Egg size and hybrid syndrome-dependent embryo mortality in Chironomus hybrids ( Diptera : Chironomidae )

Female hybrids of the cross Chironomus t. thummi 9 x Ch. t. piger 3 which are largely affected by the sterility inducing Rud syndrome were backcrossed with males of both parental strains. The aim of the study was to provide information about those egg volumes that are insufficient for a normal embryogenesis and to ascertain whether in the hybrids the lethally small egg size rep­ resents a new abnormal trait of the Rud syndrome. The egg masses obtained contain eggs of very different sizes with volumes rang­ ing from 0.5 nl to 3.49 nl. Embryo mortality is unusually frequent in those eggs of the backcrosses and of the parental strains that have volumes smaller than 1.5 nl. An egg volume of 1.5 nl represents in Ch. thummi the lower limit for those volumes that are suffi­ cient for a normal embryogenesis. Mortality increases with decreasing egg size, reaching 100% in backcross eggs with volumes of 0.99 nl and smaller. Small egg size is a new trait of the Rud syndrome affected thummi $ x piger <S hybrids. This trait is part of a postzygotic reproductive isolation barrier between thummi and piger and manifests first in the backcrosses. Most backcross eggs show volumes between 1.5 nl and 2.99 nl. Within this volume range the amount of mortality does not depend upon egg volume. Here, embryo death is great in the backcrosses but normal in the parental strains. The high frequency of embryo death in the backcrosses must be predominantly due to the action of the Rud syndrome and a second hybrid syndrome, called HLE syndrome. Since further characteristic traits of these syndromes could be detected currence of the syndromes in this generation also. Therefore, the dromes is effective in the hybrids and in their progeny as well.


INTRODUCTION
Within egg masses of Chironomus t. thummi and Ch.t. piger a correlation exists between egg size and egg num ber: The smaller the egg number the smaller the eggs (Strenzke, 1959).Up to now no information has been available about the extent to which the egg size may be reduced without lethal consequences to embryonic devel opment.During recent investigations, female hybrids of the cross direction Ch. t. thummi $ x Ch. t. piger 3 were found to produce a large scale of eggs of different sizes including lethally small egg sizes.These hybrids are af fected by the germ line specific Rud syndrome.The aber rant traits of this syndrome are no or rudimentary gonads, chromosome aberrations in germ cells, a different con densation behavior of paternal and maternal chromo somes in meiosis and a strongly reduced chiasma frequency in males (Hagele, 1984;Hagele & Oschmann, 1987;Hagele & Lachmann, 1992).The Rud syndrome operates as one efficient postzygotic reproductive isola tion barrier between thummi and piger populations if a breakdown of the prezygotic isolation mechanism has oc curred.The latter is due to a different swarming behavior of the two Chironomus subspecies (Miehlbradt & Neu mann, 1976).Another postzygotic reproductive isolation barrier is established by the HLE syndrome.This syn drome occurs also in the FI hybrids but reciprocally to the Rud syndrome in the Ch.t. piger 2 x Ch. t. thummi 3 in surviving backcross individuals, the study demonstrates the ocpostzygotic reproductive isolation mechanism of both hybrid syncross direction.The abnormal traits of the HLE syndrome are a high frequency of embryo mortality, chromosome aberrations, malformations, and a heterogeneous develop ment (Hagele, 1984(Hagele, , 1995;;Hagele et ah, 1995).
To determine the egg sizes that are too small for a nor mal embryonic development and larval hatch, we backcrossed Rud syndrome affected thummi 2 x piger 3 fe male hybrids with males of the parental thummi and pi ger strains.We compared the volumes of the backcross eggs and the frequency of the corresponding embryonic mortality with that of the hybrids and the parental stocks.The aim of the study was to ascertain whether in Chiro nomus hybrids lethally small egg sizes represent a new and abnormal trait of the Rud syndrome.If so, this trait would be the only one of the Rud syndrome that mani fests in the hybrid progeny, although it originates in the gonads of the hybrids themselves like the other Rud syn drome traits.The trait concerned is then part of the syn drome specific postzygotic reproductive isolation barrier between thummi and piger (Hagele, 1999;Hagele et ah, 1995).Investigation of the backcrosses could additionally show whether both syndromes occur in the hybrid prog eny and contribute once more to a reproductive isolation.
The chironomids provided for crossing experiments were reared under standardized feeding, temperature, aerating and lighting conditions (Hagele, 1975).Freshly hatched larvae of each egg mass were placed into a plastic dish filled with aerated pond wa ter and covered by a glass cap.The larvae were fed a paste made of powdered stinging nettles and water.The rearing tem perature was 21°C during a time schedule of 16 h of illumina tion and 8 h darkness.Hybridization was done in mating cages with the dimensions of 25 cm (I) x 25 cm (w) x 60 cm (h).For each cross 5-7 virgin females were isolated from the plastic dishes immediately after emergence and added to the mating cages to a fivefold number of earlier collected males.Female hybrids of the cross thummi $ x piger 6 were backcrossed ei ther with thummi or piger males.Egg masses were removed from the mating cage immediately after deposition.Embryonic development of the eggs within each egg mass was synchro nized by permanently aerating the culturing water.The eggs were grouped into the developmental stages EO, E1-E4 (Table 1).The stages E1-E4 correspond essentially to the ones pub lished in Keyl (1958) and Strenzke (1959).Only those egg masses were analyzed where at least one egg started embryogenesis.The eggs of each egg mass were counted and the length and diameter of the eggs was measured with an ocular microme ter adapted to the dissecting stereo microscope.For an estima tion of the egg size the egg volume was chosen.The egg volume represents the actual egg size more closely than does the egg length or the index of egg length/egg diameter.Since the eggs have an ellipsoid shape the egg volume (V) in nanoliters (nl) was estimated by the formula V = (D/2)2 x L/2 x 4/3a, where D is the egg diameter, and L the egg length (Gellert et al., 1981).Fourth instar backcross larvae were studied for the occurrence of traits of the HLE syndrome such as malformations and chro mosome aberrations.Salivary gland chromosomes were pre pared and Orcein stained and studied for FILE syndrome specific chromosome aberrations with a Zeiss photomicroscope II (Hagele, 1984).To test for the occurrence of the Rud syndrome the adults were sexed and the ovaries of the females dissected out 12 h after emergence and before egg deposition.Egg deposition occurs usually 20-24 h after emergence.Ovaries with no, few or small eggs were taken as representative for rudi mentary developed gonads caused by the Rud syndrome (Hagele, 1987).

RESULTS
Freshly laid eggs of the developmental stage EO of Chironomus I. thummi and Cli.t. piger show an average vol ume of 1.572 nl with an egg length of 283 pm and an egg diameter of 103 pm, respectively.It is not possible to de cide whether an egg in this stage will begin embryonic development or not.Hence, for the study of egg size de pendent embryo mortality, routine measurements of the eggs were started about four to six hours after egg deposi tion in the El stage when the blastoderm occurs (Miall & Hammond, 1900).Measurements were also taken at later embryonic stages E2, E3 and E4 which differ in their de velopmental time from one another as indicated in Table 1.
After backcrossing only about 9% of the thummi $ x piger 6 females lay egg masses.These egg masses con tain on average less than 10% of the 500-700 eggs nor mally present in egg masses of the parental strains.Eggs within the same egg mass of a hybrid female show, in general, abnormally large size differences (Fig. 1).Fig. 2 reveals the frequencies of the egg volumes of eggs laid by hybrid females after backcrossing with piger and thummi males and by thummi and piger females after intrastrain mating.The egg volumes were arranged into six different volume categories.A comparison of the egg volumes of the successive developmental stages El to E4 showed that volumes increased with progressive embryonic dc-Fig.1. Different egg sizes within the same egg mass of a backcrossed (Cli.t. thummi $ x Cli.t. piger 6) female with a Cli t. thummi male.The three largest eggs were in the developmental stage E3, they had volumes of V = 2.54 nl, V = 2.11 nl, and V = 1.87 nl.The two smallest eggs died, their volumes were V = 1.17 nl and V = 0.96 nl.The bar represents 100 pm.velopment (data not presented).Although the egg vol umes in the El stage differ only slightly between thummi (V = 1.891 nl) and piger (V = 1.883 nl), the average in crease of the egg volumes until the E4 stage is in piger about 24% and in thummi 38%, respectively.The 14% larger increase in thummi eggs originates presumably from the 27% larger genome of thummi in comparison to piger (Keyl, 1965).The somewhat different frequencies of the egg volumes of thummi and piger in the same egg volume categories of Fig. 2 may be due to these genomic differences.The eggs of piger and of the (thummi 9 x pi ger d) females crossed with piger males occur in the two egg volume categories with 1.5-1.99nl and 2.0-2.49nl more frequently than do the genomically larger eggs of thummi and of the (thummi $ x piger 8 ) females crossed with thummi males.In the next two volume categories with larger egg volumes an opposite situation can be ob served: The genomically smaller eggs of piger and of the {thummi $ x piger 8) 9 x piger 8 cross hold the minor ity.
Most eggs of the egg masses of thummi and piger, as well as of the backcrosses, belong to the volume category of 2.0-2.49nl (Fig. 2).In this category the average num ber of backcross eggs is lower in comparison to the eggs of the pure thummi and piger strains.However, if the categories with smaller egg volumes are considered a re verse relationship concerning the egg number of the aforementioned crosses becomes evident: Backcrosses produce smaller eggs more frequently than the parental strains.The volume category with the smallest egg vol umes only contains eggs of the backcrosses (Fig. 2).A statistical test using r x c contingency tables (Zar, 1974) demonstrates that the distribution of the egg size fre quency presented in Fig. 2 is significantly different be tween the backcrosses and the intrastrain crosses.The difference found between {thummi 9 x piger c?) 9 x pi ger 8 and piger x piger eggs is % 2 = 41.512,v = 5, p < 0.001, and between {thummi 9 x piger 8) 9 x thummi 8 and thummi x thummi eggs % 2 = 65.324,v = 5, p < 0.001.If the frequencies of the egg volumes in the four geneti cally different types of egg masses are compared with the frequencies of egg mortality in the corresponding egg masses the following dependence becomes obvious (Figs 3, 4): In those eggs of the backcrosses showing egg vol umes of 0.99 nl and smaller, embryo mortality is always 100%.Such small'egg sizes do not occur in the pure thummi and piger strains.In the next volume category with larger eggs (V = 1.0-1.49nl) embryo lethality in the backcross eggs is reduced by a factor of 0.82-0.84but shows still extraordinarily high values.Within this cate gory also eggs of the parental strains begin to appear ex hibiting a high proportion of embryo death, too.Until the volume category of 1.5-1.99nl embryo lethality is further decreasing in all egg mass types.In this category and in the next three categories with again larger egg volumes (V = 2.0-3.49nl) the amount of embryo lethality remains nearly constant in the backcrosses at a level of 65% and 50%, respectively.This is demonstrated by the small and nearly identical multiplication factors of the X in the lin ear regression equations (Zar, 1974).The equation calcu lated for {thummi 9 x piger <?) 9 x thummi 8 is Y = 3.2x-72.9,and Y = 3.8x-13.8for {thummi 9 x piger <?) 9 x piger 8 .A very similar development of embryo mor tality is also true for both the thummi and piger strains, but here mortality remains constant at a significantly lower level (Figs 3, 4).These results demonstrate that, for eggs from all four crosses, embryo mortality is not related to egg size at volumes of 1.5 nl or above.
In the four egg volume categories described above where the egg size does not further determine embryo mortality, strong differences in the frequency of embryo death between the backcrosses and the parental strains be come evident.In the parental strains the small number of embryos dying is normal and amounts to between 2% and 7% per egg volume category.In the backcrosses, however, an embryo lethality occurs that is in the average about 12 times larger than in the parental strains.Even between both backcross directions differences exist in egg mortality of more than 10%.The differences are sta tistically significant {%2 = 94.25, f = 3, p < 0.001).These observations suggest that in the backcrosses the large amount of embryo lethality may be due to the action of the Rud and the HLE hybrid syndromes.The investiga tion of the (thummi 2 x piger 3 ) $ x thummi 3 backcrosses showed abnormal traits of the HLE as well as of the Rud syndrome: HLE syndrome specific malformed larvae and salivary gland chromosome aberrations could be found, and, additionally, rudimentarily developed ova ries characteristic of the Rud syndrome (Table 2).The HLE syndrome is known to produce, in addition to its aforementioned traits, more than 75% embryo mortality in piger eggs fertilized with thummi sperm.Therefore, it can be assumed that the high embryo mortality observed in the eggs of the (thummi 2 x piger 3) females backcrossed with thummi males may at least partially also be due to the action of the HLE syndrome.In the (thummi 2 x piger d) 2 x piger 3 backcross on the other hand, nu merous rudimentarily developed ovaries were found, while HLE caused defects such as malformations and chromosome aberrations were rare or not observed (Table 2).Embryo mortality is also high but smaller in relation to the reciprocal backcross.With respect to the frequency of the damaging events it is evident that the aberrant traits of the HLE syndrome clearly occur more often in {thummi 2 x piger 3) 2 x thummi 3 than in the {thummi 2 x piger 3) 2 x piger 3 cross.An inverse relationship exists concerning the frequency of the rudimentary ova ries of the Rud syndrome.

DISCUSSION
Investigation of eggs produced by Rud syndrome af fected female hybrids after backcrossing and by the two parental thummi and piger strains showed that the small est eggs are exclusively obtained after backcrossing.Amongst those small eggs with a slightly greater volume, the eggs of the backcrosses occur more frequently than those of the pure thummi and piger strains.These fre quency differences begin to disappear when egg sizes are considered to which most of the eggs of the parental strains and of the backcrosses belong and which, there fore, represent the normal average egg sizes.Thus, it is evident that small egg volumes display a specific charac ter of the two backcrosses studied.Since the rearing con ditions were the same differences in nourishment of the females cannot be responsible for the egg size differences (Wheeler, 1996).Because 97% of the backcrossed fe males are affected by the Rud syndrome and develop ru dimentary ovaries it may be inferred that their small eggs are a result of the Rud syndrome action.Small egg sizes should therefore represent a trait of the Rud syndrome.This trait manifests itself, by contrast to the other Rud syndrome traits, first in the backcross generation but, like the other traits, originates in the hybrids (Hagele, 1984;Hagele & Lachmann, 1992;Hagele & Oschmann, 1987, 1989).
Investigation of the egg sizes and the embryo mortality showed that a relationship exists between egg size and embryo death.Starting with an egg volume smaller than 1.5 nl embryo mortality is more frequent the smaller the egg volume.This effect is independent from the genetic constitution of the eggs since it can be observed in both backcrosses as well as in the two parental strains.The smallest eggs with an embryo mortality of 100% occur only in the backcrosses.From these results it can be con cluded, that in the chironomids studied here, the egg vol ume of 1.5 nl represents a border line beneath which the frequency of successful embryonic development depends on egg size.
Taisli: 2. Frequency of traits of the FILE and Rud syndrome in the reciprocal thummi x piger hybrids and in two backcrosses.The th x th and pi x pi crosses served as a control.The HLE syndrome is here characterized by a high embryo mortality and an in creased occurrence of malformations and chromosome aberrations.A high amount of rudimentarily developed ovaries is specific for the Rud syndrome.(1984,1987,1995) and Hagele & Oschmann (1989).

Cross
Egg size dependent embryo lethality does not occur in eggs with volumes larger than 1.5 nl.This is evidenced by the fact that in this volume range a change in egg size does not further alter the frequency of embryo mortality (see also regression equations) as is the case in eggs with smaller volumes.Within the above mentioned volume range the considerable difference in the frequency of em bryo mortality between the backcrosses and the parental strains is surprising.Here, embryo mortality is about 15 times greater in the backcrosses.This means that more than 60% of the (thummi 2 x piger 8) 2 x thummi 8 eggs and about 50% of the (thummi 2 x piger 8 ) 2 x pi ger 8 eggs are lethally damaged even if the embryo death usually observed in all chironomid egg masses (2-7%) is subtracted.We suggest that this high level of damage is due to the action of the HLE and the Rud syndromes since traits of both syndromes were observed in backcross larvae and adults.In germ cells of (thummi 2 x piger 8 ) hybrid males Rud syndrome induced chromosome aberra tions occur causing partial infertility (Hagele & Oschmann, 1987).If such events also take place in the germ cells of the (thummi 2 x piger 8) hybrid females, a frac tion of non or incomplete developing eggs must be ex pected after backcrossing.This injury hits the two backcross directions to the same extent since both have the same female parents.In the (thummi 2 x piger 8) 2 x thummi 8 cross the extent of a Rud syndrome induced embryo death is enhanced by the HLE syndrome.The lat ter syndrome is already known for its egg mortality in ducing effect in the {piger 2 x thummi 8) hybrids (Hagele, 1995).A basic presupposition for the HLE syn drome seems to be the fusion of a maternal piger with a paternal thummi genome in the egg.Such an event occurs in 6.25% of the eggs of {thummi 2 x piger 8) females fertilized by thummi sperm because this percentage of hy brid oocytes receive only piger chromosomes during meiosis assuming a crossing over incidence of zero.Thus, a HLE syndrome caused embryo lethality in the {thummi 2 x piger 8) 2 x thummi 8 cross may be responsible for the 10% greater embryo death in comparison to the {thummi 2 x piger 8) 2 x piger 8 backcross, a backcross where a HLE constellation is not apparent.Whether in the latter cross the occurrence of few malformed larvae indicates a mild HLE syndrome needs to be further inves tigated.
It may be questioned whether the large amount of em bryo mortality in the backcrosses is not only due to the supposed crosses but also to parthenogenetically develop ing eggs.Parthenogenesis has occasionally been observed in eggs of five members of the genus Chironomus (Beermann, 1955;Grodhaus, 1971).In Ch. pallidivittatus and Ch.tentans it gives rise to embryos which are not viable.We never observed developing eggs laid by {thummi 2 x piger tj) hybrid females which had no mating possibility.In this study we used a fivefold majority of males in the mating cages.Mating between thummi and piger or backcrossing occured without any problems.Successful cross ing is also indicated by the observation that all the supposed backcross individuals that we analyzed for sali vary gland chromosome aberrations showed the haploid chromosome complement of those males we used for backcrossing.
Ch. t. thummi and Ch.t. piger are reproductively isolated by a prezygotic isolation mechanism, that is a different swarming behavior of the two subspecies (Miehlbradt & Neumann, 1976).In spite of this mechanism a few hybrids were found in the wild (Hagele, 1999).After a breakdown of the prezygotic isolation barrier a postzygotic isolation barrier operates on the resulting hybrids.This barrier is represented by the two nonreciprocally and in opposite cross directions occurring HLE and Rud syndromes.The Rud syndrome minimizes fertility of the {thummi 2 x piger 8) hybrids by producing rudimentary gonads, while the HLE syndrome damages the {piger 2 x thummi 8) cross by a strong embryo mortality (Hagele et ah, 1995).The data presented in this paper show that reproductive isolation factors also act upon the backcrosses and even on those adults that survived the syndrome attacks on embryo development.The surviving backcrosses may be affected by Rud syndrome induced rudimentary ovaries and by HLE syndrome caused chromosome aberrations and malformations.
The results of this study touch the controversially dis cussed question whether Ch. t. thummi and Ch.t. piger should be regarded as full species or as subspecies (Keyl & Strenzke, 1956;Strenzke, 1959;Scharf, 1973;Credland, 1973;Miehlbradt & Neumann, 1976;Lindeberg & Wiederholm, 1979).One important argument in favor of the species status is that both forms have not been found to hybridize in nature even at areas where they are sympatric.Miehlbradt & Neumann (1976) report a different swarming behavior of thummi and piger in the wild re sulting in a strong species specific reproductive isolation.If the natural swarming behavior of thummi and piger is, for example, impeded during rearing in the laboratory, both forms hybridize without any difficulty.Surprisingly, nearly all the above authors mentioned that the hybrids are fully fertile and viable.Here it must be questioned whether such a situation would occur if the two Chirono mus forms have already evolved to the species level.On the other hand, even a strong isolation of the thummi and piger swarms can be disturbed by a sudden wind drift so that individuals of both forms may contact and occasion ally hybridize.When the hybrids are fully fertile and vi able it is improbable that a supposed species status would then be maintained.Recently we detected two hybrids in nature at a geographic region in Eastern Germany where no hybrids could have escaped from laboratories.This observation, and our findings that after hybridization a second reproductive isolation mechanism occurs resulting in a strongly reduced viability and fertility of the hybrids, argues against the taxonomic classification as full species.It may indicate that this second isolation mechanism evolved to prevent along with the first one a genetic re combination between the diverging thummi and piger.Our above findings are valid for different populations since we collected our strains at France, Greece and Ger many.Cytogenetic and molecular genetic studies also in dicate a very close relationship of thummi and piger.Since the work of Keyl & Strenzke (1956) and Keyl (1965) it is known that thummi and piger are not sepa rated by inversions, as is the case between other Chironomus species.In fact, homologous thummi and piger polytene chromosomes are homosequential.In a small number of interspersed pericentric bands a size differen tiation between homologous thummi and piger bands is demonstrable on the cytogenetic (Keyl & Strenzke, 1956;Keyl, 1965) and molecular level (Schmidt, 1984).How ever, between different thummi populations similar differ entiation processes are observed in some homologous bands (Keyl, 1957;Hagele, 1970).Therefore, the cytoge netically observable differences between thummi and pi ger are not species specific.In our opinion the taxonomic status of the two chironomids should be freshly discussed under a sophisticated consideration of the cytogenetic and genetic results.We assume that both Chironomus forms are involved in a process towards species formation and that they should, therefore, further be treated as subspe cies as originally proposed by Keyl & Strenzke (1956).

%
Fig. 2. Egg sizes (volume in nl) observed in egg masses of Ch. t. thummi (= th), Ch. t. piger (= pi), and of (th x pi) females backcrossed with th and pi males.Measurements of the egg length and egg diameters were done at the developmental stage El, E2, E3 and E4.The calculated volumes are arranged in six different volume categories reaching from 0.5 nl to 3.49 nl.The frequencies of the egg volumes in per cent.

Fig. 3 .
Fig. 3. Per cent frequencies of egg volumes (in nl) and of em bryo mortality in Ch. t. thummi (= th) and in the backcross of {Ch.t. thummi 2 x Ch. t. piger 8) hybrid females with Ch. t. thummi males within six volume categories.The data were ob tained from the developmental stages El, E2, E3 and E4.

Fig. 4 .
Fig. 4. Per cent frequency of egg volume (in nl) and of em bryo mortality in Ch. t. piger (= pi) and in the backcross of (Ch.t. thummi 9 x Ch. t. piger 3) females with Ch. t. piger males in six egg volume categories.The data were obtained from the de velopmental stages El, E2, E3 and E4.