Beyond area-based design: Optimizing urban Other Effective Area-based Conservation Measures with a scale-sensitive, multi-species framework
Data files
Sep 14, 2026 version files 19.67 MB
-
All_species_EOS_for_PCA.xlsx
9.31 KB
-
CE_Pallas_s_squirrel_env_detection_data.xlsx
2.46 MB
-
EA_Amur_hedgehog_env_detection_data.xlsx
2.46 MB
-
LS_Chinese_hare_env_detection_data.xlsx
2.46 MB
-
ML_Asian_badger_env_detection_data.xlsx
2.46 MB
-
MS_Siberian_weasel_env_detection_data.xlsx
2.46 MB
-
NP_Raccoon_dog_env_detection_data.xlsx
2.46 MB
-
README.md
5.31 KB
-
SC_Spotted_dove_env_detection_data.xlsx
2.46 MB
-
TM_Chinese_blackbird_env_detection_data.xlsx
2.46 MB
Abstract
With 68% of the global population projected to reside in urban areas by 2050, conserving urban biodiversity through Other Effective Area-based Conservation Measures (OECMs) represents an urgent global priority under the Kunming-Montreal Global Biodiversity Framework. OECMs are geographically defined, non-protected areas that sustain long-term biodiversity through diverse management practices, such as community-managed forests that balance sustainable livelihoods with habitat preservation for endangered species. However, a fundamental challenge lies in the pervasive scale mismatch between conservation interventions and the specific spatial scales at which species perceive and respond to fragmented urban landscapes.
Utilizing extensive camera-trap data from 136 sites across Shanghai, we developed a novel multi-scale occupancy modeling framework. This approach quantified the scale-dependent habitat responses of eight common mammalian and avian species. We then performed Principal Component Analysis (PCA) on these results to categorize species into functional groups based on their scale-dependent response patterns.
Our results demonstrate that urban OECM effectiveness is critically scale-dependent, revealing that the majority (66.7%) of species-landscape relationships are governed by fine-scale factors (<1,000 m), with 15.6% peaking at an ultra-fine 50-m scale, highlighting the paramount importance of local habitat design.
Crucially, 51% of these relationships reversed direction across spatial scales, revealing how fine-scale habitat selection driven by natural resources can be fundamentally overridden by broad-scale anthropogenic pressures.
Building on these cross-scale response patterns, we classified urban wildlife in Shanghai into four functional groups, shifting the conservation paradigm from a one-size-fits-all solution to precision conservation strategies tailored to specific scale dependencies.
Synthesis and applications: Applying this framework, we identified specific instances within Shanghai’s potential OECM network where conservation outcomes are critically undermined by scale mismatch. Our work advocated targeted pathways for Shanghai urban biodiversity conservation, meanwhile provides a generalizable approach that explicitly integrates the scale of effect into area-based conservation planning, offering a generic approach for advancing urban biodiversity conservation under the post-2020 global biodiversity agenda.
Dataset DOI: 10.5061/dryad.4b8gthtvh
This dataset contains camera‑trap detection‑history data (split into 191 five‑day sampling segments) for 8 focal wildlife species from Shanghai megacity, together with 12 site‑level landscape covariates and 4 observation‑level detection covariates. These data support multi‑scale occupancy modelling to identify species‑specific scale of effect (SoE), and subsequent PCA‑based species functional‑group classification for urban OECM conservation planning.
Description of the data and file structure
All .xlsx files are species‑specific raw input tables for occupancy modelling. Each species file shares identical column structure.
CE_Pallas_s_squirrel_env_detection_data.xlsx: Pallas’s squirrel (Callosciurus erythraeus)EA_Amur_hedgehog_env_detection_data.xlsx: Amur hedgehog (Erinaceus amurensis)LS_Chinese_hare_env_detection_data.xlsx: Chinese hare (Lepus sinensis)ML_Asian_badger_env_detection_data.xlsx: Asian badger (Meles leucurus)MS_Siberian_weasel_env_detection_data.xlsx: Siberian weasel (Mustela sibirica)NP_Raccoon_dog_env_detection_data.xlsx: Raccoon dog (Nyctereutes procyonoides)SC_Spotted_dove_env_detection_data.xlsx: Spotted dove (Spilopelia chinensis)TM_Chinese_blackbird_env_detection_data.xlsx: Chinese blackbird (Turdus mandarinus)
Additional derived summary file for PCA:
All_species_EOS_for_PCA.xlsx: Optimal scale of effect values for each species‑landscape variable combination, input for PCA‑based functional‑group classification.
Response variable: species detection‑history columns (ob1‑ob20)
Detection‑history matrix for single‑season occupancy models:
1: Species detected within this 5‑day sampling segment0: Species not detected within this 5‑day sampling segmentNA: No valid camera‑trap survey effort for this segment
Site Covariates (Variable Categories & Names)
Landscape variables were calculated across circular buffer radii from 50 m to 5,000 m at 50‑m intervals, totalling 100 spatial scales per variable.
Column naming convention: abbreviation + buffer radius (e.g. CH50, NDVI100).
- Abbreviation - Full covariate name (unit): Description
Natural factors
- CH - Canopy height (m): Mean vertical distance from the ground to the top of the vegetation canopy within each circular buffer, unit: metres
- NDVI - Normalized difference vegetation index: Mean NDVI within each circular buffer, reflecting vegetation cover and primary productivity (unitless index, range approx. ‑1 to +1)
- WSR - Water surface ratio (%): Proportion of open water body area within circular buffers around each survey location, unit: percentage %
- WD - Waterway density (km/km²): Total length of linear water bodies (rivers and streams) per square kilometer, unit: km km⁻²
Urbanization factors
- RD - Road density (km/km²): Total length of all road classes per square kilometer, unit: km km⁻²
- BD - Building density (%): Proportion of building area within circular buffers around each survey location, unit: percentage %
- ISR - Impervious surface ratio (%): Proportion of impervious surface area within circular buffers around each survey location, unit: percentage %
Urban green space configuration factors
- NP - Number of patches: Total count of urban green space patches within circular buffers around each survey location (count, unitless)
- PD - Patch density (per 100 ha): Number of urban green space patches per 100 hectares
- MPS - Mean patch size (ha): Average area of individual urban green space patches within circular buffers around each survey location, unit: hectares
- ED - Edge density (km/km²): Total length of urban green space patch edges per square kilometer, unit: km km⁻²
- CO - Cohesion: Spatial connectivity of urban green space patches (0 to 100 %), with higher values indicating better connectivity
Detection Covariates (observation‑level, vary per sampling occasion)
- wind1 (human presence; binary 0/1)
- 1: Human activity was recorded at this camera during the occasion
- 0: No human activity recorded
- wind2 (free‑roaming cat presence; binary 0/1)
- 1: Free‑roaming cat detected during occasion
- 0: No free‑roaming cat detected
- wind3 (free‑roaming dog presence; binary 0/1)
- 1: Free‑roaming dog detected during occasion
- 0: No free‑roaming dog detected
- wind4 (season; categorical)
- spring: March‑May; summer: June‑August; autumn: September‑November; winter: December‑February
Sharing/Access information
Data available from the Dryad Digital Repository https://doi.org/10.5061/dryad.4b8gthtvh (Li et al., 2026). The R code used for the analysis is archived in Zenodo (DOI: 10.5281/zenodo.22654428), and is also maintained dynamically at GitHub (https://github.com/MinhuiLi/RCode-JAE_Urban_OECMs_SoE).
