Search for: All records

Abstract The dataset contains leaf venation architecture and functional traits for a phylogenetically diverse set of 122 plant species (including ferns, basal angiosperms, monocots, basal eudicots, asterids, and rosids) collected from the living collections of the University of California Botanical Garden at Berkeley (37.87° N, 122.23° W; CA, USA) from February to September 2021. The sampled species originated from all continents, except Antarctica, and are distributed in different growth forms (aquatic, herb, climbing, tree, shrub). The functional dataset comprises 31 traits (mechanical, hydraulic, anatomical, physiological, economical, and chemical) and describes six main leaf functional axes (hydraulic conductance, resistance and resilience to damages caused by drought and herbivory, mechanical support, and construction cost). It also describes how architecture features vary across venation networks. Our trait dataset is suitable for (1) functional and architectural characterization of plant species; (2) identification of venation architecture‐function trade‐offs; (3) investigation of evolutionary trends in leaf venation networks; and (4) mechanistic modeling of leaf function. Data are made available under the Open Data Commons Attribution License. 

The difficulty in specifying rewards for many real world problems has led to an increased focus on learning rewards from human feedback, such as demonstrations. However, there are often many different reward functions that explain the human feedback, leaving agents with uncertainty over what the true reward function is. While most policy optimization approaches handle this uncertainty by optimizing for expected performance, many applications demand risk-averse behavior. We derive a novel policy gradient-style robust optimization approach, PG-BROIL, that optimizes a soft-robust objective that balances expected performance and risk. To the best of our knowledge, PG-BROIL is the first policy optimization algorithm robust to a distribution of reward hypotheses which can scale to continuous MDPs. Results suggest that PG-BROIL can produce a family of behaviors ranging from risk-neutral to risk-averse and outperforms state-of-the-art imitation learning algorithms when learning from ambiguous demonstrations by hedging against uncertainty, rather than seeking to uniquely identify the demonstrator’s reward function. 

Summary Variation in leaf venation network architecture may reflect trade‐offs among multiple functions including efficiency, resilience, support, cost, and resistance to drought and herbivory. However, our knowledge about architecture‐function trade‐offs is mostly based on studies examining a small number of functional axes, so we still lack a more integrative picture of multidimensional trade‐offs.Here, we measured architecture and functional traits on 122 ferns and angiosperms species to describe how trade‐offs vary across phylogenetic groups and vein spatial scales (small, medium, and large vein width) and determine whether architecture traits at each scale have independent or integrated effects on each function.We found that generalized architecture‐function trade‐offs are weak. Architecture strongly predicts leaf support and damage resistance axes but weakly predicts efficiency and resilience axes. Architecture traits at different spatial scales contribute to different functional axes, allowing plants to independently modulate different functions by varying network properties at each scale.This independence of vein architecture traits within and across spatial scales may enable evolution of multiple alternative leaf network designs with similar functioning.