Apple grower pollination practices and perceptions of alternative pollinators in New York and Pennsylvania

Abstract

Pollinator declines coupled with increasing demand for insect pollinated crops have the potential to cause future pollinator shortages for our most nutritious and valuable crops. Ensuring adequate crop pollination may necessitate a shift in pollination management, from one that primarily relies on the managed European honeybee (Apis mellifera L.) to one that integrates alternative pollinators. While a growing body of scientific evidence supports significant contributions made by naturally occurring, native bees for crop pollination, translating research to practice requires buy-in from growers. The intention of agricultural extension is to address grower needs and concerns; however, few studies have assessed grower knowledge, perceptions and attitudes about native pollinators. Here we present findings from questionnaire-based surveys of over 600 apple growers in New York State and Pennsylvania, coupled with ecological data from bee surveys. This hybrid sociological and biological survey allows us to compare grower knowledge and perceptions to an actual pollinator census. While up to 93% of respondents highly valued importance of native bees, 20% growers did not know how much native bees actually contribute to their orchard pollination. Despite the uncertainty, a majority of growers were open to relying on native bees (up to 60% in NY and 67% in PA) and to making low-cost changes to their farm's management that would benefit native pollinators (up to 68 in NY and 85% in PA). Growers consistently underestimated bee diversity, but their estimates corresponded to major bee groups identifiable by lay persons, indicating accurate local knowledge about native bees. Grower reliance on honeybees increased with farm size; because native bee abundance did not measurably decrease with farm size, renting honeybees may be motivated by risk avoidance rather than grower perception of lower native bee activity. Demonstrated effectiveness of native pollinators and clear guidelines for their management were the most important factors influencing grower decision to actively manage orchards for native bees. Our results highlight a pressing need for an active and research-based extension program to support diversification of pollination strategies in the region.

Introduction

At least 35% of global food production benefits from insect pollinators (Klein et al., 2007). Bees are by far the most important pollinators in agricultural settings, and in terms of ecosystem service, contribute between $5.7 and $19 billion per year to the United States economy (Levin, 1983; Robinson et al., 1989; Southwick and Southwick, 1992; Morse and Calderone, 2000) and $217 billion per year globally (Gallai et al., 2009). Bees support human health by pollinating our most nutritious food crops (Eilers et al., 2011), for which global demands are projected to rise as developing countries become more wealthy (Aizen and Harder, 2009). With both domestic and wild bees experiencing global declines (Biesmeijer et al., 2006; Potts et al., 2010; Cameron et al., 2011; van der Zee et al., 2012; Burkle et al., 2013), explicitly incorporating pollinator well-being into farm management decisions is necessary to ensure sustainable pollination services.

Although more than 20,000 bee species have been described worldwide (Ascher and Pickering, 2013), pollination management in modern agriculture traditionally involves a single species; in the United States, this species is the European honeybee, Apis mellifera L. Commercially available, managed honeybees comprise large colonies that are readily moved into crop fields during bloom. Honeybees are especially important pollinators in large scale, highly disturbed agroecosystems that cannot support wild pollinators. Due to steady declines in honeybee populations over the past 50 yr (National Research Council, 2007) and significant colony losses due to ‘colony collapse disorder’ (CCD) (Oldroyd, 2007; vanEngelsdorp et al., 2009), it is becoming increasingly risky to rely on a single pollinator species for food production (Winfree, 2008). Indeed, if honeybees continue to decline, growers may need to diversify their pollinator portfolio to include alternative pollinators in order to sustain adequate crop pollination in the future.

Developing managed alternative pollinators could diversify pollination strategies, but evidence is also building for important crop pollination services by naturally occurring wild bees. Globally, wild bees are more efficient pollinators than honeybees (Garibaldi et al., 2013), and the diversity associated with the communities of wild bees stabilizes pollination services spatiotemporally (Kremen et al., 2004; Klein, 2009; Garibaldi et al., 2011), in a manner that provides resilience to rapid climate change (Bartomeus et al., 2011; Brittain et al., 2013). Optimizing wild bee pollination services, however, may require a shift in pollination management strategies for growers. In contrast to ordering honeybees for a few weeks, long-term efforts may be required to provide wild pollinators with semi-natural or natural areas for food and nesting resources, as well as safety from pesticides beyond the short bloom period (Watson et al., 2011; Kammerer et al., 2016a, b; Park et al., 2015; Joshi et al., 2016). Technical support and education programs will, therefore, be needed to help growers rely on a suite of pollinators and not just the honeybee (Isaacs et al., 2017). Growers may need to modify their pest management practices to accommodate pollinator health (Biddinger and Rajotte, 2015). Pesticide applications to control pests can affect pollinators throughout the growing season, even if apple is not blooming (Mallinger et al., 2015; Park et al., 2015). Even if pesticide sprays are avoided during bloom, modern systemic insecticides applied before flowering may contaminate nectar and pollen (Mogren and Lundgren, 2016). An expansion of integrated pest management (IPM) to integrated pest and pollinator management is a viable solution to managing pests and protecting pollinator health (Biddinger and Rajotte, 2015).

Understanding current grower knowledge and perceptions of alternative pollinators could inform successful outreach and encourage heavier reliance on alternative pollinators, yet only a few such studies exist worldwide (Partap et al., 2001; Kasina et al., 2009; Munyuli, 2011; Hanes et al., 2015; Gaines-Day and Gratton, 2017). Even fewer studies investigate factors affecting grower pollination strategies (Potts et al., 2011; Hanes et al., 2015). Here, we surveyed pollination practices, perceptions of alternative pollinators and willingness to implement bee-friendly management practices among apple growers in New York (NY) and Pennsylvania (PA). We coupled the grower survey data with field observations of bees to (1) compare perceived and documented importance of native bees in orchards, (2) assess grower knowledge gaps and (3) guide future extension efforts for apple pollination.

Methods

Study system

Apple (Malus domestica Borkh: Rosaceae) is an economically important crop in temperate regions of the world, including eastern North America. NY and PA rank among the four largest apple-producing states in the United States, yielding on average 1.2 and 0.5 billion pounds of fruit, respectively, and collectively worth $350 million per year (USDA NASS 2016a, b). NY's apple industry is larger with 654 growers managing roughly 40,000 acres (USDA NASS 2016a) compared with PA's 566 farms over 20,000 acres (USDA NASS 2016b).

Apple is varietally self-incompatible, meaning flowers must receive pollen from another variety to set fruit; cross-pollination is largely carried out by insects (McGregor, 1976; Free, 1993). Renting honeybees to pollinate this mass blooming crop is commonplace in North America; however, several alternative pollinators currently exist, including commercially available, managed bumble bees (Bombus impatiens) and mason bees (Osmia spp.), as well as wild bees that naturally inhabit agricultural landscapes. Field surveys of orchard pollinators over the past century indicate that wild bees are common visitors to apple flowers, particularly species in the genera Andrena, Bombus, Halictus, Lasioglossum and Osmia (Hutson, 1926; Brittain, 1935; Phillips, 1933; Gardner and Ascher, 2006; Watson et al., 2011; Ritz et al., 2012; Mallinger and Gratton, 2015; Martins et al., 2015; Russo et al., 2015; Gibbs et al., 2017). Recent empirical studies find wild bees to be as effective, if not more effective, pollinators than honeybees and contribute important pollination services in apple orchards when abundant (Ritz et al., 2012; Mallinger and Gratton, 2015; Martins et al., 2015; Blitzer et al., 2016; Park et al., 2016; Russo et al., 2017). With only a couple of exceptions, wild bee species are native to study regions, whereas honeybees are introduced from Europe.

In addition to apples, northeastern fruit growers often grow other fruit species, such as peaches, nectarines, cherries, pears and various berries. Each of these have their own pollination requirements. A general improvement of wild pollinators in a given area would also benefit these crops (Biddinger et al., 2013a).

Grower surveys

Apple growers in NY and PA were surveyed on four major themes: (1) farm and grower characteristics, (2) current pollination strategies, as well as (3) perceptions and (4) attitudes regarding native and managed alternative pollinators (see Supplementary Material for survey instruments). A total of three survey instruments were administered to apple growers: two in NY and one in PA. In NY, we contracted the United States Department of Agriculture National Agricultural Statistics Service (USDA NASS) NY field office to administer grower surveys in 2009 and 2012. Both years, USDA NASS initially mailed surveys to commercial growers (518 in 2009 and 519 in 2012), then called non-respondents by phone until they reached a minimum 50% response rate (50.6% in 2009 and 57.4% in 2012). The 2009 survey instrument comprised 16 questions, addressing the four major themes, and served as a basis for subsequent surveys (Park et al., 2010). The 2012 survey instrument asked 25 questions, which incorporated questions developed by the PA team on farm characteristics, the use of managed alternative pollinators and perceived contributions of native bees to orchard pollination. We also included questions that assessed the importance of various factors on grower decisions to implement practices that would benefit native bees. For several questions asked in 2009, we added more response categories in 2012 from which growers could choose. Spatially, survey respondents represented growers throughout NY State when compared with the proportion of growers living within specific counties (USDA NASS 2007 and 2012 census data; Table 1). Identifying data were not disclosed by USDA NASS to authors in order to ensure respondents’ privacy.

Table 1. County residence of growers who participated in New York and Pennsylvania surveys compared with government censused distributions of apple growers among counties

Data are percentages and only the top nine counties shown.

^a 2007 USDA NASS census.

^b 2012 USDA NASS census.

In fall 2010, a survey questionnaire was distributed to PA apple growers with, but a few exceptions, the same questions in the NY surveys. Questionnaires were distributed to fruit growers during extension meetings and other extension events, such as field and plant protection days; the questionnaire was also available online. A total of 73 growers responded to this survey, the majority of whom were from Adams County, the main apple production region in the state. The spatial bias of PA survey respondents, from Adams and Lancaster counties, is likely due to the proximity of these growers to meeting locations where surveys were dispensed (Table 1).

Bee surveys

In order to compare grower perceptions and knowledge of native bees to ecological reality, we included data from orchard surveys of bees conducted in NY and PA. We surveyed a total 19 farms between 2009 and 2013 in central NY (Russo et al., 2015) and nine farms between 2007 and 2013 in PA (Joshi et al., 2015, 2016). In NY, all bees observed visiting apple flowers were net-collected along standardized 15 min transects during peak bee activity with temperatures above 60°F (Park et al., 2015). Bee visitation to apple flowers in PA orchards was recorded by observation and net collection at different distances, up to 200 m, from orchard edge (Joshi et al., 2016). Cumulative observed species richness was calculated for each farm across all years of collecting (NY data from Russo et al., 2015). NY bee abundance and species richness in statistical models were based on surveys conducted in 2011 and 2012. NY orchard sizes and percent land cover (i.e., natural, agriculture, developed), within a 1 km radius of farms, were determined using ArcMap10 GIS (ESRI, 2010; see further details in data analysis). To assess drivers of bee community abundance and diversity in PA orchards, we referenced previously published surveys, for which configuration of the adjacent habitat and landscape were characterized by Fragstats 4.0 up to 1 km from orchard edge (Joshi et al., 2016).

Data analysis

Grower survey results were summarized using descriptive statistics. To facilitate comparisons among the three survey questionnaires, some NY 2012 and PA 2010 survey multinomial response variables were collapsed into fewer categories or translated to a common format. For example, several PA survey questions that provided a five-scale Likert response (always, frequently, sometimes, never, don't know) were regrouped into a three-scale response (no, yes, maybe) or redefined (e.g., very, moderately, slightly, not at all, don't know). The χ² tests were used to compare frequencies of categorical responses between years and states. Univariate analyses were used to compare means of continuous response variables among different levels of categorical factors. Non-parametric Kruskal–Wallis rank-sum tests were employed when assumptions of equal variance for analysis of variance were not met. We employed generalized linear models (GLM) to explore pre-defined relationships between farm/grower characteristics and perceptions of native pollinators. Specifically, we predicted that perceived diversity, value of native bees as pollinators and openness to relying exclusively on native pollinators would increase as acres in apple production decreased and would be highest for those farms surrounded by natural areas. These predictions are based on the strong link between healthy native pollinator communities in fields close to natural areas (Ricketts et al., 2008; Kennedy et al., 2013), and the assumption that growers may be aware of increased native bee activity in such orchards. To test the effects of farm size, state (NY or PA) and adjacent habitat (collapsed into natural, agricultural or other-mixed) on estimated number of species, we used a negative binomial GLM (Zuur et al., 2013). To test the same effects on whether native bees were considered valuable for orchard pollination (yes, no, maybe) and whether growers had considered relying exclusively on naturally occurring native bees (yes, no, maybe), we conducted multinomial logistic regressions. Only main effects of size, state and habitat were included and multinomial logistic models were not reduced.

We used descriptive statistics to summarize bee abundance and species richness recorded in orchard surveys. For NY, in parallel with grower survey analyses, we investigated effects of orchard size and surrounding natural habitat (at 1 km radius) on native bee abundance and cumulative observed species richness to see if patterns in the bee data reflected grower perceptions of bee activity and importance in orchards. Size and amount of natural habitat in the landscape were calculated using GIS (ArcMap 10, ESRI). For size, we followed property boundaries of orchards, which does not account for the fact that orchards may be adjacent to other orchards, potentially rendering actual orchard area much larger from a bee's perspective. Assessing amount of semi-natural areas within a given distance of orchard center can account for other orchards nearby; we used 1 km as our buffer radius because it has been found to be a strong predictive scale for bee response variables (Kremen et al., 2002; Watson et al., 2011). We did not want to go beyond 1 km as we wanted land cover to reflect what a grower would consider the farm's surrounding habitat. GLM and generalized linear mixed models were used to analyze diversity and abundance, respectively. In the abundance model, we included farm as a random effect since repeat collecting events occurred within a year, and because of the strong relationship between bee activity and temperature, we included log-transformed temperature as a covariate. Diversity data were pooled at the farm level, and therefore, did not include temperature or random farm effects in models with diversity as the response variable.

We conducted regressions in R software (R Core Team, 2013), using ‘MASS’ and ‘nlme’ packages (Venables and Ripley, 2002; Pinheiro et al., 2017); all other analyses were performed in SPSS (IBM Corp, 2013). For general linear models, assumptions of normality and homoscedasticity were met. For negative binomial regressions, we verified that models were not overdispersed (Zuur et al., 2013).

Results and discussion

Grower and farm characteristics

NY and PA demographics were largely similar with some notable differences in farm size, diversity and surrounding habitat (Table 2 summarizes results for questions on grower and farm characteristics). On average, growers in our study region had close to 100 acres in apple production, harboring over 15 varieties. Across states, fewer than 10% of growers had more than 200 acres in apple production. Twice as many PA growers owned orchards smaller than 10 acres than NY growers. Percent grower income derived from apple was distributed evenly among three collapsed categories: 0–25, 25–75, 75–100%, and increased significantly with orchard size (χ₃² = 164.6, P < 0.0001; Fig. 1; question only asked in NY 2012 questionnaire). Crop diversity was higher, but diversity of surrounding habitat types lower for PA growers. Significantly more PA participants grew stone fruit (χ₁² = 39.4, P < 0.001) and vegetables (χ₁² = 8.6, P = 0.003) in addition to apple. NY growers reported a full gradient of habitat types surrounding their orchards with agriculture, mixed and forest being the most common. In contrast, most PA orchards seemed to be surrounded by either forest or agriculture and little in between. Thus, we recorded a continuum of operation size and amount of natural habitat adjacent to orchards across study regions, with more specialized, commercial apple growers in NY than PA.

Fig. 1. Correlation between financial reliance of growers on apple production and amount of land planted in apple.

Table 2. Grower demographics, New York (2009 and 2012) and Pennsylvania apple 2010

s.e.m. provided with means in parentheses. Blanks indicate the questions or specific responses were not included in surveys.

In terms of pest management strategies, a majority of growers in both states reported using primarily IPM practices. Both PA and NY have well-established IPM programs, developed over the last 40 yr for apple production systems by state, land grant institutions (Kovach and Tette, 1988; Rajotte et al., 1992). By incorporating reduced-risk pesticides (Agnello et al., 2009), sex-pheromone-based mating disruption products (Joshi et al., 2011), as well as pest monitoring and forecasting tools (Damos and Savopoulou-Soultani, 2010), IPM offers reduced-risk pest management programs that are environmentally safer than conventional pest management programs in commercial fruit production (Agnello et al., 2009; Biddinger et al., 2014). Up to a quarter of growers continue to rely on conventional pest management; fewer than 10% of growers manage their orchards organically. Organic apple production is relatively rare in the study region due to high disease and pest pressure (Agnello et al., 2003). Thus, a majority of growers across the region currently follow pest management programs that encourage diligent use of pesticides, by monitoring pest pressure and spraying only when pest damage causes economic harm (Agnello et al., 2009; Jones et al., 2010); such practices have the potential to be readily modified to accommodate pollinators in commercial apple production (Biddinger and Rajotte, 2015).

Pollination strategies

Concern over reliable pollination was high among all growers. Study-wide, between 36 and 52% growers reported having experienced limited apple pollination due to inadequate visitation by pollinators (see Table 3 for results summary of questions addressing pollination strategies). Several growers commented that bad weather was an important driver of low bee activity. A particularly wet spring in 2011 and cold spring in 2012 may have contributed to higher reports of pollination limitation in 2012 as compared with 2009 in NY. Providing a Likert scale of response options (i.e., always, frequently, sometimes, never, don't know) in the two most recent surveys revealed that for the majority of growers, pollinator limitation occurs only sometimes. Only 2–5% of blossoms, compared with 80% for cherry, need to be set for a viable commercial apple crop; however, growers seek higher pollination rates to maximize fruit quality, which, ultimately, dictates market value (Westwood, 1993). Because the king bloom (the center flower of a five-flower cluster) produces the largest fruit, growers will overpollinate to ensure high set of king bloom and thin blossoms (chemically or mechanically) to avoid biennial bearing.

Table 3. Pollination strategies and concerns among New York and Pennsylvania apple growers

Recent declines in honeybee populations due to CCD were considered a threat to successful apple production by the majority of growers surveyed, but a sizeable percentage were uncertain about the impacts of CCD (Table 3). Grower concerns over the negative impacts of CCD on pollination services echo those found among blueberry growers in Maine (Hanes et al., 2015).

To our surprise, only about 50% of growers reported renting honeybees for apple pollination in both NY and PA (χ₁² = 2.1, P = 0.2; Table 3). The probability of renting bees increased directly with farm size (χ₁² = 11.8, P = 0.001, Fig. 2) and was similar between states (state × acre: χ₁² = 2.7, P = 0.1). Because number of pollinators required for adequate pollination increases directly with farm size, smaller orchards likely do not need supplemental honeybees if located near other operations that rent hives or semi-natural habitat that supports native pollinators (Park et al., 2015). Growers renting honeybees stocked their orchards at similar densities (NY: 2.5 ± 0.4 hives/ac, n = 181; PA: 1.9 ± 0.2 hives/ac, n = 44; t ₂₂₃ = 0.752, P = 0.5) and paid comparable prices per hive (NY: $63.90 ± 2.8; PA: $59.30 ± 4.40; t ₁₅₁ = 0.631, P = 0.5). In NY, regardless of farm size, a majority of growers considered honeybee rentals a minor to moderate expense. Similar results were found for PA growers with <100 acres in apple production; however, most large-scale PA growers, with 100–500 acres in apple (n = 6), described honeybee rentals to be a major expense (within PA, χ₁₅² = 26.6, P = 0.03). For most producers, hive prices may not have yet inflated to the point where growers would be economically motivated to invest in other pollinator strategies, especially if there is perceived risk in doing so. In PA, a network of growers, including large-scale operations, have demonstrated that adequate pollination can be achieved by relying on native bees alone (Biddinger, pers. obs.). Such demonstrated success is testimony that hive rentals are unnecessary in some orchards within the study region and likely inspire others to try alternative pollination even if economic benefits are not major.

Fig. 2. Proportion of apple growers in New York (NY) and Pennsylvania (PA) that reported renting honeybees for pollination increased with farm size. NY grower surveys were conducted in 2009 (NY09) and 2012 (NY12).

Grower adoption of bumble bees, as an alternative managed pollinator, was not trivial; however, use and awareness of mason bees was low. In NY and PA, 16.7 and 24.3% growers, respectively, reported using bumble bees, at least, sometimes. Few growers reported having used commercial mason bees for apple pollination, with a study-wide maximum of 8% growers in PA. Commercial bumble bees are more expensive than honeybees, but forage more reliably in cooler, early spring temperatures (Goulson, 2003). Mason bees specialize on fruit trees and are highly effective pollinators (Bosch and Blas, 1994). The native blue orchard bee, Osmia lignaria, is rarely collected in NY (Park et al., 2015) or PA orchards (Joshi et al., 2015, Joshi et al., 2016); however, the introduced Japanese horn-faced bee, O. cornifrons, is well-established throughout the east coast. There is interest in further developing O. cornifrons as an alternative managed pollinator in our study region (Biddinger et al., 2013a, b; T. Pitts-Singer pers. comm.).

Regardless of year and state, almost all (>93%) surveyed growers reported that they already considered pollinator safety when applying pesticides in orchards. Across the study region, apple is an intensively sprayed fruit crop due to intense pest and disease pressure (Agnello et al., 2009). Orchard pesticides have been shown to impact both managed and wild pollinators (Biddinger et al., 2013b; Mallinger et al., 2015; Park et al., 2015). Aside from intrinsic motivations to protect pollinators, growers have many practical reasons to be cognizant of pollinators when considering their pest management options: (1) adequate fruit set for crop production depends on adequate pollination; (2) growers often pay to have bees in the orchard, so harm to bees would be financially counterproductive; and (3) label guidelines restrict the use of insecticides during bloom when bees are most active in orchards. Judicious use of insecticides during bloom is important to reduce pesticide risk for pollinators; however, care should also be taken outside the bloom period. Pollinators are active within orchards before and after apple bloom, foraging ground cover floral resources and/or nesting. Care must also be taken when using pesticides traditionally considered safe for bees, such as fungicides and herbicides. Both fungicides and insecticides applied when apple was not in bloom decreased wild pollinator, but not honeybee, visitation and diversity in NY orchards (Park et al., 2015). The lack of a measurable response of honeybees to pesticides likely results from placing hives in orchards only during bloom and to the tendency of honeybees to forage on non-apple resources at larger spatial scales than wild bees (McArt et al., 2017). Precautions taken by growers during bloom are, therefore, inadequate to maximize pollination services by wild bees. Efforts to promote alternative bee pollination services should, therefore, raise awareness of the vulnerability of alternative pollinators to pesticide applications throughout the growing season.

Knowledge and perceptions of native pollinators

Grower estimates of bee diversity were relatively low, and half of our NY 2012 questionnaire respondents chose don't know when given this option. Choosing from a range of values (NY:1, 10, 30, 40, 100; PA:1, 10, 50, 100, 200, 300), growers estimated that apple flowers are visited by a median of ten bee species, compared with 100 and 52 bee species recorded in field surveys in NY and PA, respectively. Grower estimates, however, approached cumulative bee species richness netted within a single orchard (NY: 15–51 and PA: 10–25). Additionally, low species estimates may reflect a lay person's ability to accurately identify bees based on easily recognizable morphological groups. Bee species are commonly distinguished by characteristics only visible under a microscope, making it challenging to differentiate species in the field. For this reason, native bees are commonly lumped into morphological groups (e.g., ‘metallic green bee’) to facilitate observations of bee visitation by lay persons. Following ‘Pennsylvania Citizen Scientist Pollinator Guide’ (Donovall and vanEngelsdorp, 2008), apple bees in our study region represent 12 distinct morphological groups, which mirror median grower diversity estimates of ten species. In sum, growers demonstrated a wide knowledge gap in terms of sheer diversity of native pollinators, with many simply unwilling to guess. This gap parallels the general public's lack of pollinator literacy, having only recently considered non-honey bee species as important crop pollinators. Of the respondents who did provide diversity estimates, many possessed an accurate local, lay knowledge of the bee fauna visiting their orchards.

In biological surveys, native bee abundance and species richness were significantly and positively influenced by the amount of semi-natural area close to orchards but not by orchard size (Table 4, Joshi et al., 2016). We, therefore, predicted that grower estimates of bee diversity may be higher among growers whose orchards were surrounded by semi-natural areas (i.e., forest, mixed, meadow), but no such relationship was observed (Table 5). Nor did we observe an effect of the number of acres a grower had in apple on their bee diversity estimates. Whether we measured a lack of awareness of the bee fauna visiting orchard or, again, an inability to distinguish bee species is unclear.

Table 4. Effects of orchard size and percent semi-natural areas within a 1 km buffer of orchard on bee abundance (GLMM) and observed bee species richness (GLM) in New York State

For bee abundance [ln(y + 1) transformed], covariates temperature [ln(x) transformed] and year, as well as a random farm factor were included. All predictors but year were mean centered. Coefficients are not back-transformed. ‘–’ indicates the predictor was not included in the full model.

Table 5. Effects of state, operation size and habitat adjacent to orchards on (1) grower estimates of native pollinator diversity in apple orchards, (2) whether growers considered native bees valuable to apple pollination and (3) whether growers had considered relying exclusively on native bees

Habitat categories were collapsed into agriculture, natural and other. ‘Other’ included mixed and suburban habitats. For estimates of bee diversity, a negative binomial GLM was employed and reduced via backwards stepwise regression. Predictors that were not significant (at α = 0.05) but contributed significantly to model fit were retained. Multinomial logistic regressions were conducted on response variables with three levels: yes, no and maybe and were not reduced. Coefficients in multinomial logistic regression are log odds ratios. Significant effects are bolded.

We recorded a high appreciation for native pollinators among grower participants, but again some uncertainty about how much native bees actually contribute to orchard pollination. Native bees were viewed by 85–93% surveyed growers as valuable pollinators in orchards; this high appreciation did not change with farm characteristics or state (Table 5). Blueberry (Hanes et al., 2015) and cranberry (Gaines-Day and Gratton, 2017) growers shared similarly high appreciation for native pollinators. When asked to rate the value of native bee contributions to apple pollination (PA 2010 and NY 2012 questionnaires), over 50% of growers chose the highest possible ranking (i.e., 53.4% PA: always; 63.6% NY: very important); only 6.4 and 8.2% of NY and PA growers, respectively, reported not knowing the value of native bees for their apple orchards. Across the study region, growers estimated that native bees contribute half of orchard pollination services (NY: 45 ± 1.6%, PA: 51 ± 3.5%; t ₃₀₀ = −1.6, P = 0.1), but individual estimates ranged widely and 20% of respondents expressed that they did not know (available answer in the NY 2012 survey only). When asked whether alternative managed pollinators, such as mason or bumble bees, were important for apple pollination (PA-only question), 68% were evenly split among alternative managed pollinators being always, frequently or sometimes important; 6% reported they were never important; and 29% reported that they did not know. Grower uncertainty in the importance of non-Apis pollinators is understandable: contributions of native bees have only recently been quantified (Ritz et al., 2012; Mallinger and Gratton, 2015; Blitzer et al., 2016; Park et al., 2016), and efforts to use other managed pollinators are still new.

Over half of the study participants had previously considered relying exclusively on native pollinators (Table 3). As one might expect, more growers who did not rent honeybees (72.4%) had considered relying on native bees than those who did rent honeybees (44.2%, χ² = 63.9, d.f. = 2, P < 0.0001); and yet, almost half the growers who rented honeybees had thought about using alternative pollinators and may be especially receptive to integrated crop pollination (Isaacs et al., 2017). Whether a grower had considered relying exclusively on native pollinators was most influenced by farm size (Fig. 3, Table 5). As acreage in apple production increased, grower consideration to rely exclusively on native bees decreased (Table 5). In contrast, orchard size did not have a significant effect on bee visitation or diversity in biological surveys (Table 4). Because percent total income derived from apple production increased significantly with acres in apple production (χ² = 151.6, d.f. = 5, P < 0.0001; Fig. 2), we conclude that willingness to risk inadequate pollination decreases as apple becomes a greater source of one's income. Importance of producing high-quality fruit to increase one's crop value and meet consumer expectations likely increases with operation size. Nonetheless, even among the largest apple operations, a third of growers had considered foregoing honeybee rentals.

Fig. 3. Grower consideration to rely exclusively on native pollinators decreased with farm size. NY grower surveys were conducted in 2009 (NY09) and 2012 (NY12).

Grower openness to relying exclusively on native pollinators also depended on state and habitat types adjacent to orchards (Table 5). PA growers (67.2%) were significantly more likely to have considered relying on native bees than NY growers (2010: 51.4%; 2012: 60%). We speculate two compatible reasons for this: (1) PA growers had greater exposure to native bee crop pollination, and (2) growers accurately perceived increased native pollinator activity in orchards surrounded by semi-natural habitat. PA growers may have gained more exposure to the idea of relying on native bees from a network of growers already using such practices successfully and from increased extension activities. First, even before hive rental fees had tripled after CCD hit in 2006, a few large PA growers had successfully experimented with not renting honeybees and relying exclusively on native bees; this success provided other growers with a real demonstration that such a pollination strategy was a viable option for commercial production (Biddinger, pers. obs.). Secondly, pollinator extension in PA was likely more active, before and during the study period, due to the presence of the Center for Pollinator Research at Pennsylvania State University (http://ento.psu.edu/pollinators). Native pollinator extension was conducted by Biddinger, an established, fruit entomologist stationed at the Pennsylvania State University Fruit Research and Extension Center. Native pollinator extension in NY was largely conducted by Park, a student at the time, from 2010 to 2012, in the form of annual grower talks, a pollinator booklet, and a Department of Entomology website (entomology.cornell.edu/wildpollinators). Finally, PA respondents may have represented a more biased pool of participants given they were surveyed at extension events.

Increased PA grower openness to relying on native pollinators could also be linked to PA grower ownership of smaller orchards located near semi-natural areas (Table 2), a documented source of wild pollinators visiting orchards (Watson et al., 2011; Park et al., 2015). Amount and proximity of adjacent natural areas, primarily forest, was a strong predictor of native pollinator abundance and diversity in our field surveys in NY (Table 4) and PA (Kammerer et al., 2016b; Joshi et al., 2016). Congruently, consideration to rely exclusively on native pollinators was significantly lower for growers whose orchards were surrounded by agriculture than for those whose orchards were near natural or mixed/suburban areas (Table 5). Regardless of state differences, these results suggest an awareness among growers about the levels of native pollinator activity occurring within orchards, and a surprising openness to relying exclusively on native pollinators.

To gauge willingness of growers to enhance native pollination, we asked if they would consider low-cost land management practices that would increase native bees in their orchard. Consistent across surveys, a majority of growers indicated that they would consider such action, with 85% PA compared with 50–68% NY growers responding yes. Growers were also asked about their knowledge of and participation in federal cost-share programs, designed to aid grower efforts to create or maintain pollinator habitat. The Food, Conservation and Energy Act of 2008 provided federal funding to conserve and protect pollinators in agricultural ecosystems (Whittingham, 2011). As a result, growers receive government financial incentives and technical support to adopt pollinator-friendly production practices (Decourtye et al., 2010). In our study region, native bees are pollen foragers, and a continuous source of mixed floral resources near orchards could be important in conserving and maintaining healthy population of these bees (Kammerer et al., 2016a, b). A high proportion of growers (NY: 91%, PA: 75%) reported not knowing about government cost-share programs; of those who did, only 8% of NY and 16% PA growers were enrolled. Thus, apple growers seemed generally open to relying more on native pollinators; however, many were not aware of the resources available to them to enhance native bee habitat in their orchards.

To identify obstacles preventing growers from actively enhancing native pollinator populations in orchards, we asked growers to rank the importance of several factors (from not at all important to very important) that would influence their decision to implement land management changes for native bees (Fig. 4). Of the top three factors, proven effectiveness of native pollinators was most important, clear guidelines to implement management practices was second and environmental stewardship was third. Recent empirical studies now provide strong evidence for the effectiveness and importance of native pollinators for apple production (Ritz et al., 2012; Mallinger and Gratton, 2015; Martins et al., 2015; Park et al., 2016; Blitzer et al., 2016; Russo et al., 2017). Pollination provided by native bees depends directly on their abundance (i.e., the more bee visits the better the pollination); abundance of native bees varies across orchards due to differences in pesticide use and amount of natural area in the surrounding landscape (Mallinger et al., 2015; Park et al., 2015). In order for growers to assess the native pollination services available in orchards, accessible protocols to monitor native bee populations need to be developed (Hanes et al., 2015). Citizen science projects, such as the Northeast Pollinator Partnership (http://www.northeastpollinatorpartnership.org/), are a plausible approach to forward such efforts, by (1) gathering large amounts of data on bee abundances, and (2) making data-informed recommendations back to growers about what pollinator strategies they should adopt. Making the scientific evidence on contributions of native pollinators accessible to growers should be a primary goal of future extension efforts.

Fig. 4. Importance of factors that would influence grower decisions to adopt new land management practices to protect or enhance native pollinators.

Growers need adequate technical support and guidance to implement bee-friendly practices (Biddinger and Rajotte, 2015; Isaacs et al., 2017). Because native pollination services vary among orchards, depending on pesticide use and surrounding habitat, one cannot prescribe a single strategy that fits all. Site-specific assessment of grower pollination management options would be ideal. For example, growers with orchards near large tracts of forest could potentially rely exclusively on native bee pollination and focus on minimizing impacts of pesticides. In contrast, a grower whose orchards are surrounded by intense agriculture would focus on creating more pollinator habitat. Habitat enhancements for pollinators require technical support, to identify appropriate seed mixes and manage weeds. Growers enrolled in CRP receive technical advice from their local NRCS office, and non-profit organizations are providing increasing guidance nationally (e.g., The Xerces Society). Local extension programs with established ties to the apple industry also have potential to provide technical assistance and are poised to help balance grower needs to control pests and maximize pollination services (Biddinger and Rajotte, 2015).

Willingness of growers to manage orchards in a manner that would support native pollinators was motivated more by a sense of environmental stewardship than by cost. First, honeybee rentals were not perceived as a major expense by most growers, so if they were considering making changes, it makes sense that motivations not be financial and rather driven by concern for pollinators and their services. Secondly, the fact that cost was not one of the top three motivations may help explain grower lack of interest/awareness in government cost-share programs. Thirdly and most importantly, if a large pool of growers are motivated by environmental stewardship, they are also likely receptive to outreach and support for managing lands in more bee-friendly ways. Pollinator habitat creation at the landscape scale has been forwarded as a means by which society can increase food sustainability, by enhancing native pollination services for crops, and can conserve diversity in agricultural landscapes (Potts et al., 2011). For such a coordinated vision to become a reality, additional outreach and extension support on crop pollination are needed (Hanes et al., 2015).

Conclusions

Integrating biological and grower survey data allowed us to explore grower awareness of pollinators as well as factors influencing grower perceptions. We found overwhelming support among eastern apple growers for the importance of native bees, an openness to rely more on naturally occurring bees, and willingness to make low-cost changes to enhance native bee populations. Already 50% of growers in the study region rely on ambient pollination by not renting honeybees; even more have, at least, considered relying on native pollinators exclusively. At the same time, we documented sizeable uncertainty among growers about the effectiveness of native and alternative managed bees for apple pollination, as well as a tendency to rent honeybees to maximize production value. Uncertainty about non-honeybee pollinator effectiveness was the largest obstacle reported by growers to actively managing orchards for native pollinators. Empirical evidence for the importance of wild pollinators in apple orchards has grown within our study regions (Ritz et al., 2012; Blitzer et al., 2016; Park et al., 2016) and beyond (Mallinger et al., 2015; Martins et al., 2015). This seems an opportune time to take the scientific evidence and encourage growers to incorporate alternative pollinators into their pollination strategy. By no means do we advocate that growers abandon the use of honeybees altogether, but a more integrated pollination management approach seems viable in our study regions and may ensure growers continue to receive optimum pollination in the face of volatile honeybee supplies (Isaacs et al., 2017). Proactive steps to support alternative orchard pollinators will not only benefit honeybees and biological control agents, such as parasitic flies and wasps that require nectar and pollen as adults, but also other pollinator-dependent crops often planted nearby. With a strong history of IPM in the region, extension programs provide an existing infrastructure in which to develop technical and informational support to ensure sustainable food production systems that rely on insect pollination, like apple (Biddinger and Rajotte, 2015).