Making sense of chromosome polymorphisms in two leptysmine grasshoppers

Abstract

The touchstone of the ‘New Synthesis’ was population cytogenetics –rather than genetics – due to the abundant polymorphic inversions in the genus Drosophila. Grasshoppers were not a material of choice because of their conservative karyotypes. However, nowadays seven species of Acrididae were described for polymorphic centric fusions, five of them in South-America. Leptysma argentina and the likely biocontrol of water-hyacinth Cornops aquaticum are semiaquatic Leptysminae (Acrididae: Orthoptera), polymorphic for centric fusions, supernumerary segments and a B-chromosome. We sought to demonstrate the operation of natural selection on them, by detecting: (I) latitudinal clines; (II) regression on environmental variables; (III) deviation from null models, such as linkage equilibrium; (IV) seasonal variation; (V) comparison between age classes and (VI) selection component analyses. All of them were confirmed in L. argentina, just (I) and (II) in C. aquaticum. Furthermore, the relationship between karyotype, phenotype and recombination was confirmed in both species. Karyotype–phenotype relationship may be due to the body enlargement the fusions are associated with, along with a latitudinal transition in voltinism. Karyotype-related recombination reduction in both species may help explain all fusion clines, although there is probably more than one factor at work. No effects were noticed for a supernumerary segment in L. argentina, but it is ubiquitous and certainly non-neutral. C. aquaticum is poised for introduction in South-Africa as a biocontrol of water-hyacinths; the recent discovery of four more segment polymorphisms may imply more chromosomal markers to make sense of its genetic system.

Introduction

The field of population genetics came into being in the 1920s and 1930s thanks to the theoretical basis framed by R.A. Fisher, J.B.S. Haldane and S. Wright among many others. Their achievement was to integrate the principles of Mendelian genetics – which had been rediscovered at the turn of the century – with Darwinian natural selection. Fisher, Haldane and Wright showed that Darwinism and Mendelism were not just compatible but they worked hand in glove with each other; this finding played a key part in the formulation of the ‘New Synthesis’ and explains why population genetics came to occupy such a pivotal role in evolutionary theory (Barton, 2022).

Sturtevant's (1917) conclusion that chromosome inversions suppress recombination in Drosophila melanogaster in heterozygotic state was later corroborated by means of cytogenetical analysis (Sturtevant and Beadle, 1936); from then on, the empirical study of population genetics became not one of genes, but one of chromosome polymorphisms (Kapun and Flatt, 2019). At that time, molecular gene markers were still unavailable, but Drosophila specialists in those early times had plenty of cytological variation to work with. Drosophila pseudoobscura has five chromosome pairs, with >30 different arrangements on chromosome 3 alone, due to more than ten overlapping inversions (Fuller et al., 2017); then the number of possible karyotypes is somewhere close to 3¹⁰ = 59.049, and this is an underestimation. Hence it is not a hyperbole to affirm that the fulcrum of population genetics – which, in time, is the hard core of the New Synthesis – was population cytogenetics. The existence of chromosome polymorphisms and the assignment of different fitness values to them became the touchstone of our current idea of biological evolution.

The adaptive importance of chromosome inversions in Drosophila consisted in the suppression of recombination in structural heterozygotes (Sturtevant and Beadle, 1936). These rearrangements would capture and accumulate ‘coadapted gene complexes’, which involve epistasis (Dobzhansky, 1950). Kirkpatrick and Barton (2006) proposed instead ‘local adaptation’ as an alternative (see below). There is plenty of evidence for both scenarios, coming from steep frequency clines, correlations with environmental (climatic) factors, predictable seasonal changes in inversion frequencies, evidence of epistatic selection and so forth (see Kapun et al., 2016 and references therein).

Grasshopper karyotypes, by contrast, are rather conservative, with a prevailing chromosome number of 22 + X0/XX in males and females, respectively (Hewitt, 1979). Therefore, the study of inversion polymorphisms in this group was restricted to a few species (White, 1973). The Australian grasshopper Moraba ( = Keyacris) scurra was thoroughly studied by White and co-workers in the 1950s and 1960s, and a great deal of effort was devoted to the assignment of adaptive value to these polymorphisms (White, 1958, 1973; Lewontin and White, 1960; White and Andrew, 1960). Another inversion-bearing Australian grasshopper with consistent latitudinal variation is Caledia captiva (Groeters and Shaw, 1992; Shaw and Groeters, 1998; Shaw et al., 2011). In South America, Vaio et al. (1979) and Goñi et al. (1985) discovered several chromosome inversions in some Argentine populations of Trimerotropis pallidipennis, but the authors concluded that the inversions did not follow any environmental or geographic patterns). Later analysis (Confalonieri, 1988, 1994; Confalonieri and Colombo, 1989; Colombo and Confalonieri, 1996; see Guzmán et al., 2022 and quotations therein) showed the geographical and environmental patterns of these inversions and their stability along space and time. In chromosome inversions, local adaptation favours reduced recombination between locally adapted loci (Charlesworth and Charlesworth, 1979; Lenormand and Otto, 2000). This is a well-studied mechanism for the evolution of chromosome inversions whereby they capture two or more locally adapted alleles; later on, this high fitness chromosome spreads in the population in which those alleles are beneficial (Kirkpatrick and Barton, 2006; Charlesworth and Barton, 2018). By reducing recombination, inversions preserve the divergence between populations in the presence of gene flow, setting the stage either for permanent polymorphism or potential speciation (Navarro and Barton, 2003; Kirkpatrick and Barton, 2006).

This was the picture for population cytogeneticists up to the 1970s–1980s. But, at a time when allozyme electrophoresis, RFLPs and microsatellite polymorphisms allowed to obtain several molecular alleles which were easy to analyse simultaneously, population cytogenetics fell out of fashion because it could not keep pace with the variability evidenced by molecular research (Kirkpatrick, 2010; Fuller et al., 2019).

Centric fusion polymorphisms

Robertsonian translocations ( = centric fusions) owe their name to W.R.B. Robertson, who first discovered a V-shaped structure in a metaphase I of Locusta migratoria which was not present in all the individuals (Robertson, 1916). Polymorphic centric fusions are more frequent among mammals, the house mouse being the best studied case (Davisson and Akeson, 1993; Castiglia and Capanna, 2000; Dumas and Britton-Davidian, 2002; Merico et al., 2013; Förster et al., 2016) followed by phyllotines (Lanzone et al., 2015), Holochilus (D'Elia et al., 2015), the genus Ctenomys (Bidau et al., 2003) and the common shrew Sorex araneus (Giagia-Athanasopoulou and Searle, 2003; Basset et al., 2019). They were also thoroughly studied in populations of some fishes (Farr-Cox et al., 1996; Ostberg et al., 2013; Wellband et al., 2019), and the gastropod Nucella lapillus (Pascoe, 2006).

Among grasshoppers, McClung (1917) discovered a polymorphism for a centric fusion between the X chromosome and an autosome in Hesperotettix viridis; six more fusion polymorphic grasshopper species were discovered ever since, all of them from the New World (Hewitt and Schroeter, 1968; Colombo, 1987, 1989, 1990, 2007, 2008; Bidau and Mirol, 1988; Bidau, 1990; Remis, 2008; Taffarel et al., 2015; Castillo et al., 2018; see Colombo, 2013 for a review). Other species were omitted from this list because they are polytypic rather than polymorphic (i.e. the variation lies between, not within populations, and heterozygotes are indeed interracial hybrids), such as the remarkable wetas from New Zealand (McKean et al., 2015).

Fusions are an altogether different matter. The reduction of recombination they trigger is due both to the loss of independent segregation and because of a more distal localisation of chiasmata both in heterozygotes and in homozygotes (Colombo, 1987, 1989, 1990; Bidau, 1990; Remis, 1990; Castiglia and Capanna, 2000; Bidau et al., 2001; Dumas and Britton-Davidian, 2002). On the other hand, as mentioned above, inversions reduce recombination in heterozygotes alone. Because of this asymmetry, it is unclear whether the evolutionary forces that operate on locally adapted inversions can act on centric fusions in a comparable way (Guerrero and Kirkpatrick, 2014).

In our laboratory two acridid models have been worked out: both of them are semiaquatic leptysmine grasshoppers which are unusually polymorphic from the cytogenetic standpoint, namely Leptysma argentina and Cornops aquaticum.

Bidau and Hasson (1984) reported the survey of two contiguous populations, sited 400 m apart, of L. argentina, a grasshopper whose range is restricted to Central-North Eastern Argentina and all of Uruguay. It is unusually polymorphic for a grasshopper, with a centric fusion between chromosomes 3 and 6 of the basic complement (fusion 3/6) and a large supernumerary segment on the smallest member of the karyotype, namely chromosome 10. The number of possible karyotypes was then 3² = 9 (18 if we added the presence or absence of a B-chromosome).

Besides, C. aquaticum is an all-out Neotropical grasshopper, and it occurs between the south of Mexico (23°N) and the latitude of Buenos Aires and Punta del Este (34°S–35°S) (Adis et al., 2007). Prior to coming under our scrutiny, C. aquaticum was an already appealing semiaquatic leptysmine because it has a strict host ecological specificity (it feeds on, and lays eggs in, floating plants of the also Neotropical Pontederiaceae family, called ‘water-hyacinths’) exclusively (Adis and Junk, 2003). Moreover, after its introduction elsewhere as an ornamental, the blue-flowered water-hyacinth Pontederia crassipes became a weed that infested freshwater environments in tropical, subtropical and even temperate regions all around the world (Center et al., 2002; Coetzee et al., 2011). That's why C. aquaticum has been considered as a possible biological control of the obnoxious water-hyacinth (Oberholzer and Hill, 2001) and a great deal of effort has been devoted to study its phenology and ecology.

Mesa (1956) and Mesa et al. (1982) reported the presence of three polymorphic centric fusions in C. aquaticum in Uruguay (about lat. 31°S). Rocha et al. (2004) described its karyotype for several collection sites in the state of Pernambuco (Brazil), at lat 8°S. Cornops aquaticum bears an all-telocentric karyotype, with three large (L), five medium (M) and three small (S) pairs. The sex determination system is X0/XX and the X is M. However, the centric fusions were nowhere to be seen. In Colombo (2007, 2008, 2009), Colombo and Remis (2018), Colombo et al. (2021, 2023) and Romero et al. (2014, 2017) we found and described all three fusion polymorphisms forming clines in the lower course of the Paraná and Uruguay Rivers.

The purpose of the present contribution consists in reviewing the quest, in a material generally regarded as intractable, for a feat any population (cyto)geneticist would fancy about: to have plenty of heritable variation to work with. This was the basis for our endeavour, i.e. the assignment of differential adaptive values for the different chromosome morphs. We consider the results of this research programme on the (mostly) univoltine, more terrestrial, well-behaved L. argentina, with its synchronous generations, as a necessary step for the study of the unruly C. aquaticum, which lives on the floating mats of an aquatic plant. This quest may yield a twofold benefit, given that C. aquaticum, beyond its theoretical value, is poised for introduction in many alien environments where P. crassipes has made inroads. The knowledge of the genetic system of a biocontrol is central for its success, and the parallel clines for three centric fusions have probably much to say about it.

Results and discussion

Leptysma argentina

The aim of the present cytological research in L. argentina consisted in testing the operation of natural selection, chiefly on the centric fusion (F) and the large supernumerary segment (S) polymorphisms, and to a lesser extent on the B-chromosome. Having analysed 11 populations over a span of 7 degrees of latitude (between 27°S and 34°S) (table 1, fig. 1) a latitudinal cline for F and another for B (which forms a reverse slope) were detected. Data come mostly from already published material (Colombo, 1987, 1989, 1990, 1993, 1997, 2000; Norry and Colombo, 1999; Colombo et al., 2001, 2004); data from the Tucumán (T) population remained unpublished until now.

Table 1. Leptysma argentina: geographic and climatic variables associated with the populations studied here

Lat, South latitude; MinT, mean annual minimum temperature; MaxT, mean annual maximum temperature (in °C); Monthly rainfall, average yearly rainfall (in mm); Etp/yr, mean yearly evapotranspiration (in mm/year). Frequencies of chromosome polymorphisms for F, S and B (see text) are also indicated. N = sample size. Abbreviations of grasshopper populations are given in fig. 1.

^a Previously unpublished data.

Figure 1. Sampling area and Leptysma argentina populations included in this study: Tucumán (T), Santa Fe (SF), Los Loros (LL), Yarará Guazú (YG), El Palmar (EP), Zárate (ZA), Puerto Talavera (PT), Isla Talavera (IL), Río Luján (RL) and Pilar (P). Pie diagrams indicate the frequencies of the centric fusion 3/6 (F), the segment S and the B-chromosome (B) in dark grey. Data from T were unpublished so far.

Latitudinal clines are compatible with (but not conclusive of) the operation of natural selection (Endler, 1977). Endler (1986) listed ten criteria to detect the operation of natural selection in the wild, as follows:

1. I) Correlations of the polymorphisms with environmental (climatic) variables;

2. II) Comparison between closely related species;

3. III) Comparison between unrelated species living in similar habitats;

4. IV) Deviation from formal null-models, such as Hardy–Weinberg or gametic phase (linkage) equilibria;

5. V) Long-term study of trait frequency distribution, such as (a) long-term stability or (b) regular directional change;

6. VI) Perturbation of natural populations such as: (a) predictable seasonal changes and (b) introduction into new equivalent and non-equivalent habitats;

7. VII) Genetic demography or cohort analysis;

8. VIII) Comparisons among age-classes or life-history classes such as (a) comparison of breeding and non-breeding adults or (b) comparison of juveniles and adults and (c) fitness component analysis such as mating ability, mating preference, fecundity, viability, age of reproduction, etc.

9. IX) Non-equilibrium predictions of changes in trait distributions; and

10. X) Equilibrium predictions about trait frequency distribution.

Criteria II, III, VII, IX and X would not (likely) apply for the purposes of the present research and they were not discussed.

As for (I), besides the latitudinal clinal variation, the statistical analysis showed that:

1. i) F correlates negatively with average and maximum temperature (P = 0.0001 and 0.0002), positively with latitude (P = 0.0005) and negatively with yearly evapotranspiration (Etp yr^–1) (P = 0.0111) and rainfall (P = 0.0189). A stepwise regression analysis showed a positive regression on latitude (P = 0.00063) and a negative one on Etp yr^–1 (P < 0.00181).

2. ii) B correlates negatively with latitude (P = 0.0001) and positively with average and maximum temperature (P = 0.0002 and P = 0.0076 respectively). As for the stepwise regression analysis, B displays an inverted pattern with respect to F, with a highly significant negative selection on latitude (P = 0.0063) and a positive significant one on average temperature (P 0.0142),

3. iii) All 11 populations considered, S showed a significant negative regression against rainfall (P = 0.0189) despite the absence of any obvious geographic pattern (table 1, fig. 1). However, the IT population is very much of an outlier. A small sample from an extremely small population, in an area of intense human perturbation, it was suppressed from a second statistical analysis. As supposed, negative correlation with rainfall became highly significant (P = 0.0009).

It may seem odd that some of the polymorphisms show better regressions on latitude than on the environmental variable they ought to be related with, presumably temperature (but also rainfall in this region of South America). However, this seems to be a frequent feature of chromosomal clines of other organisms such as Drosophila (see Kapun et al., 2016 and references therein). Apparently this may indicate that the underlying factor of clinal variation is related with temperature, but not temperature as such (Krimbas, 1992).

(II) Two areas were sampled with special intensity: the northeast of Buenos Aires province (RL, O, Z, IT, PT) (lat. 34°S) and a protected area 300 km further north (‘El Palmar’, LL, EP, YG) (lat. 31.8°S). The karyotypes always adjusted to the Hardy–Weinberg expectations everywhere, but preliminary analyses in the first area suggested that

1. (i) there was a non-significant excess for S heterozygotes (BS) at the onset of the breeding season (spring), and a non-significant defect at the end (summer); the X² comparison between both samples was significant, although the frequency of S did not change.

2. (ii) By contrast, the F frequency showed a consistent increase between spring and summer samples everywhere. The following year, however, the population had the same F frequency as the precedent. The polymorphism was apparently stable, but the frequencies changed along the year. (i) and (ii) turned out to be the first evidences of seasonal selection for F and S (Endler's VI a) (Colombo, 1989).

This evidence of cyclical change was constant for over 20 years, between, 1983 and 2004 and was replicated elsewhere, i.e. in the ‘El Palmar’ area, which was geographically distant and with different karyotypic frequencies when compared to the previous one.

Gametic phase (linkage) equilibrium analyses were carried out between F and S, both in early and late spring in both populations, taking advantage of the fact that L. argentina has synchronous and non-overlapping generations (Aquino and Turk, 1997). In the El Palmar population, young individuals (spring sample) were in equilibrium; in aged ones (summer sample) gametic phase disequilibrium was significant instead, with a prevalence – according to the simulation – of FB (fused and without S) and US (unfused and with the segment) gametes (Colombo, 1993). The same result was replicated further south, with a much higher F frequency, only that linkage disequilibrium D was non-significant (Colombo, 2000). This is yet another evidence (Endler's IV), albeit indirect, of the operation of natural selection (Endler, 1986). Therefore, gametic phase equilibrium is a null model this species does not fit into.

As generations were synchronous and non-overlapping in this species (Aquino and Turk, 1997), age-class comparisons in both areas were possible. We compared the frequencies of all nine karyotypes (for F and S) at spring (young males) and summer (aged males) in both sampling regions. In the young male sample from El Palmar, karyotypes for both polymorphisms distributed independently; in aged ones the FF individuals tended to be unsegmented (BB), and the segmented (SS) ones tended to be unfused (UU); furthermore, there was always a clear excess of double heterozygotes (Colombo, 1993). The same result was obtained further south, where F displays a much higher frequency in both places, only with a different frequency (Colombo, 1993; Colombo et al., 2001). All individuals are already adult by the onset of October (spring), and at this latitude there is just one generation a year (personal observation); therefore, the change of frequency must be due to differential survival alone. According to Endler (1986), comparison between age classes, in this case juveniles and adults (Endler's VIIIb), is a fairly direct evidence of natural selection for both polymorphisms, and it may come to explain the gametic disequilibrium described above.

(IV) Coincidental correlations between environmental variables both between and within populations along the year: the evidence here is contradictory and spotty. F is inversely correlated with temperature. However, it always increases its frequency from spring to summer; F showed a positive regression on Tmax along a series of seven (not strictly consecutive) years; the regression on rainfall changes its sign in El Palmar with respect to Buenos Aires' NE. Segment S does not change its frequency along the season, it is just the heterozygotes BS who decrease among the aged males. No attempt to perform these studies on the B-chromosome was done.

(V) Selection component analyses (Endler's VIIIc) confirmed that the F dosage was correlated with adult male longevity (Norry and Colombo, 1999). The selective male longevity value of F is clearer since its frequency has always increased along the breeding season, i.e. from October to December (spring to summer) (Colombo, 1993; Colombo et al., 2001).

Moreover, a unisexual approach revealed that the F carriers had an advantage in mating selection (Colombo et al., 2001). Furthermore, F dosage also increased female mating success (Colombo et al., 2004). No effect on either phenotype or mating success was detected for segment S.

The quest for the selection target

At the time these studies were undertaken for the first time, except for a few cases (White and Andrew, 1960; Butlin et al., 1982) chromosomal change was generally deemed to be neutral for the external phenotype (Lande, 1979; John, 1983) and it was believed that only the internal phenotype (i.e. meiosis) was affected by them (John, 1983). The study of body size in all three karyotypes of F on most body size-related traits in two populations of L. argentina (Colombo, 1989, 1997) revealed that F carriers were significantly larger, and that there is a good correlation between its frequency, latitude and body size-related traits. The populations of ‘El Palmar’ (lat. 31.8°) had smaller individuals and a lower F frequency; those of Buenos Aires (lat. 34°C) have a larger body size, and a higher F frequency (Colombo, 1989, 1997). The karyotype–body size correlation was consistent and highly significant. Santa Fe individuals, one degree of latitude further north, have no F and are smaller than the others; finally, males from the norther monomorphic population of Tucumán (27°S) are tiny (data unpublished).

The selection component analyses mentioned above (Endler's VIIIc criteria) showed that F is positively selected for longevity from spring to summer, i.e. with increasing temperature; aged males are larger than young ones, the direct target of selection being thorax height (Norry and Colombo, 1999). Additionally, when mating choice was studied it was found that sexual male selection advantage is associated with a larger third femur length in males, and total body length (TL) as well as third femur length (FL) in females. However, these variables are not the actual selection targets (Colombo et al., 2001, 2004).

To summarise:

1. i) Along the latitudinal cline, F correlates negatively with temperature but positively with size.

2. ii) Within a population, at a given time, F correlates positively with body size.

3. iii) Within a population, F increases from spring to summer (i.e. its frequency correlates positively with temperature) and, as always, keeps correlating positively with body size.

This supposed fit to the ‘Bergmann rule’ would rather reflect the case of a change of voltinism along a latitudinal transect (Shelomi, 2012). There is evidence of this in Aquino and Turk (1997). Further analysis is needed to shed light on the other two phenomena.

Both longevity and mating choice would lead to an increase of F; however, along the more than 20 years of monitoring these frequencies did not modify.

No phenotypic effect could be attributed on a consistent basis to segment S, nor was it revealed to be associated to mating success. It is certainly correlated against rainfall but no geographic pattern was observed. The target of selection on S (detected through gametic phase disequilibrium and joint age class comparison with F) remains unidentified.

Effects of centric fusions on recombination

In addition to the effects of F on phenotype, there were more evident outcomes on cytological features that led to a significant reduction of recombination in F carriers, especially heterozygotes (Colombo, 1987, 1989, 1990):

1. 1) Formation of a new linkage group from the two erstwhile independently segregating chromosome pairs.

2. 2) Reduction of proximal (P) and interstitial (I) chiasma formation in the fusion trivalent, which elicits an abrupt decrease of chiasma frequency, and a strict distal (D) localisation in these elements (intrachromosomal effects) (P, I and D with respect to the centromere) (Colombo, 1990).

3. 3) Decrease in the frequency of total (T), proximal (P) and interstitial (I) chiasmata, and a slight increase in the frequency of distal chiasma frequency (D) in the fusion bivalent (but not so marked as that in the fusion trivalent). The rise in D takes place at the expense of I (both are negatively correlated) and the decrease of T is explained by a reduction of P (both are positively correlated, Colombo, 1990).

4. 4) These within-population results were mirrored by the comparison among populations that display different F frequencies (Colombo, 1989).

These results (Colombo, 1987, 1989, 1990) along with those of Bidau and Mirol (1988), Bidau (1990), Remis (1990), Colombo (2007, 2008), Taffarel et al. (2015), Martí et al. (2017) and Castillo et al. (2018) corroborated that this feature of Robertsonian polymorphisms is consistent among grasshoppers (Colombo, 2013) and beyond. Suppression of recombination in trivalents of heterozygotes and chiasma reduction in fusion bivalents of fusion homozygotes has been reported in house mice (Davisson and Akeson, 1993; Bidau et al., 2001; Dumas and Britton-Davidian, 2002; Merico et al., 2013; Förster et al., 2016), and plants (Rieseberg, 2001). Moreover, recombination is shifted towards distal parts of heterozygous metacentrics in the shrew S. araneus (Borodin et al., 2008).

Comparison between a spontaneous and a stable polymorphic centric fusion

A spontaneous fusion between pairs 5 and 7 (fusion 5/7) was found in a double heterozygote male that also conveyed the 3-3/6-6 trivalent. The number and position of chiasmata in both trivalents were compared, the 5/7 trivalent did not show any change of proximal (to the centromere) chiasma number or position, whereas the effect of the 3/6 centric fusion was conspicuous (Colombo, 1987). The higher frequency of proximal chiasmata in 5/7 trivalent when compared with the 3/6 one was associated with an increased proportion of linear orientation, thus conducing to the formation of unbalanced gametes. As both trivalents occurred in the same individual, it can be assumed that the genetic background was the same.

Hence, it was concluded that the 3/6 fusion triggered a shift of chiasma position towards the distal extremes – not so the 5/7 – and this may have been crucial in the maintenance of the polymorphism. Furthermore, the reduction of recombination between the arms of the newly arisen metacentric and its acrocentric homologues would have entailed a genetic differentiation – above all close to the centromere – between them. Rieseberg and Burke (2001) stated that rearrangements reduce recombination in the vicinity of the rupture points; in this case, this point is the centromere, where two linkage groups no longer segregate independently. This linkage disequilibrium (LD) may have encompassed genes for local adaptation that would explain the creation of a chromosomal cline such as that exposed here. Wellerenreuther et al. (2017) demonstrated the existence of such genes within the inversions of the dipteran Coelopa frigida, which shows clines for inversions in the North and Baltic Seas correlated with temperature.

Three parallel fusion clines: the case of C. aquaticum

Here we review the results of our previous work on newly collected populations of C. aquaticum on the abundant, southwardly flowing rivers Paraná and Uruguay (Colombo, 2007, 2008, 2009; Romero et al., 2014, 2017; Colombo and Remis, 2018; Colombo et al., 2021, 2023). Both large parallel rivers are born in the Brazilian rainforest and converge into the River Plate at latitude 34.5°S in the Paraná Delta, just north of Buenos Aires city. Seven populations from Argentina were sampled along the Paraná Basin since 2005, namely: Corrientes (LP, 27°S), Santa Fe (SF, 31.8°S), Rosario (RO, 33°S), San Nicolás (SN, 33.33°S), San Pedro (SP, 33.5°S), Zárate (ZA, 34°S) and Tigre (TI, 34.4°S). Three parallel fusion clines were found for fusions between L and M pairs, namely 2/5, 1/6 and 3/4. In fact, these fusions (the same ones reported by Mesa (1956) and Mesa et al. (1982)) were detected along the middle and lower course of the river; therefore, there are 3³ = 27 possible karyotypes. Fusion frequencies increase southward and downstream and the clines are steep: at the southernmost point they arrive to a frequency of 0.85 (they do not reach fixation) (Colombo, 2008). At this point the Paraná River flows into the River Plate, where there are no water-hyacinths, and consequently no C. aquaticum. In table 2a the frequencies for each fusion are shown, along with environmental variables they were correlated with, such as maximal, minimal and average temperature and yearly rainfall. Furthermore, the average fusion frequency per individual (fpi) is displayed for each population.

Table 2. Geographic and climatic variables associated with Cornops aquaticum populations from the Paraná (a) and Uruguay (b) River basins

Lat, South latitude; MinT, mean annual minimum temperature; MaxT, mean annual maximum temperature (in °C); AvgT, mean annual average temperature; Rainfall, average yearly rainfall (in mm); fusion 2/5, 1/6 and 3/4 frequencies are also indicated; fpi: average fusion number per population. Abbreviations of grasshopper populations are given in fig. 2.

^a In these populations fpi could be scored, but they were not identified due to poor fixation.

^b May also be regarded as part of the Uruguay Basin (see explanation in the text). All of the data were previously published (see quotations in the text).

Later, in co-operation with Dr M.I Remis and Lic. M.L. Romero, we set forth to sample six more populations on the parallel flowing Uruguay Basin, another oversized tributary of the River Plate, with an additional latitudinal cline, only steeper. This cline displays higher fusion frequencies when compared to the same latitude of the Paraná cline up to the GU population, which has an average number of fusions per individual (fpi) of 3.3 (table 2, fig. 2). For example, SN, on the Paraná basin, at the same latitude has an fpi of only 0.57.

Figure 2. Sampling area and Cornops aquaticum populations included in this study. Populations from the Paraná River Basin: Laguna Pampín (LP), Santa Fe (SF), Rosario (RO), San Nicolás (SN), San Pedro (SP), Villa Paranacito (VP), Zárate (ZA) and Tigre (TI). Populations from Uruguay River Basin: Monte Caseros (MC), Salto Grande (SG), Colón (CO), Concepción del Uruguay (CU) and Gualeguaychú (GU) Pie diagrams indicate the frequencies of the centric fusions 2/5, 1/6 and 3/4 in dark grey. All data were previously published elsewhere (see quotations in the text).

However, further downstream from GU, Villa Paranacito (VP) has an fpi of only 1.65. A possible explanation for this anomaly may be that the Paranacito River, sited at the deltaic inundation zone of both convergent basins, is one of the many sprawling branches of the much mightier Paraná (hence its name), and flows into the final course of the Uruguay River. It communicates both basins: the stream (and the floods) of the Paraná basin would drag the floating mats of P. crassipes in which C. aquaticum lives into the smaller Uruguay. We think therefore that the fact that VP fusion frequencies fit closer to the Paraná than to the Uruguay cline is not a shocking anomaly. In fact, if we include it among the Paraná basin populations (to which it may arguably belong), VP fits neatly into this cline on account of its latitude (table 2b).

Correlation with environmental variables

The latitudinal, parallel clines between all three fusions and both rivers set the stage for level I in Endler's scale (1986): correlation between the trait (fusions in this case) and environmental variables. First a correlation was performed between individual fusions (and fpi) and climatic variables. All three fusions and fpi were positively correlated with latitude (cf. .L. argentina) and highly and negatively so with maximum temperature (P = 0.0036 for 2/5, P = 0.0017 for 1/6 and P = 0.0037 for 3/4; P = 0.0004 for fpi) (Colombo and Remis, 2018). None of the other variables analysed displayed any significant association with the fusions. As for stepwise correlations, there were no changes: all three fusions and fpi showed a negative regression on maximum temperature (2/5: P = 0.0076; 1/6: P = 0.0026; 3/4: P = 0.0040 and fpi: P = 0.0002). None of the other variables were even allowed into the analysis.

Phenotypic effects of C. aquaticum's fusions

Romero et al. (2014) found an association between the fused 1/6 chromosome dosage and body size, which was significant for tegmina length, the provenance of the sample being the deltaic ZA and SP totalling 32 males. Colombo and Remis (2018) analysed a more robust sample of 272 individuals, and the joint analysis revealed that fusions 2/5 and 3/4 are significantly associated with larger body size-related traits such as femur and tegmen length, fusion 1/6 not showing any significant effect. Therefore, the average fusion number per individual (fpi), which can take any number between zero and 6, is positively correlated with tegmen and femur length; otherwise stated, the regression of body size on fusion number is significant (Colombo and Remis, 2018).

As fusion frequency increases southwards, and fusion number (0–6) is correlated with body size, it could be expected that body size correlates with latitude as well. Indeed, it's the way round: body size decreases with latitude and increases correspondingly with temperature and rainfall (Colombo and Remis, 2018). In this case, the within-population effect is unlinked with the difference between populations.

This finding contradicts Adis et al. (2008), who studied morphometrical variables of C. aquaticum all along a latitudinal transect between Trinidad (10°N) in the Caribbean and Punta del Este (35°S) on the River Plate's mouth. The insects were larger as the sampling site approached the Equator in this continent-wide study. The contradiction may lay on the scale of the survey. At Argentina's latitude C. aquaticum is bivoltine, but further north the number of generations a year increases, forming (seemingly) a seesaw effect. We failed to notice this effect in L. argentina because it does not occur much further north than Tucumán (27°S), where it is bivoltine and tiny, and further south is univoltine, with plenty of time for development.

We observed that, in C. aquaticum, there is a correlation between fusion dosage and body size (although fusion 1/6 does not affect body size related traits in C. aquaticum, Romero et al., 2014; Colombo and Remis, 2018).

Needless to say that the correlation of body size on fusion dosage does not mean necessarily that fusions are causing it: it is just a correlation. Secondly: if we do accept that some fusion is causing an effect, it would be too much of a leap of faith to assume that it is its function. However, it must be noted that the association between karyotype and phenotype is consistent in both species. This effect may be the target of selection (Norry and Colombo, 1999) or not (Colombo et al., 2001, 2004).

Did we demonstrate selection in these grasshoppers?

In the case of L. argentina there are many evidences of natural selection acting on the F and S polymorphisms (no matter how indirect this evidence may be) especially when we focus on the centric fusion: Endler's I, IV, V, VI and VIII criteria are comfortably met. In C. aquaticum we have three centric fusions, forming steep parallel clines, starting at the same point and always with the same frequency until they don't reach fixation because even the Paraná River was too short for them. To be sure, we have the Uruguay River besides, with a comparable cline. Too much of a coincidence. In addition, all three fusions correlate negatively and highly significantly with maximum temperature (same as L. argentina). But not much else.

Colombo (2007) reported the recombination effects of all three fusions, which were coincidental in both species, i.e. (1) formation of a new linkage group from the two erstwhile independently segregating chromosome pairs, and (2) increase of distal (D) chiasma localisation, to the expense of proximal (P) and interstitial (I) chiasma formation in the chromosomes involved in the fusion, leading to a reduction of recombination in the fusion bivalent and an almost suppression in the trivalent.

A constant feature of large centric fusion polymorphisms is that recombination between the resulting metacentric and their acrocentric homologues is greatly reduced in the resulting trivalent of heterozygotes. The reason for this is that unrestricted chiasma formation in trivalents would lead to proximal and interstitial (with respect to the centromere) chiasmata. This would entail a rise of linear orientation of the trivalent due to spatial reasons to the expense of alternate orientation that renders V-shaped trivalents and balanced gametes (Bidau and Mirol, 1988; Colombo, 2009, 2013). The ensuing reduction of recombination between the fused metacentric and the acrocentrics may provoke a partial genetic differentiation between both, especially in the regions close to the centromere, thus capturing locally adapted genes.

This differentiation between the rearranged and the non-rearranged homologue chromosomes of structural heterozygotes was often invoked as an explanation for inversion polymorphisms in Drosophila (Kirkpatrick, 2010), which capture advantageous haplotypes in linkage disequilibrium (Kirkpatrick and Barton, 2006; Charlesworth and Barton, 2018). According to this hypothesis the inversion is favoured because it prevents the breakdown of linkage disequilibrium caused by migration.

The parallel with C. aquaticum suggests an origin, somewhere in the Paraná middle course, of the Rb fusions and a local adaptation to it, and a southward migration helped by the current that drives the water-hyacinth floating meadows downstream. Upstream migration would be difficult for obvious reasons, although not impossible for a good flyer as C. aquaticum (Colombo et al., 2021).

The hypothesis of the reduced recombination is endorsed by the fact that the most downstream populations, with high fusion frequencies, are both the most genetically disparate and most genetically impoverished when a microsatellite study was performed (Romero et al., 2017). The other populations could not be distinguished by the model and were lumped together under a Bayesian approach (Romero et al., 2017). When it came to population genetics, there were just three populations: the deltaic ZA and TI, and all the other lumped together.

Being neutral, these microsatellite distributions could be taken as null hypotheses when it came to compare the morphological distribution; in this case, the morphological Bayesian approach singled out just the northern, distant and subtropical LP and SF, pooling together all the others, deltaic or not, chromosomally polymorphic or not. Therefore, the morphological pattern must reflect season length and water availability and does not seem to reflect either genetic or chromosomal patterns (Colombo et al., 2021).

Conclusions

The evidences for natural selection operating on chromosome polymorphisms in wild populations of L. argentina are based on the rejection of several null hypotheses, such as random geographical and environmental distribution of the rearrangements, gametic phase equilibrium, unsteadiness of chromosome morphs along the years (or the seasons), non-significant differences between age classes and random mating (Endler, 1986). Additional, more indirect (but concurrent) evidence for F are: consistent correlation with phenotypic variables and an inverted association with recombination, due to both the loss of independent segregation and of proximal chiasma reduction.

Cornops aquaticum is less docile. All of the evidence we got is circumstancial, such as geographical and environmental correlations, association with phenotypic variables and the well-known twofold reduction of recombination of Robertsonian rearrangements. What makes it special is the amount of centric fusions involved in the clines (second only to Dichroplus pratensis, Bidau and Martí, 2005) and the microsatellite study performed by us (Romero et al., 2017) which revealed that the most polymorphic populations (ZA and TI) had the least genetic variability and were genetically detached from the rest.

However, part of Endler's sixth criterion (1986) consists in the perturbation of natural populations by introduction into equivalent, or non-equivalent habitats: the reproduction of clinal patterns where the species did not use to live. There are many examples of it in the literature, to mention only two: D. melanogaster (Kapun et al., 2016) and D. subobscura (Prevosti el al., 1990) whose latitudinal clines from European flies reproduced elsewhere upon spontaneous transplant, even in an inverted fashion for South American populations. If C. aquaticum is introduced in South-Africa as a plague control – at roughly the same latitude studied by us in South America, i.e. 27°S and 34°S – and the cline is spontaneously reproduced, that would be the real and ultimate test of adaptive Robertsonian clines among the Orthoptera.

One more segment yet

All of L. argentina's sampled populations are polymorphic for segment S; being heterochromatic, it should bear no genes. But it's always there. It shows highly significantly negative regression on rainfall, and heterozygotes are subject to longevity selection in interaction with F. In addition, heterozygote frequency for S decreases significantly along the breeding season (see above). We cannot even start to speculate about the causes, only notice that neutrality may be an inadequate explanation for these phenomena.

Colombo (2005) reported the existence of a latitudinal cline for a massive SS polymorphism on pair 9 of C. aquaticum between 27ªS and 34ªS whose slope was contrary to that of the fusions. Furthermore, given that in our long and extensive survey of this species we frequently noticed the presence of other heteromorphic bivalents, we performed the time-honoured C-banding technique to understand this abundance of supernumerary heterochromatin as well (Colombo et al., 2023). By doing this, we detected three more C-positive polymorphic supernumerary segments in two nearby populations (ZA and VP). Now we have seven chromosomal polymorphisms in C. aquaticum to make sense of: three centric fusions, plus four supernumerary segments – at the moment. It makes 3⁷ = 2187 possible different karyotypes: not Drosophila, but not too bad for a grasshopper.