A machine learning approach to predict phyllosphere resistome abundance across urbanization gradients

Recent studies reported an increased abundance of antibiotic resistance genes (ARGs) in urban greenspaces, yet the predictability of ARG variance along urbanization gradients remains unclear. We sampled paired soil and phyllosphere samples from the same site in wetland parks along urbanization gradi...

Full description

Saved in:
Bibliographic Details
Main Authors: Rui-Ao Ma, Yi-Hui Ding, Shifa Zhong, Ting-Ting Jing, Xuechu Chen, Si-Yu Zhang
Format: Article
Language:English
Published: Elsevier 2025-08-01
Series:Environment International
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S0160412025004064
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849424170380689408
author Rui-Ao Ma
Yi-Hui Ding
Shifa Zhong
Ting-Ting Jing
Xuechu Chen
Si-Yu Zhang
author_facet Rui-Ao Ma
Yi-Hui Ding
Shifa Zhong
Ting-Ting Jing
Xuechu Chen
Si-Yu Zhang
author_sort Rui-Ao Ma
collection DOAJ
description Recent studies reported an increased abundance of antibiotic resistance genes (ARGs) in urban greenspaces, yet the predictability of ARG variance along urbanization gradients remains unclear. We sampled paired soil and phyllosphere samples from the same site in wetland parks along urbanization gradients to assess the correlations of soil and phyllosphere ARG abundance with urbanization indices. Our results revealed that the abundance of phyllosphere resistomes correlated better with urbanization gradients than did that of soil resistomes and increased along urbanization gradients. Moreover, the phyllosphere presented more ARG-MGE (mobile gene element) pairs in metagenome-assembled genomes than soil, suggesting greater transmission potential than soil ARGs. Proximity to the built area and microbial diversity were the most important factors that significantly (p < 0.01) drove the variance in phyllosphere ARG abundance. By integrating population density, land use type, landscape metrics, and air quality data into machine learning models, we predicted phyllosphere ARG abundance at a 10-meter resolution. Among the five tested algorithms tested in the machine learning models (ridge regression, K-nearest neighbor, support vector machine, and neural network), the random forest algorithm achieved the highest accuracy with the lowest root mean square error (27.24 vs. 40.79–46.79 for the other models). These results demonstrate a strong association between phyllosphere ARG abundance and urbanization indices and provide predictions of the potential ARG risk along these gradients. The heightened transmission potential in urban greenspaces may facilitate the spread of antibiotic resistance spread to human pathogens, which poses significant public health threats.
format Article
id doaj-art-32c0121af4a4486292cea475fc10c2b7
institution Kabale University
issn 0160-4120
language English
publishDate 2025-08-01
publisher Elsevier
record_format Article
series Environment International
spelling doaj-art-32c0121af4a4486292cea475fc10c2b72025-08-20T03:30:19ZengElsevierEnvironment International0160-41202025-08-0120210965510.1016/j.envint.2025.109655A machine learning approach to predict phyllosphere resistome abundance across urbanization gradientsRui-Ao Ma0Yi-Hui Ding1Shifa Zhong2Ting-Ting Jing3Xuechu Chen4Si-Yu Zhang5School of Ecological and Environmental Sciences, East China Normal University, Shanghai 200241, ChinaSchool of Ecological and Environmental Sciences, East China Normal University, Shanghai 200241, ChinaSchool of Ecological and Environmental Sciences, East China Normal University, Shanghai 200241, ChinaSchool of Ecological and Environmental Sciences, East China Normal University, Shanghai 200241, ChinaSchool of Ecological and Environmental Sciences, East China Normal University, Shanghai 200241, China; Technology Innovation Center for Land Spatial Eco-restoration in Metropolitan Area, Ministry of Natural Resources, Shanghai 200062, ChinaSchool of Ecological and Environmental Sciences, East China Normal University, Shanghai 200241, China; Technology Innovation Center for Land Spatial Eco-restoration in Metropolitan Area, Ministry of Natural Resources, Shanghai 200062, China; Corresponding author at: School of Ecological and Environmental Sciences, East China Normal University, Shanghai 200241, China.Recent studies reported an increased abundance of antibiotic resistance genes (ARGs) in urban greenspaces, yet the predictability of ARG variance along urbanization gradients remains unclear. We sampled paired soil and phyllosphere samples from the same site in wetland parks along urbanization gradients to assess the correlations of soil and phyllosphere ARG abundance with urbanization indices. Our results revealed that the abundance of phyllosphere resistomes correlated better with urbanization gradients than did that of soil resistomes and increased along urbanization gradients. Moreover, the phyllosphere presented more ARG-MGE (mobile gene element) pairs in metagenome-assembled genomes than soil, suggesting greater transmission potential than soil ARGs. Proximity to the built area and microbial diversity were the most important factors that significantly (p < 0.01) drove the variance in phyllosphere ARG abundance. By integrating population density, land use type, landscape metrics, and air quality data into machine learning models, we predicted phyllosphere ARG abundance at a 10-meter resolution. Among the five tested algorithms tested in the machine learning models (ridge regression, K-nearest neighbor, support vector machine, and neural network), the random forest algorithm achieved the highest accuracy with the lowest root mean square error (27.24 vs. 40.79–46.79 for the other models). These results demonstrate a strong association between phyllosphere ARG abundance and urbanization indices and provide predictions of the potential ARG risk along these gradients. The heightened transmission potential in urban greenspaces may facilitate the spread of antibiotic resistance spread to human pathogens, which poses significant public health threats.http://www.sciencedirect.com/science/article/pii/S0160412025004064UrbanizationPhyllosphereSoilAntibiotic resistance geneMachine learning
spellingShingle Rui-Ao Ma
Yi-Hui Ding
Shifa Zhong
Ting-Ting Jing
Xuechu Chen
Si-Yu Zhang
A machine learning approach to predict phyllosphere resistome abundance across urbanization gradients
Environment International
Urbanization
Phyllosphere
Soil
Antibiotic resistance gene
Machine learning
title A machine learning approach to predict phyllosphere resistome abundance across urbanization gradients
title_full A machine learning approach to predict phyllosphere resistome abundance across urbanization gradients
title_fullStr A machine learning approach to predict phyllosphere resistome abundance across urbanization gradients
title_full_unstemmed A machine learning approach to predict phyllosphere resistome abundance across urbanization gradients
title_short A machine learning approach to predict phyllosphere resistome abundance across urbanization gradients
title_sort machine learning approach to predict phyllosphere resistome abundance across urbanization gradients
topic Urbanization
Phyllosphere
Soil
Antibiotic resistance gene
Machine learning
url http://www.sciencedirect.com/science/article/pii/S0160412025004064
work_keys_str_mv AT ruiaoma amachinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT yihuiding amachinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT shifazhong amachinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT tingtingjing amachinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT xuechuchen amachinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT siyuzhang amachinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT ruiaoma machinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT yihuiding machinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT shifazhong machinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT tingtingjing machinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT xuechuchen machinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients
AT siyuzhang machinelearningapproachtopredictphyllosphereresistomeabundanceacrossurbanizationgradients