The botanical identification was made in the herbarium of the Colegio de Postgraduados, Campus Tabasco. Agroecosystems and productive use characterizations were performed at 75 % of the production farms through field visits (transects) and interviews with farmers. On each farm, we recorded in situ data that corresponded to the type of ecosystems, the predominant plant species, and the use and production systems (plant species, crops, animal husbandry, and fishing).

Soil was sampled and fertility parameters were measured through methods based on the Official Mexican Standard NOM-021-RECNAT-2000 (SEMARNAT 2002): pH (measured in a 1:2 soil to water ratio); electrical conductivity (EC); organic matter (OM), by the method of Walkley and Black; cation exchange capacity (CEC), texture by the Boyoucos hydrometer method; inorganic nitrogen (N); and exchangeable cations (Ca, Mg and K), extracted with 1 N ammonium acetate (pH 7.0) in a 1:20 ratio and phosphorus (P).

Sampling to quantify hydrocarbons in soil was based on NOM-138-SEMARNAT/SS-2003 (SEMARNAT 2005), for both sets of samples, i.e. soil at the surface and soil at a depth of up to 30 cm. Composite samples were taken from each site (for a total of 180 subsamples in a zigzag sampling pattern), taking into consideration the homogeneity of natural factors. The analysis for TPH was performed in the Environmental Control laboratory of the Benemérita Universidad Autónoma de Puebla (CICM-BUAP), where TPH concentration was measured by EPA method 418.1 (EPA 1986), using an FTIR Tensor 20 spectrophotometer, Bruker brand. The sample was run from 4000 to 400 cm^-1. The standard used was Altech, 418.1 and the peak area was measured to 2800 cm^-1.

After characterizing the production and analyzing the soil, we identified risk factors using reports of some studies (Teng et al. 2014). We created the following indicators: the productive diversity index (PDI_x), the productive rate risk (PR_x), and the efficient land use index (ELUI_x) for the growing of plants (p) and rearing of farm animals (a). PDI_x shows the relation between numbers of crop species or number of animal species bred on a farm (n _sp ) and the number of productive systems (n _s ) for each farm (i) within a zone with similar characteristics (equation 1). PR_x identifies the potential risk factors that are derived from the number of constraints by physical and chemical soil characteristics (n _rf ) with respect to the number of species of plants or animals (n _sp ) for each production system (n _s ) in each area (equation 2). ELUI assesses the efficiency of land use, relating PDI_x and PR_x in each production system from each community (equation 3):

P D I X = ∑ i = 1 n n s p n s i (1)

P R x =   ∑ i = 1 n n s n s p + n r f i (2)

E L U I x = ∑ i = 1 n I D P x P R x i (3)

where x represents the variable (p) relating to plant cultivation and (a) to rearing of farm animals, n _sp is the number of crop or animal species bred on a farm, n _s is the number of productive systems, n _rf is the number of potential constraints presented by soil physical and chemical characteristics, n is the number of farms, and i is the mean across the community. Dimensionless group values are interpreted as follows: the higher the numerical value, the greater the indicator, even though this depends on the kind of indicator. An increase in PDI_x and ELUI_x is considered favorable, whereas a higher value of PR would be unfavorable.

Each variable and index were analyzed using analysis of variance (ANOVA) comparisons of means by Tukey’s test (p ≤ 0.05); geochemical variables (OM, CEC, Ca, sand and TPH) and density of oil facilities were also examined through canonical-correlation analysis (p ≤ 0.05) using Statgraphics Centurion XV (Statgraphics 2006). Finally, canonical correspondence analysis was conducted with R version 3.1.2 using “ca” (R Core Team 2014) to analyze the relationship between indices in zones with or without oil facilities.

All four study zones were characterized by vegetation that is typical of alluvial plains with flood conditions with productive problems due to the high humidity conditions. Productive use was represented by cultivated pastures (Echinochloa polystachya and Brachiaria mutica) for feeding sheep. There was no observed agricultural use in remediated soil, which was characterized by mangroves (Laguncularia racemosa) and other species, such as Cyperus esculentus, Thalia geniculata and Mimosa pigra. In zones with moderate densities of oil wells (i.e., Paraiso and Ceiba), agricultural use was more evident with extensive cultivated pastures of Echinochloa polystachya and Brachiaria mutica for bovine livestock. Native species, such as Thalia geniculata, Mimosa pigra, Leucaena leucocephala, and Cyperus esculentus were also found. In Gurría, where there is no oil activity, a greater diversity of crops was noted, which included pastures of Paspalum distichum, subsistence crops (Zea mays and Phaseolus vulgaris), vegetable gardens (Cucumis sativus, Citrullus lanatus, and Cucurbita pepo), tropical fruit (Musa paradisiaca, Citrus aurantifolia Swingle, Cocos nucifera, and Manihot esculenta Crantz), and backyard livestock systems (poultry and farm animals); locals also engage in fishing activity. There were also areas with native vegetation, such as Thalia geniculata, Mimosa pigra, and other Mimosa spp. The main limiting factors were salinity and the typical conditions of low floodplains, with frequent flooding and Gleysols, therefore, Rovirosa soil is not suitable for agricultural production (Table I).

TABLE I

Zone	OF	Ecosystem description	Productive use	Production systems	Production risk factors
Rovirosa	14	Low jungle, meadow with Echinochloa polystachya (Kunth) Hitch, Brachiaria mutica (Forssk.) Stapf and flooded soils with Laguncularia racemosa (L.) C.F. Gaertn, Cyperus esculentus L., Thalia geniculata L. and Mimosa pigra L. with soil texture disturbances	Pasture grown on remediated soil. Soil remediation process without agricultural use	Livestock (sheep) only in remediated zones	TPH, frequent flooding, high salinity, petrogenic OM content, higher percentage of sand, and low levels of N, P, K
Paraíso	7	Low floodplain with frequent flooding and Gleysols. meadow with Brachiaria mutica (Forssk.) Stapf, Paspalum distichum L., Leersia hexandra Swartz, Thalia geniculata L., Mimosa pigra L., Leucaena leucocephala L., and Cyperus esculentus L.	Cultivated pasture	Livestock (cattle) with extensive grazing	TPH, frequent flooding, moderate salinity, petrogenic OM content, and higher percentage of clay
Ceiba	1	Low floodplain and meadow with Echinochloa polystachya, Paspalum distichum L., Leersia hexandra Swartz, Mimosa pigra L., and Leucaena leucocephala L.	Cultivated pasture	Livestock (cattle) with extensive grazing	Frequent flooding, moderate salinity, higher percentage of clay, and low level of K
Gurría	0	Low jungle and floodplain with frequent flooding and Gleysols with Paspalum distichum L., Panicum máximum Jacq, Thalia geniculata L., Mimosa pigra L. and Mimosa pudica L.	Native and cultivated pasture. Subsistence crops	Subsistence agriculture with basic crops (corn, beans), vegetables (cucumber, watermelon, pumpkin), fruit (banana, lemon), coconut and cassava. Subsistence livestock (poultry and farm animals)	Frequent flooding, salinity, higher percentage of clay, and low level of K
Gurría	0	Low jungle and floodplain with frequent flooding and Gleysols with Paspalum distichum L., Panicum máximum Jacq, Thalia geniculata L., Mimosa pigra L. and Mimosa pudica L.	Native and cultivated pasture. Subsistence crops	Subsistence agriculture with basic crops (corn, beans), vegetables (cucumber, watermelon, pumpkin), fruit (banana, lemon), coconut and cassava. Subsistence livestock (poultry and farm animals)	Frequent flooding, salinity, higher percentage of clay, and low level of K

OF: density of oil wells over 100 m². TPH: total petroleum hydrocarbons. OM: organic matter.

Salinity was extremely high in the soils that were located where there was a higher density of oil wells, and this was strongly correlated with TPH; we also observed sandy soil texture with very high content of OM and moderately high CEC. Petrogenic OM and high salt content in ecosystems that are susceptible to frequent flooding and poor drainage corresponded at least six risk factors for productive development. In addition, there was a lack of availability of soil nutrients. Finally, another indicator of soil disturbance was the excessive concentration of Ca, which is not typical of these areas (Table II).

TABLE II

±

pH

EC dS/m

Sand

Silt

Clay

OM

N

p

K

CEC cmol (+)/kg

Ca mg/kg

Mg

%

mg/kg

Rovirosa

5.8 ± 0.9a

56.8 ± 22a

71.8 ± 5.0a

19.1 ± 4.1b

9.1 ± 1.6a

30.3 ± 5.8a

15.3 ± 4.8a

10.9 ± 2.6a

434.0 ± 120.3a

64.9 ± 8.6a

11114.3 ± 2160.6a

3822.5 ± 634.1a

Paraíso

5.2 ± 0.2a

2.4 ± 1.3b

41.3 ± 7.3b

34.1 ± 8.4a

24.6 ± 13.1b

16.1 ± 3.3b

35.5 ± 17.8a

17.6 ± 7.5a

167.0 ± 63.5b

48.3 ± 13.7b

4816.0 ± 432.6b

3252.0 ± 594.8ab

Ceiba

5.3 ± 0.2a

5.7 ± 0.3b

17.8 ± 5.5c

43.1 ± 8.5a

39.1 ± 6.3c

5.1 ± 2.0c

20.2 ± 7.0a

13.0 ± 2.3a

268.0 ± 100ab

35.6 ± 6.4bc

2876.5 ± 596.6bc

3539.5 ± 524.4ab

Gurría

5.1 ± 0.3a

2.4 ± 1.6b

17.3 ± 7.5c

41.6 ± 4.3a

41.1 ± 10.1c

5.9 ± 1.7c

17.4 ± 4.2a

13.6 ± 4.6a

181.0 ± 67.9b

38.4 ± 4.5c

1628.0 ± 166.9c

2521.5 ± 286.3b

Values with different letters are significantly different (Tukey’s test, p ≤ 0.05).

TPH concentrations differed significantly between communities (p < 0.05), with a tendency to decrease with lower density of oil facilities (Table III). Soils contaminated with TPH were found in zones with a higher density of oil installations and were found to have values of 12 276 to 3553 mg/kg of TPH. In zones with higher contamination, soil salinity was also higher in oil-contaminated soils (56.8 dS/m) than areas without oil installations (2.4 dS/m); and was found to be the most important limiting factor. The main limiting factors in areas with moderate density of oil facilities were TPH concentrations of 1000 mg/kg, flooding zones, and moderate salinity (Table II).

TABLE III

Zone	TPH (mg/kg)	PDIa	PDIp	PRa	PRp	ELUIa	ELUI
Rovirosa	7117.5 ± 5967.5a	0.3c	0.3c	3.3a	0.3b	0.0c	0.0c
El Paraíso	1131.5 ± 2039.2ab	1.3b	1.0b	0.8b	1.0a	2.0b	1.0b
La Ceiba	138.9 ± 53.6b	1.0b	1.0b	1.0b	1.0a	1.0b	1.0b
Gurría	190. ± 1250.0b	3.2a	2.0a	0.3c	0.5b	11.2a	4.0a

TPH: total petroleum hydrocarbons, PDI_a: productive diversity index of plant cultivation, PDI_p: productive diversity index of farm animal rearing, PR_a: productive rate risk of plant cultivation, PR_p: productive rate risk of farm animal rearing, ELUI_p: efficient land use index of plant cultivation, ELUI_a: farm animal rearing.

Canonical correlation analysis indicated that the greatest diversity and production efficiency was found in Gurría (no oil installations), while highest risks were found in sites with greater density of oil installations (Rovirosa > Paraíso > Ceiba) (Fig. 2). In Gurría, the value of RP_p,a was lower, with higher PDI_p,a and IES_p,a values; this was due to the greater diversity of crops and pastures (Table II).

Fig. 2

Canonical correlation analysis with respect to analyzed soil geochemical variables (OM, CEC, Ca, Sand and TPH) indicated a high significant relationship (p < 0.01), explaining the increase of PR_p,a in the presence of greater density of petroleum facilities (Fig. 3). The negative correlation between soil geochemical variables (OM, CEC, Ca, Sand and TPH) with respect to ELUI_p,a and PDI_p,a let to identify the risk factors associated with productive development in contaminated areas, as describe above in table II.

Fig. 3

Correspondence analysis indicated that the greatest diversity and production efficiency was found in Gurría (no oil installations), whereas the highest risks were found in the zones with greater density of oil installations (Rovirosa > Paraíso > Ceiba) (Fig. 4). In Gurría, the value of RP_p,a was lower, with higher PDI_p,a and IES_p,a values, which was due to the greater diversity of crops and pastures.

Fig. 4

Monitoring in agricultural areas adjacent to petroleum facilities allows the environmental authorities to consider and focus efforts on fertility based in agroecological characteristics and land use, on which subsistence smallholders of developing countries depend (Vignola et al. 2015). In the Ecuadorian Amazon, petroleum hydrocarbons have been found in the hunting and fishing areas, endangering wildlife and indigenous populations in the region, because they are exposed to the ingestion of soils and sediments contaminated with oil (Rosell-Melé et al. 2018). This form of translocation of the contaminants is also a risk to agriculture, for example, in the rhizosphere of reeds, n-alkanes have been found in the range of C10 to C33 in the upper layers of the soil (10 cm depth of soil), through irrigation with irrigation water from the river (Tian et al. 2014).

Gleysol is mostly used for the cultivation of grasses (Saggar et al. 2001). In this case grasslands with Brachiaria brizantha (Hochst) Stapf and crops such as Zea mays L. are more tolerant to the presence of petroleum derivatives, while legumes like Neonotonia wightii Arn. can be severely affected (Gürtler et al. 2018), which would explain the presence of halophilic plants and grasses in the study area.

In this study, the presence of pastures in contaminated soils is due to their high resilience to toxic effects of TPH (Franco et al. 2004) and their tolerance to constant flooding and lack of drainage from these areas, which represents a risk for the present vegetation (Sims and Colloff 2012). In this case, the pollutant can be dissipated by water during the six months that the soil remains flooded. Zones contaminated with oil were characterized by plants belonging to the families Poaceae, Cyperaceae, and Fabaceae (García-López et al. 2006), which coincides with the data reported in this study. Echinochloa polystachya and Brachiaria mutica are perennial grass found in oil-polluted soils that contain high concentrations of petroleum, heavy metals and salts (Solís-Domínguez et al. 2007, Panta et al. 2014, Fatima et al. 2016). Laguncularia racemosa is a mangrove species found in this study, which has been reported with reductions in growth rates of up to 20 % in response to a simulated oil spill of 5 L/m² (Sodré et al. 2013), and in conditions of high salinity the efficiency of the plant in the use of nitrogen decreases between 37 and 58 % (Sobrado 2005).

The soil in this area has been reported as Mollic Gleysol (Rivera-Cruz et al. 2005), typical of alluvial plains and characterized by rich OM, high nutrient content, and clayey, silty texture (IUSS Working Group WRB 2006). But, in this study, it was found that there was a higher sand content in the transition zone of the floodplain that coincided with the oil spill in Rovirosa: an area that contains fine deposits and terrace sediments of shale and sandstone. The density of oil wells was high (equidistance of 380 m), and these were constructed with materials from fill terraces which had eroded and accumulated on the Gleysol. There have also been clean-ups of material accumulated underground, with removal of sandy materials ex situ, resulting in disturbances to soil texture.

Gutiérrez and Zavala (2002) indicated that the typical nature of these Gleysols allows for the accumulation of hydrocarbons in the groundmass, in which case, it contains a few active surfaces that promote efficient drainage and extend the leaching toxicity of oil. Mikkonen et al. (2012) found that the vertical gradient of the proportion of aliphatic and aromatic hydrocarbons increased with the depth of the soil profile, and it is, therefore, important to consider pollutant transport in the ground profile and the potential for groundwater contamination. Should this occur, surface water wells, aquatic organisms, and the food chain could be severely affected (Perhar and Arhonditsis 2014). Such situations can be aggravated by the fact that the oil originates from past spills, with petroleum that has been weathered having high molecular weight compounds (Vega et al. 2009). As a result, soils have suffered physical degradation; and the quality and sustainability of their chemical (Gallego et al. 2010), biological, and enzymatic processes (Alrumman et al. 2015) have been jeopardized. Petroleum hydrocarbon contamination is, therefore, a major constraint to agricultural production and may represent a risk to human health due to the possibility of direct contact with contaminated soil (Zhao et al. 2014), also, something that was not studied in this research was the bioaccumulation of trace elements (As, Cd, Co, Cr, Cu, Pb and Zn) in native plants (Robichaud et al. 2019), animals and sediments (Li et al. 2019), which should be analyzed in subsequent investigations.

Moreover, although OM in soil is usually a good indicator of fertility, in these soils the higher OM does not mean an improvement in soil quality, due to the ratios of C/N and C/P being unfavorable for microbial growth and enzymatic activity (Gao et al. 2013). Zavala-Cruz et al. (2005) found that high OM content in Gleysol also retains TPH fractions and may alter the solubility of phosphorus. Studies have suggested that an increase in organic waste decomposition can reduce the concentration of hydrocarbons and stimulate soil microbiota (Martín-Gil et al. 2008). In a study conducted in China, no correlation was observed between the OM content and the individual or total polycyclic aromatic hydrocarbons (PAHs) concentrations, and no significant relationships were found between the pH or total nitrogen. In this study, the zones were mainly located in agricultural regions where the artificial introduction of organic fertilizers may affect the fate of PAHs (Liu et al. 2016).

Some authors note that CEC is not affected by oil in the ground (Martínez and López 2001); however, EC can measure the salinity of soil extracts. The soil salinity is crucial to the survival of many plant species because salinity has effects on the osmotic potential of plants, with high salinity causing loss of intracellular water, reducing swelling, and increasing the accumulation of ions (sodium and chloride); this may directly interfere with internal biochemical processes. Effects range from limiting processes of germination and plant growth to posing a risk to animal feed (Masters et al. 2007). In this study, the severe salinity of the samples that were found near petrochemical facilities was probably due to water associated with extraction and storage processes, which generally have a high content of dissolved salt from geological formations (salt domes) (van Thienen-Visser et al. 2014); this reduces the potential for agricultural production. De la Garza et al. (2008) suggested that in soils with higher hydrocarbon concentrations, pH decreases but EC increases 5.6 times compared to uncontaminated soil, reducing the ability to retain Ca and K cations.

A limiting factor for agricultural production was the very high content of TPH, since 1.5 % represents a critical value for plant growth (Tang et al. 2011). High soil salinity has also been one of the main problems in contaminated soil because of negative synergistic effects on soil bacterial diversity (Gao et al. 2015). Furthermore, the specific characteristics of these soils in relation to flooding and excessive salinity levels can affect plant growth and development, affecting natural attenuation of petroleum (Tang et al. 2012), due to a reduction in cellulose decomposition. Furthermore, fertility problems arising from having chemical properties outside acceptable parameters limit availability of some nutrients (P and N) and pose a danger of oversupply of others (Na), increasing risks to both plant and animal production; however, there are plants such as the white mangrove, which has been characterized as a plant that can tolerate these conditions (Kathiresan and Bingham 2001).

The improper handling of refined petroleum products are potential sources of soil contamination that have been documented (Yamprai et al. 2014), and thus we propose indicators as simple assessments of risk factors in situ, based on analyses of information about the type of crops that have been cultivated by farmers. Land use planning and policy decision-making are crucial to avoid conflicts between the government and the local people (Duangjai et al. 2015); typical problems that have been reported in oil zones (Acuña 2015).

The productive characterization led to identifying the ecosystem in order to recognize the limiting factors from the natural and the anthropogenic conditions. In this case, flood characteristics were typical of lowland jungle, and there are even previous studies that have evaluated the zone type and its potential for ecosystem services (Namaalwa et al. 2013). The indicators recommended in this study allow productive diversity to be integrated as one of the many soil functions, providing criteria that demonstrate the deterioration in soil quality (Rodríguez and Lafarga 2011). The inclusion of more indicators, however, could assist in developing a more comprehensive view of soil functioning (Schloter et al. 2003), which will certainly have an impact on sustainable agricultural development (Volchko et al. 2013). Thus, concentrations between 250.49 and 9387.26 ng/g of PAH have been found in agricultural lands in China, representing more than 60 % of all PAHs (Liu et al. 2016).

Goodsir et al. (2019) indicate that the criteria for risk assessment of pollutants must include aspects of ecological sensitivity and socioeconomic receptors to finally decide if a remediation, intervention or monitoring program is required. Other authors have developed a multivariable index, which includes all individual compounds, based on toxicological studies (PAH and BTEX) compared with TPH, which allows simplifying the soil evaluation by reducing the number of variables (Pinedo et al. 2013). Therefore, recent studies suggest the development of evaluation methods and systems that involve the monitoring of ecotoxicity and the detection of in situ bioindicators (Shen et al. 2016), so the indicators of the present study could be a reference for the attention of sites contaminated by petroleum hydrocarbons.