sPOQ: Synthesizing Predictions on Optimality Quantitatively

Chris Muir
Last updated on Nov 20, 2019 8 min read

Interested in joining sPOQ?

I am putting together an sDiv working group proposal on optimality and leaf traits, which is due December 11, 2019. I’m looking for people from diverse backgrounds who want to participate in working group meetings (high commitment) or consult/contribute on theory/data (low commitment). If you are interested, here’s how to get involved:

  1. Read the draft proposal below.
  2. Fill out form to express interest by Friday, December 6 (scroll down). Expressing interest in no way commits you to the project.
  3. Forward to students and colleagues that might be interested.
  4. Email me if you questions or comments.

Early-career researchers, women, and underrepresented minorities are especially encouraged to fill out the Google form. The form is located at the bottom of this page and here. It takes only a minute to fill out. I will update everyone who fills out the form by email when the pre-proposal is submitted (Dec 11, 2019) and if it is invited for a full proposal (Jan 17, 2020).

Aims, scope, and scientific objectives

The sPOQ working group will quantitatively test optimality models by synthesizing theory and data on 23 putatively adaptive leaf traits and testing predictions using phylogenetic comparative methods.

Striking cases of nearly-perfect matches between organisms and their environment provide some of the best evidence for natural selection: the crypsis of leaf-life katydids hides them from predators; the mimicy of Ophrys orchids to deceives pollinators into “mating” with the flowers; corolla size and shape favors preferred pollinators with matching proboscis or beak morphology. Nearly perfect fit between phenotype and environment supports the idea that natural selection has led populations near their adaptive optimum. Are the above examples special cases or are most traits similarly optimized? What’s special about the cases above is that it is relatively easy for people to identify the target of selection (e.g. resemblence of a katydid wing to a leaf). One possibility is that other traits are also nearly optimal, but the target of selection is more difficult for people to discern. If we could identify the target of selection, we would find that other phenotypes are similarly optimal. An alternative possibility is that these cases of near-perfect adaptation are the exception rather than the rule. Most of the time, populations are not near their adaptive optimum because nonadaptive evolutionary forces prevent it. To objectively distinguish between these possibilities, we would ideally select a random sample of phenotypes, identify their optima, and measure how close natural phenotypes are to their optima.

In practice, it’s not feasible or desirable to sample traits at random, but a targeted synthesis of leaf traits may come close to objectively quantifying whether populations are near their adaptive optima. Givnish (1987) identified 23 putatively adaptive leaf traits (Table 1). For this working group, we propose to revisit these leaf traits with updated theory, data, and comparative methods. Focusing on the Givnish trait set has four advantages:

  1. it’s less biased towards confirming optimality than if we searched for published examples of optimality;
  2. since traits are putatively adaptive, it’s unlikely we will select traits with no selective value;
  3. a targeted set of a priori traits is more feasible than an open-ended search; and
  4. using traits subject to 30+ years of study ensures a large body of theoretical and empirical work to draw from.

Leaf traits in particular are well-suited to synthesis because most land plant leaves are developmentally homologous and perform comparable functions. Their fundamental ecological importance and ubiquity also means there is lots of data on leaf traits. For these leaf traits, we will synthesize theory and data to assess whether leaf trait variation is consistent with optimality models.

In particular, we will focus on quantitative predictions, not qualitative predictions. Why? For one, the latter has been done (a lot). Qualitative agreement demonstrates adaptation, but cannot show that populations are actually at their predicted optimum. Secondly, only quantitative comparisons can quantify how close or far off we are from truly understanding trait variation. If we see excellent agreement, it suggests that we have identified the major factor(s) explaining trait variation. If theoretical predictions are inaccurate or imprecise, it suggests either suboptimality or that our optimality model is bad – either result would be interesting to follow up on.

Table 1 23 putatively adaptive leaf traits identified by Givnish (1987): leaf size, leaf thickness, leaf absorptance, leaf inclination, amphistomy, stomatal conductance, mesophyll photosynthetic capacity, chlorophyll:protein ratio, chlorophyll a:b ratio, C$_4$/CAM, evergreen leaves, leaf margins, lobed leaves, drip tips, cordate leaves, compound leaves, phyllotaxis, reddish undersides, lens-shaped epidermal cells, blue iridescence, stomatal clustering, asymmetric leaf bases, anisophylly.

Methods

The sPOQ working group will meet three times to:

  1. Synthesize optimality theories, data, and identify knowledge gaps
  2. Pre-register our hypotheses, predictions, and methods on the Open Science Framework
  3. Test optimality predictions quantitatively using phylogenetic comparative methods

Meeting 1: Synthesize optimality theories, data, and identify knowledge gaps

In the first working group meeting we will assemble quantitative theoretical predictions, trait data, and phylogenies. At this time, we will assess which traits are theory deficient or data deficient, and which could be tested (Figure 1). We will also identify cases in which additional theory or data could be collected before the next working group meeting. In between meetings 1 and 2, we will clean existing data and work with our network of theory collaborators and data contributors (see below) to supplement our body of theoretical predictions and data.

decision-tree **Figure 1** sPOQ decision tree.

Meeting 2: Pre-register our hypotheses, predictions, and methods on the Open Science Framework

With theory and data in hand, the goal of the second meeting will be to pre-register a priori hypotheses, quantitative predictions, and statistical methods on the Open Science Framework. Once pre-registered, we will analyze an exemplar trait to serve as a template for remaining traits. With that template, participants will work on applying it the other traits between meeting 2 and 3.

Meeting 3: Test optimality predictions quantitatively using phylogenetic comparative methods

In the third meeting, we will work on finalizing a data paper containing trait measurements synthesized for this project. All data contributors will be included as coauthors on this paper. We will also finalize data analysis and draft a paper on our results with the aim of submitting shortly after the last working group meeting. All theory consultants and data contributors will be included on this paper.

Roles in sPOQ

Participants

Participants attend working group meetings and lead the development, analysis, and writing. The working group will consist of theory- and data- track partcipants with seperate leaders and responsibilities. The theory track will synthesize and (if necessary) extend existing theoretical predictions. One possibility is extending my leafoptimizer software to handle a wide suite of leaf optimization problems. The data track will gather trait and phylogenetic data from the literature, databases like TRY and BIEN, or through our network of data contributors.

Theory consultants

Participants will not have extensive expertise in all the theoretical areas we aim to cover. Therefore we are assembling a network of theory consultants to broaden our expertise. Theory consultants will be available during and between meetings to provide targeted expertise on specific topics, but not are not expected to carry out analyses on their own.

Data contributors

Trait databases and published papers will not have all the data we need. Therefore, we are assembling a network of data contributors who will supplement existing data with unpublished data, including cases where data from a publication is not publically available and cases where the data have never been published. This includes both trait and phylogenetic data. Data contributors are expected to provide data and metadata in a timely manner, but are not expected to contribute to data analysis.

Synthesis

sPOQ will will synthesize both quantitative optimality predictions and the data needed to test them for a diverse set of putatively adaptive leaf traits. Through this synthesis, we will first determine whether most traits even have quantitative optimality predictions (theory sufficient) and the data needed to test them (data sufficient). Among traits that are theory and data sufficient, we will be able to test whether optimality models are consistent with trait variation in nature. To my knowledge, this will be the first attempt to use a large data synthesis to test generality of optimality predictions quantitatively.

Beyond sPOQ

Comparative data are necessary, but not sufficient to infer optimality. The end goal of sPOQ is not a definitive answer, but to direct future research most productively. First, we will identify theory- and data-deficient traits where more work is needed. For theory- and data-sufficient traits, agreement between optimality models and data would strongly suggest a trait is near its adaptive optimum in most populations, but would needed to be tested using population approaches. Disagreement between model and data are even more interesting. These indicate areas to investigate potential nonadaptive evolutionary forces. They might also indicate areas where our optimality theory is missing something critical that needs to be figured out. I believe these results will inspire future theoretical, experimental, and synthesis approaches.

Form

References

Givnish TJ. 1987. Comparative studies of leaf form: assessing the relative roles of selective pressures and phylogenetic constraints. New Phytologist 106: 131–160.

sPOQ leaf phylogenetic comparative methods