Model-Based Design of Long-Distance Tracer Transport Experiments in Plants.

Abstract

Studies of long-distance transport of tracer isotopes in plants offer a high potential for functional phenotyping, but so far measurement time is a bottleneck because continuous time series of at least 1 h are required to obtain reliable estimates of transport properties. Hence, usual throughput values are between 0.5 and 1 samples h -1 . Here, we propose to increase sample throughput by introducing temporal gaps in the data acquisition of each plant sample and measuring multiple plants one after each other in a rotating scheme. In contrast to common time series analysis methods, mechanistic tracer transport models allow the analysis of interrupted time series. The uncertainties of the model parameter estimates are used as a measure of how much information was lost compared to complete time series. A case study was set up to systematically investigate different experimental schedules for different throughput scenarios ranging from 1 to 12 samples h -1 . Selected designs with only a small amount of data points were found to be sufficient for an adequate parameter estimation, implying that the presented approach enables a substantial increase of sample throughput. The presented general framework for automated generation and evaluation of experimental schedules allows the determination of a maximal sample throughput and the respective optimal measurement schedule depending on the required statistical reliability of data acquired by future experiments.

Plant phenotyping relevance

植物の長距離トレーサー輸送を機能的フェノタイピングとして測定するため、測定スケジュールとモデル解析を設計・評価し、スループット向上と推定精度を検討した方法開発研究。

Code and data availability

The paper's analysis code is explicitly stated to be publicly available on GitHub (mdate repository), and the Supplementary Material includes Supplementary Table 2 with all design definitions and fit results, which directly reproduce the paper's computational analysis of tracer transport experimental designs.