In the current issue of Science, researchers at Michigan State University demonstrate how a new virus evolves, which sheds light on how easy it can be for diseases to gain dangerous mutations.
Justin Meyer (right), MSU graduate student, led a team of researchers, including Devin Dobias, former MSU undergraduate student, that showed how new viruses evolve. Photo by G.L. Kohuth.
The scientists showed for the first time how the virus called “Lambda” evolved to find a new way to attack host cells, an innovation that took four mutations to accomplish. This virus infects bacteria, in particular the common E. coli bacterium. Lambda isn’t dangerous to humans, but this research demonstrated how viruses evolve complex and potentially deadly new traits, said Justin Meyer, MSU graduate student, who co-authored the paper with Richard Lenski, MSU Hannah Distinguished Professor of Microbiology and Molecular Genetics.
“We were surprised at first to see Lambda evolve this new function, this ability to attack and enter the cell through a new receptor – and it happened so fast,” Meyer said. “But when we re-ran the evolution experiment, we saw the same thing happen over and over.”
Ribbon diagram of the OmpF protein, Lambda's new pathway into E. coli.
This paper follows recent news that scientists in the United States and the Netherlands produced a deadly version of bird flu. Even though bird flu is a mere five mutations away from becoming transmissible between humans, it’s highly unlikely the virus could naturally obtain all of the beneficial mutations all at once. However, it might evolve sequentially, gaining benefits one-by-one, if conditions are favorable at each step, he added.
Through research conducted at BEACON, MSU’s National Science Foundation Center for the Study of Evolution in Action, Meyer and his colleagues’ ability to duplicate the results implied that adaptation by natural selection, or survival of the fittest, had an important role in the virus’ evolution.
When the genomes of the adaptable virus were sequenced, they always had four mutations in common. The viruses that didn’t evolve the new way of entering cells had some of the four mutations but never all four together, said Meyer, who holds the Barnett Rosenberg Fellowship in MSU’s College of Natural Science.
“In other words, natural selection promoted the virus’ evolution because the mutations helped them use both their old and new attacks,” Meyer said. “The finding raises questions of whether the five bird flu mutations may also have multiple functions, and could they evolve naturally?”
This week’s BEACON Researchers at Work post is by University of Texas at Austin graduate student Igor Karpov.
Igor with a pair of traditional Komi fur skis
When thinking about parts of my work that are most relevant to BEACON, several topics come to mind simultaneously. To avoid making the hard choice myself, I will briefly describe all of them, and leave the choice of what is interesting to follow up on to the reader.
The unifying theme for the projects described below is that I use of and extend evolutionary computation methods in the context of a popular and complex domain – video and computer games. The domain has several properties that make it an interesting subject of study from the perspective of artificial intelligence. First, the variety of game genres and complexities allows for a gradient of increasingly complex behaviors and adaptation approaches to be developed. Secondly, game engines have developed a good balance of complex environments and behaviors with reasonable amounts of computation and simulation speed. Finally, and perhaps most importantly, the game domain has plenty of human participation. This means both that the domain itself is interesting and challenging enough to hold our attention, and that we can study how our state-of-the-art autonomous agents do when they are interacting with human-level intelligence in its various forms.
The relative ability of bots and human players to pass for human players in the Botprize competition.
An example of the data collected from human games in the Botprize domain.
One of the most complex such domains that I will talk about is the Botprize competition. The goal of this competition is to develop a software player for a state of the art first-person game (Unreal Tournament 2004 in our case) that is behaviorally indistinguishable from a human player. To be more concrete, we have to design a bot that will fool the human players it interacts with into labeling it as a human about as often as another human player is able to do so.
To address this challenge, I have worked with a fellow BEACON researcher and UT Austin graduate student Jacob Schrum (who works on multi-objective evolution of neural network controllers for game domains) and our advisor Risto Miikkulainen, to develop UT2, a game bot that participated in the Botprize competition several times, and placed 2nd in 2010. The overall system is complex and includes a scripted behavior architecture, a module used in combat and evolved by multi-objective constructive evolution of artificial neural networks, and a module that is responsible for human-like movement that is based on playback of human examples (Believable Bot Navigation via Playback of Human Traces). The area is ripe for future work, including imitation learning from human behavior and ways of combining imitation with evolution of autonomous behaviors.
A schematic diagram explaining the human-assisted neuroevolution method. Three types of human input (advice, example traces and task shaping) are combined with an evolving population of artificial neural networks to produce desired solutions faster.
The second project that I have worked on together with Vinod Valsalam and Risto Miikkulainen, sets out to study exactly the ways in which human users can harness machine learning methods such as neuroevolution. In this human subject study, we compare manual design of game behavior and unassisted evolution of neural networks against three different types of human-assisted, interactive neuroevolution, namely evolution in the presence of task shaping, evolution with the addition of advice, and evolution with learning from examples (see Human-Assisted Neuroevolution through Shaping, Advice and Examples).
Relative time to solve the three design tasks manually, by evolving solutions automatically, and with human assistance.
Our results indicate that while the unassisted neuroevolution is a powerful game behavior design tool and outperforms manual design significantly, it can be greatly improved with the correct application of an interactive human assistance method. Further, the type of human assistance that works best depends on the task, leading to a hope of developing hybrid methods that combine the strengths of human input and of machine evolution automatically.
Finally, I will touch on a substantial open source software development project I am leading. The software is called OpenNERO: game platform for AI research and education. It is a system that includes several different game-like mods that are unified by an AI framework and support neuroevolution, reinforcement learning, search methods, planning, and potentially many others. While a description of the entire system is beyond the scope of this post, I encourage the reader to checkout our website at opennero.googlecode.com, and see some of the educational and research demos we have made available.
A color matrix representation of score differences in the OpenNERO round robin tournament. Rows and columns of the matrix are the red and blue team playing the match respectively. Redder colors mean more decisive victory for the red team and bluer colors mean more decisive victory for the blue team. Teams are ordered according to their average score across all matches played.
One of the most recent ways in which we have used the OpenNERO platform was to run the 2011 OpenNERO Tournament. This tournament, which was run as part of Stanford University’s online Introduction to Artificial Intelligence course, invited students to evolve and/or train behaviors for an RTS-like game, where their teams would compete with other submissions for the right to be called “strongest in the field.” We received 156 submissions and ran a round-robin tournament, resulting in a detailed analysis of behavior diversity and other characteristics. The infrastructure developed for parallel evaluation of games and for analysis and visualization of tournament results gives us confidence that this type of a competition can be run with a much larger number of participants, and can potentially even be used to drive the process of evolution of novel behaviors itself.
For more information about Igor’s work, you can contact him at ikarpov at cs dot utexas dot edu.
This week’s BEACON Researchers at Work blog post is by MSU postdoc Aaron Wagner.
The evolution of sociality is one of the most fascinating and productive topics in evolutionary biology. Though it is often very useful to look to social species to understand the current function, costs, benefits, and circumstances of cooperative relationships, current observable social behaviors may not reflect the selective pressures that led to their initial evolution. Instead, the observed benefits of a social strategy may be a consequence of that strategy, not the force that initially favored its evolution. For example, social grouping in carnivores is often explained as a way to increase hunting success via pack hunting. However, in many carnivores, social group size is larger than hunting group size (e.g. lions Panthera leo or spotted hyenas Crocuta crocuta), so any foraging benefits may be an effect, not a cause, of grouping.
How resource conditions (bottom) can determine if solitary living (top right) or sociality (top left) evolves.
Simple spatial groups can form through a passive process when the costs of tolerating others drop below the costs of sharing space (such as a territory). The difference between these costs is largely determined by availability of resources: when the resources in an animal’s range exceeds its needs, there is little or no cost to sharing the range, tolerance can develop, and spatial groups can form even in the absence of any benefits. Once simple groups have formed, natural selection can then act to promote the evolution of cooperative strategies and the realization of any associated social benefits. This broad construct (excess resources -> tolerance -> aggregation -> stability -> social benefits) sets the stage for developing and testing predictions about the role of resource availability and distributions in the evolution of sociality. Because simple groups must exist prior to the evolution of more complex social organizations, a fundamental part of my research focuses on uncovering the conditions that permit, promote, or preclude the evolution of simple grouping strategies and the persistence of groups.
Within the carnivores, sociality is extremely rare: 85% of carnivore species are considered solitary. For many good reasons, the vast majority of carnivore behavioral ecology research has focused on the social minority. However, several recent studies of species from the solitary majority have unveiled intriguing variations in grouping strategies and behaviors across populations. In particular, temporary groups sometimes form in these ‘solitary’ animals when resource conditions can support them. Within these groups, cooperation and social behaviors are often limited and primitive and, in the absence of active benefits to grouping, the groups themselves are often unstable. Among these ‘incipiently social’ species is the striped hyena (Hyaena hyaena), which I had the pleasure of studying for many years in Kenya.
A mother greets her nephew at her new-born cub’s den.
Before we began working with striped hyenas, they were routinely described as being strictly solitary. However, we discovered that this was not always the case. At our first site, while they almost never interacted, males were willing to share ranges while females were strictly solitary. At the second site where resources were more plentiful, males were strictly solitary but some females shared ranges with other females and often interacted with each other’s cubs at dens. Studies like this, where differences in non- or minimally-cooperative grouping strategies can be compared with differences in resource conditions, provide glimpses of the earliest stages of social evolution and hints about the constraints that were “breached” to permit the evolution of far more complex forms of sociality, like that found in the highly gregarious (and über complex) spotted hyena.
While our work with the striped hyena was enlightening (and an experience I wouldn’t trade for most anything), significant questions remain. For instance: Are permissive conditions sufficient to maintain group stability? Is group stability necessary and sufficient for the evolution of sociality? And what types of modifications in resource conditions explain variations in social strategies, including the conditions that favor immigration (moving to a new group) over philopatry (staying in the same group you were born in) as a means of group formation and maintenance?While it may be possible to address these questions via additional field studies, I have taken a very different approach… following a fairly dizzying left turn into Charles Ofria’s Digital Evolution Lab.
What first brought me to the ‘Devolab’ was an encounter with a description of the digital evolution platform Avida that Charles, his colleagues, and students have developed for and applied toward studying an impressive array of fundamental questions in evolutionary biology. When I first read about Avida, I was instantly convinced that this platform was ideal for addressing questions about the pressures and patterns underlying the evolution of tolerance, group formation, and social cooperation. In a broad sense, Avida seemed perfect for evolving ‘digital hyenas’ or, alternatively, for uncovering the conditions that lead to the evolution of carnivore-like grouping, proto-social, and social behaviors.
While cooperative behaviors have evolved in ‘avidian’ populations in the past, this occurred under conditions in which grouping was a given. That is, organisms were placed into groups, they were not ‘asked’ to form their own groups first. This was also not done in a spatial context whereby groups have physical ranges or territories in the digital world encompassing locally accessible patches of resources. My work seeks to examine exactly that process: starting with a solitary population, what distributions of resources drive the population to evolve ranging behaviors and to either defend that range as a territory or to tolerate one another’s presence?
While it is obvious to us now, having come from working with a large and complex animal that has already evolved to do it, what we did not anticipate was that in order for the organisms to ever evolve such ranging and grouping behaviors, they would also need to evolve the skills to intelligently move and navigate through their environment. Hyenas already do that… it’s something taken for granted in the field and never thought about. Since coming to BEACON, I think about it a lot! Because having intelligent and flexible navigators is so critical for addressing our original goals, much of our efforts have focused on looking at what aspects of the environment drive organisms toward evolving navigation in the first place (compared to not moving at all, or just happily running around like crazed chickens… both of which they are often quite content to do) and toward evolving use and control of sensors.
The 5 major steps in our approach toward evolving digital hyenas.
As it turns out, starting with a simple, non-moving, and blind ancestor and evolving intelligent navigation in an open ended environment like Avida is far from simple. However, we have succeeded in doing just that. We began with various environments in which the organisms were required to navigate out to a food resource and return to the nest on which they were born. The three biggest challenges that we had to overcome here were to uncover 1) the specific characteristics of the environment that would pressure the avidians to travel far from the nest, 2) the characteristics that would create pressures to evolve away from random movement, and 3) the conditions that would prevent the evolution of fixed ‘blind’ strategies. For the last one, the solution was to put the food resources in motion. For the first two, competition resulting from resource depletion and crowding does the trick: organisms suffer if they all try to feed from the same food resource, because the food runs out. Thus, they are driven to find resources farther from the nest, but the evolution of intelligent sensor use becomes more critical the farther out the organisms travel.
Avidians evolved abilities to detect distant food resources (small greys) moving in fixed orbits, navigate to and feed from them, and return home to reproduce at a central nest (large grey). Organism colors reflect the food resource the organisms are seeking or have fed from. Food resources appear speckled as the organisms deplete them. In this experiment, unlike the food resources, the central nest is not visible from long distances. The organisms here have compensated for this by using the (visible) closest orbiting resource as a landmark from which they search for the nest (note the streams of organisms heading to that lowest orbit resource, but that it is not being depleted). The noisy clouds around the food resources are a partial consequence of crowding… there are fewer cells on the resources than organisms trying to feed from them.
Once the avidians demonstrated an evolved ability to control and use their sensors and to navigate across large distances, we added a new twist to their lives: we allowed avidians to eat other avidians. In other words, we allowed for the evolution of predators from the prey population. With the same sensor capacities as before, avidians almost immediately evolved successful predation strategies in stable predator-prey populations.
Prey (greens) ‘blooms’ occur in response to seasonal shifts in resource (black) abundance and location. Greyed resources are out of season and are below the minimum the prey need to feed. Evolved from the prey, not introduced, predators (red) have adapted their sensors for use in locating prey and rapidly swarm in response to prey blooms.
Our next step is to tie these successes together with other non-spatial work in which avidians evolved abilities for territory defense as a response to local resource availability and competition. The primary hurdle here will be to evolve territory establishment and fidelity by the predators. In the process, we will ultimately be able to test the suite of hypotheses we originally targeted. Namely, under what resource conditions do predators establish ranges and alter their levels of tolerance toward conspecifics. Undoubtedly, as always, we will be learning as much about the intricacies of evolution from the development process itself as we do from the results of the final experiments…but that’s half the fun of it all!
For more information about Aaron’s work, you can contact him at apwagner at msu dot edu, or visit his website.
Call for Nominations New Graduate Student BEACON Top-Up Recruiting Fellowships
Sponsored by The BEACON Center for the Study of Evolution in Action An NSF Science and Technology Center
One goal of BEACON is to initiate and support research and training activities that involve the study of actively evolving systems and evolutionary dynamics, as well as applying these principles to solve tough computational or engineering problems. Furthermore, BEACON aims to recruit a diverse student population. To promote these goals, BEACON will be providing BEACON Top-Up Recruiting Fellowships to attract promising new Ph.D. students interested in this area to attend Michigan State University, funded by the University.
Eligibility: Top-Up Recruiting Fellowships can be used to support applicants to Ph.D. programs in all departments at MSU that conduct research in this area, with preference given to applicants who are citizens or permanent residents of the US. Any applicant nominated for a Top-Up Recruiting Fellowship must be nominated by a BEACON faculty member. In addition, the applicant must receive a 5-year support commitment from the faculty member and/or department, university or external agency (NSF, etc.). BEACON strongly encourages faculty to nominate women, students from underrepresented minorities, and persons with disabilities.
Top-Up Recruiting Fellowship Details: If an applicant is awarded a BEACON Top-Up Recruiting Fellowship, they will receive between $3,000 and $5,000 in additionalfellowship funds for each year they participate in BEACON activities, for up to a maximum of five years. If the applicant receives an NSF or other similar fellowship already providing $30,000 or more in annual support, BEACON will offer a one-time fellowship supplement of $5,000 for the duration of that fellowship.
Requirements: Students receiving this fellowship will be required to take two BEACON-related courses during their first year: one course on either evolutionary biology or computational evolution (whichever is not part of the student’s background) during Fall Semester 2012, and one project course where students work in interdisciplinary groups during Spring Semester 2013. This requirement is to support BEACON’s goal of encouraging students to pursue multi-disciplinary research. These courses are normally included in the student’s academic program.
Application Process: To nominate an applicant, please email the student’s application packet to Eric Torng (torng@msu.edu) by February 7, 2012. Please add the following two items:
A letter of nomination from the prospective advisor. This letter should describe briefly the research area in which the student is expected to work and how this is connected to BEACON. Please highlight any multidisciplinary aspects of the research. If the research will involve any of the BEACON partner universities, that should be stated.
If the candidate is a woman, member of an underrepresented minority, or a person with a disability, note that fact on the application for reporting to the NSF.
Decisions will be made within a week following the announcement of UDF/UEF recipients.
This week’s BEACON Researchers at Work post is by North Carolina A&T State University graduate student Aniesha Alford.
We all remember the events that occurred on September 11, 2001. I was a sophomore in high school, and I can still vividly recall watching the news along with my classmates as the events unfolded. Since that day, measures are continuously being taken to avoid similar tragedies. One preventive measure is a more secure identification process, specifically via the use of biometric recognition. To date, biometric recognition systems are currently used by a number of commercial and government organizations. However, there is always room for improvement. I perform research with the Center for Advanced Studies in Identity Sciences (CASIS), and my research is in a new field of study that we call Genetic and Evolutionary Biometrics (GEB). GEB is devoted to the discovery, design, and analysis of evolution-based methods for solving traditional problems within the field of biometrics. My research, in particular, focuses on the development of GEB applications to improve the performance of facial and periocular (i.e. the area around the eyes) biometric recognition.
Biometric recognition involves the use of distinct physical, chemical, and/or behavioral characteristics for automatic recognition of an individual. Examples of such biometric characteristics (also known as modalities) include the face, fingerprint, voice, and signatures. These recognition systems typically work by first using a sensor (such as a camera) to acquire a biometric sample. The newly acquired sample is then passed to a feature extractor which transforms the acquired sample into a set of unique features referred to as a feature template. Often, feature selection techniques are then applied to reduce the dimensionality of the resulting feature templates. Next, the reduced template is compared to those previously enrolled (stored) in a database. The similarity between the recently acquired and enrolled templates are then measured and used to make a decision (accept/reject an individual).
In the biometrics community, feature selection techniques have typically focused on retaining the most salient individual features (i.e. the most variant individual dimensions, the most consistent individual features, or the most discriminative individual features). However, my research proposes the use of GECs to: (a) evolve subsets of the most salient combinations of features and/or (b) weight features based on their discriminatory ability in an effort to increase accuracy while decreasing the overall number of features needed for recognition.
Three techniques have been developed and applied for facial and periocular recognition: Genetic & Evolutionary Feature Selection (GEFeS), Weighting (GEFeW), and Weighting/Selection (GEFeWS). GEFeS reduces the number of features used by evolving a feature mask (FM) that discards features that do not aid in increasing the recognition accuracy. On the contrary, GEFeW evolves a weight for each feature within a feature template based on its relevance. Our final technique, GEFeWS, is a hybrid of GEFeS and GEFeW. GEFeWS evolves a FM that discards those features that are not relevant and weights those features which are.
To test the effectiveness of these techniques, images were selected from the Face Recognition Grand Challenge (FRGC) database. Two feature extraction techniques were then applied to the facial and periocular images: the Eigenface method and the Local Binary Patterns (LBP) method. The Eigenface method, which is based on Principle Component Analysis, is a statistical dimensionality reduction technique that is used to extract only those dimensions that are necessary to efficiently distinguish images of individuals. The LBP method is a texture analysis technique which works by first segmenting an image into a grid of evenly sized regions (referred to as patches) and then analyzing the intensity changes of the pixels within each patch. GEFeS, GEFeW, and GEFeWS were then used to evolve FMs for the face-only, periocular-only, and face + periocular feature templates. The performances of these techniques were compared to performance of the feature templates without the use of GECs.
Our results showed that by fusing the periocular biometric with the face, we could achieve higher recognition accuracies than using the two biometric modalities independently. In addition, the LBP feature templates outperformed the Eigenface templates. Our results also showed that our GECs were able to achieve higher recognition rates than the baseline methods (i.e. the feature templates without the use of GECs), while using significantly fewer features. Of the three techniques, GEFeWS performed best, using less than 50% of the extracted features to achieve higher accuracies than GEFeS and GEFeW alone.
In conclusion, I am very excited about our research. It is great to be one of the pioneers in this new field of study, but it is even greater to think that one day our research could be implemented to make biometric security processes more accurate, faster, and more efficient. In addition, the potential that similar techniques may have in other areas of study are astounding. By presenting my research at conferences, I have been approached by several individuals interested in applying similar techniques to their research (i.e. tomato classification). I look forward to seeing how the skills I have gained through this research will come into play in the future. The possibilities seem endless and I believe that BEACON has prepared me for the challenge!
For more information about Aniesha’s research, you can contact her at aalford at ncat dot edu.
This week’s BEACON Researchers at Work post is by MSU postdoc Robin Tinghitella.
Last April I had the pleasure of writing the very first BEACON Researchers at Work blog post. I shared a story about how a tropical field cricket, Teleogryllus oceanicus, hitchhiked from Australia (where they’re native) through the Pacific, colonizing island after island, until they finally reached the Hawaiian Islands. It’s a great story about how humans can drive biological evolution because the crickets probably traveled on canoes with Polynesian settlers as they themselves island-hopped through the Pacific during the Polynesian expansion. For an evolutionary biologist like myself, the really intriguing part of the story is what happened once the crickets arrived in Hawaii. As a family friend announced during my wedding ceremony, the Hawaiian crickets lost their “chirpedness.” In other words, a rare mutation wiped out their ability to sing to attract mates. But how could that possibly work?
Figure 1
When organisms colonize new environments, rapid evolutionary change frequently follows. This is primarily because of two things: genetic bottlenecks and novel selection pressures. Genetic bottlenecks happen when the size of a population is drastically reduced by chance. Large populations are typically made up of lots of different types of individuals with different characteristics. In other words, they have high genetic diversity. But when a chance event happens, the individuals who live on to reproduce may not represent all of the types that were found in the original population, many genes can be lost in the process, and the genetic diversity ends up being much, much lower. When organisms colonize new locations, the same type of thing happens. Founding populations are typically made up of very few individuals who don’t necessarily represent all of the genetic diversity present in their source populations (Figure 1). This makes a lot of sense when you think about something as drastic as colonizing islands like Hawaii. Hawaii is 2000 miles from the nearest Pacific island – how many individuals would you expect to arrive there by floating on flotsam, being carried on wind gusts or hitching a ride on a canoe? My guess is not very many. In the case of the tropical field crickets, the trip across the Pacific serves as the “bottleneck” through which only a few individuals make it. Once founders arrive in a new location, they face a whole host of challenges in their new environments – novel selection pressures – things that may not have existed in their source ranges and that may favor very different characteristics than those that were advantageous in their original habitats.
Back to the tropical field crickets, then. Crickets are a classic study system for biologists who are interested in sexual signals, things like bird plumage and frog calls that are typically used by males to attract females. Even Darwin was puzzled by these sorts of showy, exaggerated traits precisely because they are so conspicuous and seem to defy natural selection. Crickets are a great model system for this sort of work because males produce two songs that are used in mating contexts – first, they sing a calling song (a long distance, high intensity song) to attract females from afar. Females are the locomotory sex, so they move through the environment (grassy fields) in search of singing males. Once they are in close proximity, males switch to producing a quiet courtship song. Both songs are thought to be required before a female will mate with a male and mating is entirely under the control of the female – it’s a classic “female choice” system in which males cannot coerce females into mating. Females even have more subtle preferences for certain aspects of the song (like the length of certain parts of the song or what proportion of singing time is filled with sound (the duty cycle)) that they use to decide which males they will mate with. In any event, the songs are really conspicuous and like many sexual signals, they often attract the attention of unintended receivers, things like predators and parasites that instead use these mating signals to locate potential hosts and prey. In Hawaii, T. oceanicus encounters a novel natural enemy that exists nowhere else in its range – it’s a parasitoid fly called Ormia ochracea that uses the crickets’ mating song to locate hosts for their maggots to live off of.
Figure 2
In an alien-like plot, this fly has drastically changed the way the crickets locate mates on two islands in Hawaii. Pregnant female flies locate signaling males and spray larvae on and around them, some of which will burrow into the body cavity of the cricket where they’ll spend the next 7-10 days literally eating the cricket from the inside out (Figure 2 – photo by J.T. Rotenberry). Hawaiian T. oceanicus, then, face a conflict between natural and sexual selection. They should sing to locate females, but singing is very risky. What happened next in Hawaii was a surprise to those of us who study this system – the crickets just stopped singing.
Figure 3
After several months of scratching our heads, we put the pieces of the puzzle together. The crickets had a rare mutation that we called “flatwing”. The mutation changes their wing morphology in a way that eliminates their ability to sing to attract mates. Male crickets typically sing using specialized ridged structures on the wings, but flatwing males are missing all of those structures (Figure 3), so they’re physically incapable of producing the songs that crickets are so well known for. The flatwing mutation probably originally appeared in just one or very few individuals, but it spread through the population to 95% of males on the island of Kauai in fewer than 20 generations. That’s one of the fastest recorded evolutionary shifts in a wild population! This discovery led to so many questions: Why did the mutation spread so quickly? What’s the advantage to being a flatwing male over a calling male? If you already guessed that flatwing crickets avoid being parasitized by the fly, you’re right. We dissected over 120 flatwing males and only one of them was parasitized. That’s a far cry from the >35% parasitization rates that were found on Kauai before the mutation appeared.
But even if flatwing crickets are safe from the fly, the rest of the story still doesn’t make sense. Silent crickets should be really unsuccessful when it comes to mating (males use song to locate females and to “convince” them to mate), so a mutation like flatwing shouldn’t be passed on to future generations. Yet this mutation spread like wildfire and there’s a thriving population of almost all silent crickets on Kauai. How do males and females find each other without the use of song? And, why do females mate with them? These questions made up the bulk of my dissertation research, and what I discovered has inspired my subsequent work to answer a basic question about female mating decisions: “why that guy?”
It turns out that male crickets sometimes use an alternative mating behavior called satellite behavior to locate mates. Instead of calling themselves, satellite males hang out near singing males and attempt to intercept females who have been attracted to the other guys. Lots of organisms like frogs, toads, and insects that signal acoustically sometimes use satellite behavior. The advantage is that satellites can avoid the energetic costs of producing songs or calls as well as the risk (of predation or parasitization) associated with singing or calling themselves. Check out this cartoon that explains how satellite behavior works. In Hawaii, flatwings behave as satellites to the remaining 5% or so of males who can call. The flies can’t find the flatwings, but flatwing males can still manage to find females! What’s more, we found out that satellite behavior existed as a behavioral option for males before the change in wing morphology. Regardless of their wing morphology, males seem to use satellite behavior most when they haven’t mated recently. We can imagine that might happen anytime the population density is low or the competition for mates is stiff. What’s really cool about this is that is suggests pre-existing behavior facilitated the loss of song, so maybe behavior plays an important role in rapid evolution. Without satellite behavior silent males couldn’t find mates and flatwing males wouldn’t have passed their flatwing genes on to future generations.
Robin "surfing" on Hawaii
Satellite behavior gets us as far as males and females finding each other, but it still doesn’t explain why females are willing to mate with a male who can’t produce the sexual signal. Past research tells us that females clearly have strong preferences, even requirements, for hearing the courtship song before mounting males for mating. Let’s think back to the very first females who colonized the Hawaiian populations of T. oceanicus to see if they can help us reason this out. Recall that the vast open ocean between Hawaii and the nearest land serves as a bottleneck that only a few crickets made it through to colonize the Hawaiian Islands. The small initial population size may also have contributed to the spread of the flatwing trait. Imagine you’re a female cricket and you land on an island somewhere in the Pacific with not very many mates to choose from. If you’re extremely choosy you may never find an acceptable mate, which means the end of your genetic line. In other words, small founding populations may favor (select for) females who are lax in their mating decisions. In mating trials with females from across the crickets’ range (Australia, Oceania, and Hawaii) we found that females from Hawaii, and Kauai in particular, were the least choosy – they mate with silent flatwing males 50% of the time whereas females from Western Australia accept them only <10% of the time. So, the process of island colonization seems to have primed the Hawaiian populations for this rapid evolutionary change. The unfussy females allowed the mutation to take hold and spread on Kauai. In another location, like Australia, even if the mutation occurred, it may not have spread because picky females would not have accepted silent flatwing males as mates. So, again, it seems we found a special role for behavior in rapid evolution. And now I have a bit more information with which to answer the question, “why that guy?”
The papers I’ve covered in this post are:
Tinghitella, R.M. & Zuk, M. 2009. Asymmetric mating preferences accommodated the rapid evolutionary loss of a sexual signal. Evolution. 63: 2087-2098.
Tinghitella, R.M., Wang, J.M.* & Zuk, M. 2009. Pre-existing behavior renders a mutation adaptive: flexibility in male phonotaxis and the loss of singing ability in the cricket Teleogryllus oceanicus. Behavioral Ecology. 20: 722-728.
Zuk, M., Rotenberry, J.T. & Tinghitella, R.M. 2006. Silent Night: Adaptive disappearance of a sexual signal in a parasitized population of field crickets. Biology Letters. 2: 521-524.
For more information about Robin’s work, you can contact her at hibbsr at msu dot edu.
This week’s BEACON Researchers at Work post is by BEACON Managing Director Danielle Whittaker.
The sense of smell is one we often take for granted in our own lives. However, even though we may not be conscious of it, odor can tell us a lot about a person, like how closely related they are to us, or how healthy they are – or how attractive we find them. Many mammals rely heavily on smell to communicate and have the anatomy to prove it: a wet nose, a large olfactory bulb in the brain, a vomeronasal organ, and lots of different scent glands. Birds, on the other hand, have a very small olfactory bulb, no vomeronasal organ, and no dedicated scent glands. For a long time people have thought that birds had little to no sense of smell. However, a growing body of research is demonstrating that birds can detect many different kinds of odors, that they produce odors themselves, and that those odors may affect their behavior.
Dark-eyed junco, pink-sided subspecies. Photo by Marine Drouilly.
I study the dark-eyed junco (Junco hyemalis), a small North American sparrow that is commonly found in backyards, especially in the winter. Juncos breed at high elevations and high latitudes, which means I get to do my fieldwork in beautiful mountain locations like Grand Teton National Park. I started working with juncos when I joined Dr. Ellen Ketterson’s lab at Indiana University as a postdoc in 2006, and still work with them today.
A junco preen gland
Like most birds, juncos have only one large sebaceous gland – the uropygial or preen gland, located just above the base of the tail. This gland produces an oil that birds spread over their feathers while preening, which helps protect the feathers from exposure to the environment and parasites, and also helps waterproof the feathers and keep the bird warm. Preen oil contains volatile compounds (like perfume!) that give the bird an odor. In collaboration with Dr. Milos Novotny and Dr. Helena Soini at the Institute for Pheromone Research at IU, I’ve been decoding the information contained in these volatile compounds. Different bird species have different compounds. For example, juncos typically have about 19 different volatile compounds in their preen oil, including several linear alcohols, methyl ketones, and carboxylic acids. Within a species, individuals vary in the relative proportion of each of these compounds. In juncos, males have higher proportions of the methyl ketones tridecanone and pentadecanone, while females have higher proportions of the linear alcohol undecanol.
Like most temperate bird species, juncos are seasonal breeders. In the winter, they produce very small amounts of these compounds, and the amounts go up dramatically during the summer breeding season. Does this change have something to do with mating behavior?
In my newest paper, I was interested in the hormonal mechanisms responsible for this seasonal change. At Dr. Ketterson’s long-term study site at Mountain Lake Biological Station, I captured male and female juncos during the first four weeks of the breeding season – a time when the birds are undergoing many physiological changes as they shift into breeding condition. As I expected, male juncos showed a steady increase in volatile compound concentration over the first four weeks of the breeding season. Females, however, showed a huge increase over the first three weeks, followed by a drop in the fourth week. What caused the drop? I looked closely at breeding data for the population that year, and found that the peak in female volatile compounds corresponded with a peak in egg-laying. I think that females may be using a strong odor to tell males that it’s time to mate.
What does bird odor have to do with evolution in action? Animals use a variety of signals to attract mates: visual, acoustic, and olfactory. When populations diverge, we often see a corresponding shift in these signals that help reinforce the isolation of the two populations by preventing mating with each other, and over time these populations can become different species. In southern California, a small population of juncos colonized the University of California San Diego college campus beginning around 1980, and over just 30 years has rapidly changed compared to their parent population in the nearby Laguna Mountains. The UCSD birds are now smaller and less aggressive, they no longer migrate, and they invest more effort in caring for offspring than the Laguna Mountain population. They look different, too – males have smaller amounts of white in their tail feathers, a trait that is attractive to female juncos. In collaboration with IU postdoc Jonathan Atwell, I found that they also smell different – male UCSD juncos have a more “female-like” odor than male Laguna Mountain juncos. I’m interested in how these different modes of signaling work together to communicate individual quality and identity to potential mates, and how they may be involved in population divergence and speciation.
Overlap in junco and white-throated sparrow volatile compounds (presence/absence data)
Sometimes, even though two species are completely different, hybridization might occur. The white-throated sparrow looks nothing like the junco, and the two species produce completely different songs – and yet, they have been known to occasionally mate and produce hybrid offspring. How could they possibly make a mistake like that? We analyzed white-throated sparrow preen oil and found that white-throated sparrows and dark-eyed juncos share many of the same volatile compounds. Perhaps their odor is the reason for the inter-species allure.
For more information about Danielle’s research, you can contact her at djwhitta at msu dot edu. For more about juncos, check out the Junco Media Project!
This week’s BEACON Researchers at Work post is by MSU graduate student Jianxun Wang.
Have you ever imagined that you could swim with robotic fish someday? This may happen in the near future. More surprisingly, these robots would be capable of monitoring the aquatic quality around you and may even save you when you are drowning. These scenes have happened a couple of times in my dreams. I am Jianxun Wang, a Ph.D student of Prof. Xiaobo Tan in the Smart Microsystems Lab at Michigan State University, and I have been working on robotic fish for more than two years.
Back when I was a senior in college in China three years ago, I already knew that I would be working with a fantastic group on the research of robotic fish. I could still remember every detail the first time I came into our lab in the Fall of 2009. John Thon, a passionate researcher in our lab as well as a teacher at Holt Junior High School, gave me a warm welcome and guided me through the existing work. Our research began with an outreach activity to stimulate interest of precollege students in the fields of science and engineering through demonstrations of fish-like robots. These robots soon turned into a much more serious set of scientific projects. Later that day, Michael Carpenter, an undergrad research assistant, and Freddie Alequin, a graduate research assistant, showed me a demo of the robotic fish working in a 15-foot-long tank holding 6,000 gallons of water in the Smart Microsystems Laboratory. This is the first time I saw a real robotic fish prototype – a shell of green plastic with a rigid tail capable of swimming straight and making simple turns. I promised myself that I could and would make substantial contributions to propel these projects forward. From then on, I have been working with John, Freddie and Cody Thon, an optimistic and accommodating undergrad research assistant, all of whom are now my close friends.
Exhibit "Swimming with Robotic Fish" at the inaugural US Science and Engineering Expo in Washington DC. (L to R): Prof. Tan, Freddie, Cody and Jianxun
One year later, we developed a 6-inch fishlike robot with gray and yellow stripes modeled after a panfish. It had a GPS unit mounted on its head, a 3D compass embedded inside the front of the body, a wireless communication component on top of the shell, and a dissolved oxygen sensor suspended from its bottom. This robotic fish was designed to patrol a pond or a lake, while collecting and sending data about water temperature and the dissolved oxygen level. The robotic fish will provide a level of spatial and temporal sensing resolution that traditional water quality measurement approaches cannot match. Thanks to Felix Adisaputra, an undergrad research assistant working with me, we had a user friendly Graphical User Interface with Labview for remotely operating the robotic fish.
At nearly the same time, I started to shift part of my research time to work on the modeling of robotic fish. The main focus of my contributions are the control of individual and groups of them. The mathematical model I developed is currently being used for evolutionary design of robotic fish, which is a collaboration with Prof. Philip McKinley’s group. There, I met Tony Clark, a smart and hardworking Ph.D student of Prof. McKinley. We work together on the challenges existing in the development of autonomous robotic fish, which include realizing high maneuverability and high energy efficiency at the individual robot level and achieving adaptive coordinated movement (such as schooling) at the group level. Live fish and evolution computation provide a source of inspiration for effectively addressing these challenges. Consequently, in this project, Liliana Lettieri and Jason Keagy (two knowledgeable research associates from Prof. Jenny Boughman’s group) and Tony and myself are working together to create autonomous robotic fish by merging bio-inspiration, evolutionary design, and experimental prototyping. In particular, Tony has shown that the dynamic models I developed for robotic fish can be used successfully in evolving waypoint-following control strategies for these robots.
Recently, Osama En-Nasr, an excellent undergrad research assistant, John, Cody and I have been developing a so-called “predator robotic fish”, which will be used to study cooperation and social behavior in stickleback fish in collaboration with Prof. Boughman’s group. The idea is to use the robotic fish as a predator to elicit animal responses, since this “predator fish” could be controlled to demonstrate many complicated and repeatable behaviors. In this prototype, I have to acknowledge a very important and impressive technique: 3D printing. Supported by an NSF grant called “Evolution Park,” we luckily have this football table size 3D printer. With 3D drawing and selection of material, this 3D printer can provide us with arbitrary three-dimensional objects that have varying stiffness within a same object, which makes it much easier to create robotic fish prototypes. In the near future, we will print different types of robotic fish just like real fish. From this semester, we have two new members joining in this special group, Jared Moore and Sanaz Behbahani, and I am sure we can make some fascinating advances in the area of robotic fish with their contributions.
For more information about Jianxun’s work, you can contact him at wangji19 at msu dot edu.
This week’s BEACON Researchers at Work blog post is by University of Washington graduate student Tasneem Pierce.
When I started as an undergraduate in BEACON a year ago, I kept hearing about Avida and how powerful it is to study evolution in action. I decided to teach myself how to use the Avida software, and I quickly discovered that there are no tutorials for biologists interested in the more complex aspects of Avida. Fortunately for me, I was in the heart of BEACON, surrounded by people who were willing to take time to teach me how to use the software. Every single one of the people in the photo below, most of whom are in Dr. Ofria’s Digital Evolution lab, helped me in some aspect of my research with Avida. Now, I am working on creating a tutorial targeted at researchers with no computational background.
Front row (L to R): Charles Ofria, Owen Pierce, and Tasneem Pierce
BEACON’s collaborative atmosphere allowed me to start an Avida project of my own. I started my Avida project when I was doing research in Dr. Lenski’s Experimental Evolution Lab. Dr. Lenski’s long-term evolution experiment studies the genetic changes in twelve populations of Escherichia coli that have evolved for over 50,000 generations. If you are wondering how long that is, I’m 23 years old, and Dr. Lenski’s experiment started just a couple of months before I was born. There have been many cool discoveries in the long-term populations, one being that six of the twelve populations have increased mutation rates. These populations are called mutator lines, as they have damaged methyl-directed DNA mismatch repair systems, which have increased their effective mutation rates by a hundred fold. A high mutation rate alone will typically be maladaptive (more mutations are detrimental than beneficial), but if a mutator causes a rare beneficial mutation, that mutator may hitchhike to fixation meaning that the mutator becomes the dominant organism in the population.
How can a higher mutation rate fix in a population? Again, it is more likely that a higher mutation rate will break something important instead of making something better. What circumstances would lead to the fixation of a higher mutation rate? Using Avida, we can identify conditions under which a population will fix a higher mutation rate if it is easy to knock out a mutation-repair mechanism, but difficult to re-evolve one.
E. coli in nature cycles between the nutrient-rich gut and the external, nutrient-limited environment. In the Lenski lab lines, E. coli starts every 24 hours in fresh media, and by the end of the day, it is in a nutrient-depleted environment. In both of these situations, the E. coli face changing environments. A strategy in a nutrient-rich environment might not be beneficial in a nutrient-limited environment and vice versa. We tried to recreate the changing environments in Avida by rewarding Avidians for task set 1 and punishing them for doing task set 2 during one cycle, and then reversing this by rewarding task set 2 and punishing task set 1 in the next cycle.
Our hypothesis was that a moderate environmental change will select for organisms with a higher mutation rate. We had a variety of environments: a static environment where all of the tasks were consistently rewarded and 6 dynamic environments where there was a toggle between rewarding and punishing traits at different rates (100, 250, 500, 1000, 1500, and 2000 updates).
The starting organisms had a divide instruction with a low mutation rate. Organisms could mutate to have a divide instruction with a higher mutation rate as compared to the starting organism (2x, 3x, 10x higher). The Avidians could evolve a higher mutation rate, but they could not re-evolve a lower mutation rate, similar to how it is much harder to fix a DNA repair mechanism once it is broken.
Our preliminary results indicate the digital organisms can fix a higher mutation rate if they are subjected to a dynamic environment. We see that the ideal environmental change is not too short (not enough time to mutate) or not too long (no incentive to change as the environment is changing slowly). As seen in the graph, populations are more likely to fix a higher mutation rate if the change in the mutation rate is smaller (ex. 2x higher mutation rate is favored over fixation of a 10x higher mutation rate). Our initial runs were 100,000 updates long. When we increased the length of the runs, we found that there is an exponential decay in the number of populations at the lower mutation rate, and gradually most of the populations will fix the higher mutation rate. It is possible that, as long-term line E. coli populations go through more generations, more of the twelve lines will fix a higher mutation rate.
I am currently a first year graduate student in the Kerr lab at University of Washington. My new Avida runs will more closely mimic the Lenski lines as the environmental change will switch to a limited resource system instead of rewards/punishments system and the population size will be limited by requiring resources for successful replication. We hope to find conditions that lead to populations fixing a mutation rate that is a hundred fold higher similar to long term E. coli lines. Stay tuned!
For more information about Neem’s work, you can contact her at neem at uw dot edu.
Dr. Deb is currently in residence at BEACON at Michigan State University, and during Fall 2011 has been teaching Introduction to Evolutionary Computation at BEACON.
Copyright 2010 Michigan State University, Board of Trustees. All rights reserved.
BEACON is a National Science Foundation Science and Technology Center.
