The study examined two areas on the margins of the basin: the Gaoquandong Anticline in the Sikeshu Depression, where a structural trap was investigated, and the eastern slope of the Fukang Depression, where lithologic and stratigraphic traps were assessed. Sampling followed regular grids, with 500 m spacing in the structural-reservoir area and 330 m spacing across the larger eastern area. Study Area A contained 256 samples, while Study Area B contained 1,351. Each soil sample was a composite of three subsamples collected 5 m apart at depths of 20–25 cm. Points affected by roads, wells, residences, lakes or other unsuitable conditions were relocated within the prescribed distance from each grid location.

After seven days of incubation at 38 °C, colonies were counted and expressed as colony-forming units per plate, or CFU. The researchers used butanol-containing selective medium to isolate butane-oxidizing bacteria, a group that can use compounds associated with light-hydrocarbon seepage as a carbon source. Preliminary dilution tests established different optimal dilutions for Study Area A, farmland in Study Area B and Gobi terrain in Study Area B. Three parallel plates, positive controls, negative controls and blank controls were used for each sample. Thresholds separating background values from anomalies were estimated through statistical and frequency-distribution methods. Study Area A had a microbial anomaly threshold of 149 CFU, while area B used 30 CFU after comparison with results from known wells.

Because farmland and Gobi soils required different laboratory dilutions, the investigators applied a correction coefficient to place their values on a comparable scale. This adjustment preserved the spatial trends, although the authors cautioned that it could not fully remove differences in shallow environmental conditions. They also constructed maps from kriging interpolation on a grid finer than the original sampling interval, while setting rules for neighborhoods with too few observations. For relationships between microbial values and well data, each well was the unit of analysis, and its microbial value was calculated from nine nearby sampling points. Regression robustness was assessed with residual bootstrapping and leave-one-out cross-validation, and spatial autocorrelation in the residuals was tested with a coordinate-based spatial-weight matrix.

Above the Gaoquandong Anticline, high and ultrahigh microbial values formed a broad, contiguous anomaly over the structural high and the location of Well GT1. Within an approximately 8 km² anomalous zone, the mean value was 209 CFU, and 91% of samples exceeded the threshold. Outside that zone but still within the broader structural contour, the mean fell to 124 CFU and only 28% of samples remained anomalous. Well GT1 yielded high-rate industrial oil and gas from a reservoir more than 5,000 m deep, beneath 500–2,000 m of regional mudstone caprock. The microbial footprint closely matched the approximately 8 km² reservoir area defined through drilling. The contrast between the structural high and lower structural positions indicates that an anticline was not uniformly hydrocarbon-bearing.

The response became less regular above lithologic reservoirs. In the eastern Fukang Depression, microbial anomalies formed three irregular, northwest–southeast bands corresponding to well zones in the Toutunhe Formation. High values occurred above the discovery wells FD2, FD5 and FD8, whereas low or background readings were found above two display wells. High-resolution seismic data helped explain why nearby wells could have sharply different signals: FD051 and FD5, less than 2 km apart, actually penetrated different lenticular sand bodies. FD5, with a nine-point mean of 48 CFU, produced oil from a lower-positioned reservoir, while FD051, averaging 22 CFU, targeted another sand body and produced water. Their apparent oil–water inversion was therefore explained by the lithologic geometry rather than by a simple relationship between structural elevation and production.

Stratigraphic traps generated a third recognizable pattern. In the eastern study area, a promising anomaly zone northeast of discovery well FD8 covered about 8 km² and contained 52 anomalous sampling points, including 30 high or extremely high values. The cluster extended eastward along the structural slope. Seismic interpretation showed that the strongest bacterial anomalies occurred above the updip pinch-out of the Qigu Formation beneath a Cretaceous unconformity. The corresponding stratigraphic trap had a closure height of about 460 m and an area of approximately 6 km². Microbial values decreased toward lower structural positions, following the distribution of higher anomalies near the pinch-out. No significant seismic anomaly was detected in the underlying Toutunhe Formation beneath this zone, illustrating how a microbial footprint may reveal a reservoir whose geometry is not immediately apparent in one seismic interval.

Taken together, the three patterns support a predominantly near-vertical model of light-hydrocarbon microseepage through microfractures. Structural reservoirs produced laterally continuous, high-intensity clusters over the anticline core; lithologic reservoirs produced weaker and more discontinuous belt-like anomalies; and stratigraphic reservoirs produced clustered or sheet-like anomalies concentrated around pinch-out zones. Reservoir geometry therefore appears to influence the surface shape of the microbial signal. The authors also warn that stacked lithologic sand bodies can produce overlapping surface anomalies, meaning that a microbial footprint may not match the outline of any single reservoir identified by seismic data. Faults and unconformities connected directly to the surface may instead carry visible, high-flux macroseepage, producing linear patterns unlike the clustered microseepage responses measured in this study.

Drilling provided a direct test of the method. Three post-survey wells drilled in low-value or background parts of the Gaoquandong structure either encountered no hydrocarbons or produced only limited oil. On the eastern slope, post-survey wells FD9 and FD10 had nine-point microbial means of 22 and 19 CFU, below the 30 CFU threshold, and yielded only minor oil. By contrast, FD022 and FD16 lay within anomaly zones, had averages of 30 and 67 CFU and encountered commercial oil and gas. Across seven post-survey wells, the microbial classifications matched the reported drilling results. The authors described this as a 100% match, while emphasizing that the small validated sample and limited geographic coverage leave substantial statistical uncertainty and that the model still requires testing in independent cases.

Hydrocarbon enrichment was the strongest measured control. For nine wells with relevant production data, the mean microbial value had a positive association with cumulative test production, with R² = 0.8035, Pearson’s r = 0.8964 and p = 0.001. The bootstrap 95% confidence interval for R² was 0.6474–0.9140. Leave-one-out cross-validation produced a predictive R² of 0.5614, with RMSE = 10.914 and MAE = 8.544, while residual analysis found no significant spatial autocorrelation. Reservoir pressure was a much weaker predictor, with R² = 0.0308, and crude-oil density was similarly limited, with R² = 0.0515. These observational relationships do not establish that microbial abundance directly causes hydrocarbon accumulation, nor do they demonstrate that pressure or oil density has no effect. They show that, within the sampled wells, enrichment was the clearest correlate of the surface signal. Pressure may still influence seepage indirectly, particularly as production alters reservoir abundance or pressure gradients.

The findings also show that sealing layers do not automatically prevent microbial detection. Strong anomalies occurred above reservoirs separated from the surface by thick mudstone, indicating that light hydrocarbons can continue upward through microfractures. The pattern may be more complex where dense or heterogeneous caprocks, especially evaporites, suppress seepage and produce false-negative results. Microbial prospecting is best understood as a screening method rather than a stand-alone imaging technique. Its reliability may also be affected by soil properties, hydrogeology, climate, surface ecology and human disturbance, none of which was comprehensively quantified here. The approach can prioritize anomaly zones for geological and seismic analysis and can be checked with surface hydrocarbon-gas measurements, but it cannot independently reveal a reservoir’s depth or stratigraphic age. Multi-depth sampling, isotope tracing, molecular analysis and validation across more independent field cases would be needed to test how widely the Junggar Basin patterns apply.

Subject of Research: Field testing of soil butane-oxidizing bacteria as indicators of oil and gas reservoirs in China’s Junggar Basin

Article Title: Controls on microbial response above typical oil and gas reservoirs in the Junggar Basin and implications for exploration

Article References: Ding, L., Wu, L., Luo, S., Wu, Y., Wu, H., & Mei, H. (2026). Controls on microbial response above typical oil and gas reservoirs in the Junggar Basin and implications for exploration. Frontiers in Microbiology, 17, Article 1979085. https://doi.org/10.3389/fmicb.2026.1979085

Image Credits: AI Generated

DOI: 10.3389/fmicb.2026.1979085

Keywords: microbial prospecting, butane-oxidizing bacteria, hydrocarbon microseepage, Junggar Basin, oil and gas reservoirs, geomicrobiology, seismic interpretation, drilling validation, reservoir enrichment, trap geometry, Controls, microbial