Hopes of dry land: Managing soils to improve fruit yield and quality in dry farm tomatoes

Progress report for GW21-224

Project Type: Graduate Student
Funds awarded in 2021: $25,243.00
Projected End Date: 08/31/2023
Host Institution Award ID: G233-22-W8615
Grant Recipient: UC Berkeley
Region: Western
State: California
Graduate Student:
Major Professor:
Dr. Timothy Bowles
University of California Berkeley
Expand All

Project Information


As farmers confront the fragility of California’s water system, a crucial question emerges: how can farmers adapt to water scarcity without jeopardizing their livelihoods? Innovative farmers have turned to dry farming, a method of growing produce with little to no external water inputs, and a radical departure from irrigated agriculture. Conversations with six such farmers in the Central Coast region reveal strong interest in research that evaluates the soil microbial and nutrient factors that can best support dry farm crops.

We propose to assess how farmers can shape soil microbial communities through fungal inoculants and nutrient management in ways that enhance dry farm tomatoes’ performance. Recent trials have established fungal inoculants as tools to alleviate crop water stress, making them ideal candidates for dry farm soils; however, high soil nutrient concentrations deter plant-fungal symbioses. Through participatory research we will measure soil nutrients and plant-fungi symbioses to determine whether inoculation can profitably improve fruit yields and quality, and under what soil nutrient conditions.

We will also help facilitate a dry farming community of practice by collaborating with participating producers and local ag advisors to organize a producer workshop, encouraging attendance among experienced and potential new dry farmers in addition to local land and groundwater policy stakeholders. Our findings will be shared through presentations at conferences, and in videos, factsheets, op-eds, and a peer-reviewed publication. Through increased awareness and improved techniques, we hope to grow the practice of dry farming, preserving water as we build the dry farm community.

Project Objectives:
  1. Coordinate field trials through on-farm participatory research to engage farmers in research design and outcomes. In Spring 2021 we began on-farm trials on seven dry farm tomato fields after having engaged in a participatory design process with dry farmers on California’s Central Coast. These farmers have a shared interest in identifying management practices that can limit blossom-end rot and increase yields, leading us to focus our study on how to manage dry farm soils to encourage fungal symbionts that show potential to improve both fruit yields and quality. The project allowed us to collect labor-intensive harvest data from the indeterminate tomato varieties grown on the fields, and to analyze soil/root samples to provide farmers with information on how to best manage dry farm soils to support beneficial fungal communities.
  2. Characterize fungal communities and quantify the utility of fungal inoculants. Our project builds on promising preliminary trial results with arbuscular mycorrhizal fungi. We set up five inoculated and five non-inoculated plots in each field to determine which native fungal species are present and deepen our understanding of the relationship between fungal diversity and improved farm production, while testing whether fungal inoculation has the potential to improve tomato performance on dry farms. 
  3. Identify soil management strategies that enhance fruit yield and quality, and evaluate their profitability. Because high nutrient concentrations can limit plant-fungi associations, we tracked soil nutrient levels (nitrogen and phosphorus, both available and total) to pair with fungal and harvest data. Of particular interest is nutrient depth, as some farmers go to great lengths to deliver nutrients deeper into the soil profile in dry farm systems. Also of interest is field history—i.e., number of years since most recent irrigation—as previously dry soils may prime native beneficial microbial communities. We will identify the soil management strategies (inoculation, nutrient depth, etc.) that show the strongest yield/quality improvements, and compare farm revenue generated by each practice to expected management costs to determine profitability.
  4. Engage Central Coast farmers and stakeholders in collaborative learning about dry farm soil management. We maintained a blog over the course of the field trials to build connections between participants and maintain engagement in the research process. In the second year of the project, after data have been analyzed, our producer workshop will allow farmers to connect with research results, one another, and relevant land and water stakeholders.
  5. Disseminate findings to build the dry farm community and public engagement. A video we produce, two conferences, a factsheet, and the producer workshop will help us communicate our results to producers, improving dry farm soil management, lowering barriers to beginning dry farmers, and introducing new farmers to the practice. A policy memo and a publication for a general audience will increase public interest in dry farming, as well as potentially shaping policy to incentivize the practice.

In September and October 2021 I will complete field trial harvests. I will then process and analyze soil, root, and fruit samples from October through Spring 2022. After samples are processed, I will analyze data to round out the first year of the project. As I obtain results from data, I will work to create the inoculation factsheet and draft the soil management video.

In the second project year, I will develop and host the producer workshop in Fall 2022/Winter 2023 and will follow up with attitude and adoption surveys that same winter. During this time I will also work on the peer-reviewed publication and middle school classroom presentation. Following the workshop, I will present at the EcoFarm Conference and CalCAN summit. I will finish the project in Spring/Summer 2023 by filming and editing the dry farm soil management video and authoring general-audience publications. Full timeline details in Table 1.


Table 1. Timeline.

Activity Academic Year 2021-2022 2022-2023
Fall, Winter, Spring, Summer F W Sp Su F W Sp Su
Objective 1: Participatory field trials
Complete on-farm field trials x              
Objective 2: Characterize fungal communities and quantify inoculation benefits
Harvest tomatoes and assess quality x              
Process and analyze root samples x x x          
Objective 3: Identify best practices for soil management
Collect and process soil samples x              
Soil nutrient analyses x x x          
Data analysis       x        
Objective 4: Engage farmers and stakeholders in collaborative learning
Maintain weekly blog x              
Workshop and survey development         x x    
Send workshop invitations         x      
Workshop day           x    
Send post-workshop survey           x    
Objective 5: Disseminate findings to dry farm community and public
Prepare and disseminate factsheet       x x      
Prepare, film, and edit video       x     x x
Prepare scholarly peer-reviewed article         x x x  
Give classroom presentation         x      
Present at EcoFarm and CalCAN           x x  
Compose policy memo and general-audience piece               x


Click linked name(s) to expand/collapse or show everyone's info
  • Thomas Broz - Producer
  • James Cook - Producer
  • Jim Leap - Technical Advisor - Producer (Educator)
  • Larry Palla - Producer
  • Joel Schirmer - Producer
  • Verónica Mazariegos-Anastassiou
  • Darryl Wong


Materials and methods:

All field and lab work will be carried out in accordance with UC Berkeley COVID protocols.


Objective 1: Coordinate field trials through on-farm participatory research to engage farmers in research design and outcomes

In Fall 2020-Spring 2021, I interviewed eleven Central Coast farmers to better understand dry farming and identify research priorities for farms in the area. All farmers either identified fungal inoculants as an area of keen interest or were excited when I suggested it as a research option, yet none had previously experimented with applying fungal inoculants on their operations. Farmers were additionally curious about broader questions of what soil health means in the context of dry farming, where field outcomes often shirk conventional wisdom about best practices. Of the eleven, eight farmers (representing six farms) had operations large enough to accommodate field trials, and all were enthusiastic to join the project.

During field visits to each farm, I collaborated with growers to develop a research plan that fits the system and addresses conditions and questions relevant to each participating farm. We decided that each farm could feasibly host ten 12-plant plots, which I then manipulated and sampled from (growers had no change to their management of these plots compared to the rest of the field). After my initial intervention to inoculate half of the plots, I sampled soils (see Objective 2) and collected yield data to give farmers desired information on the potential benefits of fungal inoculation and other soil management techniques (see Objective 3). 

Funding from Western SARE has allowed me to collect the full set of data that farmers have identified as being of particular interest: soil nutrients, inoculant efficacy, field fungal communities, yields, and fruit quality (see Objective 2). I continued communicating closely with farmers over the course of the growing season and after sampling was complete to ensure that their observations about the plots are included in data interpretation and further refinement of research questions. I also shared preliminary data with farmers, which was well-received; all farmers indicated that they would use the information provided in subsequent management decisions.


Project Team Members: Myself (YS); Advisor Tim Bowles (TB); Ag Advisor Jim Leap (JL); Producers Joel Schirmer, Thomas Broz, James Cook, Larry Palla, Barbara Palla, Verónica Mazariegos-Anastassiou, and Darryl Wong. 

Project Team Roles: 

  • Identify research questions of interest: YS, JL, Producers
  • Preliminary experimental design: YS, TB
  • Iterative experimental refining: YS, TB, Producers


Objective 2: Characterize fungal communities and quantify the utility of fungal inoculants 

Soil management shapes fungal communities, which in turn impact agricultural yields and crop quality. Practices such as cover cropping, crop rotation, and fertilization contribute to mycorrhizal fungal diversity, and the resulting communities vary in hyphal growth rate, turnover rate, spore production, and more. Interactions between these traits, host plants, and the abiotic environment influence how “helpful” a mycorrhizal community will be, with more diverse communities showing the best yield outcomes.  In addition, a history of dry soils (i.e., lack of irrigation) in a field may prime native beneficial microbial communities. Farmers are therefore interested in understanding which management practices (inoculation, rotational history, fertilization, etc.) will produce diverse fungal communities that are best suited to the crops growing in their fields. Our research explores these questions on seven dry farm fields on California’s Central Coast. These fields grow Early Girl, New Girl, and Dirty Girl tomatoes and belong to six producer collaborators (one farm is contributing two fields). 

I used a nested experimental design to account for management and biophysical differences across fields, establishing 10 plots at each field site (70 plots total). Experimental plots (12 plants each) were distributed across two adjacent rows with or without a buffer row between them, depending on the size of the field (Figure 1). Each experimental row began and ended with at least ten buffer plants, and there were three buffer plants between experimental plots. Each experimental plot ranged from 13-44 ft in length depending on the spacing between each plant, but always contained 12 plants at transplant. Buffers between plots were similarly variable in length (3-8 ft), and always contained 3 plants at transplant. Inoculation was distributed in a split-plot design to account for heterogeneity in biophysical environments across the field. To accomplish this, each of the plots in rows 2 and 4 was paired with its closest counterpart in the opposite row (a “pseudo-block”), and inoculant was randomly applied to one of the plots in each pair within three days of when farmers transplanted their first dry farm tomatoes (April-May). 

diagram of example experimental layout
Figure 1. Example experimental layout.

I inoculated with Valent’s MycoApply Ultrafine Endo, a mix of four species of arbuscular mycorrhizal fungi: Glomus intraradices, Glomus mosseae, Glomus aggregatum, and Glomus etunicatum. These species are known for having an aggressive growth strategy, which can benefit crops by establishing quickly but may also disrupt already high-functioning native communities. This inoculum is one of several commercially available mycorrhizal inoculants and was shown to decrease water stress in preliminary field trials with processing tomatoes in Summer 2020. 

I inoculated experimental plants by mixing the inoculum with water to create a slurry that can be poured at the base of each plant. Valent confirmed that this is an effective inoculation technique, though several others can be used for ease of application (e.g., drenching root balls in a water-inoculum mixture before transplanting).

Mid-way through the season, I sampled one plant per plot to collect roots, which were split into two sample sets: one for fungal colonization quantification, and one for DNA analysis to determine which fungal species are colonizing the roots. Roots for colonization were stained and dyed, and fungal presence is being quantified via microscopy. Roots for DNA analysis were ground and extracted with a kit, and DNA was quantified. These samples have been sent to the University of Minnesota's Genomic Center for Illumina MiSeq sequencing using ITS2 primers. Fungal community data will be used to determine whether the inoculated species are in fact colonizing plant roots, and for further explorations of soil health (see Objective 3).

I will use linear mixed effect regressions to model colonization as a function of inoculation, fungal species diversity (Shannon index), and their interaction, with random effects of field and pseudo-block to account for management differences and spatial heterogeneity. This analysis will allow farmers to determine whether inoculation was successful in encouraging plant-fungi interactions, and if more diverse fungal communities limit inoculation benefits. I will perform further fungal analyses with data described in Objective 3.


Project Team Members: Myself (YS), Producers, two Undergraduate Assistants (UA)

Project Team Roles: 

  • Plot setup: YS, Producers
  • Sample collection: YS
  • Root analyses (colonization and DNA): YS, UA
  • Communication of results: YS, Producers


Objective 3: Identify soil management strategies that enhance fruit yield and quality and evaluate their profitability

Though irrigated plants usually cluster their roots at the surface of the soil (close to the water source), dry farm tomatoes send their roots into deeper soils to scavenge for water and nutrients (40). It was therefore important to collect data about nutrients across the soil profile to determine which nutrient depths are most influential for dry farm tomatoes. Because high levels of phosphorus (P) and nitrogen (N) can also inhibit plant-fungus associations, it is be important to consider both the minimum and maximum nutrient thresholds for optimal dry farming performance.

Soil samples were taken at three points over the course of the field season: once at transplant (within two weeks of plant date), once mid-season (9 weeks after transplant), and once during harvest (18 weeks after transplant).  I sampled at four depths (0-15cm, 15-30cm, 30-60cm, 60-100cm) at all soil sampling events. In the lab, samples were stabilized (air dried or KCl extractions) and stored for further analysis.

WSARE funds allowed me to test soil samples for available N (nitrate and ammonium) and available P (Olsen P) via colorimetry. I also measured gravimetric water content from each soil sample to track water availability in each plot throughout the top 100cm of the soil profile. 

When fruits began to mature, I coordinated with farmers to harvest and weigh the tomatoes from each plot once per week. Funds from WSARE allowed me to hire two undergraduate students to assist with the harvest, making it possible to conduct weekly harvests for the ~10 weeks of peak production. Tomatoes from each plot were sorted into marketable, blossom-end rot, and sunburnt fruit, and then weighed. I also collected marketable fruit samples from each plot at harvests 3, 6, and 9 to test for tomato quality via fruit water content.

I will use linear mixed effect regression to model each response variable (total marketable fruit, fruit lost to blossom-end rot, fruit water content) as a function of fixed effects from inoculation, soil nutrients (N, P, C, and N:P ratio at each sampling depth), soil water content, hyphal density, field legacy (years since last irrigation), and fungal species diversity (from Objective 2). Random effects of field and pseudo-block will account for unmeasured attributes such as climate, management differences, and spatial heterogeneity in field soils. I will treat marketable and blossom-end rot fruit as repeated measures due to the 10+ consecutive weeks of sampling. I will then use Akaike Information Criteria to determine which combination of the above fixed effects provide the best-fitting model.

That best-fit model will contain vital information for farmers about which soil management practices to target for maximum yields and fruit quality. The nutrient depths that show the strongest correlations to outcome variables will indicate which depths farmers should target for fertilizer additions, and their signs may show a potential for over-fertilization. Interactions between inoculation and nutrient levels would indicate a context dependence of inoculation benefit, allowing farmers to quickly test soils in advance of inoculant application to determine its potential utility in their fields. Strong effects of fungal species diversity and/or hyphal density that do not interact with inoculation would suggest mycorrhizal contributions that cannot be supplemented with external inoculants, but must instead be fostered through on-farm diversification practices.

This study will give farmers crucial insight into the abiotic and biotic components of high-functioning soils that can boost dry farm productivity, and how these soils can be fostered through on-farm management. We will use our yield and fruit quality data to estimate potential income gains associated with these management practices, which we will compare to implementation costs (e.g., cost of inoculation) to determine profitability.  


Project Team Members: Myself (YS), Advisor Tim Bowles (TB), Producers, two Undergraduate Assistants (UA)

Project Team Roles: 

  • Sample collection: YS, UA (at harvest)
  • Fruit harvest: YS, UA
  • Soil lab analyses (N, P, C, hyphal density, gravimetric water content): YS, UA
  • Data analysis: YS, TB
  • Communication of results: YS, Producers
Research results and discussion:

After spending the past year gathering samples and analyzing them in the laboratory, I am currently in the data analysis phase and do not yet have final results to share. However, while I wait for DNA results from the sequencing facility, several preliminary patterns have emerged from the data I currently have available. 


First, fruit harvests lasted 10-20 weeks and tended to peak 3-5 weeks after the first harvest (Figure 1). These peaks coincided with when fruit quality was at its highest, as measured by fruit percent dry mass (optimal range is 8-12% dry mass by weight; Figure 2). Because there are very few conventionally farmed tomatoes in the Central Coast region due to its cool, moist climate, it is difficult to compare these numbers to what might be found in a conventional system. The cumulative marketable harvest for an individual field ranged from 175 - 542 kg per field, or 1.5 - 4.5 kg per plant, with an average of 315 kg per field and 2.6 kg per plant. Over the course of the season, on average 4.7% (sd = 11%) of the fruits harvested from each experimental plot were unmarketable due to blossom end rot.

graph of tomato yields

Figure 1. Marketable fruit yield per experimental plot in weekly harvests.


graph of fruit quality

Figure 2. Tomato fruit quality in each experimental plot as measured by percent fruit dry matter (by weight).


Second, after collecting nutrient, water, and texture data for each of the four soil depths at transplant, midseason, and harvest (see Table 1), I have begun preliminary statistical modeling. Note that these models lack fungal diversity data that will be added after it is available from the sequencing facility.




Number PCs retained

Nutrients, 0-15cm

Nitrate, Transplant

Nitrate, Midseason

Nitrate, Harvest 

Ammonium, Transplant

Ammonium, Midseason

Ammonium, Harvest 

Phosphate, Transplant

Phosphate, Midseason

N:P, Transplant

N:P, Midseason


Nutrients, 15-30cm

Nitrate, Transplant

Ammonium, Transplant

Phosphate, Transplant

N:P, Transplant


Nutrients, 30-60cm

Nitrate, Transplant

Nitrate, Midseason

Nitrate, Harvest 

Ammonium, Transplant

Ammonium, Midseason

Ammonium, Harvest 

Phosphate, Transplant

Phosphate, Midseason

N:P, Transplant

N:P, Midseason


Nutrients, 60-100cm

Nitrate, Transplant

Ammonium, Transplant

Phosphate, Transplant

N:P, Transplant



GWC, 0-15cm, Transplant

GWC, 0-15cm, Midseason 

GWC, 0-15cm, Harvest 

GWC, 15-30cm, Transplant

GWC, 15-30cm, Midseason 

GWC, 15-30cm, Harvest

GWC, 30-60cm, Transplant

GWC, 30-60cm, Midseason 

GWC, 30-60cm, Harvest

GWC, 60-100cm, Transplant

GWC, 60-100cm, Midseason 

GWC, 60-100cm, Harvest



Percent Clay, 0-15cm

Percent Clay, 15-30cm

Percent Clay, 30-60cm

Percent Clay, 60-100cm

Percent Sand, 0-15cm

Percent Sand, 15-30cm

Percent Sand, 30-60cm

Percent Sand, 60-100cm



Table 1. Soil data collected and principal components included in linear mixed models.


Due to the large number of potential covariates (nutrients and water at four depths for each of three sampling events), variables were grouped and summarized by their principal components. The groupings were nutrients at 0-15cm, 15-30cm, 30-60cm, 60-100cm, water, and texture. The variables within each group are listed in Table 1. Enough principal components were included to account for at least 70% of the variance in the data.

In preliminary yield models, harvests were treated as a repeated measure with additional random effects of field and block within field, with the 15 plot-level principal components as fixed effects. Field-level variables (e.g. timing of last water, plant spacing) were not included because models failed to converge when including both block and these variables at higher hierarchical levels (n fields = 7, constricting the degrees of freedom). Inoculation currently does not show a clear correlation with per-plant marketable yields (95% CI = [-0.022, 0.033] kg plant-1 harvest-1), total fruit (95% CI = [-0.024, 0.038] kg plant-1 harvest-1), or blossom end-rot (95% CI = [-0.0032, 0.0034] kg plant-1 harvest-1). Similar results were obtained at the plot (rather than plant) level.

Fruit quality and nutrient models were analyzed with  random effects of field and block within field. For fruit quality, fixed effects of the 15 plot-level principal components were included in addition to inoculation. Inoculation currently does not show a clear relationship with fruit percent dry mass, a proxy for quality, at any of the harvests where fruit water content was measured (for the sixth harvest, 95% CI = [-0.55, 0.14] % dry mass), nor did it have a significant correlation with any of the midseason or harvest nitrate, ammonium, or phosphate levels.

A power analysis using variances from the dataset suggests that our experimental design could detect an inoculation effect size of ~15% of a measured value (e.g. yield) with a power of 0.8. We can therefore conclude that, given our null results, it is improbable that inoculation changed yields by more than +/-15%, and the small confidence intervals centered on zero for inoculation terms suggest that the true effect was  likely less than that if present at all.

Nutrient depth, on the other hand, does show promising correlations with both yield and fruit quality. Principal components from the 30-60cm and 60-100cm show statistically clear relationships with yield and fruit quality (p-values are less than 0.05; exact values will not be finalized until fungal diversity information is added to the model), suggesting that only nutrients present below 30cm impact tomato yield in dry farm systems. These key findings will impact the timing and delivery methods for on-farm nutrient management. 

Participation Summary
8 Producers participating in research

Research Outcomes

1 New working collaborations

Education and Outreach

5 Consultations
1 Curricula, factsheets or educational tools
1 On-farm demonstrations
2 Webinars / talks / presentations
1 Workshop field days

Participation Summary:

95 Farmers participated
59 Ag professionals participated
Education and outreach methods and analyses:

The four main outreach events were:

  1. Direct outreach to farmers participating in the study to outline results from their fields
  2. A session at the Community Alliance with Family Farmers (CAFF) small farms conference 
  3. A presentation at the Lunchtime Organic Agriculture Seminar Series for Growers (hosted by UC Cooperative Extension)
  4. A farm demonstration and learning circle in partnership with American Farmland Trust (AFT).

For the direct outreach to farmers participating in the study outlining results from their fields (7 farmers total), I prepared a packet for each farm showing soil and harvest data that we collected from their field. I spent ~1 hour on each farm discussion preliminary inoculation results, nutrient management, and yield outcomes, and answered questions the farmers had about the study thus far.

For the session at the Community Alliance with Family Farmers (CAFF) small farms conference I gave a virtual presentation titled "Managing mycorrhizas on your farm" along with three coauthors to an audience of 80 farmers and agricultural professionals.

For the presentation at the Lunchtime Organic Agriculture Seminar Series for Growers I gave a virtual presentation titled "Managing mycorrhizas on your farm" along with three coauthors to an audience of 54 farmers and agricultural professionals.

For the farm demonstration and learning circle, I partnered with Caitlin Joseph at the American Farmland Trust to host an event titled "Dry Farming Techniques for Small Farm Resilience in California: A Women for the Land Learning Circle" at Brisa Ranch, one of the farms participating in the study. At the event, we discussed dry farming, did demonstrations on the farm, and invited technical assistance providers to share resources. Nine farmers and 12 agricultural professionals were in attendance.

Information Products

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.