Arizona Sky Islands: a natural laboratory for studying the EvoDevo of ants

Post by Marie-Julie Favé (MJ) – PhD student at McGill in the group of Ehab Abouheif, now postdoc at Université de Montréal and the Ontario Institute for Cancer Research.

This journey that started in Tucson many years ago finally made its tortuous way into the peer-reviewed system and got published recently here [1]. We also received some nice news coverage here! 

-----

One of my childhood cartoon heroes was Lucky Luke. For those who are not well versed in the franco-belgian universe of cartoons, Lucky Luke is a solitary cowboy living circa 1885 in the American Southwest "who shoots faster than his shadow", and whose life consisted of the recurring task of capturing the Dalton brothers, a pair of gangsters. 

When I was a little older, I dove into the universe of Sergio Leone and watched Once Upon a Time in The West more times than I can remember, imagining myself being Claudia Cardinale captured by the villain Hendry Fonda. I imagined immense hot plains of red dust, harmonicas, horses, dust balls... This was my sole and idyllic knowledge of the American Southwest when I abruptly landed alone in Tucson in August 2007 with my giant field bag as a fresh PhD student.

Fig 1: (A) Idyllic Southwest image of MJ’s childhood. (B) Less-than-idyllic landing of MJ in Tucson. This is a picture of the Speedway in the 70s when it won the "ugliest US highway" but believe me it hasn’t gotten any better since then.

Then I saw those mountains. You can actually see them quite well as the plane makes its approach to Tucson, flying right by the southern slope of Mont Lemmon in the Santa Catalinas range. I never imagined before that the desert would be punctuated with green, luxurious, high-altitude mountains harbouring such incredible biodiversity.

 Fig 2: The Sky Islands from the plane above Tucson, and from the top of the Santa Catalinas

The Arizona Sky Islands, also known as the Madrean archipelago, consist of ~45 mountains, the largest reaching ~3000m elevation, and are scattered across the US–Mexican border. They are surrounded by a dry, hot (45°C+), arid desert where we often spotted roadrunners (beep beep), rattlesnakes, giant scorpions, and other typical desert fauna/flora – as well as many of the amazing desert ant species that call Arizona their home, like Pheidole rhea (which also led to another of Ehab’s fascinating projects, [2]), Pogonomyrmex barbatus [3], and Myrmecocystus spp. (honey pot ants). However, being one of those cold-adapted Northern Canadians, I was more than happy to head, each morning, towards the higher, cooler elevations – wallowing in regret for having gulped down another Arizonan delicacy at Jerry’s breakfast place [4].

Fig 3: Cinnamon bun @ Jerry’s. The poor thing is trying to swim out of the ‘butter’ pool. I will help him with my fork.

From the dry, arid desert, every gain in elevation of a few hundred meters brings with it a novel landscape and different species. Junipers, agaves, and oak trees start to make their appearance around 1500m elevation, soon to be replaced by Douglas-firs and pinyon pines at the highest elevations, around 3000m for Mount Lemmon.  

Interestingly, Mount Lemmon was named after Sara Plummer Lemmon, a young, intellectual, activist Victorian lady who moved, in 1870, to the Wild West of gambling, prostitution and horse thievery... She became a dedicated botanist and convinced her husband to undergo – as a honeymoon, no less – a botanical expedition in the then-quite-wild Santa Catalinas! Her heritage includes getting the golden poppy named as the California state flower, and many books published under her husband's name that she unfortunately (and obviously) wasn’t given any credit for. Ok, enough tangents…

Fig 3: Cute Monomorium workers drinking honey

Monomorium emersoni, our focal species, is found throughout the middle and high elevations. To my great relief – because I needed a large number of colonies – this may be one of the easiest ant species to collect. Drive, find a good-looking spot, flip a good-looking rock, quickly suck up queens with your cool ant-sucking device, then suck up as many workers as possible to support the colony when it's back in the lab (Fig 4). Ehab and I quickly became masters of the highways that connect the Arizona Sky Islands. Over the years, and after countless hours of driving while listening to good old country music, we collected over 300 colonies, sampled from top to bottom along the slopes of five of the largest Sky Islands (Fig 5).

Fig 4: Surprised Monomorium emersoni colony when their home sweet home rock is flipped by a myrmecologist

Fig 5: (A) Sky Islands location, (B) Sky Island landscape (C) Monomorium winged queen (D) wingless queen, and (E) map of the 5 mountains we sampled with field sites in black.

Any fellow evolutionary / popgen biologist will understand that one of the first questions that I got totally obsessed with was “given this unique landscape, how are each of Sky Island M. emersoni populations related to one another?” Using classic population genetic tools and inference, we discovered, not surprisingly, that each Sky Island M. emersoni population is roughly isolated from the other ones, the large desert areas acting like barriers to gene flow between mountain ranges. However, gene flow within a mountain between elevations is possible, mostly driven by males.

We also found that gene flow within a mountain range is much greater between similar habitats than between different habitats, regardless of the geographic distance between them, pointing towards a mechanism of isolation-by-environment.  Furthermore, the Sky Island M. emersoni populations revealed an interesting demographic history: the Northern and Southern Sky Islands each form a distinct group within which contact was more extensive in the past (Fig 6). Hold onto that, we’ll come back to it in a minute.

Fig 6: Demographic inference of the contacts between Monomorium Sky Island Populations

Workers in ants are universally wingless, while most reproductive castes have wings. But it is especially common for Monomorium species to have the unusual peculiarity of having some queens born without any wings.  Some species within this genus have winged queens, while other species have wingless queens, and some have both. 

This evolutionary flexibility suggests a common developmental potential for producing either phenotype under the proper genetic or environmental conditions. Using the frequencies of wing/wingless queens along the slopes of each Sky Island, we found that a combination of habitat fragmentation and elevation best explains their distribution, with winged queens found mostly in continuous and low-elevation habitats.

Wings originate in queen larvae as imaginal discs, small synchronized clumps of cells that will become wings during metamorphosis (Fig 7). Although lacking wings as adults, wingless queens still produce these discs as larvae, but in an atrophied form called vestigial wing discs. We looked at the spatial distribution of some wing-related gene products in these vestigial imaginal discs, and found that many of them are produced and do form spatial patterns, while others are completely shut off.

Fig 7: Development of winged queens, wingless queens and workers. The last row shows dissected larvae with their imaginal discs stained with DAPI (blue). Six leg discs are observed. White arrows point to wing imaginal discs. Winged queens have 4 full wing imaginal discs, wingless queens have 4 vestigial imaginal discs, and workers do not have any wing imaginal discs.

Interestingly, we found that for some genes that are expressed in vestigial discs, there were differences in their expression in wingless-queen larvae coming from different Sky Islands. For example, we observed that the expression of mef2 – a key gene involved in wing muscle development – differs among wingless-queen larvae from different populations and shows a signature of the ancient demographic split: Mef2 shows expression differences between wingless queens from the northern and the southern Sky Islands.

We also found a gene, exd, for which wingless-queen larvae from each Sky Island exhibit population-specific expression patterns (Fig 8). This suggests that these differences arose following the complete isolation of the five Sky Island populations ~10,000 years BP and are therefore an indirect consequence of climatic warming since the last ice age.

Fig 8: The variation in Exd expression pattern across Sky Islands (Cartoons on the left column, actual expression images on the right column for evodevo geeks).  Exd expression covers the whole disc in wingless queens from the Pinals (2nd line) and the Chiricahuas (last line), whereas it is much more restricted in the other Sky Islands, the Catalinas (3rd line) and the Huachucas (4th line). This represents a pattern that likely has evolved independently in each Sky Island, as it does not recapitulate historical contacts between Sky Islands (Fig 5).

Once upon a time, about 20,000 years BP when glaciers covered most of North America, the climate was much cooler. Forests surrounded the Arizona Sky Islands, ecologically connecting each mountain range with the others. One can imagine the hundreds of streams coming from the glaciated mountain tops, the luxurious trees, ferns, and flowers… infinite possibilities for animals to move freely from one Sky Island to another, according to their dispersal abilities. It would have been a perfect habitat for a winged queen, continuous and of low elevation (since the mountaintops were covered with snow!).

Then came the warming, very slow at first but gradually accelerating – snow melt, warming, dry air, warming, desertification, forest fragmentation, aridity... the trees found refuges on each mountain at the higher elevations, which remained cool, but each refuge became isolated from the others by large areas of desert. Monomorium followed their forest. And then wingless queens, which are often produced as anomalies by ant colonies, started to become advantageous in those fragmented and higher elevations, and rose to higher frequencies independently on each mountain range as the desert now acted like a barrier between Sky Island populations.

Fig 9: Time replay of the warm-up in the Sky Islands and the evolution of the wingless-queen phenotype.

This independent evolution of wingless queens on each Sky Island occurred through recurrent interruptions of similar wing-patterning genes, and at the same time, induced the release of the unexpressed differences in some genes, such as mef-2, that accumulated in populations during or after the demographic split between the northern and southern Sky Islands ~80,000 years BP.

This natural experiment on Monomorium queens sheds some light on the predictability – or not – of changes during evolution, which is a long-standing debate in evolutionary biology [5]. We have discovered both repeated changes in genes (the same gene expression across replicate populations), and random changes (changes in gene expression unique to each population or groups of populations). Experiments that test the predictability of evolutionary change have brought insights into this debate (for example, the astonishing cell lines of Richard Lenski's group; see [6] for a review of many of them!), and the Monomorium system provided a test of those ideas in the wild, which is quite unique! In fact, this is to our knowledge the first study in wild populations to show that evolution following a major climate change occurs by a “mosaic” of both random and predictable changes. 

Fig 10: MJ and Ehab, the masters of the Sky Island highways. 



We also tried to highlight another phenomenon that hasn’t received much attention: the consequences attributable to organismal development during environmental changes. Scientists have built forecasting models about how species will react under climate change scenarios, which are necessary to predict how our landscapes, our agriculture, and our exposure to diseases will be influenced by global warming. However, most of these models are based on the current phenotypes and abilities of the species under study.

As in the case of Monomorium, developmental outcomes can be modified by a change in the environment and affect, for example here, the dispersal abilities of the individuals, which would greatly affect the predictions of such forecasting models. Our beautiful little Monomorium are probably facing a not-so-beautiful future, as their forested habitats will shrink even further under as global warming progresses in the American Southwest [7]. This habitat loss would likely select for more and more wingless queens, completely eliminating their capacity as a species for any long-range dispersal to more hospitable habitats.

Fig 11: The Abouheif lab, social and tight-knit!

One thing that this story has taught me (aside from the sad lesson that one should not abuse Jerry’s delicacies before running up mountains) is that doing integrative science is both a beautiful and an arduous task. We had to overcome several obstacles to get "integrative results" understandable to many reviewers of various backgrounds simultaneously. But the motivation and insights you get from performing such research is unmatched from any of its separate disciplines. It generates a synergy between results that cannot be achieved otherwise.

References:

[1] Favé, Marie-Julie, et al. "Past climate change on Sky Islands drives novelty in a core developmental gene network and its phenotype." BMC evolutionary biology 15.1 (2015): 183.

[2] Rajakumar, Rajendhran, et al. "Ancestral developmental potential facilitates parallel evolution in ants." Science 335.6064 (2012): 79-82.

[3] Smith, Chris R., et al. "Draft genome of the red harvester ant Pogonomyrmex barbatus." Proceedings of the National Academy of Sciences 108.14 (2011): 5667-5672.

[4] http://www.yelp.ca/biz/jerrys-restaurants-42-safford

[5] Stern, David L., and Virginie Orgogozo. "Is genetic evolution predictable?."Science 323.5915 (2009): 746-751.

[6] Elena, Santiago F., and Richard E. Lenski. "Evolution experiments with microorganisms: the dynamics and genetic bases of adaptation." Nature Reviews Genetics 4.6 (2003): 457-469.

[7] Seager, Richard, et al. "Model projections of an imminent transition to a more arid climate in southwestern North America." science 316.5828 (2007): 1181-1184.

Persistent directional selection on body size and a resolution to the paradox of stasis

By Njal Rollinson


Strong selection and abundant genetic variation appear to be widespread in nature. Adaptive phenotypic evolution should therefore be common. Yet, stasis tends to dominate the temporal dynamic of traits in natural populations. This is the paradox of stasis.

The paradox of stasis has been a recurrent theme here on the eco-evo blog. Ben Haller recently discussed his interesting work on ‘squashed stabilizing selection’ as a potential explanation for stasis (Haller and Hendry 2014). One of the main ideas of Ben’s work is that stabilizing selection is pervasive, but it is difficult to detect in natural systems. This could certainly help explain why stabilizing selection appears to be less common than expected in the wild, and hence this work represents an important step towards resolving the stasis paradox in a broad sense.

But detection limits on stabilizing selection can’t explain stasis for all types of traits: for body size, not only is selection strong, it is typically upward (i.e., in the direction of larger size). So for body size, there is an additional dimension to the paradox: not only should we expect widespread adaptive evolution, we should expect a trend towards the evolution of larger size. But as Kiyoko Gotanda points out in her eco-evo blog post, there’s no empirical support for micro-evolutionary increases in body size across a broad array of taxa, despite persistent upward selection on size (Gotanda et al. 2015). Further, there is no general trend towards the evolution of a larger size over at least one million years (Uyeda et al. 2011).

Stasis of size has been a big topic in the last decade, and it’s received a lot of empirical attention. Here’s one of the weirdest examples of stasis. In 1947, David Lack initiated a long-term study of Great Tits in Wytham Woods, near Oxford. In most years from 1947 to the present day, a team of dedicated researchers has measured the size of fledglings produced in this population, and size-specific survival has been monitored until the next breeding season. In the vast majority of years, there is a strong positive association between size at fledging and survival-to-recruitment, meaning that it is mainly the larger fledglings that actually survive and get to reproduce. Yet, despite this persistent upward selection on size, there has been a downward trend in fledging size, especially in recent years (see Garant et al. 2004, Hadfield et al. 2010).

Fig. 1: David Lack (1910 – 1973) contributed substantially to the development of life-history theory, in addition to coining the term “Darwin’s Finches”.

So why don’t larger body sizes evolve in response to upward selection? Further, if we assume that a given population is near an adaptive peak, then why is upward selection on body size so persistent in the first place? Had David Lack been aware of the size paradox, I suspect he might have had something to say about it.

Lack famously proposed that parents maximise reproductive success by trading off the fitness accrued from an increase in resource investment per offspring against the fitness losses resulting from a reduction in fecundity (see Lack 1947). In other words, there’s a trade-off between offspring quality and parental fecundity, regardless of whether parents invest energy only in yolk or whether extended parental care occurs. The most important consequence of this trade-off is that it generates parent-offspring conflict over size: the level of investment per offspring that maximizes parental fitness is lower than the level that maximizes offspring fitness. The result is that parents are always producing juveniles of a size that is (near) optimal from the parental perspective, but decidedly suboptimal from the perspective of juveniles (Fig. 2).

An important clue to the size paradox is revealed when we ignore how selection acts through parental reproductive success and instead look at how selection acts on their offspring: upward selection on size recurs in every generation. In other words, the trade-off between offspring size and number results in persistent upward selection on offspring size (Fig. 2).

Fig. 2: Parent – offspring conflict over size, and the assignment of fitness to parents vs offspring (a) A concave-downward fitness function with respect to juvenile size. (b) Stabilizing selection on juvenile size with respect to parental fitness, the result of a trade-off between juvenile size and parental fecundity (inset). The juvenile size that maximizes parental fitness, pOpt, is less than the size that maximizes offspring fitness, jOpt. Given that parents have a strong influence over resource allocation to juveniles, the phenotypic distribution of juvenile sizes, d, will more closely match pOpt than jOpt, such that the selection differential from the perspective of juveniles, s, will always be upward and will recur in every generation.

Given that our interest lies in predicting evolutionary response of body size to selection, it’s worth recasting the argument from a quantitative genetic perspective. In a quantitative genetic framework, fitness should be assigned to the individual (i.e. the “offspring” or “juvenile”, in this case) and defined as the number of zygotes produced. When fitness is assigned and defined in this way, then it becomes apparent that the individual must pay a fecundity cost for increased investment per offspring, so selection against investment per offspring exists. This matters, and it matters in a very big way: size can be determined in part by investment in offspring, so selection against investment per offspring might act as a countervailing force to upward selection on body size.

More formally, we can hypothesize from a quantitative genetic perspective that upward selection on size is balanced by selection against investment per offspring, resulting in a net selection gradient on size that is effectively zero. Equivalently, but with some simplification, we can assign fitness to parents and hypothesize that the production of many small offspring results in higher fitness than the production of a few large offspring; that is, even though size-specific mortality occurs, there are still more small offspring than large offspring after selection, such that “genes for small body size” pervade the gene pool (Fig. 3).

Fig. 3: Producing large offspring comes at a cost to fecundity, so even under positive size-selective mortality, parents that produced relatively small offspring (which go on to become relatively small adults) can leave the most descendants.

Here’s some evidence that this hypothesis is on the right track. You’ve probably read the famous studies by Kingsolver and colleagues that document strong upward selection on body size (e.g., Kingsolver et al. 2001, Kingsolver and Pfennig 2004, Kingsolver and Diamond 2011). Of note is that none of these studies differentiate between selection on adult size and selection on juvenile size (note that I am making a distinction between offspring and juvenile, where juvenile refers to the life stage of an individual). But if our thesis is correct, then upward selection on juvenile size will be stronger than upward selection on adult size, because parent-offspring conflict exists over size, and juvenile size is more strongly affected by investment per offspring than adult size. We revisited the large datasets on phenotypic selection, those originally compiled by Joel Kingsolver and colleagues, and we partitioned selection coefficients with respect to the ontogenetic stage at which they we measured. As expected, we found that selection on juvenile size was strongly directional. But for sub-adult and adult size, there is no net directional selection. This means that previous reports of upward selection on size are largely attributable to a few studies that measured selection on juvenile size (Fig. 4). This finding helps resolve why strength of selection on size in contemporary populations is too strong to match observed rates of macro-evolution: juvenile size is expected to experience strong upward selection, but its evolution is of course constrained by a trade-off with fecundity.

Fig. 4: Re-analysis of synthetic studies compiling estimates of phenotypic selection on size (Kingsolver et al. 2001, Kingsolver and Diamond 2011, Siepielski et al. 2013). Black dots are median estimates (with 95% bootstrapped confidence intervals) of selection on body size for all data. The data comprising the black dots are then divided into subgroups and re-analyzed: red dots are median selection on adult size, blue dots are median selection on juvenile size. There is no net directional selection on adult size, there is only upward selection on juvenile size.

But can parent-offspring conflict explain why size in general (i.e., juvenile size and adult size) fails to respond to persistent upward selection? This question leads to the crux of the theory, and to answer it requires a close re-examination of micro-evolutionary studies that measure selection on size.

Long-term studies on great tits, blue tits, red deer, and soay sheep demonstrate that upward selection on body size is pervasive, but evolutionary response is generally not in the direction predicted by the breeder’s equation. There is, however, a common theme among these micro-evolutionary studies: all studies measure response to selection on size traits of juveniles, such as fledgling size and birth weight, and all studies are performed on birds or mammals. Two points can be made here. First, as we already know, upward selection on juvenile size will persist because parents have a strong influence on the size of juveniles, and of course the evolution of juvenile size is constrained by the trade-off with fecundity. Second, birds and mammals invest huge amounts of energy in their offspring and exhibit protracted periods of parental care. (In some passerine birds, for example, size at fledging is actually equivalent to adult size – no structural growth occurs after fledging.) In birds and mammals, then, maternal genetic effects on size – which refers to maternal genes that influence size, such as provisioning genes – will exist for adult size, and any maternal genetic contribution to adult size may not respond to upward selection, as it is constrained to a degree by the trade-off between investment per offspring and fecundity. Some support for this idea lies in the strong correlated evolution of juvenile and adult size that is observed in birds and mammals, in contrast to a weaker correlated evolution of these two traits in groups that lack protracted periods of parental care (Fig. 5).

Fig. 5: Correlated evolution of juvenile size and adult size is strong in birds and mammals, but weaker in groups that do not exhibit protracted parental care. The relationships are between species-mean adult mass on the x axis and, on the y axis, mean weight at birth (mammals), mean egg weight (birds, reptiles, amphibians, fish), or mean seed weight (plants).

So in essence, we think a big part of the solution to the size paradox lies in seminal life-history work performed over 50 years ago by David Lack. The individual as a juvenile has high fitness if it receives lots of investment, but the same individual as an adult has high fitness if it limits investment in its offspring. Given that investment in offspring affects size in general, we expect selection against investment per offspring to balance upward selection on size. Antagonistic selection on size across life stages therefore limits the evolution of larger size, with parent-offspring conflict over size embodied in the opposing signs of the two selection gradients.

There remains huge scope for interesting empirical and theoretical work on this topic. For example, our study focused specifically on how the trade-off between size and number of offspring generates parent-offspring conflict over size, but other forms of parent–offspring conflict might also limit the evolution of body size. In species with parental care, for instance, an individual can affect its own size and the size of its siblings by competing directly for parental resources, and by manipulating parental allocation. But of course the individual is also related to its parents and siblings, and therefore the individual is expected to act selfishly only when the benefit to the individual is greater than the cost to the other parties, weighted by relatedness. Kin selection may therefore offer a complimentary avenue through which to explore how family conflicts limit body size evolution.

We’ve outlined some potential future work in the paper, which is now available online.

Rollinson, N. & Rowe, L. Persistent directional selection on size and a resolution to the paradox of stasis. Evolution. In press (Early View).

Cheers everyone; thanks for reading.


Njal Rollinson


References

Garant, D., L. E. B. Kruuk, R. H. McCleery, and B. C. Sheldon. 2004. Evolution in a changing environment: A case study with great tit fledging mass. The American Naturalist 164:E115–29.

Gotanda, K. M., C. Correa, M. M. Turcotte, G. Rolshausen, and A. P. Hendry. 2015. Linking macro-trends and micro-rates: Re-evaluating microevolutionary support for Cope’s rule. Evolution 69:1345–1354.

Hadfield, J. D., A. J. Wilson, D. Garant, B. C. Sheldon, and L. E. B. Kruuk. 2010. The misuse of BLUP in ecology and evolution. The American Naturalist 175:116–125.

Haller, B. C., and A. P. Hendry. 2014. Solving the paradox of stasis: Squashed stabilizing selection and the limits of detection. Evolution 68:483–500.

Kingsolver, J. G., and S. E. Diamond. 2011. Phenotypic selection in natural populations: What limits directional selection? The American Naturalist 177:346–357.

Kingsolver, J. G., H. E. Hoekstra, J. M. Hoekstra, D. Berrigan, S. N. Vignieri, C. E. Hill, a Hoang, P. Gibert, and P. Beerli. 2001. The strength of phenotypic selection in natural populations. The American Naturalist 157:245–61.

Kingsolver, J., and D. Pfennig. 2004. Individual-level selection as a cause of Cope’s rule of phylogenetic size increase. Evolution 58:1608–1612.

Lack, D. 1947. The significance of clutch-size. Ibis 89:302–352.

Siepielski, A. M., K. M. Gotanda, M. B. Morrissey, S. E. Diamond, J. D. DiBattista, and S. M. Carlson. 2013. The spatial patterns of directional phenotypic selection. Ecology Letters 16:1382–92.

Uyeda, J. C., T. F. Hansen, S. J. Arnold, and J. Pienaar. 2011. The million-year wait for macroevolutionary bursts. Proceedings of the National Academy of Sciences 108:15908–15913.