Periodic Reporting for period 1 - NOBLE (Novelties, Biases, and Landscapes in Evolution)
Reporting period: 2021-08-16 to 2023-08-15
How predictable is evolution? Key components of evolution (mutation, recombination, and genetic drift) are all “random” processes [1,2]. Add chance in ecological events (like a particularly large asteroid impact) and it seems like evolution should be unpredictable.
Despite this, similar traits (phenotypes) have evolved repeatedly across distantly-related groups. Examples of phenotypic convergence include the repeated evolution of multicellularity; the “tuna-shaped” body plan in fish, sharks, and ichthyosaurs; and ant-mimicry in arhtropods [3-5]. Convergent evolution is often attributed to natural selection. Lineages evolving under the same selection pressures may be pushed towards the same adaptive optima, resulting in a predictable phenotype.
Evolutionary predictability is important for medicine (how will pathogens evolve?), conservation (how will species respond to environmental change?), and our understanding of evolutionary trends (how likely was the evolution of a particular trait?).
In project NOBLE, we investigated three questions about evolutionary predictability:
Q1: Can ancestral traits predict the appearance (or loss) of a novel trait in their descendants?
Q2: Is convergent evolution driven by natural selection, or does it occur by chance?
Q3: Does division of labor (splitting functions between cells, organs, or individuals) increase phenotypic convergence?
We investigated these questions in phylum Bryozoa – a group of aquatic, colonial, suspension feeders. In particular, we focused on a novel trait found in cheilostome bryozoans called “avicularia” (Fig. 1). Avicularia are morphologically specialized colony members (like ant soldiers), and are thought to have a defensive function. Avicularia were chosen as a focal trait because the repeated evolution (and loss) of avicularia [6] provides a natural experiment to investigate the predictability of their evolution.
Conclusions: We found that while the loss of a novel trait (avicularia) was related to ancestral traits (Q1), phenotypic convergence was less predictable and could have occurred by chance (Q2-Q3).
Figure 1: Example cheilostome bryozoan colony. A: colony surface, showing colony individuals and avicularia (pink). B-C: Avicularia with closed (B) and open (C) mandibles.
References
1. T. Lenormand, D. Roze, F. Rousset, Trends in Ecology & Evolution. 24, 157–165 (2009).
2. Z. D. Blount, R. E. Lenski, J. B. Losos, Science. 362, eaam5979 (2018).
3. K. J. Niklas, S. A. Newman, J Exp Bot. 71, 3247–3253 (2020).
4. R. Motani, K. Shimada, Sci Rep. 13, 16664 (2023).
5. J. D. Mclver, G. Stonedahl, Annual Review of Entomology. 38, 351–377 (1993).
6. A. H. Cheetham, J. Sanner, P. D. Taylor, A. N. Ostrovsky, Journal of Paleontology. 80, 49–71 (2006).
Despite this, similar traits (phenotypes) have evolved repeatedly across distantly-related groups. Examples of phenotypic convergence include the repeated evolution of multicellularity; the “tuna-shaped” body plan in fish, sharks, and ichthyosaurs; and ant-mimicry in arhtropods [3-5]. Convergent evolution is often attributed to natural selection. Lineages evolving under the same selection pressures may be pushed towards the same adaptive optima, resulting in a predictable phenotype.
Evolutionary predictability is important for medicine (how will pathogens evolve?), conservation (how will species respond to environmental change?), and our understanding of evolutionary trends (how likely was the evolution of a particular trait?).
In project NOBLE, we investigated three questions about evolutionary predictability:
Q1: Can ancestral traits predict the appearance (or loss) of a novel trait in their descendants?
Q2: Is convergent evolution driven by natural selection, or does it occur by chance?
Q3: Does division of labor (splitting functions between cells, organs, or individuals) increase phenotypic convergence?
We investigated these questions in phylum Bryozoa – a group of aquatic, colonial, suspension feeders. In particular, we focused on a novel trait found in cheilostome bryozoans called “avicularia” (Fig. 1). Avicularia are morphologically specialized colony members (like ant soldiers), and are thought to have a defensive function. Avicularia were chosen as a focal trait because the repeated evolution (and loss) of avicularia [6] provides a natural experiment to investigate the predictability of their evolution.
Conclusions: We found that while the loss of a novel trait (avicularia) was related to ancestral traits (Q1), phenotypic convergence was less predictable and could have occurred by chance (Q2-Q3).
Figure 1: Example cheilostome bryozoan colony. A: colony surface, showing colony individuals and avicularia (pink). B-C: Avicularia with closed (B) and open (C) mandibles.
References
1. T. Lenormand, D. Roze, F. Rousset, Trends in Ecology & Evolution. 24, 157–165 (2009).
2. Z. D. Blount, R. E. Lenski, J. B. Losos, Science. 362, eaam5979 (2018).
3. K. J. Niklas, S. A. Newman, J Exp Bot. 71, 3247–3253 (2020).
4. R. Motani, K. Shimada, Sci Rep. 13, 16664 (2023).
5. J. D. Mclver, G. Stonedahl, Annual Review of Entomology. 38, 351–377 (1993).
6. A. H. Cheetham, J. Sanner, P. D. Taylor, A. N. Ostrovsky, Journal of Paleontology. 80, 49–71 (2006).
Investigating Q1: Organisms have limited resources for growth and reproduction. Possessing redundant traits for a non-vital function may be energetically expensive. We hypothesized that lineages with alternative defensive traits (like spines) would be less likely to evolve avicularia (and more likely to loss them).
We then created a trait database for 3200+ species of cheilostomes. Combining this trait database with a recent family tree (phylogeny) [7] allowed us to run several models of correlated trait evolution. These models show whether the presence of one trait (alternative defenses) impacts the evolutionary rate (gain or loss) of another trait (avicularia).
In contrast to our hypothesis, we found that the loss of avicularia was typically higher in the absence of alternative defenses. This suggests that multiple defenses may be worth the energetic cost and that evolution of avicularia can be predicted based on ancestral traits.
Investigating Q2: Phenotypic convergence is often assumed to be evidence of natural selection, but high levels of convergence can occur by chance [8]. Convergent mandible morphologies repeatedly occur across the cheilostome phylogeny (Figure 2), and it is unclear if this is driven by chance or adaptation.
We collected 3247 images of avicularia across 1392 species, from both museum specimens and published images. We then generated a quantitative measure of shape using outlines and landmarks (Figure 3). Next, we measured the degree of convergent evolution across species in our sample. Since closely-related species are expected to have similar phenotypes and may appear convergent, we corrected for phylogenetic relatedness [9]. To determine if the degree of empirical convergence is expected by chance, we repeatedly simulated evolution without natural selection and re-measured convergence.
We found that empirical convergence could be quite high. However, these high values of convergence were always possible in the simulations. Aaptation may not have caused convergence and avicularium morphology is relatively unpredictable. Natural selection cannot be completely ruled out, however, since some some common empirical morphologies are rare in the simulations.
Investigating Q3: Traits within organisms are usually subject to multiple selection pressures. This can be problematic if these selection pressures are in opposite directions. For example, turtle shells can be optimized for swimming speed or resistance to rushing, but not both [10]. If swim speed and defense are both important, then a turtle will have a shell that isn’t optimized for either function. Similarly, avicularia may be subject to functional tradeoffs and possess suboptimal forms.
However, division of labor can side-step these functional tradeoffs by reducing the number of functions a single unit (cell, organ, individual) must perform. While most cheilostomes only have one kind of avicularium, some can have more than four kinds in a colony. We hypothesized that the more kinds of avicularia in a colony, the fewer functions each kind of avicularium should perform, and the weaker the functional tradeoffs would be. If true, then species with more kinds of avicularia should converge on the optimal forms and lack subotpimal ones.
We examined our database of avicularium mandible shape to see if species with more avicularia had stronger clustering and no suboptimal forms. Surprisingly, we found that suboptimal forms were always present. This suggests that division of labor between different avicularia may actually be low.
Dissemination: Results of this research are currently being prepared for publication. Once published, the datasets of bryozoan traits and avicularium morphology will be freely available.
Figure 2: Cheilostome phylogeny (modified from [7]), showing independent evolutions of avicularia (circles) and convergent morphologies (silhouettes).
Figure 3: Measuring avicularium shape. A: mandible (scalebar: 30 um). B: Landmarks (black circles) and outline points (colored circles) used in the analysis. C: the resulting shape-space, showing how frequently each shape is found in the sample.
References
7. R. J. S. Orr et al., Science Advances. 8, eabm7452 (2022).
8. C. T. Stayton, Journal of Theoretical Biology. 252, 1–14 (2008).
9. K. Arbuckle, C. M. Bennett, M. P. Speed, Methods in Ecology and Evolution. 5, 685–693 (2014).
10. P. D. Polly, Integrative and Comparative Biology. 60, 1268–1282 (2020).
We then created a trait database for 3200+ species of cheilostomes. Combining this trait database with a recent family tree (phylogeny) [7] allowed us to run several models of correlated trait evolution. These models show whether the presence of one trait (alternative defenses) impacts the evolutionary rate (gain or loss) of another trait (avicularia).
In contrast to our hypothesis, we found that the loss of avicularia was typically higher in the absence of alternative defenses. This suggests that multiple defenses may be worth the energetic cost and that evolution of avicularia can be predicted based on ancestral traits.
Investigating Q2: Phenotypic convergence is often assumed to be evidence of natural selection, but high levels of convergence can occur by chance [8]. Convergent mandible morphologies repeatedly occur across the cheilostome phylogeny (Figure 2), and it is unclear if this is driven by chance or adaptation.
We collected 3247 images of avicularia across 1392 species, from both museum specimens and published images. We then generated a quantitative measure of shape using outlines and landmarks (Figure 3). Next, we measured the degree of convergent evolution across species in our sample. Since closely-related species are expected to have similar phenotypes and may appear convergent, we corrected for phylogenetic relatedness [9]. To determine if the degree of empirical convergence is expected by chance, we repeatedly simulated evolution without natural selection and re-measured convergence.
We found that empirical convergence could be quite high. However, these high values of convergence were always possible in the simulations. Aaptation may not have caused convergence and avicularium morphology is relatively unpredictable. Natural selection cannot be completely ruled out, however, since some some common empirical morphologies are rare in the simulations.
Investigating Q3: Traits within organisms are usually subject to multiple selection pressures. This can be problematic if these selection pressures are in opposite directions. For example, turtle shells can be optimized for swimming speed or resistance to rushing, but not both [10]. If swim speed and defense are both important, then a turtle will have a shell that isn’t optimized for either function. Similarly, avicularia may be subject to functional tradeoffs and possess suboptimal forms.
However, division of labor can side-step these functional tradeoffs by reducing the number of functions a single unit (cell, organ, individual) must perform. While most cheilostomes only have one kind of avicularium, some can have more than four kinds in a colony. We hypothesized that the more kinds of avicularia in a colony, the fewer functions each kind of avicularium should perform, and the weaker the functional tradeoffs would be. If true, then species with more kinds of avicularia should converge on the optimal forms and lack subotpimal ones.
We examined our database of avicularium mandible shape to see if species with more avicularia had stronger clustering and no suboptimal forms. Surprisingly, we found that suboptimal forms were always present. This suggests that division of labor between different avicularia may actually be low.
Dissemination: Results of this research are currently being prepared for publication. Once published, the datasets of bryozoan traits and avicularium morphology will be freely available.
Figure 2: Cheilostome phylogeny (modified from [7]), showing independent evolutions of avicularia (circles) and convergent morphologies (silhouettes).
Figure 3: Measuring avicularium shape. A: mandible (scalebar: 30 um). B: Landmarks (black circles) and outline points (colored circles) used in the analysis. C: the resulting shape-space, showing how frequently each shape is found in the sample.
References
7. R. J. S. Orr et al., Science Advances. 8, eabm7452 (2022).
8. C. T. Stayton, Journal of Theoretical Biology. 252, 1–14 (2008).
9. K. Arbuckle, C. M. Bennett, M. P. Speed, Methods in Ecology and Evolution. 5, 685–693 (2014).
10. P. D. Polly, Integrative and Comparative Biology. 60, 1268–1282 (2020).
Our databases represent the most complete survey of avicularium morphology undertaken to date. This will not only be useful to bryozoan scientists, but to other ecologists and evolutionary biologists.