The experiment, led by Rogério Hartung Toppa of the Federal University of São Carlos, was anything but casual. Across four seasonal periods spanning October 2023 to January 2025, the team released 1,200 worker bees, each fitted with a 2.5-millimeter RFID tag weighing about 2.4 milligrams, at increasing distances from their home colonies. The tags, glued to each bee’s thorax with non-toxic adhesive, were read by antennas mounted at the hive entrances, which logged the identity and timestamp of every returning individual. In total, 165 bees, or 13.75 percent of those released, made it home within the 35-hour monitoring window.

The numbers tell a story of steep but continuous decline rather than a hard cutoff. Return probability fell by roughly 41 percent with every additional kilometer of displacement, and the odds dropped by about 23 percent for every 500 meters. The most returns, 33 individuals, occurred at the shortest tested distance of 2.5 kilometers, with progressively fewer bees returning from 3, 3.5, 4, 4.5, 5, and 5.5 kilometers. Yet the statistical models found no evidence of a discrete homing threshold. When the researchers fitted piecewise models searching for a breakpoint between 3 and 6 kilometers, the best candidate, near 5.5 kilometers, failed to improve model fit enough to justify treating it as a true ecological boundary. Homing ability, it seems, fades gradually rather than collapsing at a fixed distance.

What makes the two 7.5-kilometer returnees so remarkable is that they represent the extreme tail of a distribution shaped by both distance and landscape. The study area in the Sorocaba Metropolitan Region of São Paulo State is a textbook fragmented environment: a mosaic of agricultural land covering 32.5 percent of the terrain, urbanized areas at 18.4 percent, remnant Semideciduous Seasonal Forest of the Atlantic Forest biome at 14.8 percent, eucalyptus plantations, pasture, temporary crops, and even mining operations. Somewhere in that patchwork, two bees with brains smaller than a grain of rice managed to reconstruct a route home across terrain that would defeat many larger animals.

The key to understanding how lies in satellite imagery. The team processed Sentinel-2 A data on the Google Earth Engine platform to derive three spectral indices: the Normalized Difference Vegetation Index (NDVI), which tracks vegetation vigor; the Normalized Difference Water Index (NDWI), which captures surface and canopy moisture; and the Normalized Difference Built-up Index (NDBI), which maps impervious urban surfaces. Combining these into a landscape resistance matrix, the researchers could model the least-cost paths a bee might take between each release point and its colony, and then ask which environmental features actually predicted who came home.

The answer was vegetation, at two different scales. In models examining the immediate surroundings of release points within 250-meter buffers, the maximum NDVI value was strongly and positively associated with return probability, meaning bees dropped into greener, denser vegetation were significantly more likely to navigate home. The same held true along the modeled corridors: higher NDVI maxima along the least-cost path increased return success, and so did a greater range of NDWI values, indicating a mosaic of wetter and drier patches along the route. Models that included these remote-sensing variables substantially outperformed distance-only formulations, and survival analyses tracking return time over 35 hours confirmed the same pattern.

Why would vegetation vigor and moisture heterogeneity matter so much to a bee’s internal compass? The authors connect their findings to decades of work on insect navigation. Bees rely heavily on path integration, an internal odometer that accumulates angular and distance information during flight, but this system accumulates error over long distances and must be corrected against external references. Vegetated landscapes provide exactly those references: textured horizons, salient tree landmarks, and stable visual contrasts that bees can memorize and match. Honeybees are known to recognize complex natural scenes, to use optic-flow snapshots for goal localization, and to calibrate their odometry against terrain contrast. In this framework, high-NDVI patches act as navigational scaffolding, allowing bees to recalibrate their path integration and reduce orientation drift as they cross hostile stretches of open field or concrete.

The moisture story is subtler but equally intriguing. A high NDWI range along a corridor does not mean the route was uniformly wet; rather, it signals the juxtaposition of contrasting surface types, riparian strips beside dry fields, forest edges against built structures. Such mosaics generate strong visual boundaries and elongated linear features, like streams and irrigation channels, that are known to serve as reliable landmarks for route learning in honeybees. Moisture gradients may also create microclimatic cues: bees possess antennal hygroreceptors that encode humidity and its rate of change, and ambient humidity is known to influence foraging activity in stingless bees. A landscape that varies in moisture, the study suggests, is simultaneously more physiologically hospitable, more visually discriminable, and richer in navigational cues than a monotonously homogeneous one.

Perhaps the most forward-looking contribution is the corridor analysis. By weighting NDVI maxima and NDWI range according to their statistical effect sizes, the team scored candidate return routes across all four seasons and identified thirteen priority corridors. A subset of these, notably the route associated with release point 45, which achieved the highest mean suitability score of 0.918, persisted as top-ranked pathways across at least three of the four periods. These were classified as core corridors: structurally stable movement routes supported by consistent vegetation cover that remain functional year-round. Others appeared only in one or two seasons, transient pathways whose suitability fluctuated with phenology and climate. The distinction matters enormously for conservation, because a static map of habitat connectivity misses exactly this temporal dimension. A corridor that works in the rainy season may fail in the dry one, and vice versa.

For a species that pollinates greenhouse tomatoes, produces honey in traditional meliponiculture, and stabilizes plant reproduction across the beleaguered Atlantic Forest, these results arrive at a critical moment. Recent modeling shows suitable habitat for M. quadrifasciata shrinking as agriculture expands into its range. The new study offers a concrete prescription: protect and restore high-vigor vegetation along the core corridors that bees already use, maintain moisture heterogeneity including riparian strips and wet-dry mosaics, and place vegetated stepping-stones within agricultural and urban matrices to extend the navigational scaffolding across large gaps. The authors are careful to note that a 7.5-kilometer homing flight is an exceptional feat, not a routine commute, and that return probability beyond roughly 5 kilometers is vanishingly small. But the fact that it is possible at all reveals a hidden reserve of navigational capacity, one that thoughtful landscape design could unlock. As pollinators worldwide face intensifying habitat fragmentation, the sight of two tagged bees threading their way home across seven and a half kilometers of Brazilian farmland is a vivid reminder that the smallest navigators may hold some of the biggest lessons for keeping ecosystems connected.

Subject of Research: Homing ability and movement ecology of the stingless bee Melipona quadrifasciata in fragmented Brazilian landscapes

Article Title: Homing ability of Melipona quadrifasciata in fragmented landscapes extends to 7.5 km

Article References: Toppa, R. H., da Silva, E. C. M., Martines, M. R., de Oliveira, C. H., de Souza, P., Arena, M. V. N., de Almeida Santos, L., & Mavinga, E. S. (2026). Homing ability of Melipona quadrifasciata in fragmented landscapes extends to 7.5 km. Discover Conservation, 3(1), Article 9. https://doi.org/10.1007/s44353-026-00080-4

Image Credits: AI Generated

DOI: 10.1007/s44353-026-00080-4

Keywords: stingless bees, Melipona quadrifasciata, homing ability, RFID tracking, pollinator movement, landscape connectivity, NDVI, remote sensing, least-cost corridors, habitat fragmentation, Atlantic Forest, conservation planning