Honey bees exhibit greater patch fidelity than bumble bees when foraging in a common environment
Abstract
Animals commonly exhibit a tendency to return to previously visited locations. Such tendency is manifested at different scales, for example, fidelity to a site or fidelity to a specific patch within a site. Although patch fidelity has important implications for the pollinators and the plants they visit, our understanding of patch fidelity, and the extent to which it varies among bee species, remains limited. Here, we used a mark–reobservation approach to compare patch fidelity and patch size preference between one bumble bee and one honey bee species foraging on patches of Medicago sativa L. Honey bees exhibited greater patch fidelity (76%) than bumble bees (47%); they were more likely to return to the patch where they were marked. Patch size affected the level of patch fidelity for bumble bees but not for honey bees. Bumble bees were more likely to return to larger rather than smaller patches. In addition to patch fidelity, we detected preference of bees for the larger patches. Bees visited the larger patches more often than the smaller patches. These results add patch fidelity to the already known repertoire of differences in foraging strategies between a bumble bee and a honey bee species. It also indicates how the simultaneous study of distinct species in a similar environment can reveal important previously undetected information about their behavioral ecology. Observed differences in patch fidelity and patch size preferences may have important implications for crop pollination and conservation habitat design.
INTRODUCTION
Mobile animals have a tendency to return to previously visited locations (Martinez et al., 2017; Morrison et al., 2021; Moura et al., 2022; Richardson et al., 2017). Site fidelity typically describes recurrent movements to specific locations within a landscape. However, resources available to foraging bees are often patchily distributed at a site, and fidelity to specific patches can increase foraging efficiency and decrease movement costs, especially when pollen and nectar resources are constant across time or space (Ogilvie & Thomson, 2016; Osborne & Williams, 2001). From the perspective of the plant, fidelity can facilitate pollination of plant species that flower sequentially over the landscape (Ogilvie & Thomson, 2016) and limit pollen movement and gene flow among patches (Rasmussen & Brødsgaard, 1992). Although the drivers behind patterns of fidelity are not completely understood, interspecific differences in memory capacity and learning ability (Chittka et al., 1999; Grüter & Ratnieks, 2011) are likely to interact with an animal's foraging environment to modulate the expression and intensity of fidelity patterns among animal species (Morrison et al., 2021; Moura et al., 2022; Switzer, 1993).
In bees, patch fidelity has received considerably less attention than flower constancy. Both behaviors describe recurrent movements, but flower constancy is usually applied to pollinators that restrict their visits to a single flower type even when other rewarding types are available (Grüter & Ratnieks, 2011; Waser, 1986). While flower constancy is widespread among insect pollinators, the preferred flower type often varies, and its selection is influenced by flower morphology, flower odor, flower color, the quantity and quality of the nectar reward, and social cues from conspecific bees (Cecala & Wilson Rankin, 2020; Franzén et al., 2009; Goulson et al., 1997; Goulson & Wright, 1998; Slaa et al., 1998). Both social and solitary bees can exhibit flower and/or patch fidelity; however, these two traits are not necessarily correlated (Franzén et al., 2009; Gary et al., 1977; Ogilvie & Thomson, 2016; Osborne & Williams, 2001). It is important to understand the degree of variation in patch fidelity among bee species because such knowledge can guide the design of effective habitats to enhance pollinator conservation (Martínez-Bauer et al., 2021).
Bees can also show preference for some patches, where they visit some patches more frequently than others. Patch size and distance between patches can affect patch selection by bumble bees (Fragoso et al., 2021) and leafcutting bees (Fragoso & Brunet, 2023a). Bumble bees tend to visit larger patches more often than smaller patches, but the impact of patch density is typically stronger than patch size, and the effect of patch size can be overcome when the landscape context varies (Bernhardt et al., 2008; Goverde et al., 2002; Heard et al., 2007; Mustajärvi et al., 2001). Visitation rates by honey bees can also be higher in larger patches (Sih & Baltus, 1987). While we have a good understanding of how pollinators choose flowers (Latty & Trueblood, 2020), the larger scale behaviors, namely, how pollinators select patches and whether patch size preference varies among bee species, are less well understood (Cresswell & Osborne, 2004; Fragoso et al., 2021; Otto et al., 2021).
In this study, we compared patch fidelity and patch size preference between the common eastern bumble bee, Bombus impatiens Cresson, and the European honey bee, Apis mellifera L., foraging in a common landscape. The experimental design controlled for patch size, and we used a mark–reobservation method to record, over two years, visitation patterns of individual bees foraging in patches of Medicago sativa L. We asked (1) whether patch fidelity varied among patches, and when it did, whether patch size affected it; and (2) whether the two bee species exhibited a preference for larger patches. Patch fidelity was defined as the proportion of times marked bees were reobserved in the patch where they were originally marked. In contrast, patch preference illustrated the proportion of the overall observations (all patches) made in a given patch. Because bumble bees tend to be less flower constant relative to honey bees (Minahan & Brunet, 2020), we predict bumble bees will also exhibit lesser patch fidelity. We hypothesize that both bee species will prefer larger over smaller patches. We consider bee characteristics that may influence patch fidelity and patch size preference and briefly discuss their potential impact on gene flow and on the design of habitats for bee conservation.
METHODS
Study site and experimental design
We conducted the study at the West Madison Agricultural Research Station, Wisconsin, USA. The experiment consisted of one center patch of M. sativa (alfalfa) surrounded by four peripheral patches. The center and two of the peripheral patches were 9.14 × 9.14 m and contained 100 plants (small), while the other two peripheral patches were 13.7 × 13.7 m and contained 225 plants (large) (Figure 1; see also Fragoso et al., 2021). One of the large and one of the small patches were located 9.14 m diagonally from the center patch, while the other large and small patches were located 18.3 m away (Figure 1). The five patches in the design were referred to, for identification purpose, as the center patch (C), the large near (LN), small near (SN), large far (LF), and small far (SF) patches. Peripheral patches of the same size were located on the same side of the center patch, in this case north (Figure 1). Seedlings of M. sativa L. were originally planted 90 cm apart within each of the five patches. Tall fescue was seeded among the plants, and among patches, to facilitate weed control. The five patches were within a 65.8 × 64 m plot surrounded mostly by corn and soybean, and alfalfa grown for hay. The alfalfa fields surrounding the experiment were cut prior to flowering and did not interfere with bee visits to the experiment. This experimental design allowed us to isolate the effect of patch size, since patch quality (flower type, resources available, etc.) and landscape context were consistent across all the studied patches.

Plant and pollinator species
M. sativa L. (alfalfa) is a perennial legume with typically high outcrossing rates (Dieterich Mabin et al., 2021), cultivated worldwide for hay or seed production. Flowers are arranged into racemes, with many racemes per stem, and stems per plant, such that each plant can bear 100–2000 flowers (Bauer et al., 2017).
The European honey bee (A. mellifera L.) and the common eastern bumble bee (B. impatiens Cresson) were used in this study. The European honey bee is an important managed pollinator in alfalfa seed production fields. The common eastern bumble bee is not used in alfalfa seed production in the United States but is commonly utilized by alfalfa breeders to make crosses in the greenhouse and for seed increase of breeding stocks. This species is the most common bumble bee species in Wisconsin.
A 10-frame deep honey bee hive with a honey super was set up at about 150 m northwest from the study site each year. Wild common eastern bumble bees visiting alfalfa flowers were used in this study, and their nests could not be located. While other pollinators can also visit alfalfa flowers in Wisconsin (Brunet & Stewart, 2010), their abundance in the patches was negligible. Individual bees were seen approaching the patches from different directions, and we rarely observed any interference between bee species.
Marking and surveying bees
During 11 days in 2020 (July 29–August 14) and 12 days in 2021 (July 8–July 28), when alfalfa was in peak bloom and pollinator activity was high, individual bumble bees and honey bees foraging in one of the five alfalfa patches were caught in a transparent, plastic vial that was immediately transferred to a cooler filled with ice packs. After the bee became immobile, it was marked with a uniquely numbered patch-dependent colored disk (Opalith queen number set from Betterbee, Greenwich, NY) glued to the dorsal surface of its thorax. After marking, a bee was released in the patch where it was captured. We aimed at marking five or more bees of each species per patch each day although this goal could not always be reached. Bee markings and observations took place at the same site over two years.
Patches were surveyed every hour between 9 am and 3 pm daily, following the first day of marking bees, for a period slightly longer than bee marking (July 29–August 31 in 2020, and July 8–August 2 in 2021). Two observers walked standard paths between adjacent rows of alfalfa plants, each starting at opposite ends of a patch. When a marked bee was spotted foraging on plants to the right or left of an observer, the time of day, patch type, and tag information on the bee were recorded. The order in which the five patches were surveyed by the two observers each hour on each day was randomly determined. These data provided the day and patch a bee was marked, together with the number of times (day and hour) a marked bee was observed and the patch(es) it was observed in throughout the duration of the study. If an individual bee was observed more than once within an observation period, a single record was used for the analysis. Surveying every hour is expected to provide independent foraging trips (the time between a bee leaving and returning to the hive); using radio frequency identification (RFID) data, Minahan and Brunet (2018) determined the average length of a foraging trip in this landscape to be less than 1 h, for both bumble bees and honey bees.
Data analysis
Bee observations
For each of the five patches, we tabulated the following variables: the number of observations of marked bees, which included multiple observations of individually marked bees; the number of patch-faithful observations, or observations that occurred in the patch a bee was originally marked; and non-faithful observations, the observations that occurred in a patch different from where the bee was originally marked. These variables were tabulated for each patch per bee species over all individual bees; we also obtained the average per individual bee by dividing over the number of marked bees reobserved in a given patch. We calculated these values for each year separately and for both years combined. In addition, we determined the number of marked bees observed, that is, marked individual bees observed at least once for each patch. All tabulations were performed using dplyr and tidyr packages in the Tidyverse collection of R packages (Wickham et al., 2019) in R version 4.0.3 (R Core Team, 2020).
Patch fidelity
Patch fidelity was quantified as the proportion of patch-faithful observations over the total number of observations of marked bees in a given patch. We quantified patch fidelity for each bee species, over combined years and independently for each year. We also examined whether patch fidelity varied among days for marked bees observed on more than one day.
To determine whether patch fidelity varied among patches, we used χ2 goodness-of-fit tests. For each bee species, we compared the observed number of patch-faithful observations in a patch to expected values if bees exhibited the same overall degree of patch fidelity in each patch. This overall degree of patch fidelity for each bee species, calculated over both years, was 76% for honey bees and 47% for bumble bees. It illustrates, for all patches combined, the total number of patch-faithful observations divided by the total number of observations (see Results section for patch fidelity). For each patch, the expected number of patch-faithful observations was therefore the total number of observations (of marked bees) in a patch multiplied by 0.76 for honey bees and 0.47 for bumble bees. We are testing the null hypothesis of lack of variation in patch fidelity among patches. Because the requirements of the χ2 test (80% of the cells must have expected values of 5 or more) were not met for the 2021 data due to its lower sample sizes, and because the χ2 results were similar for 2020, 2021, and both years combined (Appendix S1: Table S1), we present the results for both years combined. When a statistically significant difference in patch fidelity was observed among patches, we examined the potential impact of patch size on patch fidelity using a mixed linear model with patch size as a fixed effect and year as a random variable, with the lme4 package in R (Bates et al., 2015).
Patch size preference
Preference for some patches is detected when the total number of observations of marked bees differs among patches. For each bee species, we used χ2 goodness-of-fit tests to determine whether the total number of observations was equally distributed among patches, that is, with five patches, a null hypothesis of 0.20 of the observations per patch. We also examined whether the pattern of preference differed between bumble bees and honey bees using χ2 independence tests. For both the χ2 goodness-of-fit (Appendix S1: Table S2) and the χ2 independence tests (Appendix S1: Table S3), similar results were obtained when both years were combined or examined separately and we present the combined years results.
RESULTS
Bee observations
Two hundred and seven individual bumble bees and 387 honey bees were marked in the experiment, and 67.6% of the marked bumble bees, and 53.1% of the marked honey bees were observed at least once in a patch. Bee abundance was lower in 2021, when 67 bumble bees and 95 honey bees were marked, and 49 marked bumble bees and 29 marked honey bees were observed, as opposed to 2020 where 140 bumble bees and 291 honey bees were marked, and 91 marked bumble bees and 176 marked honey bees observed. On average, an individual bumble bee or honey bee was observed more than six times in this study (range 1–50 times for bumble bees, and 1–34 times for honey bees) and was observed on three different days (range 1–13 days for bumble bees, and 1–11 days for honey bees) (Appendix S1: Figure S1). While the total number of observations was slightly greater for honey bees (Appendix S1: Table S4), the number of observations per individual bee was similar between the two bee species (Appendix S1: Table S5).
Patch fidelity
Honey bees exhibited greater patch fidelity than bumble bees, and this was true over both years combined (Figure 2) and for each year examined separately (Appendix S1: Table S4). Over combined years, the overall proportion of patch-faithful observations was 76% for honey bees in contrast to 47% for bumble bees, and a similar pattern was observed each year (Appendix S1: Table S5). The pattern was consistent whether individual bees were observed on one or more days (Figure 3).


Patch fidelity varied among patches for bumble bees but not for honey bees (bumble bee: χ2 = 40.754, df = 4, p < 0.001; honey bee: χ2 = 1.2552, df = 4, p = 0.869) (Figure 2; diagonals in Appendix S1: Figure S2). Patch size affected patch fidelity for bumble bees, who were more likely to return to a patch when marked in a large patch (F1,8 = 11.87, p = 0.0009; average patch fidelity 60.5% ± 7.8 for large, and 29.7% ± 6.6 for small patches). The greater patch fidelity of honey bees relative to bumble bees, together with the differences among patches for bumble bees but not for honey bees, was also apparent when looking at individual bees (Appendix S1: Figure S3).
Patch size preference
Bumble bees and honey bees did not visit all patches at the same frequency (bumble bees: χ2 = 254.72, df = 4, p < 0.0001; honey bees: χ2 = 95.298, df = 4, p < 0.0001). Both bee species preferentially visited the larger patches and avoided the smaller patches (Table 1). The distribution of total observations among patches did, however, differ between the two bee species (χ2 = 65.561, df = 4, p < 0.0001) (Table 1).
Patch | Total observations | ||
---|---|---|---|
Observed | Expected | Difference | |
Bumble bee | |||
C | 156 | 196.6 | −40.6 |
LF | 207 | 196.6 | 10.4 |
LN | 385 | 196.6 | 188.4 |
SF | 130 | 196.6 | −66.6 |
SN | 105 | 196.6 | −91.6 |
Honey bee | |||
C | 296 | 270.8 | 25.2 |
LF | 360 | 270.8 | 89.2 |
LN | 322 | 270.8 | 51.2 |
SF | 209 | 270.8 | −61.8 |
SN | 167 | 270.8 | −103.8 |
- Note: Observed is the recorded number of observations of marked bees. Expected is the number of observations assuming they are equally distributed among the five patches (p = 0.20). The null hypothesis is no preference for patches. The differences between observed and expected values indicate preference (positive values) or avoidance (negative values) of a patch. The differences over all patches lead to the statistically significant Chisquare presented in the text (in bold). There are three small (C, SF, and SN) and two large (LF and LN) patches in the experiment.
- Abbreviations: C, center; LF, large far; LN, large near; SF, small far; SN, small near.
DISCUSSION
The European honey bee exhibited greater patch fidelity (76%) than the common eastern bumble bee (47%) when foraging on M. sativa. This pattern is robust as it was observed over all bees (Figure 2) and over individual bees (Appendix S1: Figure S3), and it did not vary with the number of days a bee was observed (Figure 3). Previous studies, which examined a single species, reported a high degree of patch fidelity for bumble bees (Ogilvie & Thomson, 2016; Osborne & Williams, 2001), or for honey bees (Gary et al., 1977; Moore et al., 2011). In the current study, unlike previous research, both bee species were observed at the same time and in the same location. Although honey bees tend to outnumber bumble bees because of differences in the size of their hives, we did not observe an abundance of either bee species in the patches and did not notice any interference between the two bee species. Each M. sativa plant bears at least 100 but more typically 1000 plus flowers, and small patches had 100 plants and large patches 225; thus, there were abundant open flowers in the patches. In addition, honey bees collect mainly nectar from M. sativa flowers, while bumble bees collect pollen in addition to nectar. Resource limitation and competition are thus unlikely to explain the difference in patch fidelity observed between the two bee species in the current study. Comparing the two bee species under similar conditions highlighted a difference in patch fidelity between the species, stressing the importance of contrasting bee species foraging under similar conditions when examining differences in patch (or site) fidelity between bee species (Grüter & Ratnieks, 2011).
Differences in patch fidelity observed among studies examining bumble bees (Ogilvie & Thomson, 2016; Osborne & Williams, 2001; this study) further stress the importance of using similar conditions when comparing bee species. While Osborne and Williams (2001) observed greater than 85% fidelity, Ogilvie and Thomson (2016) observed 67% fidelity on Delphinium flowers and 78% fidelity at the same location with Gentiana flowers, while we observed 47% fidelity for the bumble bee species. Such differences among studies could be influenced by the bumble bee species being examined (B. lapidarius: Osborne & Williams, 2001; various species: Ogilvie & Thomson, 2016; B. impatiens: this study). Differences in flower constancy have been observed among bumble bee species (Leonhardt & Blüthgen, 2012), and this pattern may extend to patch fidelity. In addition, the plant species being visited may affect the results. While Osborne and Williams (2001) used mixed plant species in each patch and Ogilvie and Thomson (2016) had two distinct plant species over time, this study used a single plant species, M. sativa. Interestingly, the reobservation rate of bees was quite low (20%) for Osborne and Williams (2001) with mixed species and much higher (63% over all species) for Ogilvie and Thomson (2016) and this study (67%) when a single species was examined at a given time. Thus, different factors may affect patch fidelity, and it is important to control for the environment when comparing bee species.
Both bee species exhibited patch fidelity. To remain faithful to a location, an animal requires reliable spatial memories that enable them to navigate complex landscapes and return repeatedly to the same site. Bees have been shown to use optic flow (retinal image motion during travel) to calculate distances (Esch & Burns, 1996; Srinivasan et al., 2000) and to rely on path integration (monitoring direction through compass information) to keep track of directions (Collett et al., 2013; Collett & Collett, 2000). They also use landmarks to pinpoint their destination (Chittka et al., 1995; Menzel et al., 1998). Both bees studied here have demonstrated their ability to return to previously visited foraging locations, so there must be other species-specific factors to explain the differences in patch fidelity observed between the two species.
Honey bees and bumble bees differ in their social organization and foraging strategies. Honey bees employ the waggle dance to recruit a large number of foragers to previously visited locations (Couvillon et al., 2014; Frisch, 1967), and they can also use trophallaxis, that is, mouth-to-mouth food exchange, to communicate the profitability of a food source after a successful foraging trip (Farina & Núñez, 1991; Wainselboim & Farina, 2003). In addition, honey bee foragers recruited to a particular patch are more likely to remain faithful to a patch than to switch to an unknown equally rewarding patch, a characteristic referred to as the “costly information hypothesis” (Beekman et al., 2004). The use of the waggle dance in honey bees, also known as the “communication hypothesis,” combined with the “costly information hypothesis,” may explain the high patch fidelity of honey bees. In contrast to honey bees that use socially acquired information to locate resources, bumble bees rely primarily on individual acquisition of information (Chittka et al., 1997). Bumble bees invest individually in sampling and learning during multiple foraging bouts to minimize travel distances and favor more rewarding locations (Lihoreau et al., 2011; Ohashi et al., 2007; Saleh & Chittka, 2007; Thomson, 1996). Bumble bees tend to be more explorative in their foraging behavior, often visiting more than one flower type during a foraging bout (Heinrich, 1979) and thus being less flower constant relative to honey bees (Leonhardt & Blüthgen, 2012; Minahan & Brunet, 2020). The more explorative behavior of bumble bees may also explain their lower patch fidelity.
An additional difference between the two bee species with respect to patch fidelity is the variation in the degree of patch fidelity among patches for bumble bees, but not for honey bees. Bumble bees were more likely to return to a patch when marked in a larger patch while honey bees were as likely to return to a patch irrespective of its size. The “costly information hypothesis” (Beekman et al., 2004) could explain the lack of variation in patch fidelity among patches observed with honeybees. If patch fidelity is favored because it reduces the risk of foraging elsewhere, as long as rewards remain profitable (Osborne & Williams, 2001), then the higher patch fidelity of honey bees suggests they may be more risk-averse than bumble bees. Interestingly, honey bees have been shown to be most risk-averse when reward volume varied and included zero rewards (Drezner-Levy & Shafir, 2007), which reflect the conditions for honey bees foraging for nectar on alfalfa flowers. Even if bumble bees are more patch faithful in larger patches, their level of patch fidelity remains lower than that of honey bees for any of the patches (Figure 2). The lower degree of patch fidelity for bumble bees relative to honey bees, and the higher patch fidelity of bumble bees in larger patches, are reflected at the level of individual bees (Appendix S1: Figure S3).
The observed differences in patch fidelity between bee species could affect their impact on plant reproduction (crop yield) and gene flow. Bees that return to the same patch can augment the visitation rates to plants in the patch, and a higher number of bee visits per plant is associated with increased plant reproductive success (Bauer et al., 2017). While greater patch fidelity may increase the reproductive success of plants in the patch, other factors affecting plant reproduction could counterbalance the effects of patch fidelity. In M. sativa, for example, despite their greater patch fidelity, honey bees trip a much lower proportion of the flowers relative to bumble bees (Brunet et al., 2019), which negatively impacts seed set and crop yield (Bauer et al., 2017).
The higher patch fidelity of honey bees could limit pollen movement and gene flow among patches (Rasmussen & Brødsgaard, 1992). Lower gene flow by honey bees, relative to bumble bees, is supported by within foraging bout data indicating a lower probability for honey bees to move genes longer distances relative to bumble bees (Fragoso & Brunet, 2023b). In natural populations, higher gene flow lowers genetic differentiation and homogenizes the genetic diversity of plant populations (Ellstrand, 2014). In agriculture, higher gene flow increases the spread of genetically engineered genes to non-genetically-engineered fields of a crop, and to feral or wild cross-compatible relatives (Ellstrand et al., 2013; Greene et al., 2015; Kalaitzandonakes & Magnier, 2013; Warwick et al., 2009). Distinct bee foraging behaviors can affect gene flow (Brunet et al., 2019), and here we add patch fidelity as a behavior that could differentially affect how distinct bee species may contribute to gene flow among plant populations.
Both bee species preferred larger patches and avoided smaller patches, as indicated by the pattern of distribution of observations among patches (Table 1). For bumble bees, a fairly large number of bees marked in the smaller patches moved to the larger patches (non-faithful observations) while the proportion of non-faithful observations was much lower for honey bees (Figure 2; Appendix S1: Figure S2). These differences in exploratory behavior between bee species suggest a greater role played by foraging experience of individual bees when bumble bees, as opposed to honey bees, select which patches to forage in. On the whole, larger floral patches attracted greater bee foraging and, in the case of bumble bees, increased patch fidelity. Larger patches provide more floral resources, and bees tend to visit more flowers in larger patches, although they visit a very small fraction of the flowers available per foraging bout (Brunet et al., 2019; Cresswell & Osborne, 2004). Foraging in larger patches may optimize energy expenditure (Fryxell, 2008), and large continuous patches can positively affect native bee and wasp reproduction (Turo & Gardiner, 2021). Future work is needed to understand whether bee preference for larger patches will translate into reproductive benefits for the bees.
In conclusion, honey bees are more faithful to previously visited patches relative to bumble bees; bumble bees are more faithful to larger patches, while honey bees are equally faithful to patches irrespective of their size. While the two bee species differed in their probability of returning to a patch, both bee species preferentially visited larger patches. Larger patches seem to benefit the plants, as the higher visitation rate often translates into greater seed set. While this study focused on monospecific patches, future studies should determine how the presence of multiple patches each with a distinct plant species, or patches with multiple plant species, may modify patch fidelity and patch preference of distinct bee species. We predict greater patch fidelity and preference of honey bees for monofloral patches and greater fidelity and preference of bumble bees for polyfloral patches. Bumble bees not only display lower floral constancy relative to honey bees (Minahan & Brunet, 2020), but they also prefer sites with high floral diversity (Martínez-Bauer et al., 2021). This study adds patch fidelity to the differences in foraging strategies previously observed between these two social bee species, with the bee species with the lower patch fidelity expected to increase gene flow between patches. It also confirms that these two bee species preferentially visit larger patches, suggesting a potential advantage to having larger floral patches when designing conservation habitats, especially if larger patches translate into reproductive benefits to bees.
AUTHOR CONTRIBUTIONS
Fabiana P. Fragoso and Johanne Brunet conceived the study, and designed the methodology and data analyses. Fabiana P. Fragoso oversaw the experiment and analyzed the data. Fabiana P. Fragoso and Johanne Brunet contributed critically to the drafts and gave final approval for publication.
ACKNOWLEDGMENTS
A. Iwan, A.-M. Lynch, C. Slawin, L. Johnson, M. Dieterich Mabin, and P. Dombrowsky assisted in the field. M. Dieterich Mabin contributed to the experimental design. L. Palmieri assisted with the figures. This project was funded by a Biotechnology Risk Assessment Grant Program, competitive grant no. 2018-33522-28707, from the USDA National Institute of Food and Agriculture to Johanne Brunet and by funds from the USDA Agricultural Research Service to Johanne Brunet. Fabiana P. Fragoso had an appointment with the Agricultural Research Service (ARS) Research Participation Program administered by the Oak Ridge Institute for Science and Education (ORISE) through an interagency agreement between the U.S. Department of Energy (DOE) and the U.S. Department of Agriculture (USDA). ORISE is managed by ORAU under DOE contract number DE-SC0014664. All opinions expressed in this paper are the author's and do not necessarily reflect the policies and views of USDA, ARS, DOE, or ORAU/ORISE.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
Open Research
DATA AVAILABILITY STATEMENT
The data (Brunet & Fragoso, 2023) are available at Dryad: https://doi.org/10.5061/dryad.zcrjdfnj4.