An Evaluation of an Experimental Model for Sympatric Speciation
Discussion
Discussion
Effectiveness of Selection
The results give very little indication that selection increased the efficiency of ovipositional choice. If selection had been successful there should have been an increase in correct laying relative to the total number of eggs laid. The results contain no indication of such a trend.
Perhaps the enormous variability of results has masked any trend that may have been present. Day to day fluctuations were large as Fig. 1 clearly shows. Furthermore a higher total number of eggs was laid by both strains in Generation 1 than in any other generation, with a very large amount of variance between the totals for the four generations (S2 = 41,977,318.66). The percentages of eggs laid correctly by each strain each day were still fluctuating as markedly in Generation 4 as in any other generation. As shown in Fig. 2 there is no detectable daily or generational relationship between the percentages laid in either strain.
In the parental selection generation one would expect that approximately equal amounts of eggs would be deposited in all ten phials, and thus the one phial selected for in each population would contain approximately 10% of the eggs laid. This was not observed. The average percentage of eggs laid correctly in the amyl alcohol population in the first generation of selection (23.46%) was more than double the expected 10%, while the average percentage of eggs laid correctly in the ammonium carbonate population (4.67%) was only half that expected. By the fourth generation of selection the percentage of amyl alcohol population eggs laid correctly was actually less than in the first generation, dropping to 19.59%. At the same time the percentage of eggs laid correctly in the ammonium carbonate population rose to 42.96% in Generation 4. The overall results do not suggest that selection for correct ovipositional choice was successful in either strain.
If no variation in preference occurred the number of eggs laid correctly should tend to correlate with the total number laid, as in Generations 2 and 4 of the ammonium carbonate strain, and Generations 3 and 4 of the amyl alcohol strain. However, correlation coefficients for correctly laid, as compared with total numbers of eggs, all showed no significant relationship between the two except for Generation 4 of the ammonium carbonate strain, where r = 0. 9083, being highly significant at the 0.1% level. While there is no significant correlation for Generation 4 of the amyl alcohol strain, the correlation coefficient (r = 0. 4184) approaches significance more closely than those for any other generation of either strain. This tends to suggest that by Generation 4 there is beginning to be a stabilisation between total numbers and correct numbers of eggs laid.
Testing Equal Amounts of Each Attractant
A very interesting difference was obtained when females bred from the last generation of the selected strains were given a choice of equal amounts of each attractant. The females from Generation 4 of the amyl alcohol strain placed 60.32% of their eggs in amyl alcohol and 39.68% in ammonium carbonate. This result is significantly different from a 50% random choice and does seem to indicate that amyl alcohol females prefer to lay their eggs in amyl alcohol. However, the ammonium carbonate page 9 females also preferred to lay their eggs in amyl alcohol, placing only 2.03% of their eggs in ammonium carbonate, and 97.97% in amyl alcohol. Yet in the selection experiment, the ammonium carbonate strain tended to a more correct choice than the amyl alcohol strain. (In Generation 4 ammonium carbonate flies chose 42.96% correctly, and amyl alcohol flies 19.59% correctly).
Other Differences Between the Two Populations
While selection was not successful in the desired direction, it was not without effect on the two populations of flies. The first generation in both cages tended to show a two day cycle of peak and trough in egg laying which was never repeated once selection began. Some physiological pattern in egg laying may have been interrupted by the artificially imposed selection.
Furthermore, some dissimilarities did arise between the two populations during the course of selection. The ammonium carbonate strain, for example, suffered a marked reduction in the number of eggs laid. Since the two strains experienced identical conditions at any one time, and the populations were of identical size, it can be assumed that the two populations should lay approximately equal numbers of eggs in any generation. This did not happen. In every generation the amyl alcohol strain laid more eggs than the ammonium carbonate strain, the difference becoming larger as the experiment progressed, with the exception of the last generation. In successive order of generations, amyl alcohol flies laid 1,216, 1,671, 3,175 and 2,519 more eggs than the ammonium carbonate flies. These differences are highly significant.
Gene Drift — Possible Explanation of the Variability
These differences between the two strains could be explained in terms of the effect of inbreeding and gene drift on the flies. With the very heavy selection pressures employed, it is quite conceivable that very low numbers of the total population actually form the parents of the next generation. Since any one female can lay up to 150 eggs at a time (West, 1951) it is quite possible that all the eggs correctly selected in one day for one strain could have come from either one or very few flies.
Differences Between Generations That Suggest Gene Drift
The two strains exhibited great variability in pupal emergence times. In Generation 3 for example, amyl alcohol flies started emerging two days before the ammonium carbonate flies. Yet in the next generation, the ammonium carbonate flies began emerging several days ahead of the amyl alcohol flies. Here again, a feasible explanation would seem to be that the bulk of selected flies came from one, or very few females, and thus, by chance, were selected for early or late emerging. The differences in emergence times cannot be explained by temperature differences during incubation of the pupae, because all pupae were reared in comparable temperatures.
Frequency of migration for the amyl alcohol flies was lower in every generation except the third (results of migration counts for both strains are given as percentages in Appendix III). This could be because of an inbreeding effect resulting in less active amyl alcohol flies.
The main problem would seem to lie in determining just how much page 10 inbreeding and gene drift are present in an experiment of this type. Inbreeding poses practical problems in affecting the viability of flies. Gene drift seems to have caused marked variability between the two strains. Furthermore, there appears to be no way of controlling the factors accidentally bred into the populations in this way. If a population that appears so well adapted can change so drastically over the course of one generation (see second section of Discussion) it would seem that gene drift can easily outweigh the effects of selection. This problem needs to be investigated, and the experimental procedure adjusted to eliminate gene drift if possible, before the method can be used efficiently.
A Further Practical Problem — Egg Attractant Effects
Early in the experiment it was noted that freshly laid eggs tend to attract other females to lay their eggs in the same place. This gives a strong clumping effect which Pimentel et al (loc. cit.) mention only briefly in the appendix to their paper. Yet should one fly lay eggs wrongly, if those eggs then attract more females to deposit their eggs in the same place, a misleading distortion of results could conceivably occur.
The experiment on egg attraction effects (see Method) was designed to test this possibility. In twelve out of sixteen trials, phials containing egg "attractant" were chosen in a significantly greater percentage than phials containing only normal egg laying medium.
While these results are not completely conclusive, they do suggest that this egg attraction effect may introduce a source of error into the breeding system, which would be difficult to control. It is pointed out in order that its effects might be weighed against the variability of results.
Nevertheless, selection should favour those flies which are attracted more strongly by the chemical attractant than by other laid eggs. If this is so, less flies should fall into the trap of being attracted to an incorrect phial, resulting perhaps in a decrease in the number of phials incorrectly selected for in each generation. Unfortunately, this trend was not detected in the results. In the amyl alcohol population, the average number of phials incorrectly selected for out of nine were 5.1, 4.0, 5.3 and 4.0 for Generations 1 to 4, whilst in the ammonium carbonate populations 5.6, 2.0, 3.8 and 2.0 phials were incorrectly selected on the average in each generation.
The experiment using phials with egg or chemical attractants (see Method) was devised in an attempt to clear up this point. In five out of six trials, more eggs were laid in the phials with chemical attractant. This result tends to suggest that selection has favoured attraction to chemicals over that to eggs by this stage of the experiment. Since the problem was not anticipated, no comparable experiment was run on the parental stocks and hence no comparison can be made.