Ep 152: A coevolutionary conundrum (with Diego Vázquez)
When a bee visits a flower, what decides the match? How can a neutral model explain community assemblages? Why are long-term ecological studies so important?
On this episode, we talk with Diego Vázquez, a community ecologist at CONICET in Mendoza, Argentina, who has spent his career mapping the networks of interaction between plants and pollinators. Textbooks still teach coevolution as tight, faithful pairs, but Diego's work tells a different story, that specialists mostly interact with generalists, and the best predictor of pollinator visits to plants turns out to be not matching traits or shared ancestry but abundance. Much of the structure ecologists prize in these networks can emerge from "neutral" models that know only who is common and when. We dig into why predicting these interactions remains hard, what his twenty-years of data on declining solitary bees is showing, and what it takes to keep long-term ecology alive in Argentina after deep cuts to science funding.
Cover art by Brianna Longo
-
Marty Martin 0:05
Hey John, let me ask you something. If I dropped you in a forest that you'd never been to before, maybe somewhere in the Andes, and I gave you a list of every plant and every pollinator in that forest, along with how common each one is when they flower, how long their tongues are, do you think you could tell me who visits whom?
John Drake 0:21
Well, I'd like to think so, Marty. We know a lot about what drives species interactions, things like morphology and phenology and abundance. But the honest answer, after decades of research, is not really. We can predict the broad outlines, which actually mostly has to do with abundance, but the details- which bee visits which flower, how often, and why? Well, those remain surprisingly hard to pin down.
Marty Martin 0:43
Yeah, and that tension between the patterns we can see and the predictions we can't yet make is at the heart of our conversation today. Our guest is Diego Vazquez, a community ecologist at Conise and Mendoza, Argentina, and the director of the Argentine Institute for Dryland Research.
John Drake 0:58
Diego has spent more than 20 years studying the networks that connect plants and their pollinators. He was one of the first to show that these networks are organized in a strikingly asymmetric way. That is, specialists don't pair up with other specialists the way textbooks had assumed. Instead, specialist bees tend to visit generalist plants, and specialist plants tend to be visited by generalist pollinators. And much of that pattern can be explained by something surprisingly simple: just how common each species is.
Marty Martin 1:25
That insight, that chance and abundance do a lot of the heavy lifting in ecological networks, has been both influential and uncomfortable for the field. It raises a hard question: if so much of what we see is just a consequence of who's common and when, how much of network ecology is about real biological processes and how much is a statistical artifact?
John Drake 1:45
Diego has tackled these questions head-on, building predictive models that combine traits, abundance, and evolutionary history. The results are honest and humbling. Predictive skill remains modest, and abundance alone often does as well as anything fancier.
Marty Martin 1:58
But that's not the whole story. Diego also runs one of the longest-running pollinator monitoring programs in South America, 20 years of solitary bee data from a nature reserve in Mendoza. And he's doing all of this while navigating deep cuts to Argentine science funding and leading his institute through that crisis that echoes what's happening here in the U.S.
John Drake 2:15
It's a big conversation: networks, prediction, pollinator decline, and the future of science in a difficult time. We're glad Diego is here to walk us through it.
Marty Martin 2:23
I'm Marty Martin.
John Drake 2:24
And I'm John Drake
Marty Martin 2:26
And you're listening to Big Biology.
John Drake 2:39
Diego, let's start at the beginning. So you went to the University of Tennessee for your PhD. You worked with Dan Simberloff there, and our listeners are going to know Dan from a recent episode. Can you tell us what it was like working with Dan and how that shaped the kind of ecologist you became?
Diego Vasquez 2:54
Dan is great. If you want just one sentence, short sentence. Yeah, he's a very intellectually stimulating person. So if you talk to him, you leave the conversation with ideas and excitement about whatever topic you've been discussing with him. He's also very supportive of his colleagues and students, especially of younger people. So, if anything, he gave me confidence in myself. So I talked to him and left the conversation feeling that what I was thinking was worthwhile, and that really helps you when you are a grad student. You don't want to go knock at his door without an appointment, though. That's something that. But yeah, he's great.
Marty Martin 3:53
Huh? How did you come to work with him? Did you guys share an interest, or did you read one of his papers before you applied?
Diego Vasquez 4:00
So I knew his name. He had. So I went to Tennessee. This is a story. I went to Tennessee to work with another professor that was Stuart Pimm, because I was really interested in food webs, and Stuart was a big name in food webs. But it didn't really work out, you know. I guess our approach and specific interests didn't really match that that much. And so I met Dan at Tennessee, and we talked a few times, and at some point I said "Well, it's probably better to switch advisors". So we discussed my ideas for a project, and he said, "Yeah, why why not? Sure." And that's how I came to work with him. And I think that that was one of the best decisions in my life because it really worked out for me, and actually brought a whole generation of Argentinian students to Tennessee. I don't know how many, but probably I don't have enough fingers in one hand to count them.
Marty Martin 5:09
Wow
John Drake 5:10
Wow, that's fantastic.
Marty Martin 5:13
So what were the interactions that you had that sort of explains the origins of your dissertation? You worked on introduce cattle and how they reshape plant pollinator interactions in Patagonia. That's not really an obvious PhD project in the Smoky Mountains of Tennessee. What drew you to that question with Dan?
Diego Vasquez 5:35
Yeah, so when I went to Tennessee, as I said, I was really interested in food webs and also conservation problems. And the first topic I thought about developing for my dissertation had to do with fragmentation, habitat fragmentation, and how that influences food webs, how the smaller fragments sort of lose the top predators, and how that cascades through the footweb. And I thought about doing a project in the Amazonian forest, and there was this project in near the city of Manaus in Brazil that had been going on for decades, and I thought it was a perfect setting for this idea. But it turned out that the top predators don't really care about fragmentation. They use both the fragments and the matrix that is sort of logged forest, pretty pretty much the same. So my ideas didn't really work out in that place. Then I thought, well, maybe I can do something like that in Patagonia. So I started looking in into the literature and talking to people that were there. Fragmentation is a problem in Chile. I thought I could work in Argentina just to contribute to science there, but then it turned out that fragmentation wasn't a problem there, and there were other problems like fire or cattle grazing in the native forest that were more of conservation concerns. And that's how I started thinking about this kind of problems. And I talked to a colleague, Marcelo Eizen, who's also a, he's a researcher in Bariloche in Northern Patagonia, and he was the one who suggested why Why don't you work with plants and pollinators, which you can also study in the same way as predator-prey interactions. And that's how I came to my dissertation project. It turns out that more than 90% of plant species in these temperate forests in Patagonia depend on animals for pollination and also for seed dispersal. So, the prospect of disruption of those interactions by grazing by exotic animals or by other types of human influence was pretty high, and then I thought, well, this is worth studying, and that's how I came to study this kind of problem.
John Drake 8:32
Yeah, what a great story, and maybe it underscores the ways in which science just kind of weaves and bobs around and never seems to follow a linear path. One of the things that you noticed in your data, once you were collecting data for your PhD, was that interactions between plants and pollinators are asymmetric. And I think what you mean by that is that specialist pollinators tend to visit generalist plants, and specialist plants tend to be visited by generalist pollinators. In some respects, that seems like maybe it should have been obvious, but it wasn't. Why did people assume that specialization was symmetric in the first place?
Diego Vasquez 9:09
I don't know why people assumed that, but I also assumed that. I thought about specialization, I was I always thought about a you know a pollinator visiting one plant which is only visited by that pollinator, and that has ecological and evolutionary implications about you know co-, so reciprocal influences that is very different from what you would expect if one of the two species is not really specialized. And I spent many hours observing flowers and recording visits by different kinds of flower visitors in this forest. And while doing that, I realized that there were some plant species that were visited by many insects, many of which I had only recorded in just that species of plant. And the same thing happened the other way around. Some pollinators were really visiting many plant species, some of which were only or mostly visited by this animal.
Diego Vasquez 10:27
So, for my PhD, I wanted to test this idea that specialists are more affected by environmental change than generalists. The idea is that if you depend on a few things, and things change, then you are sort of in trouble. But if you are talking about mutualists like pollinators and plants, or seed dispersers and plants, or whatever, then it's not just what you do what matters, but also what the other species does, and if the other species is not specialist, then you don't expect such a high impact from environmental change than if the two species are specialists. So that led me to this idea of an asymmetry or symmetry in specialization, and how that sort of made the predictions of environmental change subtler. So I wrote one chapter of my dissertation with this idea. I realized that some other people had thought about this too, but that it wasn't clear. So in my system, it was really clear that that happened a lot, but it wasn't really clear whether that happened was just a peculiarity of my system or not. I suspected that it wasn't, but it wasn't really documented.
Diego Vasquez 11:52
So I, for my postdoc that was at NCEAS, this the National Center for Ecological Analysis and Synthesis at the University of California in Santa Barbara, where John was also a postdoc. For my project there, I proposed to do a synthesis project, so that I mean a project based on literature, looking at whether how how prevalent was this asymmetric specialization, and I found just to make a long story short, I found that that is very common. I mean, all the data that I could get at the moment, this is over 20 years ago, show that this was common in communities around the world, in ecosystems around the world. And at the time, that was sort of surprising and sort of a different way of looking at specialization in interactions. That it was there, the data were there, people had looked at those data many times, but people hadn't really realized that this is how things work in these ecosystems.
John Drake 13:10
What was the reception to those ideas like when you started publishing? Were people surprised? Did you get pushback?
Diego Vasquez 13:17
No, I think people were excited about that. I wasn't the only one. So while I was doing that, another group of people. So that was Jordi Bascomte, who was also a former NCEAS postdoc, Pedro Jordano and Jens Olesen, and Carlos Melian. So the four of them. Carlos was also is also former NCEAS postdoc. So NCEAS figures prominently here in the history of ecology, so they were also working in with a different approach in similar ideas, and they came to the same conclusion. And so all this research, parallel research, I think was really had an impact in how people looked at and interactions and how communities organize. So yeah, I think people were really excited about that.
Marty Martin 14:06
This it's tricky to have a conversation about something like this because I mean I'm pulled to define words and get into the weeds about methodology, but we don't want to do that. At least we don't want to spend all of our time on that. I have to dip my toe into it a little bit and ask back to a word you used just a minute ago because it feels like this might have some important role in what we're talking about. When you say you're measuring visitation or you're sort of quantifying these visits, can you just talk about, especially in the context of this larger survey, where there were a lot of different systems with a lot of different species involved, how would we usually quantify that visitation part that sort of leads to the specialism versus generalism term?
Diego Vasquez 14:55
The way I did it was to just stand for specific amount of time in front of a bunch of flowers and record everything that visits those flowers. And you do that many times and you end up with data of frequencies of visitation of different animal species to different plant species. But you can also look at nests, bee nests, or animal nests, and find the pollen. You can also catch the animals and identify the pollen bodies. Nowadays, you can also do environmental DNA, so then then get to infer interactions from the DNA. So, for example, the DNA in the bodies of the animals, and so there's different ways. Some are plant-centered, so you focus on a plant and record what comes, or animal-centered, so you focus on an animal and record whatever it has that tells you what which plant it visited. I don't know if that answers the question.
Marty Martin 16:10
Yeah, that answers. I mean, I asked the question because I really wanted to get a better understanding of the terms generalist and specialist. I mean, that's dichotomous, but of course there's gradation, and so the what it means to be a specialist, you know, the specialism usually has to do with some very particular resource that a very particular species can exploit. Whereas generalism, the idea itself, connotes many different ways that different reasons that that particular species might be on or near or interacting with another species. So I want to sort of understand how that comes together because I want to understand what is that distribution of variation from those extremes. Like, are most species at the ends or most species in the middle?
Diego Vasquez 16:56
Yeah, it's so there's a whole gradient from extreme generalist species that visit many plant species or plant species that are visited by many animals to species that were recorded only visiting one plant species or that animals that yeah. So plants or animals that were recorded interacting with only one other species. That doesn't mean that necessarily those apparently specialized species can't interact with any other species. If you put them in a new environment, they might interact with other species, or if you sample enough, you might realize that they actually can interact with other species, and that's a problem with the data. Rare species tend to have, so we tend to record fewer interactions for rare species and for abundant species, and so we get into sampling effects in sampling artifacts. So some of the apparent specialization that we observed may be caused by sampling issues, but also you have to think that organisms sample between quotes in nature, so they encounter other species, and so the abundance and variety of species also influences with how many species they will be able to interact. And so, coming back to your question, there's a whole gradient of from generalization to specialization, and this specialization may be facultative in the same in the sense that they might interact with other species if given the opportunity, or it might be obligate because they just can't interact with other species, and that's you need to to do some to to do natural history to to be able to tell those different kinds of specialization apart.
John Drake 19:12
So you mentioned a few minutes ago about your NCEAS project to try and understand just how common this was in different kinds of networks, and then you had a paper in 2004 that showed that this asymmetric specialization is really quite pervasive. I think you found it in 18 different networks around the world. I was wondering what that means for how we think about coevolution. I mean, if specialists mostly interact with generalists, is tight pairwise coevolution of the kind that we're typically taught in ecology and evolution classes? Is that maybe actually rarer than we thought?
Diego Vasquez 19:49
Probably, but doesn't that doesn't mean that coevolution is rare. It's just different from what we thought. One way in which it can happen is that usually abandoned generalists interact with many species, but they also tend to interact frequently with other abundant generalists, and they can influence each other. There's the opportunity to influence each other a lot. We've done some studies in our system here in Mendoza, in Villavicencio Nature Reserve, where we also studied the plant-pollinator interactions. We've used some statistical methods to infer reciprocal selection among species. And our conclusion was that there was opportunity. The greatest opportunity for reciprocal selection, so for coevolution, was among the abundant generalist species who tend to interact a lot with each other. Now that doesn't really answer your question. Is reciprocal, real reciprocal specialization, rare and if so is coevolution between that kind of species also rare? Well I guess it is because at least in these kind of interactions, reciprocal specialization seems to be rare.
John Drake 21:19
Do you see anything, any evidence of that in the evolution of different characters? So I'm thinking about traits. My idea is that generalists. You said generalists tend to interact with lots of generalists, might result in some convergence toward intermediate trait values, with specialists having extreme trait values. Is there anything like that that happens?
Diego Vasquez 21:40
Yeah, yeah, I don't. I guess I don't know. That would be the answer. Yeah. So when reciprocal specialization occurs, I think, and there's examples of that that can influence each other. When that doesn't happen, I guess there's things where work differently, but I don't know.
Marty Martin 22:04
In a related vein, but in the broadest possible way, what's the most common modality? Is specialization the more common variant, or would you say generalism, irrespective of species?
Diego Vasquez 22:17
So a lot of what we call specialization is this apparent specialization that involves species interacting in our data with only few other species. And that involves many species that tend so are sort of peripheral to the network, and they tend to be connected with some abundant species that is interacting with many other species. That's very, very common. You know, this comes partly from the distribution of or similar to the distribution of abundances in communities. You tend to have a few abundant species and many intermediate or rare species. Well, that because and I guess we can talk about that too, because abundance influences so much the interactions in the community, and you have such a long tail of intermediate or rare species, and you end up having also a long tail of moderate or extreme specialists in your community.
John Drake 23:28
Yeah, so I mean, I think you've described this idea as kind of like common species accumulate more partners, just simply because they're encountered more often. Sometimes this is called neutrality. Can you say more about how you define neutrality, maybe how other people in the field think about it, how it's related to neutrality in other areas of biology, for instance, in community ecology, biogeography, or genetics?
Diego Vasquez 23:52
So yeah, neutrality has popped up in thinking in evolution and ecology many times. People, I guess, in reaction to people assuming that everything is sort of deterministic and caused by either selection at the genetic level or by niches at the ecological community level. So, for example, Hubell's neutral theory of biodiversity and biogeography assumes that the demography of species, the movement across space, and the speciation are all random. They are all stochastic. So, species individuals of any species have the same probability of giving birth, of dying, of colonizing new sites, or of speciating into another species. If you take it to the sort of interactions among species level. Then you can also think that interactions are neutral in the sense that it doesn't really matter who the individuals are, which species they belong to. They just can bump into each other and interact if they encounter each other. That's what we think about so when we talk about neutrality, that's what we mean in interactions. So it doesn't really matter your who you are, which traits you have. It's just everything is sort of dominated or driven by abundance, because if you have more individuals in your species, then you are more likely to interact with other species. Yeah, that's so that's to answer your question. You can define neutrality in terms of demography, of movement, or in the case of interaction networks, in terms of interactions, the probability of two species interacting.
Marty Martin 26:00
So if we use I mean a phrase, this is a different kind of question. If we use that as kind of a null model, and then go back to asking questions about specialization and its relationship to enduring disturbance, does it change the outcome? I mean, what fraction of this null is actually explaining what we were otherwise calling specialization?
Diego Vasquez 26:22
Yeah, so it's not just specialization, but doing this about 20 years of people looking at the structure of mutualistic networks. So the networks of species that interact through some kind of mutualism like pollination or seed dispersal, and also other kinds of networks of, for example, plant-herbivore or host-parasitoid interactions. The whole variety of interactions, people have realized that they tend to have some repeated features, so patterns that you see in these networks that are really non-random. Well, it turns out that a lot of this structure can be explained, or at least can be produced by simple models that are based in randomizing your data ,in assuming that abundant species will have a higher probability of interacting with other species. So that generates a lot of what we a lot of the patterns that we see in terms of specialization, generalization, but also in how these sort of emergent properties of the communities are generated.
Marty Martin 27:45
So, I mean, another sort of statistical question: When you're studying specialists, as you alluded to a minute ago, many of them are rare. So, to what degree do we worry about describing specialization and you know the the likelihood that that's influenced by undersampling>
Diego Vasquez 28:02
Yeah, we worry a lot about that. So, to what extent are these rare species that we sampled, we observed a few times, and we observed only interacting with a few other species, to what extent are they really specialists? One way in which we try to address that question in our studies in Villavicencio Nature Reserve is so we, for six years we went to our field sites and every week in the spring and summer we recorded the visits of pollinators to flowers. So we ended up with this dataset that includes some abundant species that tend to be generalized, and these rare species that seem to be specialists. For part of our project, we were studying using artificial nests for bees that nest in wood holes. So these are usually called trap nests, wooden trap nests, and there we can take them to the lab and identify the pollen, and then we can know which plants they are interacting with. Well, some of the specialized bee species that we had in our visitation network, when we looked at the pollen, were actually more generalized than we thought. So that answers the question that at least for some species we are overestimating specialization. So we think that they are specialists because, but that's actually because we have little information, and we are assuming that they interact with few species. But when you can amplify the information. By looking at the pollen, then you realize it's not always that case. There are some species are are really specializing on on few species, but in some cases at least this is caused by sampling issues. And I think that's one of the biggest issues in this field of research that the kind of data that we have are incomplete, and that's that poses a challenge to our inferences about the interactions in these communities.
Marty Martin 30:32
You mentioned before eDNA, environmental DNA. How much of a panacea is that going to be? I can see all sorts of reasons that in certain taxa it's not going to be helpful, but it's got to be somewhat more useful.
Diego Vasquez 30:45
So I think at least it kinda adds additional information. It won't replace. I don't think it will replace natural history or field field work, field observations, but it can complement your information. So we are doing with one of the students that is currently in my group. We are working on soil biodiversity in vineyards, and so he can identify bacteria and fungi from soil samples, and that's very useful. You can also infer some interactions between fungi and plant roots. If you have additional information, you know some you collected some of root samples and you found the some fungal species in the roots. You can so different pieces of information, including eDNA, and infer some of the interactions. So I don't think we'll just stop doing field work, stop doing natural history, and just collecting DNA and replacing everything else by that. But I think it can add very valuable information to our databases.
John Drake 32:03
So, I mean, new data streams like eDNA is one way to solve the problem. Are there statistical approaches like collectors' curves and rarefaction that you can use to adjust this sampling issue so that you can estimate? You know, you wouldn't know who the partners are, but you might know how many partners a particular species would have.
Diego Vasquez 32:22
Yeah, to a certain extent, you can. The problem with rarefaction is that you need to rarify to. So, rarefaction is a technique that you use to compare samples in, for example, in different sites. You want to know how many species you have in two different sites. So if let me explain the need for that first. You you have different you collected say you're sampling insects, and you want to know how many species you have because you suspect that some aspect of one of the sites influences the number of species, and so in one site you collect 10 individual insects, and in the and have say five species in those 10 individual insects, and in the other one you have collected with the same sampling effort you you set up your traps and you collected 100, and you have 20 species. But would you have could you have had 20 species if you had collected more specimens in the first site? You can't have more than 10 species, so you never have 20. But could you have 20 if you had collected more? So a statistical way of answering that question is not extrapolating from the first site, but making your second site rarer. So taking it to 10 individuals with a statistical technique, and then may make your samples comparable. That's rarefication.
Diego Vasquez 34:01
The problem with using rarefaction for interaction data is that some of the, for many of the species that are rare, you have very few interactions. 1, 2, 3, 4, 5, a few. And if you have to rarefy the whole network of interactions, you end up losing most of your information. So it's a challenge to use rarefaction for that. That's why we we use new models that are based on randomizing your data with some rules. For example, that you preserve the number of total interactions of each species, but then you randomize this data to generate patterns in your interactions that may tell you something about what could have generated this pattern, and excluding some other mechanisms that you suspect could be involved to see if you. End up with the same patterns or not. So it's a another way of getting to the same question and trying to remove some of the sampling issues, but without losing most of your information by making your samples too rare.
John Drake 35:23
Let's talk about prediction a little bit. Some of your recent papers have emphasized prediction. You posed a thought experiment. So, if you were dropped on a continent that you'd never visited and given a species list with abundances, phenologies, traits, phylogenetic relationships, the question is: Could you tell who interacts with whom? I mean, in a nutshell, I think you've spent 20 years trying to answer that question. I'm wondering what the honest answer is?
Diego Vasquez 35:51
Yeah, the honest answer is that we can't predict who is going to interact with whom. We can predict some broad patterns in the networks, so what's called connectants, a proportion of interactions of potential interactions that actually occur. Some sort of general patterns, but telling you which species is going to interact with which species based on all the information that you mentioned, is we are not there yet.
Diego Vasquez 36:26
And let me give you an analogy to see if we can, if this is it can help to understand the problem. Let's say that you want to predict the clothes that different people are going to use. So let's say pants. So I ask you, okay, what's your height, your waist size? So I go and buy a pant for you that you may be able to use. Just length and waist size is not the only aspect of pants that you may that may vary among different pants. So they have pockets or not. They may have zippers or buttons. The clothes maybe different degrees of softness or colors. So there's a whole complexity to even if all these traits that I mentioned, if we talk about brands, you may be able to tell differences of which brand they are based on some subtler characteristics. Even so, whether you will like and you will use a particular hand will depend characteristics of the hand, but on your own taste preferences, cultural background, your skin sensitivity, and whatever. And so, combining the two things is not really easy. And predicting which kind of pant you're going to use is not just that simple as it sounds.
Diego Vasquez 37:59
So it's the same thing for species, we can measure, I don't know, corolla depth. So how deep into the flower is the nectar that the pollinators need to reach to feed, and they have different head shapes and sizes and the length of their proboscis, the beak of the insect or the bird or whatever, but that's not the only aspect of their whole biology that will determine which species they are going to interact with. And we are talking about people and plants that we know pretty well, but then we are talking about species in communities that we barely know their names, or sometimes we don't even know their names. We know they look different from another species from which we may know the name, but what do we know about their biology? About their whatever. So it's a very complex problem that we need to solve with, in many cases, with very little information, and I think that's the root of the problem. That's why we fail. It's not that we fail completely, but we don't really do that that well at predicting at solving this problem that you were mentioning. If I drop you in an island or in a place in the world where you haven't been, and give you information about species, abundances, traits, phylogenies, we can't really predict the interactions from from that information because we we need more more than that, and we also need we need rules to combine the plants with the people, the flowers with the insects, so they they they match.
Diego Vasquez 39:44
We know which who matches with whom. What we do know is that abundance, and this has to do with neutrality, as we were discussing, is the single most influencing feature of species that can allow us to predict who is going to interact with whom. So that's better than knowing the traits or the phylogenetic relationships. But even with abundance, it's not that we are going to predict all interactions. We can predict some, or we have some degree of predictability. But it's not that great. So coming back to the question, 20 years after that, I don't think we're going to be able to predict communities. We need natural history. We need to understand species, and that takes a lot more than the data that we were discussing.
Marty Martin 40:44
Yeah, well, I want to I want to stick with that 2009 paper that I think, from Ecology, that we're we're loosely talking about because it's not only that abundance predicted connectance. There were some other really interesting patterns in that in that paper that phenology predicted other network traits, community, maybe not the individual interactions, but the community itself, and in particular nestedness. So I've thrown out a lot of jargon. First, I guess we should say that phenology is kind of the timing of life cycle events. Is it because that sort of dynamics in the abundance that it's related to these network traits at all, and then secondly, you know what is nestedness, and is that one an interesting one? Because connectance might just be a readout of abundance in a way, but nestedness is a little-that's a subtler thing, and it wouldn't seem to emerge as obviously for me, for me from abundance.
Diego Vasquez 41:36
Yeah, let's talk about phenology first. So phenology is you can think about it as the temporal distribution of abundance. So it's not just aggregate all your data for your community into one single network, but you look at different dates and when species, some species are present. So if you have that information, you are able to predict a bit more about who is going to interact with whom because you know when they were present simultaneously. It's sort of obvious, but it's a time-specific abundance if you want. So if you have that information, yeah, it helps, and you can do better at predicting the interactions.
Diego Vasquez 42:25
Regarding nestedness, nestedness is a property of matrices. Actually, of you you can have species and species, or species and sites, or wherever. But let's think about species. So in a nested network, the most specialized species are always interacting with a subset of species of the more generalized species. So, if you are a very specialized pollinator interacting with only one species of plant, then you are going to visit a plant that is also visited by your next more generalized pollinators. That leads to two things. One is asymmetric specialization, and that was what these people in I mentioned before Basconte et al. realized. And the other consequence of nestedness is that you have a a group, it's called a core network core. It's a group of abundant, generalized species that interact with each other. So it's a very dense part of a network. In any event, you can have nestedness resulting from simply neutral interactions that leads to nested patterns. So that's why we we found that yes, with abundance and and phenology, so the the time specific abundances, we could reproduce the nestedness of the that we were observing in our in our data.
Diego Vasquez 44:09
As happens with many other ecological patterns, though, you can get to the same pattern through very different mechanisms. So coming back to the neutral theory, Hubble's neutral theory. They could reproduce patterns usually observed in communities, such as the abundance distribution, which is skewed, so few abundant species, many rare species. You can get the same distribution with a completely neutral model or a completely deterministic model that assumes that niches, species differences, species traits determine everything. You get the same pattern. So the pattern doesn't really tell you what's behind it, and the same thing happens in these networks we were discussing.
John Drake 44:58
But I think you do show that, for instance, nestedness is really different than you would expect from a null model, right?
Diego Vasquez 45:06
Well, not not really. So, from these models that we were using, that were so these null models that we were using, that were using abundance and temporal distribution, they were generating levels of nestedness and connectants that were similar to those observed in the original data. And the same happens with other attributes of so other properties of these networks. You can generate these patterns in this data with very simple models that assume that species interact randomly at particular moments, and usually adding information about traits or phylogenies doesn't improve your predictive ability. I don't know if that answers your question.
John Drake 46:04
So, is the upshot then that these network properties of interaction networks, particularly in your case, that the plant-pollinator interaction networks, these things like nestedness, modularity, and so forth, they're basically statistical consequences of who's common and when?
Diego Vasquez 46:22
Yes. But that's what I was trying to say before is that that's not the only way of getting that. You could get it through other more complex mechanisms. So you could get to the same thing as happens with the log normal distribution. So the species distribution in communities you can get that through completely neutral models, or you can also get that with models based on niches. And here is the same: you can get the network properties with models that are completely neutral in terms of who interact with whom, or models that assume a few other things. So that's why looking at those patterns, only looking at those patterns doesn't really help us. You need to go beyond that to try to predict more specifically who interacts with whom, and that's when we don't are not very successful yet.
Marty Martin 47:19
Yeah, I know John wants to ask you about tapnet, but he mentioned a word a minute ago that I can't pass up because my group is interested in this for completely different reasons. He said modularity. Is that one of the network characteristics you've also ruled out? Because we weren't talking about it before. We only talked about connectance and nestedness. Modularity is also a trait that's not distinguishable from these sort of null processes.
Diego Vasquez 47:43
Yeah. So modularity is a property in which, it's modularity or compartmentalization. So the fact that these networks are organized into different subgroups of species that tend to interact within the with species within the group and very little with species in other groups. So that's modularity. Modularity has been found also in many types of networks throughout the world, and it has a consequence of sort of preventing the spread, it helps prevent the spread of perturbations in the network. So, if a species goes extinct and in a very strongly modular network, it will tend to affect only species within that module, but not it will be less likely to spread to other modules. So if you randomize your, so if you use a null model with abundance, so use your information about abundance to generate a network, you may end up with some modularity. Won't be so strong as you could expect, you could get some degree of modularity as well. So again, you can we can be if we look at these broad patterns, macroscopic patterns of community organization, you can be getting predicting the right patterns for the wrong reasons. So in other words, yes, you can get a nested network that has whatever specialists connected to journalists that has some degree of modularity or whatever. But your the mechanisms you are assuming may be very different, and you can get similar patterns. So that means that we, I guess, we are not there yet too. What we know is that abundance influences a lot, and I think we need a lot more better than. And more natural history to be able to go beyond that.
John Drake 50:04
Well, I guess this is why a couple of minutes ago you said, "Well, we have to move beyond those macroscopic patterns and actually try and understand who interacts with whom. And yeah, that's a hard problem, but you guys are working toward it. And one of your approaches is this thing you called tapnet. So tapnet, as I understand it, is a model that combines traits, this abundance information that we've talked about so much today, and phylogeny to predict the interaction frequencies. So it's pretty sophisticated tool, fit with likelihood theory, for instance. When you tested it with real data, and here I'm thinking about this hummingbird flower network in Ecuador, it actually didn't outperform a model that just used abundance alone, right? So I'm wondering, what does that tell us?
Diego Vasquez 50:52
Nothing different from what I from what I just said. So that was an honest effort with my group of German colleagues, while I was spending a year in Germany in my sabbatical. And we said, "Okay, let's try to come up with a statistical model that includes that allows you to combine abundance traits and phylogeny. Phylogeny is hard because you know the phylogenetic relationships among the group of species, the plants, say, and the animals, but how you put them together is tricky because you don't know how where they should match. But so we used a very sophisticated statistical tool that what it does is to generate different what were called latent traits. These are traits, invented traits of species, that allow you to model the traits and then see if those traits, based on the evolutionary relationships of species, allow you to improve your predictive ability. So it's a very sophisticated statistical tool, and yeah, they they so there's a newer paper that is now in second round of review in a journal that we use that method and other methods based on machine learning, different kinds of methods. So the best methods that we can find, and all the data that we could find that had abundance traits, phylogenies for plants and pollinators. And again, the conclusion is that abundance alone is is the best predictor of the best modest predictor of interaction. So we still are at a very relatively low predictive ability, but the best we can do, we do it with abundance. We don't really need to know the phylogenetic relationship of species or the traits to improve our prediction of interactions. So it's it's it's interesting it's interesting to come back to this conclusion after many years. So we we've done we we improved our data databases trying to get the best data we have on at the community level for plants and pollinators. We used a bunch of different models, but we come back to the same conclusion. There seems to be a really big component of neutrality in interactions, and also there seems to be a really big component of ignorance about how we understand interactions.
Marty Martin 53:52
Well, I want to I want to turn to the sort of practical side of where you're spending some of your time now, but I can't. I guess I want to get your take first in this space about abundance, whether you think that a lot more effort going into trying to explain disparities in abundance among species could lead us down this path to filling that gap? If abundance is powerful, and yet there's a dimension of abundance that might lend itself to explaining generalist and specialist tendencies, is there any is there any value in that? I mean, in a sense, that's kind of foundational for ecology, but probing more deliberately why there's such disparities in abundance, would that be useful?
Diego Vasquez 54:32
So, why species are have different abundances in communities?
Marty Martin 54:36
Yeah, well, I mean, so what is it that allows some species to be so abundant and therefore take such prominent roles? I mean, if it's if it's not if it's not questions beyond abundance, it seems like understanding doubling down on abundance as a phenomenon of interest. It seems to make sense.
Diego Vasquez 54:54
Yeah. Well, I guess that's one of the key questions in community ecology. Our colleague Mark Vellend, another former NCEAS postdoc, could tell you that there's basically four processes in communities that drive the patterns that we see, including the one you were referring to. So the distribution of abundances in communities. And the four processes are what he calls selection, which is differences in traits that allow species, different species, to have different niches. So it's basically trade-driven ecology. Then there is drift, which is what Hubble called the neutral demography. So random deaths and births, irrespective of who, which species an individual belongs to. Then there's movement, so migration, and then there's speciation. And we need to understand the relative contribution of these four these processes. And people are understanding more and more about how they contribute. But I think that's how we will understand community patterns. If we, on top of that, we add in different types of interactions, well, that adds a layer of complexity, but that's I think that's the way. So it's not an infinite number of processes that need that we need to understand, but these are complex processes, and understanding their relative contribution is really a challenge.
Marty Martin 56:37
Yeah, yeah, they're encompassing ideas.
John Drake 56:40
I mean, it's interesting to me. I'm trying to summarize in my own mind this conversation, but you know the way I teach population and community ecology, very often it's the species interactions explain the species abundances. But from your research, it sounds like we need to turn that on its head and say no, it's the species abundances that explain species interactions.
Diego Vasquez 57:02
Well, we published a paper that is called that refers to the chicken and egg dilemma, this chicken and egg dilemma. That's a paper that we published with two colleagues from Uruguay and Malaysia. And we were trying to. I don't know. I don't think we solved the chicken and dilemma, but our answer was that it was abundances had generated interactions more, so they influence more, the influence was greater that way than the other way around. But I don't think we really solved the dilemma, and I agree that it's probably both ways as usual, but yeah, it's something that I agree that it's an important question.
Marty Martin 57:47
Okay, well let's really turn away from the science into the sort of meta issues that allow us to do the science. So December 2023 Argentina's government made deep cuts to science funding. Salaries for researchers and grad students were frozen, and inflation's also high. So, what's that been like since 2025 for doing research in Argentina?
Diego Vasquez 58:11
Yeah, it's so. I would say I'm pathologically optimistic. So
Marty Martin 58:17
That's a good disease.
Diego Vasquez 58:19
Yeah, I tend to see the glass half full, so I would say it could be better, but it could also be worse. But it could definitely be better, as you said. Our salaries and scholarship stipends had been almost frozen. There's been some increases. increases, some raises, but it's not comparable to inflation. But also the funding, some of the big funding programs in Argentina, so the national programs like comparable to NSF. Let's call it this. There's something that it was discontinued. They called it the Agency of Promotion For Science and Technology. The programs of the agencies were, which were the largest funding programs, were discontinued. And then my institution, CONICET, which is a nationwide science and technology institution, has its own funding programs. Those were temporarily discontinued, and then they start they reopened the calls, but they redirected the funding to towards more applied. So they identified some big themes, very applied themes, and so you get a lot more funding for these topics than for, let's call, it basic science.
Diego Vasquez 59:53
So it's been very hard to keep research programs running. Fortunately, so universities have their own funding programs. They are usually much more modest. So my university, so I have positions both with CONICET and with the National University of Cuyo. This university has a funding program, but is historically been very little money. So little that many times I didn't apply for these grants because it was a lot of work for very little money. But now they sort of managed to increase the amount of grants. They are still very modest, but that's what has at least in my group helped have some funding for to go to the field and pay gas, that kind of thing. But it's very, very basic funding.
Diego Vasquez 1:00:45
At the national level, they created some other new funding programs. Like there's one big, well, big, it's not huge, but it's it's large. It's the largest funding program in terms of amounts of grants. You are required to have 15 researchers from at least three institutes in at least two regions of the country. So it sort of forces you to work in big teams, which is fine, but not everyone works in that way, and so it, in some way, eliminates diversity.
Diego Vasquez 1:01:26
Then funding for the institutions, our institute, for example, just to give you a very clear example, we haven't received yet the funding for this year. A year here is is we are now talking about fiscal year in the northern hemisphere. Here we are in the southern hemisphere. Everything starts in January. We are in the end of May. We still haven't received one peso for this year. So we are surviving in our institute with funding that we raised from services that we so we do reports on environmental assessments of different kinds of projects for the province, the province government, or for companies that need to assess their sustainability. Say if a winery wants to assess their sustainability in terms of biodiversity and environment, so we do a an assessment and we charge for that, and we've been surviving with that kind of funding. So it's a very critical situation.
Diego Vasquez 1:02:36
It could also, as I said, be worse because there's still some openings for positions. They still run the fellowship programs for PhDs and postdocs, which had been discontinued in the 90s. We had also a dark time for science and for universities. That was much worse, I would say. But this is not looking great, and there's all sorts of alarms that we need to pay attention to. And it's a very yeah tough moment to be doing science in Argentina and to be heading an institute as well, it's very challenging.
John Drake 1:03:25
I mean, you mentioned that what funding there still is, a bunch of it has been reallocated, like from basic research to applied research. What have been those areas of applied research that are being funded?
Diego Vasquez 1:03:37
So they defined some big titles. It's five, and I'm not sure if I'm going to remember the five. But it's agriculture, energy, human health, something that has to do with informatics, artificial intelligence, and there's one more I don't remember. Within each of these, you have some keywords. For example, you can have water or biodiversity or climate or whatever, and so we try to fit our research within this big title. So you may do some biodiversity research that has some application to either agriculture or energy or something, but so we are being forced to, in some way, to change to redirect our research towards to include some applied aspect. I mean, I don't think it's wrong to. I think it's really nice to be able to apply research. What I think is wrong, and I don't think the people who are thinking or or designing these programs understand, is that you can't apply all science immediately, and you can't have science that is addressing applied question, all the science to be like that. You need freedom. You need people to be driven by curiosity, so they can understand the world. So we can understand humans, the natural environment, and then from that you could apply some of that knowledge. I don't think it works if you force everyone to be trying to address very concrete, specific applied questions, and that's what's happening. In it's not universal, so we still have some funding for more basic research, but it's most of it goes in that direction.
John Drake 1:05:48
Well, and there's some obvious parallels to what's happening in the U.S. right now as well. I'm wondering, from your vantage point in Argentina, you know, if you have any sense for what these cuts are doing to the scientific enterprise in general.
Diego Vasquez 1:06:02
Yeah, so people because of the combination of salary, freezing of salaries and the lack of funding. People are either leaving science, academics, to look for a job in private sector, so quitting research or going abroad. I mean, it's not everyone doing that, but many people are doing that. So we are at that. The problem with that kind of process, we which we've had it at different moments in history because of military governments or economic crises, is that you. It's very easy. It's like going to war. It's very easy to start a war, but then ending a war, it's very hard. And it's the same thing with research programs. It's very easy to terminate it, but then recovering that program, recovering the people that know about that whole field of research, it takes a long time. It takes generations, and so I think that's a big problem. And I don't think people who are making decisions at this moment in our country realize, well they probably don't value that knowledge they don't think it's useful and that's why they are making decisions. Because I understand that we need to decrease the deficit of the state but sometimes we are talking about amounts that are not really significant in that in that regard, and so we are cutting just because of ideology. We need to cut, and so we end up making a lot of damage for saving part of a budget that is not really significant.
Marty Martin 1:08:01
Well, it's been great to talk to you today. One of the last questions that we ask our guests is: There anything that we haven't prompted you on that you wanted to cover? Whether that's sort of the state of science in the world or projects that you have planned. So, if that's something you want to talk about, we'd love to hear it. But in general, is there anything else you'd like to say?
Diego Vasquez 1:08:32
Yeah, yeah. There's something that we haven't covered, and I think it'd be useful to touch on it briefly. That has to do with my research on solitary bees and the temporal trends in their populations. And the reason why I want to talk about that is not because I want to talk about my own research, but because of the value of long-term research. So I mentioned these trap nests that we use to study these bee nests. We started using them in 2006 because we needed to study the nests to look at how plants affected the reproduction of the pollinator. And so we realized that we could use that kind of traps to these kind of artificial nests to study the nest of of these bees, and so we did some, we did that project. That led to a couple of PhD dissertation projects, and then after two or three years, I realized that if I kept doing the same over and over, I could have a long-term dataset at some point. So it was like something like my retirement project. I said, "Well, I keep doing this until I retire, and then I publish a paper. So we are getting to, next spring and summer, we are getting to collect the 20th year of data. That's the only data set that I know of time series of so the longest time series of pollinator data or bee data in South America with a standardized methodology. And that's very valuable that's allowing us to understand the population trends of these bees that there's basically nine bee species for which we have data, and eight of them seem to be declining.
Marty Martin 1:10:24
Wow
Diego Vasquez 1:10:24
That's very striking. It's this is a nature reserve where human activities are very limited. It probably has to do with, so these trends, probably have to do with climate. So we've been going through a drought. So with precipitation below the historical average for more than a decade. We suspect it has to do with that. We don't know. I can't answer that question really. But what I wanted to say about this experience is it's really valuable to collect data, long-term data. It takes effort. Sometimes, I mean, you have to find something that it doesn't take too much effort, so you can keep it going year after year without, you know, sacrificing your whole life for that, or having to sell your house to to fund the research in years in which you have you are at low funding moments like nowadays. But if you can do that, is it can be very valuable. So I think it's important to, in spite of what whatever the context, your national context, your institutional context. We need to keep research going because we can generate information. We can generate knowledge that can be very valuable for understanding the world, understanding the systems that we want to understand, in my case, these ecosystems and the interactions in these ecosystems and these species that live there. And so we need to be persistent. We need to keep getting these data and keep thinking about the processes that drive the data, and that I think that's the way to go.
Marty Martin 1:12:23
Well, Diego, thank you so much for joining us. It's been great to meet you and great to chat about your work and plans.
John Drake 1:12:28
Yeah, what a fantastic conversation.
Diego Vasquez 1:12:30
Yeah, thanks a lot for the invitation. It was an honor.
Marty Martin 1:12:49
Thanks for listening to this episode of Big Biology. If you like what you heard, tell us on Bluesky ,Threads, LinkedIn, Facebook, or Instagram, or leave a review wherever you get your podcast. We really do read them all.
John Drake 1:12:59
And if you didn't like something, we want to hear that too. All feedback is good feedback.
Marty Martin 1:13:03
Special thanks to Steve Lane, who manages the website, and Molly Magid for producing the show.
John Drake 1:13:07
Thanks also to Caroline Merriman and Cass Biles for social media support, Brianna Longo for our amazing cover art, and Clayton Glasgow, our blogger. Check out his work on our Substack.
Marty Martin 1:13:17
Big thank you as well to the College of Public Health at the University of South Florida, our Substack and Patreon subscribers, and the National Science Foundation for support.
John Drake 1:13:25
Music in this episode comes from Podington Bear and Tieren Costello.
Marty Martin 1:13:29
Thanks for listening, and we'll see you again on Big Biology.