Table 2. Test responses and overshadowing criteria for aldehyde combinations used in experiment 2.
| Ctrl A | Ctrl B | OVS | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Combination of odours (A/B) | RA(A) | RB(A) | RB(B) | RA(B) | RAB(AB) | RA(AB) | RB(AB) | RA(AB) ≠ RB(AB) (p-value) | RA(A) ≠ RA(AB) (p-value) | RB(B) ≠ RB(AB) (p-value) | Overshadowing |
| hexanal/nonanal | 93.10 | 13.79 | 96.30 | 11.11 | 100.00 | 17.24 | 96.55 | <0.001 | <0.001 | 1 | B > A |
| heptanal/nonanal | 89.66 | 27.59 | 93.10 | 75.86 | 82.14 | 64.29 | 78.57 | 0.27 | <0.05 | 0.14 | A = B (?) |
| octanal/nonanal | 85.71 | 67.86 | 96.55 | 86.21 | 92.86 | 96.43 | 100.00 | 1.00 | 0.35 | 1 | A = B (?) |
The table shows, for each odour combination, the sample size (n) and the responses of the control groups (Ctrl A and Ctrl B) and of the group trained with the mixture (OVS) in the test performed 1 h after training. Values of RX(Y) represent the percentage of ants responding with MaLER to the odour X after training to the odour Y. A difference in responses to the mixture components of the OVS group [i.e. RA(AB) and RB(AB)], first criterion for overshadowing determination, is indicated by adjusted p-values of pairwise response comparisons (sequential Bonferroni corrections after multiple McNemar's chi square tests). Fishers exact tests were used to compare responses to the mixture components of the OVS group with responses of the control groups to their training odour [i.e. RA(A) with RA(AB) and RB(B) with RB(AB)] to assess the second criterion for overshadowing determination. Response levels and statistical differences of the heptanal/nonanal combination are extracted from Experiment 1 (see Table 1). Significant differences are indicated in bold. “X = Y”: no overshadowing; “X > Y”: X overshadowed Y (where X and Y are either A and B or B and A) as shown by the lower response to Y after mixture training and by the reduction of responses to Y after mixture training compared to responses to Y after Y training. “?” indicates difficulty in determining the occurrence of overshadowing due to high levels of generalization from nonanal to heptanal and from nonanal to octanal.