Introduction

Plant breeding has historically addressed food insecurity by increasing agricultural yields and productivity1,2,3. The public-private relationships that formed the Green Revolution allowed crop yields to grow exponentially until the turn of the millennium. Globally staple crop (maize, wheat, rice, and soya) yields increased by an average of 124% from 1961–2007, one of the greatest scientific achievements of mankind, but are only predicted to increase by an average of 33% from 2007–20504. Globally, cereal yield growth rates have begun to slow in several staple crops, and public investment in agriculture has begun to decline2,5,6.

With the development of new plant breeding technologies (NPBTs) such as clustered regularly interspaced short palindromic repeats (CRISPR), zinc-finger nuclease, oligonucleotide-directed mutagenesis, cisgenesis and intragenesis, RNA-dependent DNA methylation, grafting on genetically modified (GM) stock root, reverse breeding, agro-infiltration, synthetic genomics, and others; there are uncertainties as to how these technological advancements will affect the future of public and private plant breeding industries. Public plant breeding programs often focus on important crops to society but may not be profitable due to long generation times and/or regional specialization7. Nevertheless, they conduct necessary foundational primary research and train the next generation of plant breeders8,9,10. Private plant breeding programs tend to focus on large global markets that can provide significant financial returns7.

Four significant drivers have been identified, which has resulted in plant breeding evolving from a traditionally publicly dominated sector to one that now heavily relies on the private sector11. The identified drivers were: the commercialization of agriculture, strengthening intellectual property rights, consolidation of the plant breeding industry, and the decline in official development assistance (regional, national, and international). One of the most significant concerns about these trends is that private firms have little incentive to invest in an activity whose benefits accrue over the long run (food security, for example) because plant breeding skills are not company-specific. Historically, the private plant breeding industry has proven highly effective at filtering and scaling promising discoveries originating from public-sector research. Public plant breeding research tends to excel in conducting foundational research, while industry partners are generally more proficient in product development and commercialization.

A survey of 200 public plant breeders across 18 developing countries in 2020 indicated that public breeders are increasing their focus on traits related to climate change and that climate change has exacerbated the breeders’ sense of urgency in addressing biotic and abiotic challenges12. The study also found that financial, technology transfer, and capacity-building bottlenecks remain binding, particularly for minor crops that offer limited incentives for the private sector through public-private partnerships or consortia. A study on global public rice breeding found that there are significant economic and technical obstacles to the implementation of advances in genomics, which can shorten the breeding cycle in public sector breeding programs in developing countries, further driving the wedge between the public and private breeding sectors13. A survey of plant breeders from Australia, Canada, and New Zealand found a trend of strong concern over the lack of funding and the privatization of plant breeding activities amongst public plant breeders14.

The other large-scale survey assessing the status of public and private plant breeding was conducted in United States was conducted in 1994 as part of the United States Department of Agriculture (USDA) National Plant Breeding Study and analyzed the evolution of the human and financial resources devoted to plant breeding research and development15. Human resources were measured in Science Person Years (SYs), defined as “the work done by a person who has responsibility for designing, planning, administering (managing), and conducting plant breeding research, germplasm enhancement, and cultivar development in one year and annual dollar expenditures, were derived from the data collected16. The study concluded that the number of SYs devoted to private plant breeding research and development in the U.S. was expected to grow by 32 SYs per year, while the public sector was expected to decrease by 2.5 SYs per year from 1990–1994. Graduate-level plant breeding programs reported losing 30 positions (6% of the total) from 1980 to the mid-1990s.

Since 1996, overall SYs in the United States have decreased by 10%, and 13 of 16 crop categories reported decreases in SYs in the public plant breeding sector15. It was estimated that 55% of State Agricultural Experiment Stations reported a decline in SYs allocated to the development of cultivars, and 23% experienced no change17. Similarly, 28 U.S. academic institutions had at least one cultivar development program stop within the previous ten years (2006–2016)17. The two most common reasons given for the drop of cultivar development programs were a lack of funding and the retirement of faculty members who were not replaced.

There exists evidence indicating that public and private funding streams in agriculture are to be viewed as complementary, and the private sector does not crowd out public investment18. However, there is also evidence that public and private investments tend to go toward different research sectors19. Historically, public funding to the agricultural sector has tended to be driven by the opinions of public scientists, while private research and development spending often leverages existing public research to create “value-added” extensions that focus on commercially profitable technologies that are easy to patent and/or protect via intellectual property. If this research is not complementary but rather diverging, it is imperative to better understand the ramifications of this divergence with respect to global food security.

These data provide a snapshot of the evolution of human and financial resources from the public to the private plant breeding sector. This shift does not assert a misallocation of funds, as the private industry often can innovate quicker, think more entrepreneurially, and respond to changing market conditions faster. This shift, in which human and financial capital now predominantly flow into the private sector, raises the question: Who will be left to conduct research for the public good where financial gains are difficult to capture?

This study’s impetus was to query public and private plant breeders’ opinions globally regarding the direction they believe each sector is moving. Responses have been collected from plant breeders on over 40+ crops across the globe and include both the public and private sectors. The result is an aggregate perspective of strengths and challenges that can help better understand the trajectory of public and private plant breeding. Historically, public plant breeding programs have long-term, integrated approaches to plant breeding research, germplasm enhancement, and variety development (all three legs of the plant breeding stool), but as funding is reduced, the future is less certain as to areas of focus. Our survey builds off those before it8,15,16,20,21,22 and asks the next logical question, if a divergence between private and public breeding is taking place, what aspects are illustrative of that divergence? It could be that the future role of public plant breeding is to train the next generation of private plant breeders and simply build capacity in the industry. However, knowledge of plant breeding as an impure public good may not always be produced at the socially optimal level by private firms. The public plant breeding sector will yield the largest social returns if it focuses on research directed at carefully identified “problem areas” with clear public good components10.

Here, we survey 583 plant scientists working on 40-plus crops globally to elicit future convergence/divergence of the public and private plant breeding sectors. Specifically, we investigate the comparative roles of public and private plant breeding sectors in fostering innovation, with a particular focus on identifying systemic constraints and facilitators of innovation in the rapidly changing global plant breeding environment. Using a Best-Worst scaling approach and 12 breeding attributes, we find that private and public breeders believe that the public sector should focus on climate change (a public good) and the private sector should focus on disease resistance and yield enhancement. Perhaps most interestingly, our results suggest that only 12% of private plant scientists surveyed believed that public sector breeding should focus on primary research, with most respondents saying that the primary job should be educating the next generation of plant breeders. Combined, this presents a concerning evolution in which the public sector would pivot its focus to education while the private sector continues its focus on traditional traits; thereby leading to a potentially large underutilization of breeding resources targeting climate change resiliency.

Results

Public/private plant science comparison

The market shares highlighted in Table 1 are derived from the RPL subset results for the public and private sectors given in Supplementary Table 2 for the pooled (plant breeders and plant scientists) sample. The pooled RPL results are presented in Supplementary Table 2. The results are relative to the omitted attribute, seed delivery, and indicate that all specified attributes are statistically significant (P < 0.01). The RPL models for each subset (public/private sector, public/private plant breeders, and region) are given in Supplementary Tables 3–6. Market/preference shares, in this sense, can be viewed as what each respondent thought the primary focus of each sector (public/private) should be in the future. Climate change for biotic/abiotic stress is the largest preference share (57.55%) in the public sector, while disease resistance holds the largest preference share (29.82%) in the private sector. These results make financial sense as the private sector can likely obtain patents for disease-related issues, and the opportunity for financial returns is anticipated to be higher. In contrast, climate change and its effects will receive the most attention from plant breeders in the public sector, which is likely a function of forward-thinking for the public good, which may not materialize in economic returns or welfare gains in the immediate future.

Table 1 Preference shares (in %) of focus attributes and their differences derived from the Random Parameter Logit Model
Full size table

Significant differences (P < 0.01) exist between the sectors for climate change, market value-added enhancement of attributes, and seed delivery. The difference between public and private sector preference shares regarding climate change is 32.78% (57.55% and 24.77%, respectively). This is likely due to climate change being viewed as a public good where progress is more long-term, and knowledge gains can be transferable across companies as plant breeders move between employers. Other differences are significant but small in magnitude; the differences for market value-added and seed delivery are less than 5%.

Public/private plant breeding comparison

When analyzing just plant breeders, there is consistency between the views of plant scientists (all observations with plant breeders removed) and the plant breeding subsample, in terms of their thoughts on what the primary focus of each sector (public/private) should be in the future. Like the pooled sample, climate change for biotic/abiotic stress is the largest preference share (57.30%) for public breeding, while disease resistance holds the largest preference share (34.08%) in the private sector, according to plant breeders. Similarly, the second largest areas of focus were disease resistance (public breeding) and climate change for biotic/abiotic stress (private), identical to the pooled sample. The only difference between the preference shares of the plant science vs the plant breeder subsample is that in the pooled sample, there was a statistical difference (P < 0.01) between Market Value-Added Enhancement of Attributes, where the plant science sample indicated a higher (4.63%) market share for the private sector. The other difference between subsets was that in the plant-breeding sample, the private industry had a statistically (P < 0.1) higher market share (5.49%) for breeding for pest resistance.

Regional comparison

The results were derived from the RPL models in Supplementary Tables 5 and 6. Regional head-to-head comparisons and the difference between a specific region and the rest of the world were also of interest to this study, given the vast differences in the research areas of focus for plant breeders globally. Preference shares were calculated for the rest of the world, which included all regions except the one used for comparison. For example, preference shares were calculated for the world (without Africa) to find differences between Africa and the rest of the world. This resulted in three additional subsets (all responses excluding North America, all responses excluding Europe, and all responses excluding Africa). These values are given in Table 2.

Table 2 Percentage differences between preference shares of future research focus attributes by region derived from the Random Parameter Logit Model
Full size table

Climate change was internalized as the greatest area of focus across all regions. The preference shares for climate-focused research were over 50% for North America (61.07%) and Europe (54.95%). Disease resistance has the second-largest area of focus across all regions, except for Africa, where yield enhancement had the second-largest preference share (20.01%). The African result is expected as increasing food security via yield enhancement is likely the thrust of many breeders’ focus in Africa, whereas the rest of the world may focus more on yield maintenance breeding. The largest difference (P < 0.01) was between North America and the rest of the world, where the preference share for climate change is 13.16% larger in North America (Table 2). Interestingly, significant differences (P < 0.05) regarding CRISPR are minimal (<1.5%) and only occur in half of the comparisons.

Climate change

In total, 79% of respondents chose climate change as the most important focus area at least once during the BW choice sets. Ideally, we would have included each biotic and abiotic stress in the BW choice options, but given the number of potential choices, this would have resulted in a survey that would have been too long, given the fact that there was no compensation for completion. When looking at the plant breeding subset, 84% of public plant breeders and 63% of private plant breeders chose climate change as the most important focus area at least once. These respondents were asked if they believed biotic or abiotic stress was the most important for future attention. The percentages for these responses are given in Table 3. When looking at the participants who chose climate change as most important at least once during the Best Worst (BW) block, 67% said that abiotic stress was more important, and 33% chose biotic stress as the most important. In the plant breeding subsample of participants, 75 and 25% of public plant breeders choose abiotic and biotic, respectively, compared to 64 and 36% of private plant breeders, respectively. Abiotic stress was chosen as “most important” more often than biotic stress across all subgroups. Drought stress was the leading attribute across all subgroups, with an average of 73%, followed by heat stress, with an average of 20%. Increased presence/severity of pathogens due to climatic change was selected as the area that necessitated the greatest research priority regarding biotic stress, followed by the introduction of pathogens due to climatic change. (Supplementary Table 7).

Table 3 The percentage of respondents who selected climate change as the most important area of future research at least once.
Full size table

Public sector breeding moving forward

The results of the questions about what type of research all respondents felt the public sector should focus on the most in the future are highlighted in Tables 4 and 5. “Educating the next generation of plant scientists” had a similar response rate in the public (31%) and private (35%) sectors as well as within the plant breeding subsample where public breeders (36%) and private breeders (38%) thought this should be the primary focus of the plant breeding sector moving forward. This would indicate that approximately a third of public and private plant breeders thought the primary goal of the public sector moving forward should be education and not primary research.

Table 4 What survey respondents thought the public plant breeding sector should focus on moving forward (in percentages)
Full size table
Table 5 Results from survey respondents’ opinions on the impacts of potential public breeding budgetary shortfalls.
Full size table

A total of 35% of private sector respondents (32% of private sector plant breeding respondents) said that “becoming more integrated into research to support the private industry” was the most important for the public sector to focus on, while only 15% of the public sector participants (18% of the public sector plant breeding respondents) selected this choice. Of particular concern is that only 12% of the private industry respondents (13% of private plant breeding respondents) indicated that the public sector should “focus on primary research to enhance plant science.” While each appointment is myopic, a safe assumption is that most public academics who teach in the plant science discipline also have a significant research component to their appointment. These results show the schism concerning how active the public sector should be in conducting primary research moving into the future.

Another concern is that across all sub-groups, a minimum of 90% (92% of public and 93% of private) of respondents indicated that budget cuts in the public sector are likely true in their region/sector/crop (Table 5). Compounding this is that 48% of the total sample (52% of Africa-focused respondents) believed that budget cuts in public breeding programs threaten global food security. Interestingly, when respondents were asked if cuts in public breeding was not a concern but likely a function of the private sector simply increasing funding in breeding sectors, the overwhelming answer was no. Only 16% of the private sector (and 16% of private plant breeders) thought that cuts in the public sector should not be a concern and that the private sector could fill that gap (Table 5).

Discussion

Plant breeding has been pivotal in improving global food security and reducing poverty. As the world faces a growing population and increasing climatic stresses, there are concerns as to how the shifting dynamics of the plant breeding industry will affect future food security. As the private industry plays a more significant role in traditionally dominated public breeding spaces (rice, wheat, and others), a shift is unfolding, shifting from breeding for global food security to breeding for profit. While food security and profitability are not mutually exclusive, the marketability of some breeding traits may service producers in high-income countries more than producers in low-income countries.

Three themes were consistent for private and public breeders throughout this study: disease resistance, climate change for biotic/abiotic stress, and yield enhancement being the focus areas in the future. While there is a clear consensus that these areas are paramount, the relative rankings of importance differ across subsectors of breeding. In the private sector, the ordered focus of future research was to target disease resistance, climate change for biotic/abiotic stress, and yield enhancement. The ordered preference for future research in the public sector targeted climate change for biotic/abiotic stress, disease resistance, and yield enhancement. Our results indicate that respondents thought the public sector should place a greater weight (32.78% in the pooled sample and 36.69% in the plant breeding subsample) on climate change. This is likely because climate change research, while in clear need, holds many characteristics consistent with developing public goods, where progress can be slow and financial returns small. Concerns are then raised regarding the magnitude of impact the public sector will be able to make in the future, given the diminishing human and financial resources characterizing the industry8,15,16,17,21.

The focus on climate change remained a constant concern across all regions. Other important focus areas varied by region between disease resistance, yield enhancement, and pest resistance. The data shows that yield enhancement will be important in Africa (20.01% of the future preference share). The importance of yield enhancement, coupled with 47% of respondents working in Africa reporting that 0% of their budget comes from the private sector, leads to concerns about public-private relationships in Africa moving forward to fight food insecurity.

Given capital requirements, this study also explored the concern that NPBTs would contribute to the schism between public and private plant breeding sectors. The preference shares for CRISPR (a leading NPBT) with respect to the sector, crop type, and region do not exceed 2.34%, and the differences are not significant (P < 0.01) among any of the comparisons. This would suggest, at least in the case of CRISPR, that NPBTs will not create a division between the public and private sectors. It should be noted that CRISPR targets other traits in the survey (yield enhancement, disease resistance, etc.), and participants could have used these traits as a proxy for CRISPR.

Climatic change for biotic/abiotic stress was a top priority among all sectors, crop types, and regions, suggesting that it will warrant significant attention from plant breeding programs in the future. Additional questions on climate change show drought stress and increased presence/severity of pathogens are the most important focus areas for abiotic and biotic stress, respectively. The results support that climate change is a problem within plant breeding that will demand attention moving into the future.

When asked if “budget shortfalls or uncertainty ‘endangered or severely constrained’” the respondent’s ability to conduct meaningful research, 92% of total respondents (95% and 92% of private and public plant breeders, respectively) said that this was true in their region. Additionally, 51% of public sector respondents (52% of public plant breeders) agreed that this was a threat to regional/global food security, while only 34% of the private sector (29% of private plant breeders) considered budget shortfalls a threat. This difference is stark, where it appears that the private sector believes that while budgets are falling for the public sector, the private sector will be able to fill the food security gap left by diminishing funding to the public sector. While this may be true in many high-income countries where profits drive agricultural research and development in the private sector, it begs the question of the feasibility of the private sector filling the research void in many regions across the low-income world, where profit margins could be much thinner. Assistant professors in public institutions are in the greatest position to take academic risks through primary research and generate public goods like those targeted at mitigating climate change through plant breeding with little short-run financial return19.

Compounding potential problems is that only 12% of private sector respondents (13% of private plant breeders) said that the primary focus of public sector plant breeding should be primary research, with most respondents saying that the primary job should be educating the next generation of plant breeders. If this shift unfolds, there would appear to be a research gap regarding breeding for climate change resiliency in the future. With the private sector acknowledging the importance of research dedicated to climate change but, on the other hand, believing that the role of the public sector should be to educate instead of conducting primary research, the gap seems evident. If the public sector only focuses on educating plant scientists, climate change and its externalities could receive less attention, given the lack of opportunity for the private sector to receive returns on new developments12. Breeding for climate change externalities has contributed to the selection in plant breeding programs through increased genotype by environment interactions, potentially offsetting the genetic gains23. Private breeding programs would likely have an aversion to funneling money to a breeding program that would reduce potential genetic gains. This result is surprising and troublesome as the public sector is the primary research component of many low and middle-income countries battling food insecurity24.

Historically, private breeders tend to focus on creating better varieties of widely grown, high-value crops like soya and maize. The important task of improving lower-value crops, such as plantains and cassava, has historically been conducted by scientists at public universities and laboratories run by federal and state agencies. If the future of public breeding is not in primary research, as preferred by the private breeders in this survey, society will need the private sector to fill these roles.

Public breeders often focus on long-term research in which the payoff may require many years of work, often by many individuals across various areas of the public sector. After development and proof of concept by the public sector, the new products are often commercialized by the private sector with little or no return of funding to the public sector. The future of plant breeding should not be competition between sectors but rather finding complementary areas for collaboration, not annihilation, to ensure global food security. In his 1970 Nobel lecture, Norman Borlaug said, “There are no miracles in agricultural production. Nor is there such a thing as a miracle variety of wheat, rice, or maize which can serve as an elixir to cure all ills of a stagnant, traditional agriculture,”25 which is why the future of plant breeding likely needs a robust private as well as public breeding sector to thrive under the pressures of population growth and a changing climate.

There are a few limitations of this analysis that are important to mention, the first of which pertains to the geographically unbalanced nature of the survey responses. The majority of responses were from North America (31%), Europe (26%), and Africa (21%), likely associated with the survey only being provided in English. Asia and South America have large, diverse, active private and national plant breeding programs, which are underrepresented in this study. In addition, only 20% of responses were from the private sector (15% of the sample were private breeders), which may underrepresent the actual percentage of private plant breeders. Obtaining email addresses for private breeders was challenging, but future research should focus on obtaining a more balanced sample. Ideally, more interactions using combinations of sector, field of science, crop focus, and region would have been explored, but given the lack of degrees of freedom (lack of observations), they could not be estimated. Results suggested that only 12% of the private sector thought that the public sector should conduct primary research, and 35% believed the public sector should focus on educating the next generation of plant scientists. In academics, primary research and education are not easily disentangled, specifically at the graduate level. There is the possibility that some respondents viewed education and primary research as so intertwined they selected education as a catch-all for both. Because the word “education” is so broad, what is not known is how many participants considered education and primary research as a joint function. Previous research has indicated to have a well-functioning plant breeding sector that, the public and private industries need to collaborate in the education of the next generation. Universities can offer basic education in plant breeding, with the private sector providing practical and field breeding experience in a commercial program, offering a seamless experience for the next generation of students14. Lastly, it also stands to reason that there could be some inherent bias in respondents’ answers to this survey as their perspective - and their priorities as well – is potentially focused around their respective areas of research and/or funding support.

Methods

Data collection

This study utilized an online survey to target plant scientists globally. The survey was distributed to scientists working in a wide range of plant science fields, including pathology, physiology, entomology, and/or plant breeding. A diverse representation of plant science fields and geographic locations was crucial to obtaining a global perspective on the future of plant breeding. Contacts were gathered through websites of plant scientist platforms, plant science societies, universities, and private companies globally. Targeted plant breeding email addresses were initially gathered from plant breeding institution websites such as CGIAR, NARES, Universities, private industry and other large public breeding institutions. Following this, we used key words “public plant breeding” and “plant breeding” from the Web of Knowledge to obtain additional email addresses, specifically for private plant breeders who often don’t share their contact information on their companies’ websites. Many plant scientists (non-plant breeders) were captured in this search from being coauthors with a plant breeder or who conducted research relative to plant breeding. As such, this study disaggregated results from plant scientists (non-plant breeders) and plant breeders. The sample was skewed towards the public sector as many private companies in the plant breeding/science arena do not provide email addresses for their employees. In addition, the Web of Knowledge database was used to obtain plant scientists’ contacts who had conducted research in the broad field of plant science. The final distribution list consisted of 6,294 distinctive email addresses.

The initial survey was sent out on April 4, 2022, and a reminder email was sent two weeks later to those who had not completed the survey. The survey closed on May 4, 2022, with 822 responses; however, incomplete responses (28%) or those who indicated they had only obtained a high school academic level (2%) were removed from the data set. The final data sample consisted of 583 completed responses. A total of 408 useable responses resulted from the initial email, and an additional 175 useable responses were collected from the follow-up email.

Qualtrics was used to administer the survey via email distribution. University of Arkansas Internal Review Board (IRB) approval was obtained for the survey on 03/08/2020 (protocol number 2102314838). Participants were provided with a written consent form they had to agree to before the beginning of the survey, which stated that it was voluntary and that they could quit the survey at any time. The survey was anonymous, and participants received no payment for completion.

Survey methodology

At the beginning of the survey, each respondent was asked about their academic level, primary research sector (public or private), primary research field (plant breeding, physiology, pathology, entomology, other [asked to specify]), primary region of research focus (Africa, Asia, Europe, Oceania, North America, South America, or no location focus), and primary crop of research focus where major crops were considered to include wheat, maize, soybean, rice, potatoes, cassava, sorghum, millet, yams, and plantains. Minor crops include vegetables, fruits, legumes, and others. Only one choice could be selected for each question, in an attempt not to disentangle results for a plant scientist who worked in multiple crops and regions. While not ideal, as many plant scientists work across regions, from a modeling standpoint, it was not feasible to weight areas of focus by region as the weights could be heterogeneous across attributes (crop type, for instance). Throughout the survey, questions were customized to the respondent’s sector of focus. Questions that contain “(public/private)” were programmed to autofill based on the respondent’s indicated sector.

Best/worst scaling

This study aims to elicit what plant breeders feel the future direction is for the public and private sectors and determine which regions/crops/breeding attributes each sector should focus on moving forward. The Best-Worst Scaling (BWS) approach was used in an attempt to answer these questions as it provides a higher discriminatory rate between items compared to traditional rating scales (such as Likert) in which respondents can allocate the same level of importance to multiple items26. The simplicity of the questions allows for more robust data collection with less burden on respondents27,28. Additionally, the BWS method has been shown to provide better outcomes than other value measuring methods that employ rating or ranking scales, such as Rokeach’s Value Survey, Kahle’s List of Values, Mitchell’s Values and Life Styles, and Schwartz’s Values scale27,29.

The BWS question blocks were created using a Nearly Balanced Incomplete Block Design. There were eight BWS questions, each displaying six of the twelve different attributes. Respondents were asked to select which attribute should receive the most attention and least attention in the future, given their primary sector (public or private). Each attribute appeared four times throughout the block. The order of the questions and the attribute choices within each question were randomized to control for any effects from the order of choice sets28.

The 12 focus attributes that participants could select between were: 1- disease resistance30,31, 2- climate change for biotic/abiotic stress32,33,34, 3- yield enhancement35, 4- pest resistance36, 5- market-value added enhancement of attributes32,37, 6- yield by environment genomic prediction38, 7- digital agriculture39,40 8- CRISPR36,41,42, 9- seed delivery43,10- plant-based protein44, 11- biofortification37,45, and 12- bio-physical/deterministic modeling46. While not an exhaustive list of potential focus areas, given that the participating experts were not compensated for their time, brevity was key for a robust completion rate. There is endogeneity among these research areas, i.e., climate change can have spillover effects for pest resistance. While there was potential overlap, participants saw each choice set and were asked to choose the research area that was the “most” and “least important” for their respective field and thus could choose between the two. Importantly, given the survey design, in most choice set options, participants were not picking between two endogenous issues, but rather one versus the other five research areas of focus in a choice block. Supplementary Fig. A is an example of a BWS question used in the survey for a participant who said they primarily worked in the public sector. It stands to reason that there could be some inherent bias in respondents’ answers as they likely think their respective area of focus is the most important. We tried to mitigate this bias by stating that participants needed to answer the questions about their respective sectors, public/private, and not their own research programs.

Climate change

After the BWS section of the survey was complete, follow-up questions were included to gather more specific information regarding attributes encompassing a wide range of topics. If the respondent selected climate change for biotic/abiotic stress as a “most important” topic of focus at least once, they were asked a follow-up question to specify between abiotic and biotic stress, asking them to specify which type of stress was most important for their respective sector to focus on. The abiotic choices were drought stress47, heat stress48, salinity stress49, flood stress32, toxin stress50, and cold stress32. Only one answer choice could be selected. The biotic choices were increased presence/severity of pathogens (fungi, bacteria, etc.) due to climatic change51, increased presence/severity of insects due to climatic change52, increased presence/severity of weed pressure due to climatic change52, the introduction of new pathogens (fungi, bacteria, etc.) due to climatic change53, and introduction of new weed species due to climatic change54.

Public sector breeding moving forward

Two additional questions regarding the future of public plant breeding were included at the end of the survey. The responses to these questions elicit the views of all the plant scientists and provide the potential for comparisons between the sectors, major crops/minor crops, and across regions. The first asked what the primary role of public-sector plant breeders should be in the future. The choices were: 1- educating the next generation of plant scientists, 2- conducting primary research to enhance plant science, 3- becoming more integrated into research to support private industry, and 4- filling the gap on research in which the private industry shows no/little interest. The impetus of this question revolves around the recent schism between public and private plant breeding, leaving the future role of public plant breeding unclear9,11,55.

A survey20 conducted of 278 U.S. public breeding programs showed a decrease in full-time employees, consistent with past studies8,15,16,21. This could be a function of how universities have invested more in basic research rather than cultivar development. We presented20 findings to the participants and sought to explore if their findings were a concern and in what capacity. We provided the following statement:

Many public programs reported that budget shortfalls or uncertainty “endangered or severely constrained” their ability to conduct meaningful research. A recent 2020 study of 278 public breeding programs in the United States indicated that 21.4% of public breeding programs reported a decline in full-time employees (FTE) over the past five years.

Respondents could choose from the following: 1- This is not a concern as it is likely a function of the private sector simply increasing funding in breeding sectors once dominated by the public sector, 2- This is a trend likely true globally, 3- This is a threat to regional/global food security, and 4- This trend is likely not true in my region. Participants could select more than one answer choice.

Survey responses

The survey resulted in 583 complete and usable responses for the empirical analysis. The descriptive statistics of the response pool, including academic level, sector, field, primary crop focus, and primary region of work, are highlighted on Table 6. The majority (483 or 83%) of participants hold a Doctor of Philosophy (Ph.D.) or Juris Doctor (J.D.) in a plant science discipline. Public plant breeders made up 29% of the total sample, while private breeders encompassed 15%. Respondents were predominantly active in the public sector (80%), which is expected given the limited access to email addresses available from private companies. The breakdown by the specific primary crop that respondents worked on is shown on Supplementary Table 1. There were thirteen crop options and an “other” catchall category. The “other” option was the most selected choice, consisting of 35% of the public and 34% of the private sector responses, respectively. This is also consistent across North America (35%), Europe (45%), and Africa (29%); however, rice was the most common crop of focus in Asia (30%) and South America (55%).

Table 6 Characteristics of survey participants in percentages
Full size table

Econometric analysis

Data analysis was conducted using a discrete choice framework, consistent with the random utility theory56, and estimated random parameter (RP) maxdiff models. This assumes that respondents simultaneously choose the pair of best and worst items that maximize the difference between the two selected items on an underlying scale of importance57.

Assuming that there are \(J\) items listed in each choice set \(t\), then the number of possible pairs is \(J(J-1)\). The observable level of importance of the item \({j}\) on the underlying scale is defined as \({\lambda }_{i}\), while the unobservable level of importance of respondent \({i}\) is given by \({I}_{{ij}}={\lambda }_{j}+{\varepsilon }_{{ij}}\), where \({\varepsilon }_{{ij}}\) is the random error term. The maxdiff model follows the random utility theory48, where the probability that respondent \(i\) selects item \({j}\) as the best and item \(k\) as the worst in choice set \(t\) equals the probability that the difference in utility of the selected items (\({I}_{{ij}}\) and \({I}_{{ik}}\)) is greater than all the other possible differences \(M=J\left(J-1\right)-1\) in the choice set. Assuming that the error terms are independently and identically distributed (iid), the probability is represented by the multinomial logit model.

$${P}_{{ijkt}}=exp\left({\lambda }_{{ijt}}-{\lambda }_{{ikt}}\right)/\mathop{\sum }\limits_{l=1}^{J}\,\mathop{\sum }\limits_{l=1}^{J}exp\left({\lambda }_{{ilt}}-{\lambda }_{{imt}}\right)-J$$
(1)

The estimated \({\lambda }_{j}\) represents the importance of item \(j\) relative to some item that is normalized to zero for identification purposes. In this analysis, “Seed Delivery” was normalized to zero.

To account for heterogeneity in preferences for the various attributes, we estimate a random parameter logit (RPL) model. The RPL model assumes that the estimated parameters \({\lambda }_{j}\) are distributed according to a multivariate normal distribution. The model uses simulation methods (maximum simulated likelihood estimator) to provide estimates of mean and standard deviation for each coefficient.

To ease interpretation, preference shares (an estimate of how much importance each attribute has over the others or the probability that an attribute is picked as more important than another) for each attribute were then estimated using the RPL model as follows:

$${S}_{j}={e}^{\hat{{\lambda }_{j}}}/{\sum }_{k=1}^{j}{e}^{\hat{{\lambda }_{l}}}$$
(2)

A distribution of 500 preference shares for each item \(j\) was drawn from a multivariate normal distribution created using the means and standard deviations estimated for each item \({j}\) from our RPL model. The bootstrapping method58 was implemented to create these distributions.

Subsets of the survey data were created based on sector (public/private), crop (major/minor), and region (N. America, S. America, Europe, Africa, Oceania, and Asia) to elicit comparisons. For example, to compare the preference shares of the public and private sectors, our data was separated into two subgroups (\({S}_{j,{public}}\) and \({S}_{j,{private}}\)) and separate RPL models were estimated accordingly. The overall quality of preferences estimated using the two RPL models was tested using the likelihood ratio test (LRT)59. The LRT requires estimating an RPL model on the pooled sample while controlling for potential differences in scales across the two data sets. The LRT is based on the test statistic in Eq. 3.

$$\lambda =-2[{L}_{\mu }-({L}_{{private}}-{L}_{{public}})]$$
(3)

where \({L}_{\mu }\) is the log-likelihood value from the estimation of the RPL model using the pooled model, \({L}_{{public}}\) is the log-likelihood value from the estimation of the RPL model using the public sector subgroup, and \({L}_{{private}}\) is the log-likelihood value from the estimation of the RPL model using the private sector subgroup. The test statistic is asymptotically chi-squared distributed with \(K+1\) degrees of freedom, where \(K\) is the number of coefficients estimated in the three models.

If the null hypotheses of overall equality of preferences between the public and private subgroups are rejected, we test if the differences in pairs of preference shares (\(\Delta S=\,{S}_{j,{private}}-{S}_{j,{public}}\)) are statistically significant using the Poe test60. When multiple comparisons were made, we used adjusted p-values to ensure the significance levels were 95% or greater.

This process was repeated for major/minor crops and regional subgroups. Regional comparisons were made between two continents and then by comparing each continent to the other regions. Based on the number of responses, Africa, Europe, and North America were the only continents used individually. The comparisons were public vs. private plant scientists, public plant breeders vs. private plant breeders, and geographical regions.