A machine learning enabled approach to assess trade-offs between growth and stress tolerance in Pooideae grasses following domestication

Plant domestication may create trade-offs between growth and stress tolerance, raising concerns about yield stability in future climates. Previous studies have found limited direct evidence for such trade-offs, often focusing on weakened defenses associated with higher growth rates. Trade-offs can also occur when traits optimized for favorable conditions perform less efficiently under stress. Deciphering these mechanisms is crucial for maintaining growth in changing environments. We examine one key aspect of vegetative growth, leaf elongation, in six species of grasses. We use a machine learning-enabled pipeline to extract cell dimensions and positions from leaf microscope images to study cell kinematics. We find that domesticated plants generally have longer leaves, larger division zones, and higher cell production rates. While no clear trade-off is observed between domestication and drought response in final leaf length, a trade-off occurs in development; wild species exhibit a smaller decrease in the elongation zone size under drought compared with domesticated species. This pattern points to compensatory mechanisms, such as extended elongation duration or increased cell production, mitigating drought effects in domesticated plants. These nuanced trade-offs associated with domestication highlight the importance of robustly phenotyping developmental and physiological traits, possibly informing breeding strategies to enhance crop resilience in future climates.

Methods
The species studied are listed in Table 1. Both domesticated wheat varieties (Triticum turgidum L. var. Durum 'Langdon' and 'Svevo', Triticum aestivum 'Chinese Spring') and wild emmer wheat (Triticum turgidum L. subsp. dicoccoides 'Zavitan' and 'Vavilovii') were subjected to a pre-germination treatment at 32°C dry conditions for 7 days, followed by 4°C dry conditions for 2 days. Domesticated oat (Avena sativa cv. 'Sang'), wild oat (Avena insularis 'BYU 207'), domesticated barley (Hordeum vulgare 'Morex'), wild barley (Hordeum vulgare L. subsp. spontaneum), and Brachypodium distachyon (Bd21) seeds were placed at 4°C in water for 7 days to synchronize germination. After cold treatment, all seeds were soaked in Gibberellic Acid for 1 day to promote uniform germination. Seeds were then planted in Kord square pots (the HC Companies, OH, USA) filled with 70g of Profile porous ceramic rooting media (Profile Products, Buffalo Grove, IL).

Common name
Scientific name
Accession identifier
Name symbol
Domestication status
ploidy

wheat (wild emmer)
Triticum turgidum ssp. dicoccoides
Zavitan
Wheat-Z-Wild
Wild
4x

wheat (bread wheat)
Triticum aestivum
Chinese Spring
Wheat-CS-Domesticated
Domesticated
6x

wheat (durum wheat)
Triticum turgidum L. var. durum
Langdon
Wheat-L- Domesticated
Domesticated
4x

wheat (durum wheat)
Triticum turgidum L. var. durum
Svevo*
Wheat-S- Domesticated
Domesticated
4x

wheat (wild emmer)
Triticum turgidum L. subsp. dicoccoides
Vavilovii*(GRIN:PI 352322)
Wheat-V-Wild
Wild
4x

oat
Avena sativa L.
Sang (NPGS:PI386171)
Oat- Domesticated
Domesticated
6x

oat
Avena insularis
BYU 207
Oat-Wild
Wild
4x

barley
Hordeum vulgare L. subsp. spontaneous
(GRIN:PI 662181)
Barley-Wild
Wild
2x

barley
Hordeum vulgare
Morex
Barley- Domesticated
Domesticated
2x

purple false brome
Brachypodium distachyon
Bd21
Bd21
Wild
2x

*The two species are only used for leaf elongation in this study, while the rest are used for both leaf elongation and cell elongation.

Before planting, the dry weight (DW) of each pot and its field capacity (FC) for water were recorded. FC was determined by saturating the pot with water and allowing it to drain overnight under gravity, with the weight difference between the saturated and dry pot used as the basis for subsequent soil dry-down control. The plants were grown in growth chambers (Biochambers FXC-9) with controlled conditions: 60% humidity, 500 µmol photons m-2 s-1 light intensity, and a temperature regime of 25°C during the day and 20°C at night (16h/8h day/night cycle). Initially, the plants were bottom watered every other day with tap water (pH 5-6), supplemented with DYNA-GRO GROW Liquid Plant Food 7-9-5 (Dyna-Gro, Richmond, CA, USA) until the appearance of the second leaf. From the appearance of the second leaf stage, control plants continued to be watered nightly without fertilizer, while drought treatment plants were subjected to controlled soil dry-down. Soil water content in the drought treatment group was reduced to 45% FC, and then water content maintained between 45-55% by watering at 10 a.m., 4 p.m., and 10 p.m. daily. 

The third leaf was regularly checked at each watering. For both control and drought-treated groups, four to six replicates were used to measure the entire process of the third leaf’s elongation, and an additional four to six replicates were harvested on the fifth day after leaf appearance (approximately at the steady state of elongation) for microscopy. 

The third leaf of plants used for leaf elongation were photographed every four hours using a Raspberry Pi camera. Leaf length was measured from the leaf base to tip using ImageJ, with a length scale correction applied based on a predefined standard to adjust for the perspective of each image.

We used a second set of plants for leaf cell measurements. We carefully unwrapped an emergent third leaf from older leaves, and dissected as close as possible to its connection to the seed. This is done on the fifth day after the appearance of the third leaf, which corresponds to a steady state of leaf elongation as observed in the whole-leaf measurements, above. At this stage, the ligule is within 1mm of the leaf base. Dental impression material (Defend Impression Material, IL, USA ) was applied to the base region (for leaves > 50mm) or whole leaf (<50mm) of the abaxial surface to create an imprint of the leaf. Thereafter, a thin layer of nail polish was applied to the imprint, then dried, and placed onto a microscope slide. Microscopy images of imprints are taken with Zeiss Axiolab 5 with Axiocam 208 CCD camera (Carl Zeiss AG, Germany) with 10x magnification. A sequence of images were taken from the base of the leaf towards the tip with >20% overlap between images.

This dataset contains leaf elongation dataset, and original microscopy images of the leaf surface imprints.