Transformer Adaptation for Plant Bioelectric Potential Classification: A Scoping Review and Analogical Transfer Framework

Plant bioelectric potential (BEP) signals offer a non-destructive route for monitoring plant responses, stress, and physiological state in smart agriculture. However, evidence on BEP-aware Transformer classification remains fragmented and sparse, and many candidate mechanisms are reported in adjacent physiological or time-series domains rather than validated directly on plant electrical recordings. This scoping review synthesized 182 studies following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR), spanning BEP and plant electrophysiology, Transformer-based time-series classification, and transferable mechanisms from non-BEP physiological and general time-series research. The final corpus comprised 18 direct BEP/domain studies, 13 Transformer/time-series bottleneck studies, and 151 analogical patch-source studies. The review searched six electronic sources (IEEE Xplore, MDPI, ScienceDirect, SpringerLink, Taylor & Francis Online, and Scopus) using common eligibility criteria and a common publication cutoff. The review identifies recurring Transformer bottlenecks related to tokenization, locality, trend-noise separation, frequency representation, channel dependency, limited domain generalization, class imbalance, and computational cost. It also formulates BEP-specific signal frictions caused by weak amplitude, noise, drift, event irregularity, electrode variability, and cross-plant or cross-setting differences. The novelty of this review lies in integrating these model-side bottlenecks, BEP-specific frictions, and transferable adaptation mechanisms into a bottleneck-friction-patch mapping framework for future BEP-aware smart sensing systems. Within the coded corpus, the most frequent mapped relationship was between generalization limitations and cross-domain variability, pointing to validation shift as a key design concern. Because most adaptation mechanisms are drawn from non-BEP physiological and general time-series studies, the proposed framework should be read as a hypothesis-generating design map rather than a set of validated BEP solutions. The findings indicate that future BEP classification studies should prioritize standardized datasets, cross-plant and cross-setting validation, transparent data/code availability, and deployable model designs for agricultural monitoring.