Investigation of a Low-external-input Sustainable Rice Production System to Identify Ecosystem Services Towards Adoption Costs and Benefits

Final report for GS17-166

Project Type: Graduate Student
Funds awarded in 2017: $16,476.00
Projected End Date: 02/28/2019
Grant Recipient: Mississippi State University
Region: Southern
State: Mississippi
Graduate Student:
Major Professor:
Dr. Beth Baker
Mississippi State University
Expand All

Project Information

Summary:

Rice is the staple food for more than half of the world’s population and has the ability to support more people per unit of land area than wheat or corn because rice produces more food energy and protein per hectare than other grain crops. However, with the human population projected to reach 8.5 billion by 2030, there are major concerns about the sustainability of rice production practices because of its major role in consumption of natural resources, namely water and soil. There is a critical need to identify sustainable production practices that are economically feasible and minimize adverse environmental effects. The purpose of this project is to investigate a potentially sustainable rice production system in the Mississippi Alluvial Valley (MAV) that uses ecological principles to enhance environmental quality and economic gain at the field scale. We hypothesize that the annual flooding of rice fields to create water bird habitat will benefit soil health and water quality and increase avian biodiversity, as well as provide agronomic benefits to the farmer. Soil health, avian biodiversity, and water quality will be quantified to determine the profitability of implementing this system. Proof of concept at the field scale will provide a framework for other producers within the MAV to adopt similar management methods, ultimately improving the overall integrity of soil, water, and environmental quality as well as the farmer lifestyle.

Project Objectives:
Objective 1: Determine the impact of agronomic practices on soil health in rice production systems: conventional vs low-external-input. Based on the farmer reports of fertilizer input, literature reports of impacts of bird habitat use, and tillage practices, we hypothesize that soil health, in terms of structure, microbial diversity, and nutrient availability, will be greater at the low-external-input production site.

Objective 2: Determine potential drawbacks of low-external-input rice cultivation. Based on literature reports of migratory bird habitat use, and preliminary soil pathogen and bird use data, we hypothesize that greater pathogen detection will be correlated with bird abundance and flooded low-external-input production fields.

Objective 3: Determine effects of farming approach of water quality and runoff in rice productions systems: Conventional vs low-external-input. Based on farmer reports of water recapture methods and previous data from continuous flooded rice fields, we hypothesize that water runoff from low-external-input production will be lower in quantity, but higher in quality.

Objective 4: Quantify adoption and implementation economic costs and benefits to producers. Based on producer documentation of costs associated with system implementation in comparison to calculated benefits (yield impacts, economic gains of using less fertilizer, hunting lease potential, intrinsic value), we hypothesize that the low-external-input system will provide overall economic stability and profitability with long-term adoption.

While agricultural land has been identified as suitable wintering habitat for several avian guilds, this project is innovative and original because the use of this habitat toward developing an ecologically driven, integrated agriculture system has not been documented. We expect this work to result in documentation of management methods which will provide a foundation for sustainable information and technology transfer to producer peer groups, researchers, and policy makers. These outcomes are expected to have a positive impact on the sustainability of rice agriculture and protection of natural resources by providing scientific linkages between sustainable rice agriculture and ecosystem services in the Mississippi Alluvial Valley. The results will be translational at the producer scale, while ecosystem services will be applicable toward the restoration of natural functions and enhancement of wintering habitats for water birds. If adoption benefits outweigh costs, implementation at other rice production sites can spearhead the growth of a conservation paradigm shift.

 

 

Cooperators

Click linked name(s) to expand
  • Dr. John Brooks
  • Dr. Ray Iglay
  • Dr. Brian Davis

Research

Materials and methods:

Objective 1: Soil Health

Study areas were subjected to two different types of agronomic management, conventional rice production and LEISA management. To eliminate sources of weather and soil variation, the LEISA (n = 1) and conventional (n = 1) farms were 10 miles apart in Tallahatchie and Leflore County, respectively. The LEISA system uses slotted board riser pipes to control field flooding that occurs through precipitation during the non-growing season, creating a wetland-like environment. Overwintering waterbirds use these fields to rest and forage. The conventional system represented a typical continuous rice system of the region, which includes rolling post-harvest rice stubble into soil and leaving fields fallow during the non-growing season (i.e., winter). Unlike most conventional systems in the region, experimental field flooding occurred on both farms from fall 2017-spring 2018.

Flooding treatment was assigned according to landowner preference and categorized as LEISA Flooded (LF), LEISA Non-flooded (LN), Conventional Flooded (CF), or Conventional Non-flooded (CN). Ten fields from each category were selected for study, each ranging approximately 10-40.5 hectares (24-100 acres), with the exception of one LN field of 52 ha. Non-flooded fields under LEISA management were a consequence of field elevation, compared to other fields that were zero grade. Given these circumstances, 4 LN fields were available for this study.

Soil sampling was conducted post-harvest in November 2017 and just prior to planting in March 2018. Within each field, one sample was taken for every 4.05 ha (10 acres). To eliminate location sampling bias, soil grids were laid over field maps.  Grid coordinates were generated using stratified random sampling methods. GPS coordinates were extracted from the grid coordinates as sampling locations for each field. Individual samples were assumed to represent the micro-ecosystem of the area, and when aggregated, provide representation of field scale conditions (Patil et al. 2013).

Triplicate soil cores (3.8-4 cm diameter x 10 cm depth) were collected at each GPS location within 10 cm of each other and placed within a sterile plastic bag. Soil core samplers were sterilized with a 90% ethanol prior to sampling at each location. Soil samples were stored on ice (~ 4 oC) and then transported back to the Mississippi State University Water Quality Laboratory. Cores taken from each sample location within each field were combined into a composite sample from which two subsamples were used, one each for soil quality analysis and microbial analysis.

Soil Testing

Chemical and Physical Properties

One sub-composite sample was transferred to a 473 ml (1 pint) container and sent to Mississippi State University’s Soil Testing Laboratory for analysis of pH, Phosphorus (P), Potassium (K), Calcium (Ca), Magnesium (Mg), Sulfur (S), Sodium (Na), Cation Exchange Capacity (CEC), and percent organic matter (OM%).  All testing followed Mississippi State University Extension Service Soil Testing Laboratory guidelines. One gram of dried and crushed soil was sent to the United States Department of Agriculture to test for total percent nitrogen and carbon. Soil total C/N analysis was determined using an automated dry combustion Vario Max Cube Elementar C/N analyzer.

Microbial Properties

Gravimetric moisture content was calculated by drying soils at 105 °C for 24h. Soil preparations and heterotrophic plate counts followed methods modified from Zerzghi et al. (2009). Heterotrophic plate count  bacteria (HPCs) was determined from soil samples using 10-fold serial dilutions of 0.85% saline (EMD Chemicals Inc. Gibbstown, NJ) followed by plating on ½ R2A media (1/2 concentration R2A) (Difco Co., Sparks, MD) and incubation at 30 °C for 14 days to assess aerobic HPCs. A second set of ½ R2A plates was incubated anaerobically, using an Anoxomat anaerobic jar system with the default anaerobic gas setting to assess anaerobic/facultative anaerobic HPCs. This was to account for the anaerobic conditions created over flooded soils. Cultural heterotrophic counts for fungi were made using a spread plate technique on Sabouraud Dextrose Agar from 10-fold serial dilutions and then incubated at room temperature for 5 days. Cultural Gram-negative bacterial counts were made using a spread plate technique on MacConkey agar from 10-fold serial dilutions and then incubated at 35 °C for 24 hr. Cultural Gram-positive bacterial counts were made using a spread plate technique on Mannitol salt agar from 10-fold serial dilutions and then incubated at 35 °C for 48 h. All counts were reported as colony forming units per g (CFU g-1).

Dehydrogenase activity was determined using a procedure modified from Chu et al. (2007) to indirectly measure soil microbial activity, or the microbial oxidoreduction process and oxidation of organic substance (Maier et al. 2009). Two grams of sieved field-moist soil was mixed with 0.05 ml of 10% glucose and pre-incubated at 30°C for 24h. Following pre-incubation, samples were mixed with 0.02g of CaCO to adjust for acidic soil. A 1% triphenyltetrazolium chloride (TTC) with 10% glucose solution was mixed with soil in sealed test tubes. The tubes were incubated at 35 °C for 34 h followed by methanol extraction and quantification of water-insoluble red dye triphenylformazan (TPF) using a BioTek Synergy HTX multi-mode reader at 485 nm. A TPF standard curve, ranged from 0.05 to 0.50, was prepared to quantify samples and reported as µg TPF g-1.

Fecal Matter Estimates

Bird Surveys

Trial bird survey methods were conducted in December 2016. Bi-weekly point counts of all birds within fields were made on foot, counting and identifying all species observed within the boundary of the field, including birds standing on field banks and field edges. However, observer presence substantially biased data, among other obstacles associated with the initial sampling effort. For example, because of the close proximity of one field to the next, disturbed birds often flew to the next closest field resulting in recounting individuals at each field instead of accurately estimating bird use. Therefore, no-glow infrared camera traps (Stealth Cam G42NG) were placed in fields in order to eliminate problems associated with surveyor influence over bird presence. 

Cameras were placed at the midway point of the southern-most side of each field, 15 m in from the field edge, mounted to T-posts 91 cm above ground level and secured with plastic coated wire at a 90° angle facing north. One camera was placed in fields 20 ha or less in size. Fields > 20 ha received two cameras with the second camera placed at the northern-most side of the field, 15 m in from field edge and facing south. One LN field was > 40 ha and received three cameras, one at the southern-most edge, the northwest corner facing southeast, and the northeast corner facing southwest. A marker was placed 30.4 m directly in front of each camera as a distance reference point.

Cameras were programmed to capture one picture hourly during the non-growing season (November 3, 2017 through March 15, 2018). Data cards were retrieved and replaced once every month. Three cameras were stolen from three CN fields and two CF cameras fell over in January 2018. Additionally, two LF cameras stopped recording pictures because of technological glitches in December 2017, resulting in a period of missing data.  Because of this, pictures from these fields were only available for a partial season.

Images from camera traps were downloaded and opened in GNU Image Manipulation Program (GIMP) to quantify birds in fields (Mattis and Kimball, 2018). An image layer was created that clearly defined the 242.316m² sampling area directly in front of each camera based on the reference point and standardized camera placement. The image layer was overlaid on each photo taken, and all birds in the defined area were counted and categorized as goose, duck or other. Birds identified as duck or goose were recorded for each hour in each field.  Method and individual counter bias were evaluated by repeat counts of a subset of available pictures by two observers. Comparison of counts between observers indicated no significant observer bias. Average geese and duck use per day and ha (geese/duck/day/ha)based on number of active cameras was used for analysis.

Fecal Estimates

Total fecal inputs to fields were based on dry weights of bird droppings per day. For simplicity of calculation, weights were taken from past research regarding dropping estimates of Canada goose (Branta canadensis ; Terres 1980) and mallards (Anas platyrhynchos ; Sanderson and Anderson 1978).  These species were assumed to provide an average dropping rate of their respective categories (Fleming et al., 2001; Manley, 2008; Sanderson and Anderson, 1978; Terres and National Audubon Society., 1980; Zhang and Lu, 1999).  Average daily geese and duck densities were multiplied by 81.6 g (geese; Terres 1980) or 27.0 g (ducks; Sanderson and Anderson 1978) to get average total dry grams of fecal input to field per day per ha by geese or ducks over the non-growing season, respectively. Geese and duck fecal input per day per ha over the non-growing season were added together for final estimates of fecal matter input to fields (Table 1).

Statistical Analysis

Prior to analysis, soil parameters were checked for general multivariate frequentist test assumptions including but not limited to homogeneity of variance, multivariate normality, and outliers. Nonmetric multidimensional scaling (NMDS) was performed on all measured soil parameters and fecal matter estimates using Bray distances, three dimensions and a maximum of 100 random starts in R package “vegan” (R Core Development Team, 2016). Treatment groups (LF, LN, CF, CN) were used as categorical variables to delimit 95% confidence ellipse plots. Vectors on plots represented correlations between NMDS axes and soil variables. Bray dissimilarity distances (999 permutations) were used in an Analysis of Distance Matrices (ADONIS) as a robust equivalent to a multivariate analysis of variance (PerMANOVA) with fecal matter and treatment group serving as a predictors. P-values were adjusted with post-hoc pairwise comparisons using Bonferroni correction.

 Parameters demonstrating influence in treatment comparisons on NMDS graphs and meeting normality assumptions were tested for differences in treatment groups with analysis of covariance (ANCOVA) using type III sums of squares in program R (R Core Development Team, 2016). Separate individual models were developed to investigate the effect of predictor variables of bird use (continuous) and treatment group (categorical) on response variable. The fall 2017 measurement was used as a covariate (continuous) to account for initial differences between fields. Model fit was assessed using adjusted R², Akaike information criterion (AIC), Cook’s D outlier detection, and distribution of residuals. Tukey’s post-hoc test was performed on the final model for multiple comparisons of means.

Parameters demonstrating influence on the NMDS graph but violating normality and/or homogeneity of variance assumptions were analyzed with the nonparametric rank-based ANCOVA equivalent using packages ‘Rfit” and ‘npsm’ in program R (McKean and Kloke, 2014). Model fit was assessed by robust R² values. Nonparametric models were reported and subjected to post-hoc Jonckheere-Terpstra distribution free ordered alternatives test to examine the hypothesis that at least one strict inequality existed. Additionally, Tukey-Kramer pairwise comparisons were performed post-hoc on the nonparametric one-way design, excluding fall measurement covariate and fecal matter predictor.  

Objective 2: Pathogens

Composite soil samples were tested for the following pathogenic or fecal indicator bacteria using modified methods from Brooks et al. (2009, 2010):  Enterococci, Clostridium perfringens, Salmonella, Campylobacter and Escherichia coli. Prior to assay, a 10 g aliquot was mixed in 95 ml sterile saline and homogenized via stomacher, whereby 10 fold serial dilutions were utilized for assays. Enterococci was membrane filtered through a 0.45 μm filter (Millipore; Billerica, MA) and subsequently transferred to m-Enterococcus agar (Neogen; 48 h at 35 °C). Filters suspected of containing Enterococcus were subsequently transferred to bile-esculin agar (Neogen; 1 h at 35 °C). Typical esculin hydrolyzing, black-haloed colonies were suspected as Enterococci.

Clostridium perfringens was also membrane filtered and transferred to CP Chromoselect agar (Sigma-Aldrich, MO) and incubated anaerobically at 44.5 °C for 16 h. Prior to filtration, each sample aliquot was heat-shocked at 70 °C for 10 min. Following anaerobic incubation, plates with suspected colonies were exposed to aerobic conditions for 1 h at 44.5 C. Colonies that turned “mucus-green” or blue/green were presumed C. perfringens. Five percent of colonies were transferred to 5% sheep-blood tryptic soy agar (HealthLink; Boca Raton, FL). Colonies exhibiting a double-zone of hemolysis were presumed C. perfringens.

E. coli were membrane filtered and transferred to mTEC agar. Plates were held at 35 °C for 2 h then transferred to 44.5 °C for 24 h. Filters were suspected of containing E. coli if they contained bright yellow colonies. Colonies were transferred to MacConkey agar and incubated at 35 °C for 24 h.

Salmonella and Campylobacter were assayed via presence/absence enrichment because of expected low levels in the soil. Salmonella assays consisted of incubating aliquots in tryptic soy broth (Neogen; 24 h at 35 °C), followed by transfer to Rappaport Vasilidales R10 semisolid broth (42 °C for 24 h), and then transferred to Hektoen Enteric agar (Neogen; 42 °C for 16 h). Black-centered, blue-green colonies were considered Salmonella positive. Campylobacter enrichment consists of Campylobacter enrichment broth incubated microaerophillically (35 °C for 4 h) then transferred to 42 °C for 44 h. Aliquots were transferred to Preston Agar (Neogen-Accumedia) containing 5% horse blood at 42 °C for 48 h (Hema-Resources; Aurora, OR) and checked for growth (Brooks et al., 2010). 

Prior to analysis, pathogen data was log transformed then checked for general multivariate frequentist test assumptions including, but not limited to, Barlett’s test for homogeneity of variance, examination of q-q plots and Shapiro’s test for multivariate normality, double zeros in pathogen detection and Cook’s D outlier exploration. Enterococci, Clostridium perfringens, and Escherichia coli data were found to violate the assumptions of normality and could not be corrected.  Salmonella and Campylobacter were not detected in soil samples and thus were not included in statistical analysis.

The null hypothesis of no differences among treatments by pathogen load (Clostridium perfringens and Escherichia coli.) was tested with analysis of variance using distance matrices (ADONIS) in program R’s vegan package as the multivariate response variable (Anderson, 2001). Fecal matter and treatment were used as predictor variables with 999 permutations and Euclidean distances (Oksanen, 2012). Only one Enterococci positive was detected out of the 34 fields, and therefore not included in the multivariate model.

The alternative hypothesis of no difference or lesser incidence of individual pathogen detection in fields was also tested. The non-parametric rank-based ANCOVA in program R’s “Rfit” and “npsm” was applied to Enterococci, Clostridium perfringens, and Escherichia coli, respectively, with the fall measurement as a covariate and fecal matter and treatment fields as predictors. Data were simultaneously run with parametric statistics. If both tests returned the same results, the more robust parametric results were used for interpretation and reported in results. Tukey’s post-hoc test was performed on individual final models for multiple comparisons of means.

 

Objective 3: Water Quality

Forty liters of topsoil (approximately 1 liter from each field) were collected from four study fields, LN, LF, CN, CF. Gravimetric moisture content was calculated by drying soils at 105°C for 24h. Groundwater was added to each soil type until the soil was homogenized into a pourable slurry state.  The slurry soils were distributed by treatment into cylindrical microcosm containers, filled to approximately half of the containers volume, and placed into an environmental chamber to dry to their original moisture content.

Once microcosms were all within 10% of their original soil moisture, they were organized into a randomized block design, with each block containing a microcosm of LF, LN, CN, CF and an empty (E ) control. Four liters of groundwater experimentally reformulated into moderately hard reconstituted water was mixed to contain “high” concentrations of nitrogen and phosphorus (5.67 mg/L and 6.37 mg/L). Treatment water was added to microcosms.

The microcosms were then placed in environmental chambers with temperatures set to mimic winter weather in the Mississippi Delta, approximately 15°C high and a 4°C low. At pre-determined intervals (24, 48 and 168 hours), 30 mL of water from each microcosm was sampled and tested for changes in nutrient concentrations. Samples were analyzed for inorganic and TN in liquid samples, using HACH TNT 880 s-TKN™ kits. Inorganic phosphorus (PO₄) will also be analyzed using HACH TNT 843 and 845 Phosphorus™ kits. Percentage change in nutrient concentrations between time periods were calculated. Preliminary descriptive statistics indicated non-normality of data, which was confirmed with both Kolmogorov-Smirnov and Shapiro-Wilk tests. Because of this, non- parametric Analysis of Variances (ANOVAs) were used to assess nutrient reduction differences between farm treatments at each specified time interval. Post hoc (P) analysis significant alpha value was set at 0.05.

Objective 4: Economics

Participating farmers were asked to complete an economic questionnaire indicating the following:

o   Fertilizer type, application rate/acre

o   Herbicide type, application rate/acre

o   Seed type, cost/acre

o   Planting cost/acre

o   Equipment used, cost/acre

o   Labor/acre

o   Rice yield in 2017

Data was compiled and organized so that LEISA and Conventional farms costs could be compared and net gain per acre was determined to assess economic stability.

 

Research results and discussion:

Objective 1: Soil Health

Results of the 2017-2018 study in the Mississippi Delta yielded data supporting the impact of wintering migratory birds’ fecal matter inputs on the microbial activity and diversity in rice soils, and the nutrient and structural additions migratory birds provide to rice soils.

Experimentally flooded rice amassed approximately 2.8 times more fecal inputs in LF fields than in the LN fields. First-year flooded fields from the conventional farm (CF) collected 2.5 times more fecal matter than their non-flooded counterpart. LF fields had statistically greater fecal matter inputs than the conventional fields. Previous studies have shown that birds prefer rice fields that leave standing rice straw over the non-growing season because it increases that mount of available waste rice in comparison to fields that practice rice stubble incorporation (Kross et al., 2008). As the LEISA farm practices no-till, this will likely increase their bird activity.

This study correlates fecal matter inputs to a positive increase in microbial activity and diversity in rice soils. Estimates of LF’s mean fecal matter inputs ranged approximately 1.5-2.3 kg of dry fecal matter/day/ha compared to CF fields ranging from 0-1.1 kg of dry fecal matter/day/ha. This stark difference aligns with the response seen in anaerobic HPC means of field types and NMDS ordination of soil health indicators, depicting fecal matter to be a potential major driving factor of LF’s difference from the other field treatments. 

Measures of microbial activity through the dehydrogenase assay also confer with these findings, showing the LEISA fields’ mean response twice as active as conventional fields. Bird fecal matter supplies N from uric acid excreted in the ammonium form, which in chicken litter can be as high as 21% N (Hadas and Rosenberg, 1992). Although Manny et al. (1975) estimates a 4.38% N per bird dropping, Post et al. (2008) reports an individual goose can contribute 3.15g N to an area over a 24h period (Post et al., 2008). Using the Post et al. (2008) estimate as a low threshold and the Hadas and Rosenberg (1992) estimate as the upper threshold to calculate the potential N available, LF could be collecting anywhere between 5.0-48.9 kg N/ ha per season.

The Mississippi State University Extension’s Rice Growers Guide recommends to farmers a total application rate of 133-201 kg/ha of nitrogen fertilizer. Theoretically, the upper end of the LF fecal input estimate could replace almost 1/3 of the total recommended nitrogen fertilizer application, providing economic incentive to the farmer to winter flood. The significantly higher %TN content found in LF fields when compared to the other treatment fields correlates with the estimated fecal inputs per field, suggesting that bird activity on those fields is impacting N loads. The significant increase in N in the LF fields as compared to LN, both of which practice the same post-harvest no-till, further points to bird activity effecting N levels. The statistically similar %TN in CF compared to CN suggests that one year of flooding and bird activity is not enough time to build up N reservoirs in the soil to have an impact in soil fertility.  Further studies should observe the change in soil characteristics in fields flooded with migratory birds present over multiple seasons.

Both pre- and post- winter flooding results show LF fields with significantly higher %C than CN, significantly different at the p=0.10 level from CF, but not statistically higher than LN. Rice straw is 40% carbon by dry weight and has a high C:N ratio (Greenland, 1984). The LEISA system practices no-till management, which is known to increase soil organic C and mimic the process of soil carbon sequestration in natural systems, as shown by the high %C mean value of 2.11% in LF fields (Sylvia et al., 2005). Bird activity could also contribute to observed results of %C in experiment fields. Water birds in rice fields break up standing rice straw and incorporate it into the soil as they forage for food (Manley, 1999). Higher densities of birds will increase the contact rice stubble has with soil and consequently increase %C. Moreover, goose fecal matter is 75-85% dry C, increasing the amount of C input to fields as bird levels rise (Manny et al., 1975). Using mean LF fields as an example, birds could contribute somewhere between 144.47-1639.93g/ha of carbon per LF field per day. Thus, the largest bird response reported in LF fields can partially account for the significant increase in %C in the LF fields. 

While the soil C results indicated a positive effect that LF system is having on carbon sequestered in soil, and consequently overall soil health, the management style is predicted to have the highest methane emissions than the other treatment fields, which would be a substantial environmental drawback of the LEISA system. Rice production creates a carbon conundrum that consists of a series of tradeoffs between soil health and greenhouse gas emissions that neither the LEISA nor conventional system resolve. It is possible that intermittent drainage of rice fields could help mitigate the effects of emissions; however, this will likely have a large impact on the habitat and soil contributions of water bird communities in flooded fields (Haque et al., 2017). Further research on the interaction of this complex system is needed to make future environmentally conscience management decisions.

Objective 2: Pathogen Detection

Soil samples from the study fields were below detection limits (approximately 10/g or 100/10g) for Salmonella or Campylobacter, two pathogens that are known to have detrimental effects on public health (Hubálek and Hubá, 2004; Waldenström et al., 2002b). Incidence of E. coli and Enterococci were at levels well below the EPA standard of fecal contamination, with the greatest prevalence of E.coli occurring in conventional non-flooded (CN) rice fields (Mean= 2.8 CFU/g dry soil) and only one detection of Enterococci occurring in conventional flooded (CF) fields (mean= 1.7 CFU/g dry soil) (EPA, 2012). The low detection rate of the aforementioned indicators is a positive sign that winter flooded fields for waterfowl may not be a problem in the LEISA systems.

Clostridium perfringens, a spore forming, gram-positive bacterium, was detected at significantly higher levels in LF fields than those of CN, and while not statistically significant, observably higher levels than that of LN or CF fields. The difference in detection of C. perfringens loads between LF and CN correlates strongly with bird fecal matter inputs, with LF’s mean fecal matter weight being 7.4 times larger than that of CN (Table 2) and mean C. perfringens 6.3 times larger in LF than CN (Table 2). It could be argued that C. perfringens needs anaerobic conditions for enrichment and thus would not thrive in well-aerated soil; although it is a spore forming bacterium and thus would survive for prolonged periods of time. However, the LN fields, also an aerobic environment, yielded a mean of 1883.64 CFU/g dry soil, almost 3.8 times higher than CN while additionally supporting 2.6 times more birds on its’ fields, indicating that flooding conditions alone are not responsible for this significant difference in C. perfringens presence (Table 2, 3). Some bacterial indicators are better suited for monitoring than others, and C. perfingens may not be ideal indicator for this system because the environmental conditions promote its survival. However, while it may fail as an indicator for public health, it does show that the fields with bird activity are being influenced by fecal deposition.

Objective 3: Water Quality

Preliminary results of ANOVA for time period 0- 24 hours showed no significant differences for TN (P>0.05). Results for TIP showed significant differences between farms (P=.001). Post-hoc pairwise comparison results for TIP showed LF to be significantly different from E (p=0.004) and LN to also be significantly different from E (p=.012). Kruskall-Wallis test showed significant difference between both TN (p=.001) and TIP (.004) for time period 24-48 hours. Post-hoc pairwise comparison results for TN showed LF to be significantly different from CF (p=0.036) and E (p=0.001). Pairwise comparisons of TIP also showed LF to be significantly different from E (p=.002). No significant differences between farm treatments were found for time period 48-168 hours for TN or TIP (P>0.05).

Results moderately supported hypotheses that LEISA soils (LF and LN) would improve nutrient removal from overlying water, as significant differences between LF and the control (E) were observed after 24 hours for TIP, and significant differences were observed after 48 hours between LF and E and CF for TN and between LF and E for TIP. Summary statistics also indicate greater mean and median nutrient reductions after 48 hours for LF and LN. Results indicate that, in LEISA soils, most nutrient removal occurred between 0 and 48 hours, whereas nutrient reduction in conventional soils increased during the 48-168 time interval.

We would expect the LEISA soils (LF and LN) to impact nutrient concentrations differently due to the minimal tillage practices, and addition of organic wastes via avian defecation that have been applied to these soils, which will increase microbial activity as well as organic fertilizer, during their life history.  Unlike LEISA soils, the conventional soils (CF and CN) have a life history of regular tillage. Intensive tillage practices fracture the soil, disturbs soil structure, and can accelerate surface runoff and soil erosion (Al-Kaisi, Hanna, Tidman). Nutrient cycling is facilitated by soil organisms, which require adequate environmental conditions to thrive. Increases in TIP in certain samples are likely a result of anaerobic conditions causing dissolution of P from soil back into the overlying water.

In general, the results moderately support the hypotheses, however there were soils came from two farms and replication was limited to 5 replicates of each treatment. Potential future researchers may consider utilizing intact soil cores, considering different time intervals, and investigating more complex modelling of the data. Results of ANOVA analysis are preliminary. We anticipate applying a mixed model analysis to include an additional water treatment to thoroughly investigate data.

Objective 4: Economics

The conventional farm reported fertilizer, chemicals, seed and application costs as a lump sum per acre, while LEISA priced each category out ($41.24, $30, $15, $24, respectively). While individual categories weren’t available to compare between farms, LEISA reported spending less, leading one to conclude that they use less fertilizer and chemicals, as well as spend less on actual planting. LEISA also reported spending less on equipment operation than the conventional farm ($90 vs $134.92), but more on labor per acre ($35 vs $18.83). LEISA reported an average yield of 150 bu of rice/acre, while conventional reported 192 bu/acre. Thus, at $4.60/bu, LEISA sold an acre’s worth of rice for approximately $690 and conventional for $883.20 in 2017. Factoring in farming expenses, however, LEISA’s net gain was approximately $100 more per acre of rice than the conventional farm ($454.75 vs $353.45). There are some specific economic parameters that are important to calculating exact cost estimates that producers did not have recorded, limiting depth of the economic analysis. These are preliminary analysis and we intend to reanalyze and supplement with published cost-estimate averages where available to finalize data. Tables & Figures; References; Poster Presentation

Participation Summary
2 Farmers participating in research

Educational & Outreach Activities

Participation Summary

2 Farmers
Education/outreach description:

Four presentations were given at various scientific meetings, one thesis presentations was successfully defended, one thesis was completed, and four manuscripts are being prepared for publication. The presentations and thesis are listed below.

Firth, A. G. (Author & Presenter), Baker, B. (Author), Brooks, J. P., Smith, R. (Author), (April 2, 2019). “Ecological Agriculture Application with Winter Flooding.” Oral Presentation. Mississippi Water Resources Conference, Mississippi Water Resources Research Institute, Jackson, MS. Abstract Accepted.

Firth, A.G., 2018. Investigation of a low-external-input sustainable rice production system to identify ecosystem services towards adoption costs and benefits (Thesis).  Mississippi State University Library.

Firth, A.G. (October 2, 2018). “Investigation of a low-external-input sustainable rice production system to identify ecosystem services towards adoption costs and benefits.” Thesis defense, Mississippi State University.

Firth, A. G. (Author & Presenter), Brooks, J. P. (Author), Baker, B. (Author), (July 30, 2018). “Ecological Agriculture: Application of Winter Flooding.” Poster. Soil and Water Conservation Society Meeting, Soil and Water Conservation Society, Albuquerque, NM.

McKnight, C. (Author & Presenter), Baker, B. (Author), Firth, A. G. (Author), Hall, K. D. (Author), Hamid, K. (Author), (July 26, 2018). “Investigation of soil management practice impacts on water quality.” Poster. Mississippi Academy of Sciences, Mississippi Academy of Sciences, Bost Extension Center.

Firth, A. G. (Author & Presenter), Baker, B. (Author), Brooks, J. P., (April 2, 2018). “Ecological Agriculture Application with Winter Flooding.” Poster. Mississippi Water Resources Conference, Mississippi Water Resources Research Institute, Jackson, MS.

Project Outcomes

2 New working collaborations
Knowledge Gained:

During the course of this project, the value of having a one-on-one interaction with our farmer participants was crucial to both the success of the experiment and for productive communication of information. Farmers know more about their land than any outsider may begin to gleam. In order to establish new practices, or to learn about the reasons behind landowners’ current practices, we had to create a relationship where each party felt comfortable sharing their values and objectives. It appeared that most of the time, farmers had already started adopting sustainable practices or were very interested in doing so, but were looking for consistent and reliable information to support their management ideas.

It also became clear that landscape specific research will yield the best results when designing sustainable solutions. There isn’t a “catch-all” solution to the challenges we face with the environment. What worked on the LEISA farm may not work on all rice farms throughout the LMV simply because of logistics. However, the basic principles that LEISA employed, i.e. evaluating the landscape as a whole and making use of seasonal patterns, is the most effective way to inspire sustainable change.

This project encouraged interdisciplinary collaboration, which we have come to rely on when dealing with systems research. It also led us to pursue further questions and created additional opportunities with individuals and organizations in the state. 

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.