 BW gain + 0.877 egg mass was adjusted to individual measures of FI. The residuals of the regression line, adjusted for the effect of the hatch, were used to estimate RFI. Factorial Correspondance Analysis. Correspondance analysis is an inductive exploration of a data set for finding patterns in the data when several traits have been recorded for each individual (Benzecri, 1992). By reducing the number of primary traits into a smaller number of factors that are combinations of the original variables and that account for the most variance, the redundancy in the original data set is eliminated, and true associations between traits may be found. Two applications of correspondance analysis are available: principal component analysis for normally distributed traits and factorial correspondence analysis (FCA) for categorical traits. In the present work, the distribution of TI (Figure 1) was not readily transformable, and the distributions of several production traits were very asymetrical (Table 1) so a FCA was carried out on the data set with the SPAD 4.0 software (Lebart et al., 1999). First, raw values were corrected for the effects of the hatch (production-related traits) and technician (TI) by using, respectively, the GLM and NPAR1WAY procedures from the SAS program library (SAS Institute, 1988). Then, Spearman rank correlations were estimated between all variables taken 2 × 2, and for each pair of redundant variables (rs >
1841
TONIC IMMOBILITY AND PRODUCTION IN JAPANESE QUAIL TABLE 1. Elementary statistics on the 27 variables measured for the 355 Japanese egg laying quail and repartition of the data into low and high categories for the second factorial correspondance analysis Category Trait Tonic immobility, s BW at 35 d, g Body temperature at 35 d, °C Body temperature at 31 wk, °C Feed intake in 24 d, g Residual feed intake in 24 d, g BW gain in feed test, g Mean egg weight, g Mean yolk weight, g Mean albumen weight, g Mean shell weight, g Ratio yolk:albumen Age at first egg, d Total egg number Soft-shell eggs, % Broken eggs, % Double yolk eggs, % Defective eggs, % Clutch length, d Number of pauses Months in test at first egg, mo Maximum 1 mo egg number Age at peak egg production, mo A1 B1 C1 tinf2
Mean ± SD
Skewness
Kurtosis
Minimum
Maximum
Low if trait ≤
High if trait >
± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
0.13 0.25 0.28 0.32 0.10 0.84 −0.34 0.32 0.14 0.26 0.50 0.60 2.70 −1.40 9.24 3.89 8.49 5.58 4.02 1.77 9.64 −2.21 1.72 −0.76 2.60 3.42 0.40
−1.33 0.19 0.52 −0.42 0.05 1.79 0.66 0.34 0.08 0.49 0.98 1.37 13.75 1.84 110.0 21.32 102.8 45.63 31.29 3.02 91.63 16.10 2.51 6.35 10.57 17.41 0.66
17 147.5 40.7 40.9 453.4 −96.4 −41.5 9.07 2.73 5.14 0.66 0.43 37 112 0 0 0 0 2.1 0 1 18 2 2.904 0 0 1
300 238.8 43.0 42.4 858.0 183.3 19.4 15.27 5.15 9.30 1.25 0.78 79 415 29.4 11.4 4.8 11.3 47.9 70.5 2 30 15 35.23 3.035 0.986 11.85
100 180.4 — 41.40 — −34.4 — 11.23 — — — — — 320 — — — 0.2 — — — — — — — 0.05 —
259 199.2 — 41.84 — 34.2 — 12.69 — — — — — 370 — — — 0.5 — — — — — — — 0.10 —
178.9 189.0 41.66 41.60 658.8 0.0 −3.79 11.93 3.92 7.13 0.88 0.55 44.1 333.0 0.89 0.80 0.12 0.61 8.1 12.2 1.0 27.1 4.25 20.765 0.493 0.112 5.23
87.7 16.6 0.38 0.32 68.3 473.3 9.19 1.04 0.43 0.68 0.09 0.06 4.9 58.4 2.2 1.3 0.36 1.0 4.2 14.1 0.10 1.1 2.81 3.206 0.382 0.120 1.58
1 A, B, and C are coefficients of the Wood egg-laying curve: ENt = A tB e−Ct, where EN is the egg number, and t is the age expressed in 2-wk periods. 2 tinf is the age at peak egg production estimated from the Wood curve.
0.7), only 1 trait was included in the analysis. Finally, for each trait, the data were assigned to 1 of 3 categories (high, low, and mean) defined (Table 1) in such a way that each category was equally frequent. For 2 variables, only 2 classes could be defined. A first FCA analysis was performed with 14 nonredundant variables and explained only 55% of the total variation in the first best 10 axes. Consequently, a second FCA with a subset of 8 variables representing the different types of traits (fear, egg production, feed consumption, and egg quality), but with only one or two variables per type, was performed to obtain more significant FCA axes.
RESULTS AND DISCUSSION Briefly, in the first FCA (data not shown), axis 1 (accounting for 8.3% of the total variation) distinguished between poor (low EN and high C) and good (high EN and low C) layers where C is a coefficient of the Wood egg-laying curve. Axis 2 (7.4% of the total variation) compared late (older AFE and low A) and early (younger AFE and high A) layers, and axis 3 (6.4% of the total variation) distinguished between small females laying small eggs (low BW and low EW) and large females laying larger eggs (high BW and mean EW) where A is a coefficient of the Wood egg-laying curve. Results for these 3 axes were not surprising, because in the Wood model (Timmermans, 1973), C is an indicator of the decline of egg production and A
relates to the level of early egg production, and positive correlations between BW and egg size have been reported extensively in quail (e.g., Sefton and Siegel, 1974; Strong et al., 1978; Minvielle et al., 1999a). TI appeared as the predominant variable for the first time in the eighth axis (4.4% of the total variation) and seemed to explain only a minor part of the total variability. In the second analysis, the first 5 axes accounted for 45% of the total variation, and for each axis, the variables explaining a major part of the variability are listed with their contribution to the axis (Table 2). In Figure 2, axis 1 distinguished between poor (low EN and high C) and good (high EN and low C) layers, as in the previous analysis, and an association (inside the ellipses) among RFI, BW, and EW could be observed 3 times (low BW with low EW and mean RFI, mean BW with mean EW and low RFI, and high BW with high EW and high RFI). It indicated that the average-sized, egg-laying F2 quail used feed most efficiently, which is in partial agreement with the small negative correlations between BW and RFI reported in chickens (Bordas and Me´rat, 1981; Luiting and Urff, 1991; TixierBoichard et al., 1995). TI contributed only 1.2% to the variation of FCA axis 1 and remained close to the origins of the axes in Figure 2. A better spread of the 3 classes of TI was obtained with secondary axes 3 and 4 in Figure 3, in which a 3-variable association (inside the ellipses) among TI, defective eggs, and BT could be visualized 3 times. A similar but indirect
1842
MIGNON-GRASTEAU AND MINVIELLE TABLE 2. Contributions of the first 5 axes of the factorial correspondence analysis (FCA), and of the main variables1 on each axis, to the total variation Axis Rank 1 2 3 4 5
Variation (%)
List of the main2 variables (% contribution to the axis)
12.5 9.2 8.3 7.7 7.2
C (36.7); EN (36.2); DEF (17.0) EN (22.8); BW (19.6); RFI (18.5); C (18.2); EW (16.8) BW (27.7); EW (25.3); EN (10.5%) TI (23.0); BW (19.0); DEF (15.7); RFI (12.9); BT (12.5); C (11.6) BT (36.3); TI (20.0); EW (19.3)
1 C = coefficient of the Wood egg-laying curve; EN = total egg number; DEF = percentage of defective eggs; RFI = residual feed intake; EW = mean egg weight; TI = duration of tonic immobility; and BT = body temperature at 31 wk. 2 Explaining more than 10% of the variance of FCA axis.
relation between stress and BT was reported previously by de Jong et al. (2002) who observed lower BT in restricted-fed, and therefore chronically-stressed, broiler chickens than in control birds, but their observation resulted likely from the added effects of stress and lower FI. There was also some indication of a negative association between RFI and extreme classes of TI, maybe in relation to the stress susceptibility, as evaluated by TI, of high emotivity quail. Indeed, in chickens, birds with low RFI are less active physically than high RFI birds (Katle et al., 1984; Gabarrou et al., 1998). Finally, the widest spread of F2 quail for TI was obtained with axes 4 and 5 (Figure 4). All other traits, however, were located away from TI in this figure, and no association was visible. Concomitant differences in TI and performances have been reported for several line comparisons (Mills et al., 1994; Jones et al., 1997; Minvielle et al., 2002), and they have
been sometimes interpreted as evidence for the existence of a relation between those traits (Jones and Hocking, 1999). Associations between traits, however, can only be tested on animals with similar genetic backgound. Overall, the present results obtained within an F2 population of Japanese quail showed that TI measured early in life was not strongly associated with any of the main poultry production traits, in agreement with the preliminary indications reported by Minvielle et al. (2002). Using egg shell quality as an example, it has been proposed that stress might be needed to trigger the association with TI (Mills et al., 1994). In the present experiment, no special stress was imposed, but laying quail were kept in an environment (lodging, feed, and contact with man) similar to that of commercial quail egg production in France. Consequently, including TI as an overall indicator of fearfulness in a selection programme should have little impact on the genetics of the
FIGURE 2. Spread of the variables according to the first 2 axes of the factorial correspondance analysis. Filled diamond = tonic immobility (TI); filled square = BW; open triangle = body temperature (BT); circle = residual feed intake (RFI); multiplication sign = egg weight (EW); addition sign = egg number (EN); double cross = coefficient of the Wood egg-laying curve (C); and open diamond = defective eggs (DEF). The size of the symbols increases with their mean contribution to the 2 axes. Consistent associations among traits across categories are circled.
TONIC IMMOBILITY AND PRODUCTION IN JAPANESE QUAIL
1843
FIGURE 3. Spread of the variables according to the axes 3 and 4 of the factorial correspondance analysis. Filled diamond = tonic immobility (TI); filled square = BW; open triangle = body temperature (BT); circle = residual feed intake (RFI); multiplication sign = egg weight (EW); addition sign = egg number (EN); double cross = parameter of the Wood egg-laying curve (C); and open diamond = defective eggs (DEF). The size of the symbols increases with their mean contribution to the 2 axes. Consistent associations between traits across categories are circled.
production traits, at least in a well-managed production unit because the level of fear of humans might be related negatively with egg production (Hemsworth and Barnett, 1989). Interestingly, a similar experiment with TI and fear-
related behavior traits (Mignon-Grasteau et al., 2003) has also shown that TI was not associated with the other traits. Consequently, TI might not be a good choice as a unique variable of stress response to be included among the selec-
FIGURE 4. Spread of the variables according to the axes 4 and 5 of the factorial correspondance analysis. Filled diamond = tonic immobility (TI); filled square = BW; open triangle = body temperature (BT); circle = residual feed intake (RFI); multiplication sign = egg weight (EW); addition sign = egg number (EN); double cross = parameter of the Wood egg-laying curve (C); and open diamond = defective eggs (DEF). The size of the symbols increases with their mean contribution to the 2 axes.
1844
MIGNON-GRASTEAU AND MINVIELLE
tion objectives, and the association between stress response and production will need to be studied further by using several other stress-related traits.
ACKNOWLEDGMENTS The technical assistance of Jean-Louis Monvoisin (INRA, Jouy-en-Josas), David Gourichon and Chantal Moussu (INRA, Nouzilly) is gratefully acknowledged. The authors thank Jean-Michel Faure and Andrew Mills (INRA, Nouzilly) for kindly providing quail from their LTI line to start this experiment.
REFERENCES Benoff, F. H., and P. B. Siegel. 1976. Genetic analyses of tonic immobility in young Japanese quail (Coturnix coturnix japonica). Anim. Learn. Behav. 4:160–162. Benzecri, J. P. 1992. Correspondance Analysis Handbook, Dekker, New-York. Besbes, B., V. Ducrocq, J. L. Foulley, M. Protais, A. Tavernier, M. Tixier-Boichard, and C. Beaumont. 1993. Box-Cox transformation of egg-production traits of laying hens to improve genetic parameter estimation and breeding evaluation. Livest. Prod. Sci. 33:313–326. Bordas, A., and P. Me´rat. 1981. Genetic variation and phenotypic correlations of food consumption of laying hens corrected for body weight and production. Br. Poult. Sci. 22:25–33. de Jong, I. C., S. van Voorst, D. A. Ehlhardt, and H. J. Blokhuis. 2002. Effects of restricted feeding on physiological stress parameters in growing broiler breeders. Br. Poult. Sci. 43:157– 168. Gabarrou, J. F., P. A. Geraert, N. Franc¸ois, S. Guillaumin, M. Picard, and A. Bordas. 1998. Energy balance of laying hens selected on residual food consumption. Br. Poult. Sci. 39:79–89. Gallup, G. G., Jr. 1979. Tonic immobility as a measure of fear in domestic fowl. Anim. Behav. 27:316–317. Hemsworth, P. H., and J. L. Barnett. 1989. Relationships between fear of humans, productivity and cage position of laying hens. Br. Poult. Sci. 30:505–518. Jones, R. B. 1986. The tonic immobility reaction of the domestic fowl: A review. World’s Poult. Sci. J. 42:82–96. Jones, R. B., and J. M. Faure. 1982. Open-field behaviour of male and female domestic chicks as a function of housing conditions, test situations, and novelty. Biol. Behav. 7:17–25. Jones, R. B., and P. M. Hocking. 1999. Genetic selection for poultry behaviour: big bad wolf or friend in need? Anim. Welfare 8:343–359. Jones, R. B., D. G. Satterlee, and H. L. Marks. 1997. Fear-related behaviour in Japanese quail divergently selected for body weight. Appl. Anim. Behav. Sci. 52:87–98. Jones, R. B., D. G. Satterlee, and F. H. Ryder. 1992. Open-field behavior of Japanese quail chicks genetically selected for low or high plasma corticosterone response to immobilization stress. Poult. Sci. 71:1403–1407. Katle, J., H. B. Bentsen, and B. O. Braastad. 1984. Correlated traits with residual feed consumption. Pages 136–138 in Proceedings of the 17th World’s Poultry Congress and Exhibition, Helsinki, Finland. Lebart, L., A. Morineau, T. Lambert, and P. Pleuvret. 1999. Syste`me SPAD. CISIA-CERESTA, Montreuil, France.
Luiting, P., and E. M. Urff. 1991. Residual feed consumption in laying hens: 2. Genetic variation and correlations. Poult. Sci. 70:1663–1672. Mignon-Grasteau, S., O. Roussot, C. Delaby, J. M. Faure, A. Mills, C. Leterrier, D. Gue´mene´, P. Constantin, M. Mills, G. Lepape, and C. Beaumont. 2003. Factorial correspondence analysis of fear-related behaviour in Japanese quail. Behav. Proc. 61:69–75. Mills, A. D., and J. M. Faure. 1986. The estimation of fear in domestic quail: correlations between various methods and measures. Biol. Behav. 11:235–243. Mills, A. D., and J. M. Faure. 1991. Divergent selection for duration of tonic immobility and social reinstatement behavior in Japanese quail (Coturnix coturnix japonica) chicks. J. Comp. Psychol. 105:25–38. Mills, A. D., K. M. Herron, M. M. Bain, S. E. Solomon, and J. Faure. 1994. Eggshell quality in Japanese quail Coturnix Japonica genetically selected for high or low levels of fearfulness. Pages 292–293 in Proceedings of the 9th European Poultry Conference, Glasgow, UK. Minvielle, F., J. L. Coville, A. Krupa, J. L. Monvoisin, Y. Maeda, and S. Okamoto. 2000a. Genetic similarity and relationships of DNA fingerprints with performance and with heterosis in Japanese quail lines from two origins and under reciprocal recurrent or within-line selection for early egg production. Genet. Sel. Evol. 32:289–302. Minvielle, F., E. Hirigoyen, and M. Boulay. 1999a. Associated effects of the roux plumage color mutation on growth, carcass traits, egg production, and reproduction of Japanese quail. Poult. Sci. 78:1479–1484. Minvielle, F., A. D. Mills, J. M. Faure, J. L. Monvoisin, and D. Gourichon. 2002. Fearfulness and performance related traits in selected lines of Japanese quail (Coturnix japonica). Poult. Sci. 81:321–326. Minvielle, F., J. L. Monvoisin, J. Costa, A. Frenot, and Y. Maeda. 1999b. Changes in heterosis under within-line selection or reciprocal recurrent selection: an experiment on early egg production in Japanese quail. J. Anim. Breed. Genet. 116:363–377. Minvielle, F., J. L. Monvoisin, J. Costa, and Y. Maeda. 2000b. Long-term egg production and heterosis in quail lines after within-line or reciprocal recurrent selection for high early egg production. Br. Poult. Sci. 41:150–157. SAS Institute. 1988. SAS User’s Guide: Statistics. SAS Institute Inc., Cary, NC. Sefton, A. E., and P. B. Siegel. 1974. Inheritance of body weight in Japanese quail. Poult. Sci. 53:1597–1603. Skinner-Noble, D. O., R. B. Jones, and R. G. Teeter. 2003. Components of feed efficiency in broiler breeding stocks: is improved feed conversion associated with increased docility and lethargy in broilers? Poult. Sci. 82:532–537. Strong, C. F., Jr., K. E. Nestor, and W. L. Bacon. 1978. Inheritance of egg production, egg weight, body weight and certain plasma constituents in Coturnix. Poult. Sci. 57:1–9. Timmermans, M. P. F. C. A. 1973. The statistical and genetical significance of the application of mathematical models to explain egg production curves in poultry. Arch. Geflu¨gelkd. 37:37–45. Tixier-Boichard, M., D. Boichard, E. Groenveld, and A. Bordas. 1995. Restricted maximum likelihood estimates of genetic parameters of adult male and female Rhode Island Red chickens divergently selected for residual feed consumption. Poult. Sci. 74:1245–1252.