In a paper published in The American Naturalist in 1999, Craig Benkman demonstrated the existence of a selection mosaic with coevolutionary hotspots in a population of the Rocky Mountain lodgepole pine. Where red squirrels are present, they drive selection on lodgepole pine cone structure. In locations where they are absent, red crossbills are more abundant and are part of an evolutionary arms race with lodgepole pine. Eighteen years after the paper was published, I spoke to Craig Benkman about how he got interested in this study system, his memories of field work, and what we have learnt since about selection and coevolution in the lodgepole pine system.
Citation: Benkman, C. W. (1999). The selection mosaic and diversifying coevolution between crossbills and lodgepole pine. The American Naturalist, 153(S5), S75-S91.
Date of interview: 8 February 2017 (via Skype)
Hari Sridhar: What was your motivation to do this particular piece of work? Your research on crossbills goes back all the way to your PhD, which was in the early 80s. Could you place this particular piece of work in the context of all the work you’ve done earlier?
Craig Benkman: Prior to this paper, I’d focused on trying to understand the patterns and causes of crossbill diversity. It was pretty clear that the diversity of conifer cones and seeds was driving crossbill diversity. Crossbills were adapting to the variation among these conifers. Before I started my PhD, my Master’s research was on red squirrels and their role in influencing cone evolution, with Clark’s nutcrackers as a seed disperser and the squirrels as antagonists – a seed predator. And I knew from other work that squirrels had a strong impact on conifer cone evolution. Only a year or two before this paper, I came to the realization that crossbills could actually influence the evolution of cones. Up until then, I thought that squirrels were driving the evolution of cones, and crossbills just followed along. They’re kind of ornaments on a tree. Crossbills were pretty interesting to study but they didn’t have a big impact. And then visiting the multiple different mountain ranges with and without squirrels, made it obvious something was going on with crossbills. Visits to the South Hills especially, and the Cypress Hills, brought that home.
HS: Stepping back a bit, how did you first get interested in crossbills?
CB: When I was an undergraduate in the late 70s, and then also a beginning graduate student, there was a lot of interest in foraging theory. My PhD advisor Ron Pulliam was one of the architects of foraging theory. There was also a lot of interest in resource competition. But, I thought, one of the big problems was that the best papers in those fields were ones that could actually measure resource availability in the field. Yet, very few studies, or in very few systems, can you do that – nectarivorous birds, Galapagos finches – ground finches that eat seeds in the dry season -and crossbills, I think, are another one. Conifer cones are very discreet resources. You can measure crossbill feeding rates, how many calories crossbills get per unit time, because you can count the number of seeds consumed per second, weigh the seeds, and convert them into calories. And there’s not a lot of variation in quality of the seeds in terms of specific caloric value or chemical defenses. And then you could bring the conifer cones to the lab and ask questions you couldn’t ask in the field. And crossbills are extremely tame. So I thought it was a nice system where you could go back and forth between the lab and the field while quantifying resource availability. And I think, over time, it’s proven to be a very good system.
HS: At the time when you decided to work on crossbills were you already familiar with their natural history? Were you a birdwatcher from a young age?
CB: I wasn’t a birdwatcher from a young age. However, I did spend a lot of time in the mountains and in conifer forests when I was young. My passion for birds started in college. Then in the summer before starting my PhD, I helped a friend with his PhD, and crossbills were quite common in that forest. I got to know them a little bit more. So, I had some feel for crossbills and their habitats, and knew enough to believe they could be interesting to study.
HS: How did you pick the various sites you chose to work in?
CB: As I mentioned earlier, in the late 80s I was trying to understand the diversity of crossbills. By looking through the forestry literature I was able to predict which conifers could support a specialist, in terms of their seed availability and seed production. Rocky Mountain lodgepole pine was one of them. And so I was doing a lot of reading about lodgepole pine. And the cones of lodgepole pine are very similar throughout their range from the Yukon to Colorado, except in the Cypress Hills. The Cypress Hills was intriguing because, a few years earlier, I was in Ottawa, measuring white-winged crossbills. And I met the curator who was retired, Earl Godfrey, who wrote the Birds of Canada. I told him about my ideas on Newfoundland crossbills and red squirrels, and he said you should go to the Cypress Hills because there’s a very large, very common crossbill there and red squirrels were absent. I measured the birds that he collected in the late 1940s, and didn’t think about it until years later, and then realizing, all these lodgepole pine cones are quite different, they were rather large and so were the crossbills. So that’s what drew me to that area. There were also ranges to the south without squirrels, so I visited them too. That’s when I began to think that crossbills might be playing a role in cone evolution.
HS: At one point in the manuscript you say that the population represented by the red crossbills collected by Godfrey may be extinct. Do we know more about this population now?
CB: We don’t know. I’m a little embarrassed about that because we haven’t followed up on it. But I have gotten toe pads from all the crossbills – different forms throughout the world – Himalayas, Vietnam, so forth and so on – for which we haven’t been able to get blood samples in the wild, or tissue samples. We’ve also requested 10 toe pads from crossbills collected in the Cypress Hills. We’re waiting for them to be sent to us, and then we’re hoping to do a lot of DNA sequencing. Then we will know to what extent the Cypress Hills crossbills are genetically distinct from other forms. And if that’s the case – if they are genetically distinct – that’ll be a great motivation to go back to try to find the Cypress Hills crossbill. I’ve been there a few times, and several students of mine have been there too, but we just don’t see many crossbills. It is very unlike the South Hills where crossbills are abundant or what Godfrey described in the Cypress Hills in the 1940s, presumably because red squirrels were introduced there in 1950.
HS: Do the sites that you worked in for this study continue to be your primary research sites today?
CB: We go back every year to the South Hills. I’ll be going there this summer. And so, yes. But the other sites, I haven’t visited most of them in over 10 years.
HS: Has the South Hills changed a lot from the time you worked there for this study?
CB: No, it hasn’t changed a whole lot, unlike a lot of mountains around here where bark beetles have killed the majority of lodgepole pine and spruce. Beetle kill hasn’t been that bad in the South Hills, which actually appears to be related to the absence of red squirrels. But that’s a longer story.
HS: Could you give us a sense of what this work involved? What was your daily routine during the study? Did you have people to help you?Were you staying in the field when you were doing this work?
CB: A lot of the initial cone sampling was done by Bill Holimon, who was a former student of mine, my first Master’s student. We use a 10-meter pole with a clipper on the end to cut and collect cones. I’m not strong enough to do it by myself, but several of my students are. I’ve visited all these sites, and we just camp, usually in a campground or just out somewhere in the forest, and we work all day. And then, often, in the afternoon, we’ll go for a hike and write up field notes. And then repeat that day after day until we’re done. And then go to another site.In the South Hills, I usually camp there for about 10 days to two weeks every year. And then my students, depending on the projects they’ve been doing, will sometimes stay several months. So I will be there in the beginning and then they will stay longer. And then I’ll maybe visit the sites at other times of the year as well.
HS: Where were the foraging experiments conducted?
CB: Those were all conducted on campus at New Mexico State University in Las Cruces. We had two rooms in the animal care facility, and when I first got there I built aviaries, just like I’d done in previous places, with hardware cloth and wood. I’d spend a weekend building them. Sometimes I’d have help, but usually I just did it myself.
HS: Do you know what those facilities are used for today?
CB: I have no idea.
HS: Could we go over the list of people you acknowledge to find out a little more of how you knew them and how they helped?
CB: Yes, I can. Kristie Edwards, Matt Millen and Adrian Tedeschi were undergraduate students who I hired to measure pine cones. Because we gather lots of cones and it takes a lot of time to measure them, I often hire students to help. Chris Milensky measured cones, but he also joined me in the Cypress Hills when we did our first cone collections there, looking at selection by squirrels as well. So we camped there in the fall, and then we also visited the Sweetgrass Hills – another range without red squirrels – and sampled cones there. That trip was the one that really got me thinking that crossbills influenced cone evolution. Bill Holimon, who I already mentioned, did a lot of the cone collections, and also conducted the foraging experiments in New Mexico. I don’t remember if I was involved in those foraging experiments. Other folks were individuals who gave permission to conduct fieldwork. Dick Hutto provided survey data of crossbills in the Rocky Mountains, which allowed us to compare our data from the South Hills to the Rocky Mountains. Chris Smith had the heritability data for lodgepole pine cones that was unpublished and it remains unpublished. Chris had conducted classic work on co-evolution between squirrels and lodgepole pine. Trudy Nicholson was recommended to me from a colleague. She’s an artist that researchers at the Smithsonian Institution often hire to make scientific illustrations and she was great. She did Figure 2.
HS: What about Figure 1?
CB: Well, I just used drawings from very old books that didn’t have copyrights. Pine cones were from a pine book. I have never seen the original for the bills. I just have a Xerox from a colleague that I used.
HS: E. Greene.
CB: Erick Greene. He’s a professor at the University of Montana. I’ve known Erick for a long time.He does a lot of song recordings, and he kindly offered to make spectrograms for me. He did it in a few minutes, whereas it would take me days.
HS: You thank J Trotter for arranging a site to capture crossbills.
CB: Oh, yeah. Jack Trotter. He lives in Twin Falls, Idaho. He did Breeding Bird Surveys, which are conducted across North America every year. He did one survey every year that barely extended into the pine forest in the South Hills. Even though he spent very little time during the survey in pine, he found crossbills every year. This was useful to know after my first visit to the South Hills. It turned out Jack’s in-laws had a cabin in the South Hills and crossbills came down to a water faucet outside. So, Jack showed me the location and that’s where I went to catch them.
HS: You thank Dolph Schluter for providing some software
CB: He developed the software for the pursuit projection approximations that we used. I think Dolph, actually, sent me the software. I had difficulties downloading it from his website or getting a latest version. I just don’t remember the details – it was so long ago.
HS: J.N. Thompson.
CB: John Thompson was the one who invited me to participate in a symposium on the geographic mosaic of co-evolution for the American Society of Naturalists. The society has a vice-presidential symposium every year, and John was the vice president. He was very encouraging. I thought my work didn’t fit very well in the symposium but John convinced me otherwise. He is also a big advocate of my work, which I very much appreciate.
HS: You say this was part of a working group at the NCEAS. Tell us more about that.
CB: John Thompson was part of this working group, but I am less certain how it was related to the symposium and resulting publications.
HS: Was all the work presented at the symposium put together in this issue of the journal?
CB: Yes. Everyone needed to send John their manuscript prior to the conference. Every year The American Naturalist usually publishes a supplemental issue devoted to a vice-presidential symposium.
HS: Did this paper have a relatively smooth ride through peer review?
CB: I don’t remember there being a lot of changes because of the reviews; for better or worse. This paper was written more as a story, not in a standard journal format. But the reviewers didn’t mind.
HS: In the section titled “Replicate and Rapid Coevolution”, you cite a personal communication from M. Sauchyn.
CB: Sauchyn is a paleobotanist who worked in a bog in the Cypress Hills, reconstructing past plant communities. I was provided unpublished information on the occurrence and relative abundance of lodgepole pine pollen and macrofossils over time, which provided historical context for crossbill-pine co-evolution. I think I also cite a paper that Sauchyn had written.
HS: Later you also mention a personal communication from O.K. Davis.
CB: Oh, yes, he did his PhD in the Albion Mountains adjacent to the South Hills. I just happened to be giving a seminar at the University of Arizona in Tucson, where he’s a faculty member. I was able to meet with him during my visit. I knew his papers and we had corresponded, but then I was able to talk to him in person, which was very fortunate.
HS: In another place in the manuscript, again on Page S85, you mention some unpublished work. I was just curious about whether that was published later as a separate paper. You say,“the ponderosa pine crossbill (type2 of Groth 1993) is common in Rocky Mountain ponderosa pine (Pinus ponderosa ssp. scopulorum) forests and has a bill and a groove in its palate for husking seeds (husking groove) that approximate the optima for foraging on these pines.”
CB: Yes. In 2001, I published a paper with Bill Holimon and Julie Smith as co-authors that included data consistent with bill size approximating the optimum. In 1993, I had published such data for the groove in the palate used to husk seeds.
HS: Do you remember how long you took to write this paper, and when and where you did most of this writing?
CB: I was in Las Cruces, at New Mexico State. The symposium was in June 1998 in Vancouver, British Columbia, and we needed to send John Thompson our completed manuscripts beforehand. If I recall correctly, sometime during the winter when I had started to work on the manuscript, I wrote John that my research wasn’t a good fit for the symposium. I was worried because speciation rather than gene flow was the outcome. But John convinced me otherwise. So, it must have been over about a four-five month period in the spring that I wrote the paper. I think I included crossbill survey data that Holimon gathered in the field in early June, less than a month before the conference. It wasn’t a drawn-out paper because of the deadline.
HS: This is a single author paper, as are many of your other papers. I was wondering whether that is because you like to work on your own. Is that something that you’ve consciously tried to do in your career?
CB: Although I like to work with my students, I’m fairly independent. I am more of an introvert. I like to walk around in the woods by myself. I used to backpack by myself for weeks. So, yes, I often end up working by myself, in part because I often get my inspiration when I am by myself, but I wouldn’t say it is something I consciously or intentionally do.
HS: At the time when the paper was published, do you remember how it was received in academia? Did it attract a lot of attention?
CB: I know that my presentation went over well at the meeting. I think the paper was well-liked too. I don’t think there were any write ups about it, like in Science magazine or something like that. The simple story and one figure in particular made it easy to visualize and understand. I think it has attracted attention over the years as a good example of the geographic mosaic of co-evolution, and of co-evolution itself.
HS: Are you referring to a particular figure in the paper?
CB: Figure 2 showing the cones, the bills, geography, and then with a little bit of text to that, you have a story. It’s used for teaching on some websites. I think it’s a compelling story. And it seems the basic story hasn’t changed. We’ve looked at a lot of different things since and the pieces still fit. And then we’ve looked to replicate it in other systems to see if it’s general, and it seems that the basic story applies widely in crossbills.
HS: What kind of impact do you think this particular paper has had on your career?
CB: Oh, I think it has been tremendously important. A few years before this paper, I was thinking of shifting to work on lizards or toads! But finding indications of co-evolution and of speciation in the South Hills has driven much of my subsequent research and that of my students.
HS: Why were you considering moving to a different study system?
CB: Well, I was living in southern New Mexico in a desert, and I really like lizards and I was interested in conducting studies on communities of lizards and on toads, on tadpoles especially. And I was far removed from good crossbill habitat. Moreover, crossbills, because most are nomadic, can be hard to find and very hard to study. Crossbills are abundant and resident in the South Hills so I could go there year after year and always find them.
HS: Did this paper also have a strong influence on the trajectory of your research immediately afterwards? Did you follow up this work in any way? And, over the long term, has the path of your research shifted in any way because ofthis study and how important it turned out to be?
CB: Yes, it has affected my research. We followed up this work multiple times. Just a couple of years ago, we wrote a paper looking at selection exerted by crossbills and other predispersal seed predators in the South Hills. The one criticism of our earlier work was that we hadn’t measured selection exerted by crossbills in the South Hills. We had measured selection exerted by crossbills in a similar situation in the Little Rockies in Montana, but not in the South Hills. And so we measured selection in the South Hills, in part because we were also interested in documenting the replicated nature of selection. And then, this fall, we published a paper on the genomics of New World crossbills following up on the speciation component. Quite surprising to me, the South Hills crossbill is the most genetically distinct red crossbill in New World. So distinct that it is the one form of red crossbill in North America that is now recognized as a distinct species, which I think is very interesting because this supports the original speciation argument in my 1999 paper. So, yeah, co-evolution and the patterns of selection exerted by crossbills on pines and responses in terms of bill evolution, and then speciation, are the two main areas of research that I’ve focused on.
HS: Do you have a sense of what this paper mostly gets cited for?
CB: I think it is cited mostly as an illustration of the geographic mosaic of co-evolution, or co-evolution. Not so much for speciation, because we really didn’t have any direct evidence of reproductive isolation. But I think, increasingly, it will be cited more for the speciation component based on our behavioral and genetic work.
HS: Today, 18 years after the paper was published, would you say that the main conclusions from this paper still hold true,more or less?
CB: Yes. I’d worked on the system for a long time. When you work on a system a long time, you get to know a lot about it, so that even before you have complete data sets you have a good hunch of the outcome. All the roads were leading to Rome, and they were converging on the story I told in the paper. I guess we shouldn’t be really surprised. If it had been a new system I was working on, and I didn’t know the biology as well, then I think it’s more likely that my initial conclusions wouldn’t have held up. That is an advantage of long-term studies. I think their conclusions are more likely to stand the test of time.
HS: If you were to redo this study today, would you change anything about it?
CB: What I would redo are things that we subsequently did and published. For example, we have measured selection exerted by crossbills on lodgepole pine cones in the South Hills, and we have been able to estimate the optimal bill depth for foraging on cones in the South Hills to compare to the observed bill depths. Before we had not measured selection exerted by crossbills or been able to predict an optimal bill depth and test for reciprocal adaptation by crossbills. So yes, we would change it, but we’ve been changing it gradually over time, as we fill in the blank areas, or strengthen the areas where information was incomplete.
HS: In one place, you say, “Crossbills have evolved stouter bills in the Cypress Hills and in the South Hills and Albion Mountains in apparent response to the increase in putative defenses directed at crossbills.” And then, you say,“These differences in bill structure are likely heritable.” Do we know more about the heritability of bill structure today?
CB: We do for crossbills in general, but not for crossbills in the South Hills or Cypress Hills. Ron Summers and colleagues in Scotland published a paper in 2007 in which they determined the narrow-sense heritability of bill depth for captive-bred crossbills. The heritability of bill depth was about 0.6 or 0.7.Rather high, much along the lines of Darwin’s finches and a lot of other birds. I think it’s safe to assume that these crossbills have similar levels of heritability.
HS: In another place you say, “Whether crossbills in the Cypress Hills  or the South Hills  have evolved the optimal bill depth for foraging on their respective cones is unknown; additional feeding experiments are needed on South Hills crossbills with sufficiently large bills to detect where feeding efficiency levels off.” Have such experiments been done?
CB: Yes. In a 2001 paper in Evolution we published such feeding performance data, and in a 2003 paper, I published some preliminary data showing phenotypic selection on crossbills in the South Hills that were consistent with the predictions based on feeding performance.
HS Where did you obtain the study skins of the Cypress Hill crossbills on which you made measurements?
CB: The National Museum in Ottawa. I visited there in 1988 to measure white-winged crossbills, which I was focusing on for fieldwork at the time. But after telling the curator at the museum –Earl Godfrey–a bout my ideas on the Newfoundland crossbill, which I had been thinking about that summer, he recommended that I look at the Red Crossbills he collected in the Cypress Hills in the 1940s. They were large-billed and seemingly resident like in Newfoundland. So, I measured them without any idea that they might become important to me.
HS: You say, “Perhaps the similarity of the crossbills and lodgepole pine in the South Hills and Cypress Hills is the result of the colonization of, for example, the more recently forested Cypress Hills by crossbills and pine from the South Hills rather than an example of convergence. This scenario is doubtful, given that genetic studies indicate that lodgepole pine in the Cypress Hills are derived from a source well north of the South Hills (Wheeler and Guries 1982a). If, as is likely, cone evolution in each area was gradual from a Rocky Mountain–like ancestor, then gradual and convergent evolution of resident crossbill populations in each area is the most plausible scenario. This, however, should be tested with genetic studies.” Have such genetic studies been done?
CB: We have genetic data for the pines that we published a few years ago, which is consistent with independent evolution of the pine east and west of the Rocky Mountains. I mentioned getting toe pads from the Cypress Hill crossbills from the 1940s. But we haven’t done the genetic studies using those samples yet. But given that South Hills crossbills are so sedentary, it’s extremely unlikely that these crossbills rather than nomadic crossbills colonized the Cypress Hills.
HS: You say, “One reason is that there are trade-offs so that costs eventually outweigh benefits. Based on the estimated costs and benefits in terms of individual cones (fig. 6), cones should continue to evolve in response to selection from crossbills. However, this point assumes that the only costs relate to individual cones, yet other costs may be involved. Heavier cones, for example, might cause the slender branches to break or bend so that photosynthesis is compromised.Without more complete measures of trade-offs,I am unable to differentiate between a co-evolutionary arms race that is ongoing versus one that has stabilized.” Later on, did you figure out better ways to measure such trade-offs?
CB: We still haven’t measured the trade-offs directly. I have tended to believe that the arms race in the South Hills has not stabilized, but that is more a belief than based on unambiguous data.
HS: You highlight the need for studies of mating behavior to determine the extent of reproductive isolation. Have such studies happened?
CB: Yes. When I wrote this paper, I had a graduate student, Julie Smith, who had just finished her Master’s. And when I found this system, I told her about it, obviously, and she got very excited. We thought that measuring the extent of reproductive isolation would be a great question to address in the South Hills. And she did such a study, and what she found is quite remarkable. Using a measure of reproductive isolation where zero represents random mating between different forms of crossbills and one represents complete reproductive isolation, Julie found a value of 0.999 or nearly complete reproductive isolation over a two-year period. We found the same in four subsequent years. And then, what’s really cool is, as I mentioned, the genetic data that we’ve just published this fall shows that these crossbills are monophyletic and genetically distinct. The genetic data show that the high level of reproductive isolation that we measured over 6 years has been occurring for many years.
HS: You say, “Divergent selection, therefore, is a double-edged sword. Divergent selection is critical in creating a selection mosaic but erodes the selection mosaic if it promotes reproductive isolation and speciation”. Today, given all that we have learnt since this paper, would you say it the same way?
CB: If I rewrote that, I would maybe qualify it, in the sense of what I interpret as a geographic mosaic of co-evolution. To me, John Thompson describes a situation where you have populations where there’s gene flow between them. In this case, I’m just basically saying that once you get these populations cleaved off and no longer interbreed, there’s no potential for gene flow, and then it’s no longer the selection mosaic, in the sense of the geographic mosaic theory of co-evolution.
HS: Have you ever read this paper after it was published?
CB: Yes, I have. I read it when I was writing other subsequent papers. I’ve skimmed parts of it when we were working on other systems. But I’m not sure if I’ve read it in the previous 10 years. And I actually read the whole paper yesterday! I should do that more often.
HS What struck you most about it when you read it yesterday? Do you feel that your writing has changed in anyway?
CB: I’m not sure. I don’t know what struck me the most. I think the one part that I would have changed, that I was the most uncomfortable with, was the section on remixing and gene flow and maladaptation. I felt that that was a little too speculative. Subsequent work supports all the other parts quite well, but I probably shouldn’t have included that part. But it was my attempt to address all the components of the geographic mosaic. I cringed when I read it.
HS: Would you count this paper as one of your favorites?
CB: Oh yes. I think back to the time when I discovered these things, and finding the South Hills and going to them and collecting pine cones and then realizing, wow, I think I found a new crossbill species. That just brings back this discovery element. And then also realizing when you’re returning home that I might be working on this system for the rest of my career. There’s just so many cool things to think about.
HS: What would you say to a student who was about to read this paper today? Would you guide his or her reading in any way and point him or her to subsequent papers they should read along with this? Would you add any caveats to their reading?
CB: Well, I would say some of what I said earlier. This was the first in a series of papers. If you’re interested in, let’s say, the speciation component, there are other papers to read. If you’re interested in co-evolutionary components, there are other papers that I would recommend to read. I wouldn’t hold this up as a model paper, but I think it does tell a story that has largely stood the test of time. I think it could have been said more succinctly, and I think there are other ways to have said it or just show it, but then some of those datasets were not available. I think there are probably stronger papers that have been published since for the different components, just because we had more time to reflect and gather more data to more convincingly test some of these ideas.