Globally widespread woody encroachment into grass-dominated ecosystems has substantial consequences for carbon (C), nitrogen (N), and phosphorus (P) cycles. Despite its significance as an essential macronutrient, however, little is known regarding potential changes in the sulfur (S) cycle. We quantified S concentrations, stoichiometric relationships, and δ³⁴S values in the plant-soil environment to investigate landscape-scale changes in the S cycle following grassland-to-woodland transitions in a subtropical savanna. Plant tissues of woody species had significantly higher S concentrations and δ³⁴S values than those of herbaceous species, resulting in a landscape-scale correspondence between spatial patterns of S and δ³⁴S in surface soils and vegetation distribution, with higher S and δ³⁴S in soils beneath woody patches. These patterns were more subtle at soil depths > 5 cm. Woody plants had higher N:S ratios but comparable P:S ratios relative to herbaceous species, which contributed to contrasting spatial patterns between N:S and P:S ratios in surface soils. Sulfur in surface soils increased proportionally less relative to N, but proportionally more compared to P. Our findings indicate that grassland-to-woodland transitions amplify landscape-scale S dynamics, especially in surface soils, and create a S-enriched environment that enables woody plants to acquire sufficient S relative to demand to support their continued productivity and proliferation.

This dataset was collected at the Texas A&M AgriLife La Copita Research Area (27˚40' N, 98˚12' W) in southern Texas, USA. Briefly, a 160 m × 100 m landscape consisting of three major landscape elements (i.e., grasslands, clusters, and groves) was established on an upland portion of this site. We further subdivided this landscape into 10 m × 10 m grid cells (160 cells in total), and georeferenced each corner of each grid cell based on UTM coordinates system. In July 2014, we selected two randomly located sampling points within each 10 m × 10 m grid cell, yielding a total of 320 sampling points across this landscape. At each sampling point, we collected two adjacent soil cores (2.8 cm in diameter and 5 cm in depth).  All soil samples were passed through a 2 mm sieve to remove coarse organic matter and dried at 65 °C for 48 hrs prior to elemental and isotopic analyses.

 In addition, to characterize the elemental and isotopic composition of vegetation inputs, live leaves and fine roots were collected from 16 dominant plant species located within the 100 x 160 m study area. Samples were collected from four individual plants per species. Furthermore, for each landscape element, we selected five locations for collecting litter and fine roots. At each location, litter was collected within a 25 cm × 25 cm frame and sorted into twigs (i.e., coarse woody debris with a diameter < 1 cm) and leaf litter for groves and clusters. After removing the litter layer, one soil core (7 cm in diameter and 5 cm in depth) was collected from each location and washed through a 2 mm sieve to obtain fine roots. All plant and litter samples were carefully washed with deionized water and then dried at 65 °C for chemical analyses.

Oven-dried plant tissues, litter, and soil samples were ground to fine powder. All samples were analyzed for N concentration using a Costech ECS 4010 elemental analyzer, and for P concentrations using the lithium fusion-molybdenum blue method. In addition, all samples were combusted to SO2 for determination of total S concentrations and δ34S values using a Thermo Scientific EA IsoLink CNSOH elemental analyzer interfaced via a ConFlo IV with a Delta V Advantage isotope ratio mass spectrometer.

The dataset is comprised of an excel file with different tabs showing measurements for soil and plant samples.