The research location was in the Changqing oilfield (34° 20′N, 107° 10′E) in Qingyang County, Gansu Province, Northwest China. It has a temperate continental monsoon climate with an approximate average annual precipitation of 470 mm and an average annual temperature of 10.7 °C (Liang et al., 2011). Soil types are yellow loamy soils (Yuan et al., 2022).

The sampling spots are located at the 2nd Oil Production Plant of the Changqing oilfield. The area where we collected soil samples is bare land without vegetation cover. In May 2022, soil samples were collected using 2-cm diameter augers within the area of 1 square kilometer surrounding the pumping well in the Changqing oilfield (Supplementary Figure S1). Contaminated (S2, S3, and S4) and uncontaminated (S1) soils were sampled from three directions along pumping wells. The site at S1 was not affected by production activities and served as the control. The separation between each sampling site in the same direction was over 100 m. In each sampling site (10 m × 10 m), 10 soil core samples (0–10 cm) were randomly selected to form one soil sample. Three replicate samples were obtained from sampling sites in all three directions at equal distances from the pumping well site and transported to the lab in an icebox. After being passed over a sieve of 2 mm to discard visible organic debris, soil samples were grouped into three subsamples for the extraction of DNA, the analysis of ALP activity, and the determination of soil physicochemical properties, respectively.

A test of soil physicochemical properties was carried out as described by Bao (2008). Soil water content (SWC) and pH were measured by the gravimetric and electrode methods, respectively. Extraction of dissolved organic carbon (DOC), available nitrogen (AN: NO₃⁻ and NH₄⁺), and available phosphorus (AP) from soils was performed using deionized water, 2 M KCl, and NaHCO₃ (pH 8.5), respectively. Extraction of total phosphorus (TP) from soils was performed using an acid solution (H₂SO₄-HClO₄). Soil total nitrogen (TN) was measured by the dry combustion method. Soil organic carbon (SOC) was estimated by oxidation with dichromate. Extraction of total petroleum hydrocarbons (TPHs) in the soil was performed using an acetone and dichloromethane mixture (1:1, V: V; Li et al., 2020). The contents of TPHs were assayed by the gravimetric method after solvent evaporation using a nitrogen blowing instrument (Yuan et al., 2022). The activities of ALP (E. C.3.1.3.1) were measured using a fluorometer (Synergy^H4, BioTek, United States) as described by Saiya-Cork et al. (2002). The Supplementary material provided detailed methods for the determination of ALP activity.

Microbial DNA was obtained as described by Li et al. (2018). Assessment of phoD gene abundance was performed by quantitative PCR using a thermocycler analyzer (Eco™, Illumina, United States) with primers for F733/R1083 (Chen et al., 2019). Assessment of the phoD-harboring bacterial diversities and community structures was performed with a sequencing platform of Illumina MiSeq PE300. Sequence data were placed in the database of the NCBI Sequence Read Archive (SRA) under the accession number PRJNA908779. The Supplementary material provided detailed methods for quantitative PCR and Illumina MiSeq sequencing.

Merged sequences were obtained by processing the raw sequence using FLASH (Version 1.2.11) and then filtering it with QIIME (Version 1.9.1) to discard ones of low quality. The clustering of sequences into operational taxonomic units (OTUs) was executed at a 97% similarity (Edgar, 2013). The classification of each sequence was performed on the basis of information from the Functional Gene Database (FunGene), applying a 0.75 confidence threshold (Maidak et al., 2001; Chen et al., 2019). The coverage indices (Supplementary Table S2) and the rarefaction curves (Supplementary Figure S2) indicated that the data volumes of the sequencing reads were reasonable. Mothur (version 1.30.2) was used to calculate the α-diversity indices of the phoD-harboring bacterial community (Simpson, Shannon, Abundance-based Coverage Estimator, and Chao). The relative abundance of the phoD-harboring bacterial community was the ratio of the sum of OTUs belonging to a homogeneous phylogenetic group to the whole OTU number.

Values were means ±standard deviations (n = 3). In order to satisfy the criteria of normal distribution, the abundances of the phoD gene have been log-transformed. One-way ANOVA and Duncan’s test were used to analyze the differences in physicochemical property parameters (TPH, SWC, pH, SOC, TN, TP, DOC, AN, and AP), gene abundance, diversity indices, phoD community relative abundance, and ALP activity among the sampling sites. The relationships between gene abundance, diversity indices, ALP activity, and physicochemical parameters were investigated by SPSS (version 27.0, IBM SPSS, Inc.) using Pearson’s correlation with a 95% confidence interval.

The following analysis was performed using R (Version 3.3.1). Changes in phoD community structures were evaluated using a Bray–Curtis distance matrix through principal coordinate analysis (PCoA). Significant differences in the community composition of phoD among oil-contaminated soils were assessed on the basis of permutational multivariate analysis (PERMANOVA) with the “adonis” function (Anderson, 2001). Correlations between the community structure of phoD and physicochemical property factors were identified using canonical correspondence analysis (CCA) and the Mantel test by the “vegan” package.

The indirect and direct effects of physicochemical properties and phoD community characteristics on ALP activity were examined using structural equation modeling (SEM), based on the method of maximum likelihood in AMOS 26.0 (IBM, Chicago, United States). Non-significant χ² tests (p > 0.05), root mean square error of approximation (RMSEA, p< 0.05), and goodness-of-fit index (GFI, > 0.9) were considered acceptable. The bootstrapping procedure was applied to extend sample size (1,000 bootstrap samples) due to the small sample size (Li D. et al., 2023). The first axis of the PCoA score was applied to proxy phoD community structures (Li et al., 2021).

Across the four sampling sites, phoD gene abundance varying from 3.7 × 10⁶ ~ 4.7 × 10⁷ g⁻¹ dry soil was significantly different (Figure 1). The phoD gene abundance at site S1 was approximately 13 times higher than those at site S4 (p < 0.05). Positive correlations were found between SWC, NH₄⁺, and AP contents and phoD gene abundance, and negative correlations were found between pH, NO₃⁻, and TPH contents and phoD gene abundance (p < 0.05; Table 1).

The Chao and Shannon indices were used to estimate the phoD-harboring bacterial community richness and diversity, respectively. High values of the Chao and Shannon indices indicate the high community richness and diversity, respectively. Both the Chao and Shannon indices were similar among S1, S2, and S3, all of which had significantly larger values than those in S4 (p < 0.05; Figures 2a,b). The Chao and Shannon indices were positively related to SWC and NH₄⁺ contents but were negatively related to nutrient stoichiometry (AN/AP and DOC/AP), TPH, and NO₃⁻ contents (Table 1).

The compositions of the phoD-harboring bacterial community were similar in S2 and S3 and were separated from the sites of S1 and S4 (Figure 3a). PERMANOVA analysis showed different phoD-harboring bacterial community composition in oil-contaminated soil at Changqing oilfields (Figure 3a). The CCA showed that soil physicochemical properties together accounted for 51% of the variance in the composition of this phoD-harboring bacterial community composition (Figure 3b). The key soil physicochemical property parameters affecting shifts in the composition of the phoD-harboring bacteria community were the SWC, TN, TP, SOC, AN, and TPH contents and nutrient stoichiometry (SOC/TN and DOC/AN; Figure 3b; Supplementary Table S2).

The predominant phoD-harboring bacterial communities detected in all samples included Actinomycetia, Alphaproteobacteria, Rubrobacteria, Planctomycetia, Betaproteobacteria, Gammaproteobacteria, Thermoleophilia, and Gloeobacteria (Figure 4). The most abundant functional groups were Actinomycetia among the phoD-harboring bacterial communities and increased from 13% in S1 to 48% in S4 (p < 0.01; Figure 4). Both Thermoleophilia and Rubrobacteria abundance increased from the sites of S1 to S4. By contrast, the Planctomycetia showed a decreasing tendency from the sites of S1 to S4. Compared to S1, Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria increased in the sites of S2 and S3 and then decreased in S4 (Figure 4). Actinomycetia, Thermoleophilia, and Rubrobacteria abundances were positively related to TPH, SOC, and DOC contents but were negatively related to NO₃⁻, AN, TN, AP, and TP contents (p < 0.05; Supplementary Figure S3). Both Alphaproteobacteria and Betaproteobacteria abundances were positively related to NO₃⁻, AN, TN, AP, TP, and DOC but were negatively related to TPH contents (Supplementary Figure S3).

The activities of ALP in the contaminated sites (90 ~ 924 nmol h⁻¹ g dry soil⁻¹) were significantly lower than those in the uncontaminated site (1,550 nmol h⁻¹ g dry soil⁻¹) (Figure 5). ALP activity was positively associated with SWC, AN, and AP contents; phoD gene abundance; and Betaproteobacteria abundance but was negatively associated with TPH content, phoD community structure, and Rubrobacteria abundance (p < 0.05; Table 2). The SEM explained 62% of variations in the activities of ALP (Figure 6). The comprehensive factors, including SWC, AP, and AN contents, phoD gene abundance, and community structures, exerted effects on the activities of ALP. The activities of ALP were directly affected by AN (path coefficient = 0.54), SWC (path coefficient = 0.27), phoD gene abundance (path coefficient = 0.25), and phoD-harboring bacterial community structures (path coefficient = −0.29) and were indirectly affected by TPH and AP content.

We compared the phoD gene abundance and phoD-harboring bacterial community diversities in oil-contaminated soils with three levels of TPH content to undisturbed pristine soils. The TPH content of the soils was ranked as follows: S1 < S2 ≈ S3 < S4 (p < 0.05; Supplementary Table S1). The results demonstrated that phoD gene abundance and phoD-harboring bacterial community diversities were negatively impacted by the TPH content, as obvious declines in phoD gene copies and Chao and Shannon indices in the contaminated soils were observed. The correlation analysis also illustrated negative relationships between functional gene abundance, α-diversity indices, and TPH content (Table 1). These findings were inconsistent with our first hypothesis, which suggested an increased population of the phoD-harboring bacterial community in the soils contaminated with oil. A decrease in abundance and diversity could be attributed to the toxic oil components (Nejidat et al., 2023; Yu et al., 2024). The findings were similar to previous studies by Li et al. (2022a), who suggested that petroleum pollution can significantly reduce bacterial and fungal abundance in the Changqing oilfield. However, oil pollution greatly enhanced bacterial abundance, richness, and biodiversity in the Huabei and Daqing oil fields (Liao et al., 2015). Megharaj et al. (2000) showed that a stable microbial abundance was observed in the low petroleum pollution group (< 2,100 mg TPH kg⁻¹ soil). This study found a considerable decrease in microbial abundance at the sampling sites of S2, S3, and S4, where the TPH concentration ranged from 554 to 999 mg kg⁻¹. In contrast to previous studies, a relatively low concentration of petroleum would trigger a substantial reduction in the number of microorganisms involved in the phosphorus cycle. Furthermore, both the Chao and Shannon indices did not show differences at the sites of S1, S2, and S3, suggesting that a higher threshold concentration of TPH was essential to result in changes in diversity compared to absolute abundance at this field site.

Spatial heterogeneity of soil physical and chemical characteristics leads to notable differences in soil microbial community abundance and diversity responding to oil contamination in regions with different geographic locations (Liu et al., 2009; Sutton et al., 2013; Song et al., 2025). It was found that SWC showed a positive correlation with phoD gene abundance and phoD-harboring bacterial community diversities (including the Shannon and Chao indices) at the study sites (Table 1). Water is widely recognized as a major influencing factor for the activities of soil microorganisms, having a crucial contribution to the distribution of phoD-harboring bacterial communities in uncontaminated ecosystems (Li W. et al., 2023; Xu et al., 2025). Generally, an increase in TPH content in soils could enhance soil hydrophobicity that reduces SWC and water storage capacity (Supplementary Table S4), and this causes a further limitation of the connectivity between nutrient-rich soil micropores and immobilizing soluble substrates in the soil matrix (Khamehchiyan et al., 2007; Devatha et al., 2019; Marañón-Jiménez et al., 2022). In this study, we present the observation that a decreased phoD gene abundance in the petroleum-contaminated soils was associated with decreased soil water and nutrient contents (AN and AP) despite abundant carbon sources. Moreover, the Changqing oilfield is situated in arid and semiarid regions, which receive very low annual rainfall. The water availability in this area, prone to drought or scarce water resources, has a major effect in dictating changes in the phoD-harboring bacterial community in oil-contaminated soils. The phoD gene abundance was positively dependent on NO₃⁻ content and negatively dependent on the NH₄⁺ content, possibly because P and N transformation processes were found to be highly related, as reported by several previous researchers (Liu S. et al., 2020; Xu et al., 2022). Furthermore, the Chao and Shannon indices varied significantly depending on the ratios of AN/AP and DOC/AP in the soils post-oil contamination (Table 1), indicating that these bacteria carrying the phoD gene directly regulated species richness and community diversity rather than their proliferation in reaction to the soil nutrient stoichiometry imbalance, which was caused by a great quantity of carbon input originating from hydrocarbon.

There were considerable variations in phoD-harboring bacterial community composition from oil-contaminated soil compared to undisturbed pristine soil (Figure 3), suggesting that the oil contamination concentration is a major factor reshaping the distribution of phoD microorganisms in the Changqing oilfield. Therefore, the high similarity of phoD-harboring bacterial community composition at the sites of S2 and S3 may be explained by the non-significant TPH content in these two sites. Apart from THP content, other soil physicochemical properties, i.e., SWC, TN, TP, SOC, and AN contents and nutrient ratio, were also important factors influencing distribution of the community structures. The above results indicated that the inputs of hydrocarbon significantly altered the soil physicochemical properties (Supplementary Table S4) and consequently altered phoD-harboring bacterial community compositions (Zhou et al., 2017).

Although the abundance and diversity of the overall phoD-harboring bacterial community have declined, the community composition has undergone a dramatic change, with the enrichment of tolerant groups and the decline of sensitive groups within it. In this study, it was observed that the relative abundance of Actinobacteria, Rubrobacteria, and Thermoleophilia increased from the sites of S1 to S4. Both Rubrobacteria and Thermoleophilia detected in this study belong to a separate subline of Actinobacteria. Previous studies reported that Actinobacteria were prevalent in oil-contaminated environments and were the dominant group of phoD-harboring microbes within different soil environments (Huang et al., 2021; Lu et al., 2022). These findings suggested that Actinobacteria may be strongly adapted to the environment contaminated by petroleum in the study sites. The resource competition theory highlighted that species with low trophicity are the only ones that are selected by resource limits (Tilman, 1982). The most abundant Actinobacteria were observed in oil-contaminated soils, which may be due to the fact that these bacteria have a trophic lifestyle similar to oligotrophic microorganisms (Dai et al., 2020; Wei et al., 2021). We indeed observed that Acidobacteria showed a negative correlation with the contents of N and P in the soils, which further supports our second hypothesis that N and P nutrients are key factors influencing phoD-harboring bacterial community composition. In general, oil contamination affects microbial survival and growth in soil, causing changes in microbial community composition by forcing native soil microbes to adapt to the new environment and elevating the enrichment of hydrocarbon-tolerant microorganisms (Calvo et al., 2019).

By contrast, Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria were categorized in the Proteobacteria phylum, which was primarily derived from copiotrophic microorganisms (Dai et al., 2020). Compared to oligotrophs, copiotrophic bacteria had a fast growth in response to available C concentration (Luo et al., 2019) and highly demanding N and P nutrients (Dai et al., 2020). In this study, the Proteobacteria phylum showed positive correlations with DOC, AN, and AP contents. The high level of available C input derived from hydrocarbons led to the enrichment of the copiotrophs (Proteobacteria) under low contamination stress, while the growth of these groups is also dependent on other nutrient availability (e.g., N and P) under high contamination stress. In the present study, the observed relative abundance of Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria tended to increase and then decrease from the sites of S1 to S4. This trend confirmed the previous observation that microbial population increased as TPH concentration increased, possibly due to the adaptation of soil microorganisms to lower-dose TPH (Guo et al., 2012).

This study found that ALP activities in contaminated soil were lower than those in uncontaminated soil, which was opposite to our first hypothesis. Previous studies by other authors reported that either a significant decline or increase in microbial activity appeared in the soils after oil pollution (Meng et al., 2018; Chakravarty et al., 2022; Song et al., 2025). Low levels of P availability would evoke the expression of phosphatases to meet requirements for bacterial growth in infertile soils, resulting in an increase in ALP activities (Liu et al., 2021; Luo et al., 2021). The suppression observed in our study, however, was most likely caused by the inhibition of beneficial microbial populations due to oil contamination (Telesiński et al., 2018; Chakravarty et al., 2022). A toxic effect for petroleum hydrocarbon decreases the total abundance of the phoD-harboring bacterial community, as it reduces the availability of substrate for microorganisms (as described above), thereby ultimately impeding the enzymatic reactions (Chakravarty et al., 2022). In this study, a significant decline in ALP activity was observed as TPH concentration increased, which was directly affected by the decreased abundance and altered phoD-harboring bacterial community compositions. Surprisingly, phoD-harboring community compositions had negative impacts on ALP activity, which could be attributed to different lifestyle characteristics that the species in phoD functional groups usually possess (e.g., copiotrophs and oligotrophs). The overall ALP production capacity of this new community was lower than that of the original community following soil oil contamination, although the unit efficiency of some of its groups might be very high.

Among the different phoD-harboring bacterial community members, the activity of ALP showed positive and negative correlations with the relative abundance of copiotrophic Betaproteobacteria and oligotrophic Rubrobacteria, respectively (p < 0.05). Xu et al. (2025) have reported that not all microorganisms harboring the phoD gene play a crucial role in the synthesis of ALP. Xu et al. (2025) and the current findings suggest that total alkaline phosphatase production was the result of the interaction among different taxa that harbor similar amino acid sequences but possess contrasting lifestyle characteristics. Furthermore, both Betaproteobacteria and Rubrobacteria were in low abundance, only accounting for under 4% of all phoD gene sequences, and were considered to be rare taxa. Both Betaproteobacteria and Rubrobacteria, two rare microbial taxa, played crucial roles in maintaining soil ecosystem functions, according to a growing number of studies (Xu et al., 2022). Our previous studies also observed that rare microbial taxa were more closely associated with the activities of ALP than the abundant ones (Liu S. et al., 2020). The present findings further highlighted the importance of rare phoD-harboring microorganisms in the promotion of ALP activity from oil-contaminated soils. Betaproteobacteria have been reported to be found and isolated from the majority of contaminated soils, and many of them were confirmed as petroleum degradation agent pioneers. This finding infers that some Betaproteobacteria, as petroleum hydrocarbon degraders, are capable of mineralizing organic phosphorus to release bioavailable phosphorus. Therefore, increasing the abundance of Betaproteobacteria that engage in both hydrocarbon degradation and organic-P mineralization in oil-contaminated soils could improve soil nutrient conditions and benefit the survival and proliferation of these microbes. Our study demonstrates that the presence of culturable Betaproteobacteria will facilitate future in situ bioremediation of oil-contaminated soils in the Changqing oil field.