Insectes soc. 47 (2000) 271– 279 0020-1812/00/030271-09 $ 1.50+0.20/0 © Birkhäuser Verlag, Basel, 2000
Insectes Sociaux
Research article
Testing genetic variance hypotheses for the evolution of polyandry in the honeybee (Apis mellifera L.) P. Neumann 1, 2 and R.F.A. Moritz 2 1 2
Department of Zoology and Entomology, Rhodes University, Grahamstown 6140, South Africa, e-mail:
[email protected] Molekulare Ökologie, Institut für Zoologie, Martin-Luther-Universität Halle-Wittenberg, Kröllwitzerstr. 44, 06099 Halle/Saale, FRG, e-mail:
[email protected]
Received 11 October1999; revised 31 January 2000; accepted 15 February 2000.
Summary. Colony size, honey yields and colony levels of infestation with Varroa jacobsoni of 30 queenright honeybee (Apis mellifera) colonies with naturally mated queens were evaluated over a two-year period. Workers taken from each colony were genotyped at four DNA-microsatellite loci to determine the level of polyandry. All queens mated with more than 10 drones (mean number of observed patrilines = 17.7 ± 5.23). We found significant correlations between colony size and honey yield and between colony sizes of two subsequent years. Analyses of variance revealed a strong impact of the breeding lines on the tested phenotypic traits. The impact of polyandry on colony honey yields was weak (p < 0.05, not significant when applying a Bonferroni adjustment) and 8% of the phenotype was determined by the effect of polyandry. The contribution of polyandry to colony size (0.25%) or levels of infestation with Varroa jacobsoni (0.09%) was even weaker in both test years. Likewise, we could not find any averaging effect of polyandry on the honey yield, size nor parasite load of honeybee colonies. Our data set does not resolve the question, whether polyandry and genetic diversity causes more productive colonial phenotypes. If colony level selection is an evolutionary force for polyandry, the effects are hard to detect in man-kept colonies headed by naturally mated queens. Key words: Apis mellifera, evolution, honeybee, performance, polyandry.
Introduction Honeybees, Apis, show an exceptionally high level of polyandry (Boomsma and Ratnieks, 1996; Fuchs and Moritz, 1999). Through the developments in DNA microsatellite technology (Estoup et al., 1993, 1994) detailed studies have
revealed matings of honeybee queens with more than 25 drones (Moritz et al., 1995, 1996; Oldroyd et al., 1996; Neumann et al., in prep.). Because of the high variance for polyandry within Apis mellifera, ranging from one to 45 reported matings (Neumann et al., in prep.), this species should be a prime system to empirically test the various hypotheses for the evolution of polyandry. Many hypotheses and several potential mechanisms have been proposed to explain the evolution of polyandry in social insects (Page, 1980; Crozier and Page, 1985; Sherman et al., 1988; Ratnieks, 1990; Crozier and Pamilo, 1996; Boomsma and Ratnieks, 1996). Most recent theoretical and empirical studies focused on the genetic variance hypotheses (Keller and Reeve, 1994) which predict fitness gains through decreased intracolonial relatedness resulting from multiple mating. Genetically based task specialization on colony phenotypes has led to several versions of the division of labor hypothesis for the evolution of polyandry (reviewed by Robinson, 1992). Polyandry should ensure a broader variety of genetic specialists (Crozier and Page, 1985; Oldroyd et al., 1997; Fuchs and Moritz, 1999) and thus allow for a more efficient division of labor through task specialization of individual workers (Crozier and Page, 1985; Calderone and Page, 1991; Oldroyd et al., 1992a; Dreller et al., 1995). The parasite and pathogen model predicts that genetic variance reduces the susceptibility of colonies to parasites and pathogens (Hamilton, 1987; Sherman et al., 1988; Hamilton et al., 1990; Shykoff and Schmid-Hempel, 1991a, b). Under this hypothesis polyandry in social insects has evolved as an adaptive response to high parasite and pathogen loads. A few studies have addressed this issue. Either instrumentally inseminated queens and/or unnaturally small colonies were used (Oldroyd et al., 1992b; Fuchs and Schade, 1994; Page et al., 1995; Fuchs et al., 1996), which may not represent conditions in full size colonies with naturally mated queens. Moreover, these studies yielded contradictio-
272
P. Neumann and R.F.A. Moritz
nary results. Whereas Oldroyd et al. (1992b) and Fuchs and Schade (1994) found increased performance of bee groups with a higher genotypic diversity, Fuchs et al. (1996) reported on a decreased performance of colonies with a higher degree of genetic variation between workers. Page et al. (1995) did neither find a decreased nor an increased performance in a study of full sized colonies. Instead, they reported that colonies with higher levels of genotypic diversity are more likely to express an average rather then a superior colony phenotype. Thus, genotypic diversity may reduce phenotypic variance in honeybees. But all these studies used artificially inseminated queens. Clearly, artificial insemination is unlikely to interfere with sperm composition. However, it may result in different colony phenotypes compared to colonies with naturally mated queens as shown by Wilde (1989), who found that colonies with instrumentally inseminated queens produced significantly more brood. In this study we investigated the performance of honeybee colonies headed by naturally mated queens and use DNAmicrosatellites to determine polyandry. We measured three traits that directly affect colony performance: honey yield, disease load and colony size. Methods Colony phenotype data We estimated three phenotypic traits of colonies which can be regarded as being closely linked to colony fitness: 1) honey stores which are essential for surviving periods of dearth appear to have direct impact on fitness; 2) parasites and diseases can obviously reduce colony fitness; 3) colony size, because only large colonies have enough workers to produce viable swarms and large numbers of drones during the reproductive season. Colonies with a larger work force might per se be more likely to buffer environmental changes than smaller colonies (Crozier and Page, 1985). The performances of 89 queenright A. m. carnica colonies headed by naturally mated queens were determined by routine performance testing (2 years) at the apiary Schwarzenau, Germany (Bayerische Landesanstalt für Bienenzucht, 1994/95). The colonies were from ten different breeding lines each consisting of four to ten sister-queen colonies. All queens were raised at the same time of the year and naturally mated at four mainland honeybee mating apiaries with a variable number (10–42) of strong drone producing colonies. For both testing years, total annual levels of infestation with the ectoparasitic mite Varroa jacobsoni were quantified per colony by counting and adding the total number of dead mites in the hives after each of three treatments with the acaricide Perizin®. The total infestation levels were obtained when the performance-tested colonies were free of brood from mid to end of November. Honey yield was determined by weighing honey frames before and after honey extraction in early June and in mid July and by estimating residual winter honey stores. Colony size was determined using the amount of sealed worker brood frames in May. All colonies were headed by the same queens during the two-year testing period and all colony phenotypes were evaluated in both years. DNA isolation and microsatellite analysis At least 40 workers were sampled from each of 30 randomly chosen performance-tested colonies. The samples were stored in 75% ethanol until DNA extraction from individual workers (Beye and Raeder, 1993; Neumann et al., 1999a, b, c). We used four DNA-microsatellite loci for
Colonial phenotype and polyandry in honeybees genotyping (Estoup et al., 1993, 1994). Multiplex PCR of two pairs of loci (A43/B124, A76/A107) was done according to standard protocols (Estoup et al., 1993, 1994; Neumann et al., 1999a,b,c). Amplification products were separated on 6% polyacrylamide sequencing gels. M13mp18 control DNA sequencing reactions were run on the same gel as size standards. Microsatellite alleles were scored as fragment lengths in base pairs. Genotype analysis and number of observed patrilines The genotypes of the queens and their drone mates were derived from the genotypes of the worker samples. The queen was considered to be homozygous if one allele was present in every worker of the colony. The queen was assumed to be heterozygous for two alleles if every worker carried one of the two alleles. The paternal alleles were those not transmitted by the queen. In case a queen genotype could not be unambiguously derived from the worker offspring, we used appropriate pedigree information (putative mother genotype of the tested sister queens) to exclude allele combinations. As a rule, we chose the queen genotype yielding the lowest number of observed patrilines. Drifted workers were identified using maternity testing as described by Neumann et al. (1999c). Individuals not sharing a maternal allele at each of the tested microsatellite loci were excluded from further data analysis in this study. Data analysis The observed number of patrilines may severely underestimate the actual number due to small sample sizes. Therefore, we estimated the number of patrilines present in the colony according to Cornuet and Aries (1980):
,
1 E (k) = k – k – 1 – 3 k
n
(1)
where E (k) = expected number of patrilines in the colony k = number of equally frequent patrilines n = sample size. We then followed Oldroyd et al. (1997) and numerically evaluated k by substituting E (k) with our observed number of patrilines (no) and the worker sample sizes for n. To test for potential sample size effects on our estimates of the number of effective males (me) we calculated me following two different approaches: 1. The relatedness-based approach of Estoup et al. (1994) which can be biased due to small sample sizes. We estimated relatedness using the pedigree coefficients of relatedness G (Pamilo and Crozier, 1982) between all possible worker dyads in the sample (either 0.25 or 0.75) and calculated the arithmetic mean to obtain the average intracolonial coef– ficient of relatedness G . We then calculated the number of effective males (me) using the equation of Chevalet and Cornuet (1982). 2 me = 02 – 4G – 1
(2)
where me = number of effective males – G = average intracolonial relatedness. 2. We also calculated the effective number of males using a non-relatedness-based estimate. According to Starr (1984) me can be expressed as the reciprocal of the sum of the male contributions (∑ pi2): 1 me = 8 ∑ pi2
(3)
Insectes soc.
Vol. 47, 2000
Research article
Following Pamilo (1993) the non-sampling detection error can be corrected by calculating: 1 me = 8 ∑ pi2
(4)
where yi = observed proportional contributions of the queen-mates N = sample size. The important estimate in this context is the effective number of males because the GV hypotheses are based on the intracolonial relatedness. We tested with analyses of variance for breeding line effects on the honey yield and on colony levels of infestation with Varroa jacobsoni. For this analysis all performance tested sister-queen colonies for which data was available were considered (N = 89 colonies). We tested for potential correlations between all investigated parameters. The impact of polyandry on colony performance was estimated using partial correlations (corrected for both colony size 1994, 1995 and worker sample size) between the phenotypic data and our estimates of polyandry for the 30 genotyped colonies. The partial correlations for colony size were only corrected for worker sample size. Since we wanted to exclusively extract the impact of polyandry on the colony phenotypes from the data, we eliminated effects of the genetic variance among the breeding lines on the colony phenotype by trans-
273
forming the data as standardized deviations from the strains’ mean. Thus, for each colony and each tested phenotypic trait the deviation from the mean of its corresponding breeding line divided by its standard deviation was used for further analysis. Similarly, the differences between the actual phenotypic values of both years were divided by the mean of the corresponding breeding line to determine the phenotypic variability during the testing period. We used this approach to test for the homeostatic effect of genotypic diversity on colony performance found by Page et al. (1995). Since the high number of comparisons may cause significant differences only by chance, we used a Bonferroni procedure (Zar, 1996) to adjust the significance levels. All statistical tests and analyses were two-tailed and performed using the SPSS© statistical package.
Results Colony phenotype data The colony phenotype data for the years 1994 and 1995 are given in Table 1. The honey yields ranged from 16.5kg to 59.9kg. No significant differences between the two testing years were detected (Mann-Whitney U-test: U = 76.5,
Table 1. Genotype and phenotype data of the tested A. m. carnica colonies. The number of observed patrilines (no), estimated number of patrilines (k), the effective number of males (me), the worker sample sizes (N) and the colonial phenotypes for the years 1994 and 1995 are given. All colonies – were headed by the same queens during the two-years testing period (BL = breeding line, G = average intracolonial relatedness as defined by Estoup et al., 1994, A–J = breeding lines) Colony
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 18 19 20 21 22 23 24 25 26 27 27 28 29 30 x ± s.e.
BL
A A B B B C C D D E E E F F F F G G G G H H H I I I I J J J
no
18 11 18 21 15 20 14 19 17 11 17 18 13 22 20 25 28 26 17 20 10 11 18 14 15 14 26 10 12 26 17.7 0.95
k
23.18 11.31 22.69 27.47 20.75 21.32 14.73 25.68 21.93 12.83 21.93 21.54 14.63 33.05 30.05 15.31 34.8 54.55 23.17 24.04 10.41 25.66 27.14 17.24 20.75 18.18 51.04 10.41 13.82 48.63 23.94 2.05
me
18.1 10.10 17.5 27.44 21.67 17.59 14.13 15.98 20.67 9.52 19.84 14.96 14.09 31.07 31 15.32 29.06 33.06 12.79 14.84 9.36 28.88 25.38 18.9 23.1 19.12 28 7.52 10.64 39 20.01 1.47
G
0.28 0.3 0.28 0.27 0.27 0.28 0.28 0.28 0.27 0.3 0.28 0.28 0.29 0.27 0.27 0.28 0.27 0.27 0.29 0.28 0.3 0.27 0.27 0.28 0.27 0.28 0.27 0.32 0.3 0.26 0.28 0.02
N
34 39 35 39 26 58 24 38 32 24 32 38 31 30 32 38 56 35 30 42 32 22 29 28 22 26 36 32 27 39 33.53 1.52
Varroa infestation [number of mites]
Honey yield [kg]
Colony size [brood frames]
1994
1995
1994
1995
1994
1995
161 154 520 385 472 375 399 320 393 234 173 182 231 283 268 301 149 186 149 178 319 285 446 324 160 322 398 358 188 188 288.37 19.67
286 311 1355 402 206 432 951 512 95 378 388 718 631 559 492 326 952 1050 110 175 417 263 705 426 1039 375 255 1753 525 249 544.53 69.66
34.6 29.4 38 37.7 34.8 49.3 32.8 27.7 26.1 57.3 44.6 44.4 40.2 49.5 56.1 34.8 54.1 54.2 50.4 33.4 56.5 20.2 41.6 54.5 53.7 46.8 54.4 53.8 45.2 59.9 43.87 1.98
40.5 30.3 32.5 33.4 46.1 41.7 33.7 38.3 51 43.5 42.6 16.5 38.4 27.3 34.5 41.4 55.6 47.4 40.8 34.6 53.4 38.9 52.6 41.6 44 41.7 46.3 54 43.5 55.4 41.38 1.62
6 6 7 9 6.5 9 5.5 5.5 4 8 6.5 8 7 7 8 7 8 8 9 6.25 8 7 7 8 9 7 9 10 7 7 7.34 0.24
8 7 7.5 7.75 7 9 6 7 8 9 8 6 8 6 8 8 9.5 10 9 7.5 10 6.5 10 8.5 9 7.5 9.5 10 8 9.5 8.16 0.23
274
P. Neumann and R.F.A. Moritz
Colonial phenotype and polyandry in honeybees
Table 2. Correlation r-matrix for the colony phenotype data of all tested honeybee colonies in the years 1994 and 1995 (N = 89). Colony size, colony level of infestation with Varroa jacobsoni, honey yield of colonies and variance for the tested phenotypic traits during the two-years testing period were considered. The adjusted significance level is a = 0.01. Significant correlations are indicated with ** for p < 0.01 and *** for p < 0.001
Colony Size Varroa Infestation Honey Yield Phenotypic Variance
1994 1995 1994 1995 1994 1995 Honey Size Varroa
Colony size
Varroa infestation
Honey yield
Phenotypic variance
1994
1995
1994
1995
1994
1995
Honey
Size
Varroa
1 0.50** –0.01 0.27 0.51** 0.06 –0.08 0.14 0.17
1 0.25 0.15 0.70*** 0.77*** –0.23 –0.18 0.13
1 0.23 0.08 0.21 –0.09 –0.07 0.08
1 0.14 –0.03 –0.11 0.06 0.33
1 0.44 –0.12 –0.16 –0.03
1 –0.10 –0.23 0.06
1 0.36 –0.36
1 –0.03
1
p > 0.27). The colony levels of infestation with V. jacobsoni were highly variable ranging from 95 to 1753 mites. The total annual levels of infestation were significantly higher in 1995 than in 1994 (Mann-Whitney U-test: U = 224.5, p < 0.001). Colony size ranged from four to ten brood frames. Colonies were significantly larger in 1995 (Mann-Whitney U-test: U = 297, p < 0.03). The tested breeding lines differed significantly in their honey yields (ANOVA: p 0.05). Thus, larger colonies did not result in more homeostatic colony phenotype expression for the tested traits. Polyandry data The polyandry data for the tested colonies is given in Table 1. A total of 1290 nestmate workers were genotyped and assigned to patrilines. The number of observed patrilines (no) per queen ranged from 10 to 28 with a mean of 17.7 ± 5.23. The number of estimated matings (k) was higher ranging from 10.41 to 54.55 with a mean of 23.95 ± 11.25. The two different approaches for estimating the effective number of males obtained identical results for all colonies (only one data set shown in Table 1). This indicates that our sample sizes were sufficiently large to precisely determine the effective number of males. The number of effective males (me) ranged from 7.52 to 39 with an average of 20.1 ± 8.05. The – average intracolonial relatedness G ranged from a minimum of 0.26 to a maximum of 0.32. Impact of polyandry on colony phenotypes We found a close to significant correlation after Bonferroni adjustment (p = 0.032) between the honey yields and our
estimates of paternity (Fig. 1a). We did not detect any significant correlation between the number of observed patrilines, the estimated number of matings nor the effective number of males with the parasite load (Fig. 1b). Likewise we did not find any significant correlation between our estimates of polyandry and colony size (Fig. 1c) and the year to year variation of the tested phenotypic traits (Table 3). Potential interactions with other parameters There is a significant correlation between the number of observed patrilines and worker sample size (Table 4). However, we did not detect any significant correlation between sample size and effective number of males nor the estimated number of matings. This also indicates that our sample sizes were sufficiently large to precisely determine the effective number of males. Less than 5% of the workers were identified as drifted individuals from other colonies. As reported elsewhere (Neumann and Moritz, 1997; Moritz and Neumann, in prep.; Neumann et al., 2000) there were no significant correlations between our estimates of polyandry and the level of drifting workers and between the level of drifting workers and colony honey yields, sizes and levels of infestation with V. jacobsoni. Similarly, there were no significant correlations between our estimates of polyandry and the number of drone producing colonies at the different mating apiaries (Neumann et al., 1999a) nor were there significant differences in the level of polyandry between the tested breeding lines (Neumann et al., in prep.).
Discussion Our data set does obviously not resolve the question whether extreme polyandry has evolved through colony level selection. We found no significant correlations between the tested phenotypes of honeybee colonies and the level of polyandry of queens. Our best estimate from the data is that colonies with a more genetic diverse work force do show an increased
Insectes soc.
Vol. 47, 2000
Research article
275
Table 3. Correlation matrix (simple correlations, r-matrix) between the phenotypic variance for the investigated traits and the estimates of polyandry during the two-year testing period (k = estimated number of patrilines, me = effective number of males). Colony size, colony levels of infestation with Varroa jacobsoni and honey yield of colonies were considered. No significant correlations are found (N = 30 colonies)
a
b
Phenotypic variance
k
me
Honey Size Varroa
–0.14 0.18 0.19
0.05 0.25 0.07
Table 4. Correlation matrix (simple correlations, r-matrix) between the estimates of polyandry (no = observed number of patrilines, k = estimated number of patrilines, me = effective number of males) and worker sample size (s. Table 1). Thirty colonies were considered. The adjusted significance level is a = 0.017. Significant results are indicated with * for p < 0.01 Polyandry estimate
N
no k me
0.58 * 0.27 0.16
honey yield. Polyandry contributed to 8% of the total phenotypic variation. The contribution of polyandry to colony size (0.25%) or levels of infestation with Varroa jacobsoni (0.09%) was weaker in both test years. Likewise, we could not find any averaging effect of polyandry on the honey yield, size nor parasite load of honeybee colonies. The effects of other sources of variation such as breeding lines or year to year variance were clearly factors substantially affecting colonial phenotypes. However, given the observed correlation between honey yield and polyandry to be true and causative, an effect of 8% may definitely be sufficient to favor polyandry of queens if it outweighs the costs associated with multiple matings. Indeed, several aspects may have masked a relationship in our analysis, which we address in the following sections.
c Figure 1. Performance of colonies in the years 1994 and 1995 and estimates of polyandry. The figures show the correlations for the effective number of males (me). The Bonferroni adjusted significance level is a = 0.017. a) Relative honey yields. Close to significant correlations were found for the effective number of males (me , r = 0.28, p = 0.032), for the observed number of patrilines (no , r = 0.29, p = 0.027) and for the number of estimated patrilines (k, r = 0.24; p = 0.073). b) Relative colony levels of infestation with V. jacobsoni. No significant effect was detected (no: r = 0.09, p = 0.5; k: r = 0.01, p = 0.97; me: r = 0.03, p = 0.81). c) Relative colony size. No significant effect was detected (no: r = 0.03, p = 0.82; k: r = 0.12, p = 0.38; me: r = 0.05, p = 0.73). The performance data of all genotyped colonies for both testing years is considered (N = 60)
Genotype data Our methods for genotyping might be not precise enough for testing the hypotheses. However, the reliability of the DNA microsatellite technique employed for honeybees (Estoup et al., 1994) has been confirmed using instrumentally inseminated queens (Neumann et al., 1997). The probability for non-identification errors of patrilines due to identical drone genotypes was lower than 2% in our sample as a result of the high degree of heterozygosity of the used microsatellite loci in the tested honeybee population (Neumann et al., 1999a). The limited sample sizes per colony might have affected our estimates, as discussed in previous papers (Moritz et al., 1995; Oldroyd et al., 1995, 1996, 1997). Small sample sizes preclude the detection of rare patrilines (Cornuet and Aries, 1980; Boomsma and Ratnieks, 1996). However, the approach
276
P. Neumann and R.F.A. Moritz
of Starr (1984) and Pamilo (1993) of calculating the effective number of males (which is corrected for small sample sizes) yielded identical estimates compared to the relatedness based approach of Estoup et al. (1994). Moreover, our estimates of the effective number of males are not correlated with sample sizes. This suggests that sampling errors did not substantially effect our estimates. Phenotype data The procedures for evaluating colony performance may have lacked precision. However, the large number of tested colonies and the repeated measures reduce the consequences of unlikely weighing and counting errors in the performance test routine. Potential effects resulting from drifting individuals between colonies might interfere with the genotypic composition. However, as we have shown elsewhere (Neumann et al., 2000), drifted workers were identified in the samples and had no quantitative impact on evaluating colony phenotype data. Colony sizes and honey yields are positively correlated in our sample as expected from routine beekeeping experience (Sachs, 1964). This is a further indication that the colony phenotype characteristics were properly evaluated. Queen sources Our queens were from pre-selected commercial honeybee breeding stock which were mated on mating apiaries where only few drone producing colonies were available. Thus, although we tested naturally mated queens, the actual degree of genotypic diversity may be less than under true natural conditions. On the other hand, our queens were from ten different breeding lines and were mated to four different drone populations, thus genetically resembling truly independent replicates. Natural mating situation As a result of the natural mating situation all queens showed polyandry levels above ten drones. This is a systematic problem because about 90% of the influence on intracolonial relatedness occurs below this level. Therefore our sample of tested queens may suffer from a variance reduction for intracolonial relatedness (not necessarily for polyandry). If this is the case, it might well be that all the queens in our sample show at least the minimum level of polyandry necessary for a functional colonial phenotype. If we then measure fitness directly or indirectly via fitness parameters we suffer from the problem that we cannot find any correlation between fitness and polyandry regardless of how precisely our methods of investigation might be. However, if we want to explain the evolution of extreme polyandry (exceeding 10 matings), data with low degrees of polyandry may be less relevant and point to a different evolutionary mechanism than the one in question.
Colonial phenotype and polyandry in honeybees
A factor, which may complicate the results, could be the direct individual fitness of queens. It might well be that the most vigorous queens yield the most vigorous colonies and that they also mate more, thus yielding an apparent causal relationship between polyandry and colony performance. Although genetic variance among queens within the tested breeding lines was rather low since they were all sisters, we can certainly not exclude that this factor had an impact on our results. From an evolutionary biological point of view this aspect is however less important if we address the problem of the evolution of extreme polyandry. Why should ”vigorous” (i.e. fit) queens mate with more males if this is associated with costs? Irrespective of the queen’s direct “vigor”, natural selection should favor individuals which optimize their mating strategy. Relation between evaluated phenotypic traits and colony fitness The fitness of honeybee colonies clearly is the lifetime number of surviving swarms and of drones produced. Colony phenotype characteristics enhancing or reducing the likelihood of producing viable swarms and large numbers of drones are the cues to colony fitness. In light of the genetic variance hypotheses, these should be enhanced under larger genotypic variability. This could be either in terms of superior performance of more genetic diverse colonies or in terms of a reduced phenotypic variance (Page et al., 1995) assuming average colony phenotypes have a superior chance of reproduction. Since we did not measure the fitness of our colonies directly in terms of swarms produced or successful matings of drones, our interpretation basically suffers from the problem of how our observed colony traits correlate with fitness under natural conditions. For example if there is an optimal phenotype for colony size (Allee et al., 1949) and too large colonies form a fitness disadvantage our interpretation based on a linear relationship between size and fitness looses explanatory power. In particular the management techniques, simply by providing an above average nesting opportunity, might interfere with the evaluation of the true fitness under natural conditions. The tested traits should be carefully interpreted, maybe just as “tokens” of fitness (Page et al., 1995) that are associated in some way with natural colony survival and reproduction. We may also be misled in our conclusion if fitness and tested trait have extreme non-linear relationships. Cost aspect: is polyandry selectively neutral in honeybees? The mating system of the honeybee makes it difficult to quantify the cost aspect for polyandry. Obviously, polyandry can only evolve if the fitness benefits exceed the costs of additional matings. In our study the colonies might have encountered conditions too favorable to reveal the full potential benefit of polyandry. On the other hand, the high genetic variance component suggests that the impact of polyandry is weak in comparison to the breeding line effect. Given that
Insectes soc.
Vol. 47, 2000
there are only small costs of polyandry in honeybees one might expect only small benefits resulting from high polyandry. These may be difficult to measure under field conditions at the level of a colony’s phenotype. On the one hand, it has been assumed that multiple mating of queens in highly polyandrous species has such minimal costs, that there is virtually no selection against it. On the other hand, it has been argued that polyandry increases the time and/or number of the honeybee queen’s mating flights and therefore increases the risk of queen loss (Moritz and Southwick, 1992). For example, polyandry in honeybees has been assumed to be costly as a result of sexually transmitted diseases or harsh weather conditions (Moritz et al., 1995; Oldroyd et al., 1996) and high predation rates at drone congregation areas (Oldroyd et al., 1997). So far, only the risk of honeybee queen losses after mating flights has been evaluated. Unfortunately, the results are highly variable and range from 1.04% (Ratnieks, 1990) up to 96% (Paradeau, 1955). The latter figure may not be very common in light of the weak effects of polyandry found in our study. In fact, assuming a 96% risk of queen loss one would expect an enormous selection pressure to abandon mating altogether and switch to parthenogenetic reproduction. This has been argued as one hypothesis to explain the evolution of thelytoky in the Cape honeybee, A. m. capensis (Moritz, 1986). Neumann et al. (1999a, b) found that under extremely isolated or harsh island conditions the number of honeybee queen matings can be significantly lower compared to more moderate conditions. This might reflect higher risks of mating. Although the current picture of the cost aspect for polyandry in the honeybee remains obscure, we do not expect polyandry to be completely free of costs. Therefore, we consider it unlikely that polyandry is completely selectively neutral. Parasite-Pathogen hypothesis Another limitation of the present study might be that Varroa jacobsoni is an inappropriate organism to test the parasite model: 1. V. jacobsoni and Apis mellifera do not represent a wellestablished host-parasite-system like V. jacobsoni and Apis cerana. Thus, a higher resistance against parasites through a more genetic diverse worker force may not be detected, because there is hardly any resistance at all for once infected A. mellifera colonies. However, this argument does not hold, since we found genetic variance for colony levels of infestation among breeding lines, with some lines being more resistant than others. 2. In our study the colonies were treated with acaricides such as Perizin® to determine the number of mites per colony. This may mask potential effects of polyandry on infestation under natural conditions particularly in the second test year. In spite of these obvious shortcomings, our findings are in line with Woyciechowski et al. (1994) and Page et al. (1995) who compared colony levels of infestation with sacbrood, Ascosphaera apis and Penibacillus larvae (Page et al., 1995) or Nosema apis (Woyciechowski et al., 1994) of
Research article
277
singly inseminated queens versus multiply inseminated queens and found no significant differences. Moreover, Fuchs et al. (1996) inseminated queens with mixed sperm of one or several breeding lines and found that line-mixed colonies showed consistently lowered performance with respect to Varroa infestations. This result is also contrary to the theoretical expectation of enhanced disease resistance with increased genetic variations between workers. According to Kraus and Page (1998) the parasite and pathogen model is in general unlikely to explain polyandry in the social insects (but see Sherman et al., 1998). In light of our results, we can neither reject nor accept the parasite-pathogen hypothesis (Sherman et al., 1988). Division of labor hypothesis a) Averaging effect hypothesis There have been numerous reports of genetic influences on division of labor in honey bee colonies, but the impact of worker genotypic diversity on the level of the colony phenotype remains unclear. Page et al. (1995) argue that the averaging effect of genotypic variability on colony phenotypes may have selective advantages, making colonies less likely to “fail” because of inappropriate colony responses to changing environmental conditions. In their experiments, colonies with greater genotypic diversity did not have an increased fitness but had a reduced phenotypic variance. However they used artificially inseminated queens and drones from different breeding lines which might have interfered with the results. Moreover, as Page et al. (1995) pointed out, they analyzed the effects of worker genotypic diversity on the phenotypes of honey bee colonies during a critical phase of colony development, the “nest initiation” phase. Kolmes et al. (1989) presented data suggesting that genotypic variability may be important when colonies are under environmental stress. Likewise, Louveaux (1966) found that similarly performing sister queen colonies showed distinct phenotypic differences when transferred to a new different environment. We do not find any averaging effect of polyandry in our sample. This may result from our mature, unstressed and well managed colonies having different requirements. Thus, high levels of polyandry may have an averaging effect (Page et al., 1995) during certain phases of colony development such as the post swarming phase or after a change in environmental conditions. b) Genetic specialist hypothesis In spite of all the limitations and potential masking effects, we found a close to significant correlation between the honey yield and our estimates of the degree of polyandry. Therefore, we consider our results to be consistent with the findings of Oldroyd et al. (1992b) and Fuchs and Schade (1994), who found increased performance of colonies headed by queens with a higher level of polyandry. In no case did we obtain negative correlation estimates between polyandry and colony fitness. We have clearly no evidence to reject that genotypic diversity resulting from a naturally mated queen may be important for colony productivity.
278
P. Neumann and R.F.A. Moritz
If the increase of honey production is truly a consequence of polyandry it is unlikely to be a result of the rather small change of intracolonial relatedness resulting from the variation of polyandry found in this study. The effect may more likely point to the theory of Fuchs and Moritz (1999) which is not restricted to low polyandry levels. We conclude that the current picture of the evolution of extreme polyandry in honeybees cannot be unambiguously resolved. A paucity to find clear fitness advantages in polyandrous colonies is complemented by an extreme wide range of mating risk estimates. At the current state we would either expect the cost for polyandry to be extremely low or we have missed a critical piece in the benefits resulting from intracolonial genetic variance. Further studies focusing on the cost aspect may be rewarding to contribute to our understanding of the evolutionary proximate factors selecting for extreme polyandry in honeybees. Acknowledgments We thank D. Mautz and the Bayerische Landesanstalt für Bienenzucht for allowing us to take the bee samples and for kindly providing performance data. We wish to thank A. Kühl and K. Gladasch for technical assistance. We are grateful to R. Page for stimulating discussions about polyandry and division of labor. We also thank M. Beye, R. Crozier, H. R. Hepburn, P. Kryger, F.L.W. Ratnieks, J.P. van Praagh and anonymous referees for valuable comments on earlier versions of the manuscript. Financial support was granted by a Rhodes University Fellowship to PN and by the DFG to RFAM.
References Allee, W.C., A.E. Emerson, O. Park, T. Park and K.P. Schmidt, 1949. Principles of Animal Ecology. Philadelphia, Saunders, USA, pp. 350. Beye, M. and U. Raeder, 1993. Rapid DNA preparation from bees and % GC fractionation, Bio. Tech. 14: 372–374. Bayerische Landesanstalt für Bienenzucht, 1995. Leistungsprüfung 1993-1995, Prüfhof Schwarzenau, Abschlußbericht. Bayerische Landesanstalt für Bienenzucht, Erlangen, Germany, pp. 78. Boomsma, J.J. and F.L.W. Ratnieks, 1996. Paternity in eusocial Hymenoptera. Phil. Trans. R. Soc. B. 331: 947–975. Calderone, N.W. and R.E. Page, 1991. The evolutionary genetics of division of labor in colonies of the honeybee (Apis mellifera). Am. Nat. 138: 69–92. Chevalet, C. and J.M. Cornuet, 1982. Etude théorique sur la sélection du caractère “production du miel” chez l’abeille. I . Modèle génétique et statistique. Apidol. 13: 39–65. Cornuet, J.M. and F. Aries, 1980. Number of sex alleles in a sample of honeybee colonies. Apidol. 11: 87–93. Crozier, R.H. and R.E. Page, 1985. On being the right size: male contributions and multiple mating in social Hymenoptera. Behav. Ecol. Sociobiol. 18: 105–115. Crozier, R.H. and P. Pamilo, 1996. Evolution of Social Insect Colonies. Sex Allocation and Kin Selection. Oxford University Press, Oxford, New York, Tokyo, 306 pp. Dreller, C., K. Fondrk and R.E. Page, 1995. Genetic variability affects the behavior of foragers in a feral honeybee colony. Naturwissenschaften 82: 243–245. Estoup, A., M. Solignac, M. Harry and J.M. Cornuet, 1993. Characterisation of (GT)n and (CT)n microsatellites in two insect species: Apis mellifera and Bombus terrestris. Nuc. Acid Res. 21: 1427– 1431.
Colonial phenotype and polyandry in honeybees Estoup, A., M. Solignac and J.M. Cornuet, 1994. Precise assessment of the number of patrilines and genetic relatedness in honeybee colonies. Proc. R. Soc. Lon. B. 258: 1–7. Fuchs, S. and V. Schade, 1994. Lower performance in honeybee colonies of uniform paternity. Apidol. 25: 151–171. Fuchs, S., R. Büchler, S. Hoffmann and K. Bienefeld, 1996. Non-additive colony performances by inseminating queens with mixed sperm of several carniolan breeder lines. Apidol. 27: 304–306. Fuchs, S. and R.F.A. Moritz, 1999. Evolution of extreme polyandry in honeybees (Apis mellifera). Behav. Ecol. Sociobiol. 45: 269–277. Hamilton, W.D., 1987. Kinship, recognition, disease, and intelligence: constraints of social evolution. In: Animal Societies: Theories and Facts (Y. Itô, J.L. Brown and J. Kikkawa, Eds.). Japan Science Society Press, Tokyo, pp. 81–102. Hamilton, W.D., R. Axelrod, R. Tanese, 1990. Sexual reproduction as an adaptation to resist parasites (A review). Proc. Nat. Acad. Sci. 87: 3566–3573. Keller, L. and H.K. Reeve, 1994. Genetic variability, queen number and polyandry in social Hymenoptera. Evolution 48: 694–704. Kolmes, S.A., M.L. Winston and L.A. Ferguesson, 1989. The division of labor among worker honey bees (Hymenoptera: Apidae): the effects of multiple patrilines. J. Kans. Entomol. Soc. 62: 80–95. Kraus, B. and R. Page, 1998. Parasites, pathogens and polyandry in social insects. Am. Nat. 151: 383–391. Louveaux, J., 1966. Les modalités l’adaptation des abeilles (Apis mellifica L.) au milieu naturelles. Ann. Abeille 9: 323–350. Moritz, R.F.A., 1986. Two parthenogenetical strategies of laying workers in populations of the honeybee, Apis mellifera (Hymenoptera, Apidae). Entomol Gener. 11: 159–164. Moritz, R.F.A. and E.E. Southwick, 1992. Bees as Superorganisms. Springer Verlag, Berlin, 395 pp. Moritz, R.F.A., P. Kryger, G. Koeniger, N. Koeniger, A. Estoup and S. Tingek, 1995. High degree of polyandry in Apis dorsata queens detected by DNA microsatellite variability. Behav. Ecol. Sociobiol. 37: 357–363. Moritz, R.F.A., P. Kryger and M.H. Allsopp, 1996. Competition for royalty in bees. Nature 384: 31. Neumann, P., K. Fondrk, R.E. Page and R.F.A. Moritz, 1997. Testing the reliability of DNA-microsatellites in instrumental inseminated queen-honeybees (Apis mellifera L.). In: Soziale Insekten, IUSSITagung Graz 1997, Graz, 1997 (K. Crailsheim, A. Stabenhteiner, Eds.). Graz, p. 66. Neumann, P. and R.F.A. Moritz, 1997. Drifting of workers and drones as an evolutionary factor in the honey bee Apis mellifera (L.). In: Proceedings of the sixth congress of the ESEB, Arnhem, The Netherlands, pp. 120. Neumann, P., R.F.A. Moritz and J. Van Praagh, 1999a. Queen matingfrequency in different types of honeybee mating apiaries. J. Apic. Res. 38: 11–18. Neumann, P., J. Van Praagh, R.F.A. Moritz and J. Dustmann, 1999 b. Testing reliability of a potential island mating apiary using DNA microsatellites. Apidol. 30: 257–276. Neumann, P., R.F.A. Moritz and D. Mautz, 1999c. Using DNA microsatellites for maternity testing in honeybees (Apis mellifera). Apidol. 30: 505–512. Neumann, P., R.F.A. Moritz and D. Mautz, 2000. Colony evaluation is not affected by the drifting of drone and worker honeybees (Apis mellifera L.) at a performance testing apiary. Apidol. 31: 67–79. Oldroyd, B.P., T.E. Rinderer and S.M. Buco, 1992a. Intra-colonial foraging specialism by honey bees (Apis mellifera L.). Behav. Ecol. Sociobiol. 30: 291–295. Oldroyd, B.P., T.E. Rinderer and S.M. Buco, 1992b. Effects of intracolonial genetic diversity on honey bee (Hymenoptera: Apidae) colony performance. Ann. Entomol. Soc. Am. 85: 335–343. Oldroyd, B.P., A.J. Smolenski, J.M. Cornuet, S. Wongsiri, A. Estoup, T.E. Rinderer and R.H. Crozier, 1995. Levels of polyandry and intracolonial relationships in Apis florea. Behav. Ecol. Sociobiol. 37: 329–335.
Insectes soc. Vol. 47, 2000 Oldroyd, B.P., A.J. Smolenski, J.M. Cornuet, S. Wongsiri, A. Estoup, T.E. Rinderer and R.H. Crozier, 1996. Levels of polyandry and intracolonial genetic relationships in Apis dorsata (Hymenoptera: Apidae). Ann. Entomol. Soc. Am. 89(2): 276–283. Oldroyd, P.B., M.J. Clifton, S. Wongsiri, T.E. Rinderer, H.A. Sylvester and R.H. Crozier, 1997. Polyandry in the genus Apis, particularly Apis andreniformis. Behav. Ecol. Sociobiol. 40: 17–26. Page, R.E., 1980. The evolution of multiple mating behaviour by honey bee queens (Apis mellifera L.). Genetics 96: 263–273. Page, R.E., G.E. Robinson, M.K. Fondrk and M.E. Nasr, 1995. Effects of worker genotypic diversity on honey bee colony development and behavior (Apis mellifera L.). Behav. Ecol. Sociobiol. 36: 387– 396. Pamilo, P., 1993. Polyandry and allele frequency differences between the sexes in the ant Formica aquilonia. Heredity 70: 472–480. Pamilo, P. and R.H. Crozier, 1982. Measuring relatedness in natural populations: methodology. Theor. Pop. Biol. 21: 171–193. Paradeau, C., 1955. Sélection et Élevage de Reines, Union Nationale de l’Apiculture Française, Paris, pp. 160. Ratnieks, F.L.W., 1990. The evolution of polyandry by queens in social Hymenoptera: the significance of the timing of removal of diploid males. Behav. Ecol. Sociobiol. 26: 343–348. Robinson, G., 1992. Regulation of division of labor in insect colonies. Annu. Rev. Entomol. 37: 637–665. Sachs, H.G., 1964. Vergleich der Honigerträge von Bienenvölkern der Carnica-Rasse bei unterschiedlicher Raumumgebung und unterschiedlicher Volksstärke. Z. Bienenforsch. 7: 47–52.
Research article
279
Sherman, P.W., T.D. Seeley and H.K. Reeve, 1988. Parasites, pathogens and polyandry in social Hymenoptera. Am. Nat. 131: 602–10. Sherman, P.W., T.D. Seeley and H.K. Reeve, 1998. Parasites, pathogens, and polyandry in honey bees. Am. Nat. 151: 392–398. Shykoff, J.A. and P. Schmid-Hempel, 1991a. Parasites and the advantage of genetic variability within social insect colonies. Proc. R. Soc. Lon. B. 243: 55–58. Shykoff, J.A. and P. Schmid-Hempel, 1991b. Genetic relatedness and parasite-mediated selection on the genetic composition of groups. Behav. Ecol. Sociobiol. 28: 371–6. Starr, C.K., 1984. Sperm competition, kinship and sociality in aculeate Hymenoptera. In: Sperm Competition and the Evolution of Animal Mating Systems (R.L. Smith, Ed.) Florida Academic Press, Orlando, pp. 428–459. Wilde, J., 1989. Development and productivity of honeybee colonies with naturally and artificially inseminated queens. In: Proc. XXXIst Int. Congr. Apiculture, Warsaw, Poland, August 19–25, 1987. Apimondia Publishing House, Bucharest, pp. 442–444. Woyciechowski, M., R. Król, E. Figurny, M. Stachowicz and M. Tracz, 1994. Genetic diversity of workers and infection by the parasite Nosema apis in honey bee colonies (Apis mellifera). In: Les Insectes Sociaux (A. Lenoir, G. Arnold and M. Lepage, Eds.). Paris, Université Paris Nord, p. 347. Zar, J.H., 1996. Biostatistical Analysis. 3rd edition. Prentice-Hall, Inc., Upper Saddle River, New Jersey, USA. 662 pp.
