Land sparing or sharing: A paired stream study to evaluate the relationship between land use configuration and macroinvertebrate diversity in streams
Data files
Jul 24, 2026 version files 813.02 KB
-
LSLS_PairedStreams_Clean_Data_.csv
9.02 KB
-
LSLS_PairedStreams_Clean_Data_metadata.csv
2.10 KB
-
LSLS_PairedStreams_Difference_Data_.csv
10.37 KB
-
LSLS_PairedStreams_Difference_Data_metadata.csv
3.15 KB
-
LSLS_PairedStreams_Rawdata_AFDW.csv
22.20 KB
-
LSLS_PairedStreams_Rawdata_chlorophyll.csv
25.66 KB
-
LSLS_PairedStreams_Rawdata_masterdata.csv
25.25 KB
-
LSLS_PairedStreams_Rawdata_metadata.csv
8.45 KB
-
LSLS_PairedStreams_Rawdata_nutrients.csv
20.09 KB
-
LSLS_PairedStreams_Rawdata_sediment_richness_data.csv
2.45 KB
-
LSLS_PairedStreams_Rawdata_sedimentsheet_median.csv
33.17 KB
-
LSLS_PairedStreams_Rawdata_sedimentsheet.csv
27.53 KB
-
LSLS_PairedStreams_Rcode.Rmd
26.45 KB
-
LSLS_Taxa_Data_RAW_Accumulation_curves.csv
82.94 KB
-
LSLS_Taxa_Data_RAW_metadata.csv
1.62 KB
-
LSLS_Taxa_Data_RAW_Species_ID.csv
309.27 KB
-
Python_code.pdf
122.53 KB
-
Python_NMDS_for_R_.csv
3.10 KB
-
Python_NMDS_for_R_metadata.csv
808 B
-
README.md
7.08 KB
-
Taxa_data_wide_pairs_.csv
38.65 KB
-
Taxa_data_wide_pairs_metadata.csv
968 B
-
Taxa_data_wide_Python_NMDS_.csv
29.47 KB
-
Taxa_data_wide_Python_NMDS_metadata.csv
705 B
Abstract
Increasing food production often comes at the cost of habitat loss, which contributes to declining biodiversity. Consequently, balancing global food production with biodiversity conservation is a growing challenge, particularly as agricultural production expands and intensifies to meet the needs of a growing human population. The land sharing–land sparing (LSLS) framework presents two contrasting strategies for managing the trade-off between food production and biodiversity conservation. Land sharing involves low-intensity agriculture being interspersed within heterogeneous landscapes that retain natural habitats, whereas land sparing concentrates high-yield agriculture in some areas to allow separate areas to remain undisturbed as natural habitat designated for conservation. While LSLS has been studied extensively in terrestrial ecosystems, its relevance for freshwater ecosystems, despite their exceptional biodiversity and vulnerability, remains largely unexplored. Here, we evaluate how aquatic macroinvertebrate biodiversity is influenced by LSLS land use configurations in agricultural landscapes of Pennsylvania, USA. We compared macroinvertebrate species richness, evenness, abundance, and community composition between paired stream reaches—one flowing through a land sharing landscape configuration and the other through a land sparing configuration. The paired study design ensured streams were matched for physical habitat characteristics and most water quality variables known to influence macroinvertebrates, thus allowing us to isolate any potential impacts of land configuration per se. We found no significant differences in macroinvertebrate species richness, evenness, abundance, or composition between paired streams flowing through land sharing versus land sparing configurations. The only differences identified were small changes in the abundance of collector-filterers that are likely to be biologically insignificant. As one of the first empirical tests of the LSLS framework in aquatic ecosystems, these findings suggest that the spatial configuration of agricultural land use, whether integrated with or spatially separated from natural habitat, may have limited influence on macroinvertebrate biodiversity in streams, cautioning against overgeneralizing terrestrial conservation strategies for use in freshwater ecosystems.
Dataset DOI: 10.5061/dryad.b8gtht7t0
Description of the data and file structure
Dataset DOI: 10.5061/dryad.b8gtht7t0
Dataset Overview:
These datasets contain the data and analytical files required to reproduce the analyses presented in Padda and Cardinale 2026, which evaluated whether land sharing and land sparing landscape configurations influence stream macroinvertebrate biodiversity in Pennsylvania, USA.
Data were collected from 36 paired stream reaches (72 total streams) distributed across multiple HUC-8 subbasins within U.S. EPA Level III Ecoregions 67, 69, and 70. Each pair consisted of one stream embedded in a land sharing (dispersed agriculture) landscape and one in a land sparing (clustered agriculture) landscape. Spatial configuration was quantified using Moran's I within a 2.0 km² buffer.
This repository contains raw environmental data, processed analytical datasets, macroinvertebrate community matrices, NMDS outputs, metadata files describing variables, and R/Python code used in the analyses.
Files and variables
File Descriptions:
LSLS_PairedStreams_Clean_Data_.csv
Processed site level dataset used for statistical analyses after quality control and standardization.
Variable definitions are provided in LSLS_PairedStreams_Clean_Data_metadata.csv.
LSLS_PairedStreams_Clean_Data_metadata.csv
Metadata describing every variable in LSLS_PairedStreams_Clean_Data_.csv, including descriptions, units, data types, and notes.
LSLS_PairedStreams_Difference_Data_.csv
Paired-difference dataset (Sharing-Sparing) for biodiversity metrics and environmental variables.
Variable definitions are provided in LSLS_PairedStreams_Difference_Data_metadata.csv.
LSLS_PairedStreams_Difference_Data_metadata.csv
Metadata describing every variable in LSLS_PairedStreams_Difference_Data_.csv.
LSLS_PairedStreams_Rawdata_masterdata.csv
Raw field measurements collected at sampled stream reaches.
Variable definitions are provided in LSLS_PairedStreams_Rawdata_metadata.csv.
LSLS_PairedStreams_Rawdata_sediment_richness_data.csv
Sediment measurements and associated site level sampling information (i.e., richness).
Variable definitions are provided in LSLS_PairedStreams_Rawdata_metadata.csv.
LSLS_PairedStreams_Rawdata_sedimentsheet.csv
Raw Wolman pebble count data collected at sampled stream reaches. Each group of four columns represents one stream reach. The column headers identify the EPA Level III ecoregion, bridge identifier, and HUC-8 subbasin for each stream reach, while the second header row identifies the sediment sample number (1–4). Each row within the table represents an individual pebble selected during the Wolman pebble count, and each cell contains the diameter of that particle in millimeters (mm). Values represented by the symbol ">" indicate particles larger than the largest sediment size class measured (>180 mm).
Variable definitions are provided in LSLS_PairedStreams_Rawdata_metadata.csv.
LSLS_PairedStreams_Rawdata_sedimentsheet_median.csv
Contains the median substrate particle size (D50) calculated from the raw Wolman pebble count measurements for each sampled stream reach. The column headers identify the EPA Level III ecoregion, bridge identifier, and HUC-8 subbasin for each stream reach. Each value represents the median particle diameter in millimeters (mm) calculated from the individual particle measurements collected during the Wolman pebble count. These median values were used as the representative substrate size for each stream reach in subsequent analyses.
Variable definitions are provided in LSLS_PairedStreams_Rawdata_metadata.csv.
LSLS_PairedStreams_Rawdata_nutrients.csv
Raw nutrient measurements.
Variable definitions are provided in LSLS_PairedStreams_Rawdata_metadata.csv.
LSLS_PairedStreams_Rawdata_chlorophyll.csv
Raw chlorophyll-a measurements.
Variable definitions are provided in LSLS_PairedStreams_Rawdata_metadata.csv.
LSLS_PairedStreams_Rawdata_AFDW.csv
Raw ash-free dry weight (AFDW) measurements.
Variable definitions are provided in LSLS_PairedStreams_Rawdata_metadata.csv.
LSLS_PairedStreams_Rawdata_metadata.csv
Metadata describing variables contained in the raw environmental datasets.
LSLS_Taxa_Data_RAW_Species_ID.csv
Long-format macroinvertebrate abundance dataset.
Variable definitions are provided in LSLS_Taxa_Data_RAW_metadata.csv.
LSLS_Taxa_Data_RAW_Accumulation_curves.csv
Species accumulation curve data used to evaluate sampling completeness.
Variable definitions are provided in LSLS_Taxa_Data_RAW_metadata.csv.
LSLS_Taxa_Data_RAW_metadata.csv
Metadata describing variables in the raw macroinvertebrate datasets.
Taxa_data_wide_pairs_.csv
Wide-format site × taxa abundance matrix.
Variable definitions are provided in Taxa_data_wide_pairs_metadata.csv.
Taxa_data_wide_pairs_metadata.csv
Metadata describing variables in Taxa_data_wide_pairs_.csv.
Taxa_data_wide_Python_NMDS_.csv
Wide-format abundance matrix used for Python NMDS analysis.
Variable definitions are provided in Taxa_data_wide_Python_NMDS_metadata.csv.
Taxa_data_wide_Python_NMDS_metadata.csv
Metadata describing variables in Taxa_data_wide_Python_NMDS_.csv.
Python_NMDS_for_R_.csv
Two-dimensional NMDS ordination coordinates exported from Python for use in R.
Variable definitions are provided in Python_NMDS_for_R_metadata.csv.
Python_NMDS_for_R_metadata.csv
Metadata describing variables in Python_NMDS_for_R_.csv.
Missing Values:
Some datasets contain empty cells representing unavailable observations. Empty cells were intentionally retained rather than replaced with placeholder values (e.g., "NA") because the accompanying R and Python scripts were developed using the original data structure, and modifying these values could interfere with reproducibility of the analyses.
Unless otherwise noted in the associated metadata files, blank cells indicate that data were not available for that observation. Missing values may occur because: a measurement could not be obtained in the field or laboratory, a sample was unavailable or unsuitable for analysis, or the variable was not applicable for that observation. Individuals wishing to import these data into statistical software may treat blank cells as missing values (e.g., NA in R).
Code files:
LSLS_PairedStreams_Rcode.Rmd
R Markdown file containing all analyses, statistical models, and figure generation.
Python_code.pdf
Python code used to calculate Bray–Curtis dissimilarity and perform NMDS.
Code/software
All tabular data are provided as CSV files. Statistical analyses were conducted in R (version 4.4.3) and Python (version 3.13). The accompanying R Markdown and Python files reproduce the analyses described in the manuscript.
