Assembly history modulates vertical root distribution in a grassland experiment

Abstract

The order of arrival of plant species during assembly can affect the structure and functioning of grassland communities. These so‐called priority effects have been extensively studied aboveground, but we still do not know how they affect the vertical distribution of roots in the soil and the rooting depth of plant communities. To test this hypothesis, we manipulated the order of arrival of three plant functional groups (forbs, grasses and legumes) in a rhizobox experiment. Priority effects were created by sowing one functional group 10 days before the other two. Rhizoboxes in which all functional groups were sown simultaneously were used as controls. During the experiment, the total visible root length and the mean and maximum rooting depth of plant communities were monitored using image analysis and a new methodological approach using deep learning (RootPainter) for root segmentation. At harvest, we measured aboveground (community and species level) and belowground (community level) biomass, and assessed the vertical distribution of the root biomass in different soil layers. At the community level, all scenarios where one functional group was sown before the other two had similar shoot and root productivity. At the species level, two forbs (Achillea millefolium and Centaurea jacea) benefited from arriving early, and one legume (Trifolium pratense) had a disadvantage when it was sown after the grasses. Priority effect treatments also affected the vertical distribution of roots. When grasses were sown first, plant communities rooted more shallowly (lower mean and maximum rooting depth) than when forbs or legumes were sown first. In addition, roots moved down the soil profile more slowly in grasses‐first communities. Our results highlight that plant functional group order of arrival in grassland communities can affect the vertical distribution of roots in the soil and this may have implications for species coexistence.

Exploring priority and year effects on plant diversity, productivity and vertical root distribution: first insights from a grassland field experiment

“…Given that grasses tend to root more superficially than forbs (Bakker et al, 2019(Bakker et al, , 2021Chen et al, 2020), sowing grass species first may have increased root colonisation and interspecific competition in the topsoil, which could have made it more difficult for laterarriving forbs and legumes to grow in deeper soil layers, thus making root distribution of the entire community more shallow. The fact that data from this field experiment and another controlled experiment in rhizoboxes (Alonso-Crespo et al, 2023) support a strong effect of PFG order of arrival on root distribution without affecting root productivity strongly suggests that previous observations that legumes-first communities have a lower standing root length density in the topsoil (Weidlich et al, 2018) may actually be due to differences in vertical root distribution, and not to differences in root productivity. To better understand the mechanisms behind plant order of arrival effects on root distribution and their ecological consequences for plant communities, additional work is needed to measure root distribution at the species level, for instance using molecular techniques based on DNA sequencing (Wagemaker et al, 2020).…”

Special Issue: Root traits and functioning: from individual plants to ecosystems

“…Such belowground trait diversity may in turn explain vegetation dynamics. Using a novel deep‐learning approach to detect roots and analyze root growth images from a rhizobox, Alonso‐Crespo et al (2023) show that the vertical distribution of roots depended on which functional group was arrived first within a plant community. This may in turn determine competitive outcomes among plant species, and thus determine the overall coexistence and exclusion of species within plant communities.…”

Automated Seminal Root Angle Measurement with Corrective Annotation

“…Our proposed root angle measurement method offers speed, accuracy (depending on root visibility) and is freely available as open-source software with a user-accessible GUI. The segmentation component of our pipeline was implemented using RootPainter [34], which previous studies have demonstrated to be a flexible, accessible and efficient way to train segmentation models for a variety of root datasets [33, 3, 11, 12, 2, 21], including for images obtained from rhizobox experiments [1, 4].…”

“…Where y A is the vector of seminal root angles for n plants. Model ( 2) is mostly the same as (1), but an extra effect Y ϕ was added where Y is the design matrix capturing random effects for the combined effect of seed number, replicate, and block, ϕ ∼ N (0, Iσ 2 ϕ ) is the vector of associated random effects, and the variance σ 2 ϕ is treated the same as described above. The BVS approach applied to marker effects β in ( 1) and ( 2) assumes a mixture of two normal distributions with a small variance for unselected/unimportant markers and a large variance for selected/large effect markers.…”