A loss function to evaluate agricultural decision-making under uncertainty: a case study of soil spectroscopy.

Modern sensor technologies can provide detailed information about soil variation which allows for more precise application of fertiliser to minimise environmental harm imposed by agriculture. However, growers should lose neither income nor yield from associated uncertainties of predicted nutrient concentrations and thus one must acknowledge and account for uncertainties. A framework is presented that accounts for the uncertainty and determines the cost-benefit of data on available phosphorus (P) and potassium (K) in the soil determined from sensors. For four fields, the uncertainty associated with variation in soil P and K predicted from sensors was determined. Using published fertiliser dose-yield response curves for a horticultural crop the effect of estimation errors from sensor data on expected financial losses was quantified. The expected losses from optimal precise application were compared with the losses expected from uniform fertiliser application (equivalent to little or no knowledge on soil variation). The asymmetry of the loss function meant that underestimation of P and K generally led to greater losses than the losses from overestimation. This study shows that substantial financial gains can be obtained from sensor-based precise application of P and K fertiliser, with savings of up to £121 ha-1 for P and up to £81 ha-1 for K, with concurrent environmental benefits due to a reduction of 4-17 kg ha-1 applied P fertiliser when compared with uniform application. Supplementary Information: The online version contains supplementary material available at 10.1007/s11119-022-09887-2.

Comparing the effect of different sample conditions and spectral libraries on the prediction accuracy of soil properties from near- and mid-infrared spectra at the field-scale.

The prediction accuracy of soil properties by proximal soil sensing has made their application more practical. However, in order to gain sufficient accuracy, samples are typically air-dried and milled before spectral measurements are made. Calibration of the spectra is usually achieved by making wet chemistry measurements on a subset of the field samples and local regression models fitted to aid subsequent prediction. Both sample handling and wet chemistry can be labour and resource intensive. This study aims to quantify the uncertainty associated with soil property estimates from different methods to reduce effort of field-scale calibrations of soil spectra. We consider two approaches to reduce these expenses for predictions made from visible-near-infrared ((V)NIR), mid-infrared (MIR) spectra and their combination. First, we considered reducing the level of processing of the samples by comparing the effect of different sample conditions (in-situ, unprocessed, air-dried and milled). Second, we explored the use of existing spectral libraries to inform calibrations (based on milled samples from the UK National Soil Inventory) with and without 'spiking' the spectral libraries with a small subset of samples from the study fields. Prediction accuracy of soil organic carbon, pH, clay, available P and K for each of these approaches was evaluated on samples from agricultural fields in the UK. Available P and K could only be moderately predicted with the field-scale dataset where samples were milled. Therefore this study found no evidence to suggest that there is scope to reduce costs associated with sample processing or field-scale calibration for available P and K. However, the results showed that there is potential to reduce time and cost implications of using (V)NIR and MIR spectra to predict soil organic carbon, clay and pH. Compared to field-scale calibrations from milled samples, we found that reduced sample processing lowered the ratio of performance to inter-quartile range (RPIQ) between 0% and 76%. The use of spectral libraries reduced the RPIQ of predictions relative to field-scale calibrations from milled samples between 54% and 82% and the RPIQ was reduced between 29% and 70% for predictions when spectral libraries were spiked. The increase in uncertainty was specific to the combination of soil property and sensor analysed. We conclude that there is always a trade-off between prediction accuracy and the costs associated with soil sampling, sample processing and wet chemical analysis. Therefore the relative merits of each approach will depend on the specific case in question.

Predicting the growth of lettuce from soil infrared reflectance spectra: the potential for crop management.

How well could one predict the growth of a leafy crop from reflectance spectra from the soil and how might a grower manage the crop in the light of those predictions? Topsoil from two fields was sampled and analysed for various nutrients, particle-size distribution and organic carbon concentration. Crop measurements (lettuce diameter) were derived from aerial-imagery. Reflectance spectra were obtained in the laboratory from the soil in the near- and mid-infrared ranges, and these were used to predict crop performance by partial least squares regression (PLSR). Individual soil properties were also predicted from the spectra by PLSR. These estimated soil properties were used to predict lettuce diameter with a linear model (LM) and a linear mixed model (LMM): considering differences between lettuce varieties and the spatial correlation between data points. The PLSR predictions of the soil properties and lettuce diameter were close to observed values. Prediction of lettuce diameter from the estimated soil properties with the LMs gave somewhat poorer results than PLSR that used the soil spectra as predictor variables. Predictions from LMMs were more precise than those from the PLSR using soil spectra. All model predictions improved when the effects of variety were considered. Predictions from the reflectance spectra, via the estimation of soil properties, can enable growers to decide what treatments to apply to grow lettuce and how to vary their treatments within their fields to maximize the net profit from the crop.

Evidence for functional state transitions in intensively-managed soil ecosystems.

Soils are fundamental to terrestrial ecosystem functioning and food security, thus their resilience to disturbances is critical. Furthermore, they provide effective models of complex natural systems to explore resilience concepts over experimentally-tractable short timescales. We studied soils derived from experimental plots with different land-use histories of long-term grass, arable and fallow to determine whether regimes of extreme drying and re-wetting would tip the systems into alternative stable states, contingent on their historical management. Prior to disturbance, grass and arable soils produced similar respiration responses when processing an introduced complex carbon substrate. A distinct respiration response from fallow soil here indicated a different prior functional state. Initial dry:wet disturbances reduced the respiration in all soils, suggesting that the microbial community was perturbed such that its function was impaired. After 12 drying and rewetting cycles, despite the extreme disturbance regime, soil from the grass plots, and those that had recently been grass, adapted and returned to their prior functional state. Arable soils were less resilient and shifted towards a functional state more similar to that of the fallow soil. Hence repeated stresses can apparently induce persistent shifts in functional states in soils, which are influenced by management history.

Distinct respiratory responses of soils to complex organic substrate are governed predominantly by soil architecture and its microbial community.

Factors governing the turnover of organic matter (OM) added to soils, including substrate quality, climate, environment and biology, are well known, but their relative importance has been difficult to ascertain due to the interconnected nature of the soil system. This has made their inclusion in mechanistic models of OM turnover or nutrient cycling difficult despite the potential power of these models to unravel complex interactions. Using high temporal-resolution respirometery (6 min measurement intervals), we monitored the respiratory response of 67 soils sampled from across England and Wales over a 5 day period following the addition of a complex organic substrate (green barley powder). Four respiratory response archetypes were observed, characterised by different rates of respiration as well as different time-dependent patterns. We also found that it was possible to predict, with 95% accuracy, which type of respiratory behaviour a soil would exhibit based on certain physical and chemical soil properties combined with the size and phenotypic structure of the microbial community. Bulk density, microbial biomass carbon, water holding capacity and microbial community phenotype were identified as the four most important factors in predicting the soils' respiratory responses using a Bayesian belief network. These results show that the size and constitution of the microbial community are as important as physico-chemical properties of a soil in governing the respiratory response to OM addition. Such a combination suggests that the 'architecture' of the soil, i.e. the integration of the spatial organisation of the environment and the interactions between the communities living and functioning within the pore networks, is fundamentally important in regulating such processes.

On the origin of carbon dioxide released from rewetted soils.

When dry soils are rewetted a pulse of CO2 is invariably released, and whilst this phenomenon has been studied for decades, the precise origins of this CO2 remain obscure. We postulate that it could be of chemical (i.e. via abiotic pathways), biochemical (via free enzymes) or biological (via intact cells) origin. To elucidate the relative contributions of the pathways, dry soils were either sterilised (double autoclaving) or treated with solutions of inhibitors (15% trichloroacetic acid or 1% silver nitrate) targeting the different modes. The rapidity of CO2 release from the soils after the drying:rewetting (DRW) cycle was remarkable, with maximal rates of evolution within 6 min, and 41% of the total efflux over 96 h released within the first 24 h. The complete cessation of CO2 eflux following sterilisation showed there was no abiotic (dissolution of carbonates) contribution to the CO2 release on rewetting, and clear evidence for an organismal or biochemical basis to the flush. Rehydration in the presence of inhibitors indicated that there were approximately equal contributions from biochemical (outside membranes) and organismal (inside membranes) sources within the first 24 h after rewetting. This suggests that some of the flux was derived from microbial respiration, whilst the remainder was a consequence of enzyme activity, possibly through remnant respiratory pathways in the debris of dead cells.

Defining and quantifying the resilience of responses to disturbance: a conceptual and modelling approach from soil science.

There are several conceptual definitions of resilience pertaining to environmental systems and, even if resilience is clearly defined in a particular context, it is challenging to quantify. We identify four characteristics of the response of a system function to disturbance that relate to "resilience": (1) degree of return of the function to a reference level; (2) time taken to reach a new quasi-stable state; (3) rate (i.e. gradient) at which the function reaches the new state; (4) cumulative magnitude of the function (i.e. area under the curve) before a new state is reached. We develop metrics to quantify these characteristics based on an analogy with a mechanical spring and damper system. Using the example of the response of a soil function (respiration) to disturbance, we demonstrate that these metrics effectively discriminate key features of the dynamic response. Although any one of these characteristics could define resilience, each may lead to different insights and conclusions. The salient properties of a resilient response must thus be identified for different contexts. Because the temporal resolution of data affects the accurate determination of these metrics, we recommend that at least twelve measurements are made over the temporal range for which the response is expected.

Landscape scale estimation of soil carbon stock using 3D modelling.

Soil C is the largest pool of carbon in the terrestrial biosphere, and yet the processes of C accumulation, transformation and loss are poorly accounted for. This, in part, is due to the fact that soil C is not uniformly distributed through the soil depth profile and most current landscape level predictions of C do not adequately account the vertical distribution of soil C. In this study, we apply a method based on simple soil specific depth functions to map the soil C stock in three-dimensions at landscape scale. We used soil C and bulk density data from the Soil Survey for England and Wales to map an area in the West Midlands region of approximately 13,948 km(2). We applied a method which describes the variation through the soil profile and interpolates this across the landscape using well established soil drivers such as relief, land cover and geology. The results indicate that this mapping method can effectively reproduce the observed variation in the soil profiles samples. The mapping results were validated using cross validation and an independent validation. The cross-validation resulted in an R(2) of 36% for soil C and 44% for BULKD. These results are generally in line with previous validated studies. In addition, an independent validation was undertaken, comparing the predictions against the National Soil Inventory (NSI) dataset. The majority of the residuals of this validation are between ± 5% of soil C. This indicates high level of accuracy in replicating topsoil values. In addition, the results were compared to a previous study estimating the carbon stock of the UK. We discuss the implications of our results within the context of soil C loss factors such as erosion and the impact on regional C process models.

The short-term effects of prescribed burning on biomass removal and the release of nitrogen and phosphorus in a treatment wetland.

Nutrient removal by constructed wetlands can decline over time due to the accumulation of organic matter. A prescribed burn is one of many management strategies used to remove detritus in macrophyte-dominated systems. We quantified the short-term effects on effluent water quality and the amount of aboveground detritus removed from a prescribed burn event. Surface water outflow concentrations were approximately three times higher for P and 1.5 times higher for total Kjeldhal nitrogen (TKN) following the burn event when compared to the control. The length of time over which the fire effect was significant (P < 0.05), 3 d for TKN and up to 23 d for P fractions. Over time, the concentration of soluble reactive phosphorus (SRP) in the effluent decreased, but was compensated with increases in dissolved organic phosphorus (DOP) and particulate phosphorus (PP), such that net total P remained the same. Total aboveground biomass decreased by 68.5% as a result of the burn, however, much of the live vegetation was converted to standing dead material. These results demonstrate that a prescribed burn can significantly decrease the amount of senescent organic matter in a constructed wetland. However, short-term nutrient releases following the burn could increase effluent nutrient concentrations. Therefore, management strategies should include hydraulically isolating the burned area immediately following the burn event to prevent nutrient export.

Assessment of the spatial distribution of soil properties in a northern Everglades marsh.

Florida Everglades restoration plans are aimed at maintaining and restoring characteristic landscape features such as soil, vegetation, and hydrologic patterns. This study presents the results from an exhaustive spatial sampling of key soil properties in Water Conservation Area 1 (WCA 1), which is part of the northern Everglades. Three soil strata were sampled: floc, upper 0- to 10-cm soil layer, and 10- to 20-cm soil layer. A variety of properties were measured including bulk density (BD), loss on ignition (LOI), total phosphorus (TP), total inorganic phosphorus (TIP), total nitrogen (TN), total carbon (TC), total iron (TFe), total magnesium (TMg), total aluminum (TAl), and total calcium (TCa). Interpolated maps and model prediction uncertainties of properties were generated using geostatistical methods. We found that the uncertainty associated with spatial predictions of floc, particularly floc BD, was highest, whereas spatial predictions of soil chemical properties such as soil Ca were more accurate. The resultant spatial patterns for these soil properties identified three predominant features in WCA 1: (i) a north to south gradient in soil properties associated with the predominant hydrological gradient, (ii) areas of considerable soil nutrient enrichment along the western canal of WCA 1, and (iii) areas of considerable Fe enrichment along the eastern canal. By using geostatistical techniques we were able to describe the spatial dynamics of soil variables and express these predictions with an acceptable level of uncertainty.

Spatial patterns of labile forms of phosphorus in a subtropical wetland.

Phosphorus (P) has been identified as the key constituent defining wetland productivity, structure, and function. Our goal was to investigate the spatial patterns of total P and three labile forms of P (labile organic, inorganic, and microbial biomass P) across a subtropical wetland located in east-central Florida, the Blue Cypress Marsh Conservation Area (BCMCA), and link spatial patterns to ecosystem processes. The wetland received a continual input of nutrients primarily from the south and intermittently from the west and east, respectively, which ceased in the mid-1990s. Since then the marsh system has been undergoing natural succession. We used (i) ordinary kriging to characterize the spatial patterns of total P and labile P forms across the wetland, (ii) local, moving spatial correlations to investigate relationships between total P and labile P forms, and (iii) a clustering technique to link the identified spatial patterns to biogeochemical processes. The spatially explicit analyses revealed patterns of total P and labile P forms as well as changing relationships between variables across the marsh. We were able to distinguish P-enriched areas from unaffected ("natural") areas and intermediate zones that are currently undergoing change as P is mobilized and translocated. We also identified areas that are at risk, showing a shift toward a more P-enriched status. Our results improve our understanding of P and its labile components within a spatially explicit context.

Response of biogeochemical indicators to a drawdown and subsequent reflood.

Temporal oscillations in hydrology are a common occurrence in wetlands and can result in alternating flooded and drained conditions in the surface soil. These oscillations in water levels can stimulate microbial activities and result in the mobilization and redistribution of significant amounts of carbon (C), nitrogen (N), and phosphorus (P). The goal of this study was to experimentally simulate a drawdown and reflood of marsh soil from a nutrient-enriched site and a reference site of a wetland (Blue Cypress Marsh Conservation Area, Florida). The goal was to better understand the changes in biogeochemistry and microbial activities present in these soils as a result of hydrological fluctuations. Measurements of dissolved reactive phosphorus (DRP), ammonia, and nitrate in the floodwater indicated significantly higher (alpha = 0.05) NH(4)(+) and DRP fluxes from the nutrient-enriched site; floodwaters in the cores from both sites contained significant NO(3)(-) concentrations (9.6 mg N L(-1)), which was rapidly consumed over the core incubation period (30 d). Water level drawdown and reflooding initially stimulated the soil microbial biomass, methanogenic rates, and extracellular enzyme activities (acid phosphatase and beta-glucosidase). The anaerobic microbial metabolic activities (CO(2)) where initially significantly (alpha = 0.05) enhanced by the reflood, resulting in roughly equivalent rates as the aerobic respiratory activities (CO(2)), presumably as a function of the high water column NO(3)(-) levels. This study illustrates that the reflood event in the hydrological cycles in a wetland can significantly stimulate the activities of hydrolytic enzymes and microbiological communities in these soils.

Adaptation of the CAS test system and synthetic sewage for biological nutrient removal. Part I: development of a new synthetic sewage.

A new synthetic medium has been developed for routine use in laboratory-scale sewage treatment simulation and biodegradation tests, such as OECD guideline 302A & 303A or ISO method 11733. The new medium, Syntho, was designed to meet the following objectives: 1) to be more representative of real sewage than the existing standard OECD synthetic sewage, 2) the COD:N:P ratio and mineral composition must allow a good degree of biological nutrient (N, P) removal, and 3) the medium should result in stable unit operation, including good sludge settling and minimal need for control actions. The IAWQ Activated Sludge Model No. 2 (ASM2,) was used to help design the medium and predict reactor performance for different possible media compositions. The results obtained with Syntho indicate that Continuous Activated Sludge (CAS) units with or without nutrient removal can be operated routinely on this feed. The new medium was also characterized by means of a respiration test. The different influent fractions applied in the model were validated, and a respiration profile indicated that Syntho is a close approximation of real sewage.

Adaptation of the CAS test system and synthetic sewage for biological nutrient removal. Part II: design and validation of test units.

A global increase in biological nutrient removal (BNR) applications in wastewater treatment and concern for potential effects of anthropogenic substances on BNR processes resulted in the adaptation of the Continuous Activated Sludge (CAS) laboratory test system (cf. guideline OECD 303A or ISO 11733). In this paper two novel systems are compared to the standard CAS unit: the Behrotest KLD4 and a University of Cape Town system (CAS-UCT). Both are 'single sludge' systems with an anoxic/aerobic and an anaerobic/anoxic/aerobic configuration, respectively. They both can simulate the essential processes of full-scale BNR installations. The units where fed with a specially designed synthetic sewage, Syntho (cf. Part I of this study), or its precursor BSR3 medium. The performance of the two new units was benchmarked against the standard CAS system in terms of carbon/nitrogen/phosphorus (C/N/P) removal, as well as primary biodegradation of the surfactants linear alkylbenzene sulfonate (LAS) and glucose amide (GA). Both systems allow to easily achieve stable excess N- and P-removal. Experimental C/N/P removal data compared closely with simulations obtained with the IAWQ Activated Sludge Model No. 2 (ASM2), and with full scale BNR plants with a similar configuration. In both units the effluent concentrations of the surfactants tested were significantly reduced in comparison to the standard CAS system (up to 50% less). No adverse effects on BNR were noted for the test surfactants dosed at 400 microg/l together with an overall surfactant background concentration in the feed of ca. 20 mg/l. The proposed systems hold potential to complement the standard CAS system for situations where advanced sewage treatment plants with BNR need to be simulated in the laboratory with minimum effort.