Biology subjects

Tsutsumida, N.

Publications and source records attributed to Tsutsumida, N..

3 recordsLinked to original sources

bioRxiv2026

A tabular foundation model rivals established species distribution models from presence-only records

Predicting where something occurs from records of where it has been observed, without reliable records of absence, is a common problem in science. Species distribution models (SDMs) such as MaxEnt address it in fields from biological conservation and climate-change assessment to the mapping of disease vectors and archaeological sites, but must be tuned for each dataset. Tabular foundation models, pre-trained on synthetic prediction tasks, predict new datasets without per-dataset training. Only one study has tested them as SDMs, and it fine-tuned on survey data from the evaluated regions. We evaluated tabular foundation models on the standard SDM benchmark (225 species in six regions, with independent presence-absence surveys), under the protocol used to compare 21 methods, and on 120 virtual species with known distributions. Without tuning, a tabular foundation model, TabPFN 3.5, equalled the best published approach, a five-model ensemble (mean AUC 0.732). Fine-tuned on presence-background tasks from other regions, it became the best single model on the benchmark (0.736), outperformed MaxEnt (0.728) and boosted regression trees after correction for multiple comparisons, and was better calibrated than MaxEnt, although six regions were too few to establish the advantage for new regions. Fine-tuning on presence-absence surveys of other species helped only when they came from the evaluated regions. On virtual species, it recovered simple additive relationships less well than regression-based methods but complex ones at least as well when records were unbiased, and was worse with environmentally biased records. Tabular foundation models are competitive SDMs without per-dataset tuning, but they are sensitive to environmentally biased records.

ecology↗

bioRxiv2023

Spatiotemporal Mapping of Cherry Blossom Blooming by Semi-Automatic Observation System with Street-Level Photos

The importance of floral phenology as a critical indicator of regional climate change and ecosystem services is widely recognized. The annual blooming of cherry blossoms is a nationally celebrated event in Japan, and historical phenological records have been used to document regional climate change. The cultural ecosystem services provided by this phenomenon are important as they not only signal the arrival of spring but also offer a picturesque spring landscape. Despite its importance, constructing a spatiotemporal record of cherry blossom blooming is challenging due to the limited coverage of traditional stationary observations. To address this issue, citizen-based observation programs and remote sensing applications have been implemented; nevertheless, these strategies are still limited by infrequent and insufficient observations throughout space and time. To compensate, we developed a flower detection model for geographically and temporally dispersed street-level photos that may be used as the core component of a semi-automatic observation system. Specifically, we developed a customized YOLOv4 model for cherry blossom detection from street-level photos obtained through Mapillary, one of the social sensing data repositories. The detection model achieved an overall accuracy, recall, and precision of 86.7%, 70.3%, and 90.1%, respectively. By using observation coordinates and dates attached to Mapillary photos, we mapped the probability of cherry trees blooming in a spatial grid of dimensions 10 m x 10 m on a daily basis. With sufficient observations, start, peak, and end of blooming were estimated through time series analysis. A case study conducted at Saitama Universitys main campus in 2022 confirmed the possibility of mapping the presence of cherry blossoms and their blooming timing automatically. Since our approach relies solely on geotagged street-level photos that can be taken by anyone with no prior knowledge of cherry tree species identification, we anticipate that it will be easier to build blooming records over space and time than conventional stationary observations or citizen-based observation programs. This novel approach also has potential applications for detecting other species as well.

plant biology↗

bioRxiv2022

Mapping cherry blossoms from geotagged street-level photos

Floral phenology is useful information as an indicator on climate change and ecosystem services, however its observation is not straightforward over space and time. Satellite remote sensing and official and volunteer-based in-situ observations have been conducting, but the long-term and accurate data collection is challenging due to the insufficient quality and quantity of observations and the lack of financial and human resources to sustain. Here, we demonstrate a flower detection model from street-level photos, which can be the core function of a semi-automatic observation system to tackle those issues above. We detected cherry blossoms by this model from geotagged images with the observation date, obtained from Mapillary, which is one of social sensing data sources, and mapped dates of flowering in a study site, Aizuwakamatsu, Japan in April 2018. This approach enables us to collect floral phenology information semi-automatically as a data-driven approach. It is expected to collect a large number of observations with a certain level of quality by avoiding human-induced biases for the observations.

ecology↗