Phenology of Honey Bee Swarm Departure in New Jersey, United States

Phenology of Honey Bee Swarm Departure in New Jersey, United States Abstract Departure of swarms from honey bee (Apis mellifera Linnaeus (Hymenoptera: Apidae)) nests is an important reproductive event for wild honey bee colonies and economically costly in managed bee colonies. The seasonal timing of swarm departure varies regionally and annually, creating challenges for honey bee management and emphasizing the potential for swarming behavior to be affected by plant-pollinator phenological mismatch. In this study, we first document variability in the timing of swarm departure across the large and heterogeneous geographical area of New Jersey over 4 years using 689 swarm-cluster observations. Second, hypothesizing that honey bee colonies adaptively tune the timing of swarm departure to match floral food-resource availability, we predicted that growing degree-days could be used to account for regional and annual variability. To test this idea, we used local weather records to determine the growing degree-day on which each swarm cluster was observed and tested for differences among climate regions and years. The state-wide mean swarm cluster date was May 15 (± 0.6 d), with moderate but significant differences among the state’s five climate regions and between years. Use of degree-day information suggests that local heat accumulation can account for some climate-region differences in swarm-departure timing. Annual variation existed on a scale of only several days and was not accounted for by growing degree-days, suggesting little adaptive tuning of swarm-departure timing with respect to local heat accumulation. phenology, swarm, growing degree-day, climate, temperature Swarm departure in honey bees (Apis mellifera L.) is the culmination of a complex colony fission process involving the buildup of worker population, rearing of replacement queens, preparation of the mother queen for flight, swarm departure, and monogyny restoration (summarized in Winston 1987, reviewed by Tarpy and Gilley 2004, Gilley and Tarpy 2005, Grozinger et al. 2014). Swarm departure results in significant economic loss for beekeepers by reduction of colony worker population, interruption of the brood cycle, and potential failure to re-queen. The timing of swarm departure is an adaptive decision by a honey bee colony (Seeley and Visscher 1985) that takes place on two timescales. In the short term, the day and hour of swarm departure is chosen to avoid poor weather (Henneken et al. 2012) and prevent fighting between the mother queen and daughter queen (Gilley and Tarpy 2005). In the long term, on a seasonal time scale, the timing of swarm departure is chosen to minimize the likelihood of colony death from late frost and maximize the likelihood of matching colony growth to bloom times of important forage plants (Seeley and Visscher 1985). We focus here on the timing of swarm departure on this seasonal scale because it is most directly relevant to swarm management practices and is most likely to be affected by mismatch between plant and pollinator phenologies caused by climate change (‘phenological mismatch,’ see Stenseth and Mysterud 2002). The seasonal timing of swarm departure has been documented for a number of locations around the world, including many cities and locales within North America (Mitchener 1948, Burgett and Morse 1974, Fell et al. 1977, Page 1981, Seeley and Visscher 1985, Villa 2004). Informal comparison of these studies suggests substantial variation in swarm-departure timing on a national scale (e.g., among central California, upstate New York, and southeastern Louisiana); formal meta-analysis is complicated by differences in sampling-area size and sampling approach. Annual variation also appears substantial (e.g., Villa 2004) but has not often been fully reported. Overall, current documentation of swarming phenology does not seem adequate to detect potential future phenological mismatch resulting from spatial and temporal temperature changes, which are likely to affect swarm-departure timing due to the dependence of this process on floral bloom timing as described below. More information is needed across large regions for longer periods of time to enable us to detect phenological mismatch and to overcome the challenges regional and annual variability creates for honey bee management. Effective swarm prevention typically involves identifying a causal factor leading to swarm departure, developing a procedure to disrupt that factor, and determining when to apply the disruption procedure (e.g., Wright 2002a,b). Optimal timing can be determined from knowledge of the typical earliest swarm departure date together with an estimate of the number of days that elapse before the causal factor results in swarm departure. Precise estimates of swarm timing are important for success because early intervention can be as unsuccessful as late intervention. Regional variation compounds these challenges by making specific management solutions difficult to apply in other locations. A likely driver of the seasonal timing of swarm departure is the bloom timing of key forage plants, which are a necessary prerequisite for the growth and reproduction of a honey bee colony. A strong relationship between swarming and the flowering of major honey bee plant genera was found in a meta-analysis of 23 regions of Africa (Hepburn and Radloff 1995). The bloom timing for a particular plant species can be predicted with some accuracy by measuring accumulated thermal energy as the cellular physiology that drives plant growth is temperature dependent (Higley et al. 1986). Accumulation of thermal energy can be measured using degree-days (sometimes referred to as growing degree-days [GDD]), which is calculated most simply for each day by subtracting a chosen base temperature (the standard used for many applications is 10°C [50°F]; Herms 2004) from the daily average temperature. Degree-days summed during the course of the growing season provide a metric of cumulative thermal energy that has been applied to pest insect development for decades (e.g., Mussey and Potter 1997), is important in studies linking climate change to plant-pollinator phenology (e.g., Hodgson et al. 2011), and has even been extended to growth in fish (Neuheimer and Taggart 2007). Given the dependence of honey bee colony growth and development on plant bloom timing, it is reasonable to hypothesize that there is a relationship between the timing of honey bee swarm departure and degree-day accumulation. The goals of this study are 1) to document the timing and variability of swarm departure across a large and geographically heterogeneous area and 2) to investigate the relationship between growing degree-days and the timing of swarm departure. Traditional approaches to documenting swarm departure are to closely monitor a small number of hives for swarming activity (e.g., Simpson 1959, Simpson and Moxley 1971, Seeley and Visscher 1985) or to rely on observations contributed by local beekeepers (e.g., Mitchener 1948, Burgett and Morse 1974, Fell et al. 1977, Villa 2004). Modern communications and database tools allow expansion of the latter approach to address phenological (e.g., Henneken et al. 2012) as well as public safety issues (e.g., Renato et al. 2009) related to honey bee swarming behavior. Here, we seek to address the first goal using data from a targeted effort to catalog swarm-cluster observations across the state of New Jersey between 2013 and 2016. To address the second goal, we use local weather records to determine the degree-day of each swarm departure, predicting that this measure of local heat accumulation will help account for variation in swarm-departure timing. Methods Data Collection Observations of swarm clusters were reported by observers across the state of New Jersey. These observations were submitted to the ‘New Jersey Swarm Report,’ a long-term project created and managed by members of the New Jersey Beekeepers Association. Since the project’s establishment in 2013, beekeepers have been urged by New Jersey Beekeepers Association leadership to make a report any time they encountered a swarm. Participants were able to submit their observations via the Swarm Report website or via email to the project’s coordinators. The information requested for each swarm was as follows: date captured or observed, swarm location address (including ZIP code and county), size of swarm (small/medium/large), and height of swarm (less than 6 feet/between 6 feet and 10 feet/higher than 10 feet) (where 6 feet = 1.8 m and 10 feet = 3.0 m). Between 1 January 2013 and 1 January 2017, 755 swarm clusters were reported. The raw data from the Swarm Report, which is not available via the website and has not been previously published, was communicated to us by the Swarm Report project coordinators. The analysis presented here uses all swarm observations between 1 January and 15 July of each year. Swarms reported after 15 July likely represent the beginning of a second minor cycle of swarming in the late summer, which we consider a distinct phenomenon from both biological and apicultural perspectives. A small number of incomplete swarm reports, which did not contain sufficient information for this analysis, were also excluded from the dataset, as were any obvious duplicate reports. The total number of swarm observations in the final dataset was 689. We assume that each reported cluster represents a unique swarming event, following a swarm departure from either a feral nest or a managed hive. False-positive identifications may exist in these data but are unlikely as the majority of observers are beekeepers and/or associated with the New Jersey Beekeepers Association, plus the details requested for a report are likely to dissuade false or inaccurate reports (e.g., mistaking a wasp nest for a honey bee swarm). Analysis Variables For each of these 689 swarms we determined the following variables. Calendar Day The calendar day number (from 1 to 365) was determined for each swarm. The year 2016 was a leap year, which was taken into account when converting reported date to calendar day. Degree-day The local cumulative degree-day on which each swarm cluster was observed was calculated using public information tools available from Northeast Regional Climate Center, based on data from weather stations across New Jersey. We used the NRCC CLIMOD 2 web tool to determine, based on reported ZIP code, the identity of the weather station closest to each swarm. If the nearest weather station had more than five missing daily records during the swarming period (a rare occurrence), then the next closest weather station to the swarm cluster was used. Degree-days were calculated for each weather station by the CLIMOD 2 web tool using the average-temperature method (growing degree days) with a base temperature of 10°C (50°F), and accumulation beginning on January 1 of each year. The degree-day for each swarm was defined as the cumulative degree-day on the reported date at the closest weather station to the swarm cluster. Climate Region The State of New Jersey, considered by the United States Geological Survey as a part of the Mid-Atlantic Region, consists of 22,600 km2 of land, ranging from sea-level to 550 m, includes a large coastal plain (~50% of area), significant forest area (~40% of area), and some agricultural land (~20% of area). Five climate zones are defined by the Office of the New Jersey State Climatologist (following Ludlum 1983): North, Central, Southwest, Pine Barrens, and Coastal. These zones we considered to be the most likely boundaries influencing the timing of reproductive swarming. In addition, the size of these regions allowed for sufficient sample sizes for comparative analysis. Full characterization and maps of the boundaries of these climate zones is available from the Office of the New Jersey State Climatologist (Rutgers University) and Ludlum (1983). Because the boundaries may be useful for readers of this study, approximate boundaries of these zones are given in Table 1. Each swarm was assigned to one of these five climate regions based on the location of the weather station nearest to the observed swarm cluster. Table 1. Boundaries of the climate regions used in this study Region name Description of boundaries Counties included North The inland portion of the northern half of the state Sussex, Warren, Morris, Hunterdon; western halves of Passaic and Somerset Central The coastal portion of the northern half of the state and the central inland portion of the state Bergen, Essex, Hudson, Union, Middlesex, Mercer; eastern halves of Passaic and Somerset Coastal The coastal border, ~20 km wide, of the southern half of the state Cape May; eastern halves of Monmouth, Ocean, and Atlantic Southwest The western portion of the south half of the state, bordering the Delaware Bay and Delaware River Western halves of Burlington, Camden, Glouchester, Salem; southwestern half of Cumberland Pine Barrens The interior of the southern half of the state Central Burlington; eastern halves of Camden, Glouchester, Salem; northeastern Cumberland; western halves of Monmouth, Ocean, and Atlantic Region name Description of boundaries Counties included North The inland portion of the northern half of the state Sussex, Warren, Morris, Hunterdon; western halves of Passaic and Somerset Central The coastal portion of the northern half of the state and the central inland portion of the state Bergen, Essex, Hudson, Union, Middlesex, Mercer; eastern halves of Passaic and Somerset Coastal The coastal border, ~20 km wide, of the southern half of the state Cape May; eastern halves of Monmouth, Ocean, and Atlantic Southwest The western portion of the south half of the state, bordering the Delaware Bay and Delaware River Western halves of Burlington, Camden, Glouchester, Salem; southwestern half of Cumberland Pine Barrens The interior of the southern half of the state Central Burlington; eastern halves of Camden, Glouchester, Salem; northeastern Cumberland; western halves of Monmouth, Ocean, and Atlantic View Large Table 1. Boundaries of the climate regions used in this study Region name Description of boundaries Counties included North The inland portion of the northern half of the state Sussex, Warren, Morris, Hunterdon; western halves of Passaic and Somerset Central The coastal portion of the northern half of the state and the central inland portion of the state Bergen, Essex, Hudson, Union, Middlesex, Mercer; eastern halves of Passaic and Somerset Coastal The coastal border, ~20 km wide, of the southern half of the state Cape May; eastern halves of Monmouth, Ocean, and Atlantic Southwest The western portion of the south half of the state, bordering the Delaware Bay and Delaware River Western halves of Burlington, Camden, Glouchester, Salem; southwestern half of Cumberland Pine Barrens The interior of the southern half of the state Central Burlington; eastern halves of Camden, Glouchester, Salem; northeastern Cumberland; western halves of Monmouth, Ocean, and Atlantic Region name Description of boundaries Counties included North The inland portion of the northern half of the state Sussex, Warren, Morris, Hunterdon; western halves of Passaic and Somerset Central The coastal portion of the northern half of the state and the central inland portion of the state Bergen, Essex, Hudson, Union, Middlesex, Mercer; eastern halves of Passaic and Somerset Coastal The coastal border, ~20 km wide, of the southern half of the state Cape May; eastern halves of Monmouth, Ocean, and Atlantic Southwest The western portion of the south half of the state, bordering the Delaware Bay and Delaware River Western halves of Burlington, Camden, Glouchester, Salem; southwestern half of Cumberland Pine Barrens The interior of the southern half of the state Central Burlington; eastern halves of Camden, Glouchester, Salem; northeastern Cumberland; western halves of Monmouth, Ocean, and Atlantic View Large Statistical Analysis To investigate the influence of climate region and year on the calendar day of swarm clustering, we used the General Linear Model with climate region (a = 5) and year (a = 4) as fixed factors. Swarm cluster calendar day deviated significantly from a normal distribution (Anderson–Darling Normality test, A2 = 2.980, P < 0.001), thus was first transformed with λ = −0.5, the suggested optimal transformation from the Box–Cox procedure (Sokal and Rohlf 1995), to yield approximately normal distributions with equal variances among both climate region and year. Tukey’s 95% confidence intervals with a group-wise error rate were used to evaluate all pair-wise comparisons within climate regions and within years. Identical analysis, but with a λ = 0.0 transformation, was conducted using the degree-day of swarm clustering as the response variable. Excluded from these analyses were observations from the Pine Barrens climate region in 2015 because an insufficient number of swarms were reported in that region in that year to give reliable parameter estimates (n = 4; for all other region-year combinations n ≥ 17). This exclusion resulted in a final sample size of 685 swarm cluster observations. Statistical analyses were performed using Minitab 17 (Minitab, Inc., State College, PA). Results Swarm Cluster Calendar Day A histogram of the calendar day of swarm clusters for all 4 years across all five regions is shown in Fig. 1. The overall mean calendar day was 135 (15 May) with a SD of 15.8, SE of 0.6, median of 133, minimum of 92 (2 April), and a maximum of 187 (6 July). The mean swarm cluster calendar day for each region and each year is reported in Table 2. Table 2. Mean swarm cluster calendar day (± SE) for each climate region and year Year Region 2013 2014 2015 2016 North 143 ± 2.6 (n = 34) 146 ± 3.0 (n = 27) 145 ± 2.2 (n = 23) 133 ± 2.2 (n = 125) Coastal 140 ± 2.9 (n = 17) 141 ± 2.1 (n = 33) 140 ± 2.7 (n = 18) 134 ± 2.6 (n = 32) Central 133 ± 3.6 (n = 26) 137 ± 1.9 (n = 57) 139 ± 3.4 (n = 18) 131 ± 2.2 (n = 66) Southwest 135 ± 4.4 (n = 25) 141 ± 2.4 (n = 31) 133 ± 2.8 (n = 17) 129 ± 1.9 (n = 75) Pine Barrens 129 ± 2.2 (n = 24) 135 ± 2.5 (n = 18) a (n = 4) 122 ± 3.0 (n = 19) Year Region 2013 2014 2015 2016 North 143 ± 2.6 (n = 34) 146 ± 3.0 (n = 27) 145 ± 2.2 (n = 23) 133 ± 2.2 (n = 125) Coastal 140 ± 2.9 (n = 17) 141 ± 2.1 (n = 33) 140 ± 2.7 (n = 18) 134 ± 2.6 (n = 32) Central 133 ± 3.6 (n = 26) 137 ± 1.9 (n = 57) 139 ± 3.4 (n = 18) 131 ± 2.2 (n = 66) Southwest 135 ± 4.4 (n = 25) 141 ± 2.4 (n = 31) 133 ± 2.8 (n = 17) 129 ± 1.9 (n = 75) Pine Barrens 129 ± 2.2 (n = 24) 135 ± 2.5 (n = 18) a (n = 4) 122 ± 3.0 (n = 19) aToo few observations were made to determine a reliable estimate of the mean. View Large Table 2. Mean swarm cluster calendar day (± SE) for each climate region and year Year Region 2013 2014 2015 2016 North 143 ± 2.6 (n = 34) 146 ± 3.0 (n = 27) 145 ± 2.2 (n = 23) 133 ± 2.2 (n = 125) Coastal 140 ± 2.9 (n = 17) 141 ± 2.1 (n = 33) 140 ± 2.7 (n = 18) 134 ± 2.6 (n = 32) Central 133 ± 3.6 (n = 26) 137 ± 1.9 (n = 57) 139 ± 3.4 (n = 18) 131 ± 2.2 (n = 66) Southwest 135 ± 4.4 (n = 25) 141 ± 2.4 (n = 31) 133 ± 2.8 (n = 17) 129 ± 1.9 (n = 75) Pine Barrens 129 ± 2.2 (n = 24) 135 ± 2.5 (n = 18) a (n = 4) 122 ± 3.0 (n = 19) Year Region 2013 2014 2015 2016 North 143 ± 2.6 (n = 34) 146 ± 3.0 (n = 27) 145 ± 2.2 (n = 23) 133 ± 2.2 (n = 125) Coastal 140 ± 2.9 (n = 17) 141 ± 2.1 (n = 33) 140 ± 2.7 (n = 18) 134 ± 2.6 (n = 32) Central 133 ± 3.6 (n = 26) 137 ± 1.9 (n = 57) 139 ± 3.4 (n = 18) 131 ± 2.2 (n = 66) Southwest 135 ± 4.4 (n = 25) 141 ± 2.4 (n = 31) 133 ± 2.8 (n = 17) 129 ± 1.9 (n = 75) Pine Barrens 129 ± 2.2 (n = 24) 135 ± 2.5 (n = 18) a (n = 4) 122 ± 3.0 (n = 19) aToo few observations were made to determine a reliable estimate of the mean. View Large Fig. 1. View largeDownload slide Histogram of observation day for 685 spring swarm clusters reported across the state of New Jersey from 2013 to 2016. The median cluster date was 13 May (Day 133). Fig. 1. View largeDownload slide Histogram of observation day for 685 spring swarm clusters reported across the state of New Jersey from 2013 to 2016. The median cluster date was 13 May (Day 133). Both climate region and year explained a highly significant amount of variance in swarm cluster calendar day (climate region: F = 10.37; df = 4, 677; P ≤ 0.001; year: F = 21.00, df = 3, 677; P ≤ 0.001). Analysis of potential interaction effects between climate region and year was possible only for four of the five climate regions due to missing data for the Pine Barrens in 2015; no significant interaction effects between climate region and year were observed (F = 0.88; df = 9, 608; P = 0.545). Mean swarm cluster calendar day is shown in Fig. 2 for each climate region (panel a) and each year (panel b). These mean days translate into the following average dates for each region: North = 18 May (SD = 15.5 d), Coastal = 18 May (SD = 13.0 d), Central = 14 May (SD = 16.5 days), Southwest = 13 May (SD = 17.1 days), and Pine Barrens = 8 May (SD = 12.3 days). Tukey’s Simultaneous Tests among all levels of year revealed significant differences only between 2016 and all other years (P ≤ 0.001 for all three comparisons); all other pair-wise comparisons were not significant. Tukey’s Simultaneous Tests among all levels of climate region revealed significant differences between North and Central, North and Southwest, North and Pine Barrens, and Coastal and Pine Barrens (P ≤ 0.001); all other pair-wise comparisons were not significant. Fig. 2. View largeDownload slide Mean swarm cluster day (a) for each climate region and (b) for each year, with error bars representing the SEM. Both region and year were significant predictors of swarm calendar day (GLM; df = 4, 3, 677; both P ≤ 0.001). Letters indicate pair-wise comparisons (Tukey’s simultaneous tests), where levels of each factor that share a letter were not significantly different from each other at a group alpha of 0.05. Fig. 2. View largeDownload slide Mean swarm cluster day (a) for each climate region and (b) for each year, with error bars representing the SEM. Both region and year were significant predictors of swarm calendar day (GLM; df = 4, 3, 677; both P ≤ 0.001). Letters indicate pair-wise comparisons (Tukey’s simultaneous tests), where levels of each factor that share a letter were not significantly different from each other at a group alpha of 0.05. Swarm Cluster Degree-Day The overall mean swarm degree-day was 179.4°C·d (323°F d), with a SD of 118.3°C d (213°F d), SE of 4.5°C d (8.1°F d), median of 142.8°C d (257°F d), minimum of 16.6°C d (30°F d), and a maximum of 778.9°C d (1,402°F d). The largest difference between years was 82.2°C d (2015 vs. 2016) and smallest difference was 19.4°C d (2013 vs. 2014). Both climate region and year explained a highly significant amount of variance in swarm cluster degree-day (climate region: F = 5.02; df = 4, 677; P = 0.001; year: F = 13.71; df = 3, 677; P ≤ 0.001). Mean swarm cluster degree-day is shown in Fig. 3 for each climate region (panel a) and each year (panel b). Tukey’s Simultaneous Tests among all levels of year revealed significant differences between 2015 and 2013 (P ≤ 0.001), 2015 and 2014 (P = 0.034), 2015 and 2016 (P ≤ 0.001), and 2014 and 2016 (P ≤ 0.001); all other pair-wise comparisons were not significant. Tukey’s simultaneous tests among all levels of climate region revealed significant differences between Southwest and Central (P = 0.024), Southwest and North (P = 0.002), and Southwest and Pine Barrens (P = 0.002); all other pair-wise comparisons were not significant. Fig. 3. View largeDownload slide Mean swarm cluster growing degree-days (GDD) (a) for each climate region and (b) for each year, with error bars representing the SEM. Both region and year were significant predictors of swarm GDD (GLM; df = 4, 3, 677; both P ≤ 0.001). Letters indicate pair-wise comparisons (Tukey’s simultaneous tests), where levels of each factor that share a letter were not significantly different from each other at a group alpha of 0.05. Fig. 3. View largeDownload slide Mean swarm cluster growing degree-days (GDD) (a) for each climate region and (b) for each year, with error bars representing the SEM. Both region and year were significant predictors of swarm GDD (GLM; df = 4, 3, 677; both P ≤ 0.001). Letters indicate pair-wise comparisons (Tukey’s simultaneous tests), where levels of each factor that share a letter were not significantly different from each other at a group alpha of 0.05. Discussion The overall mean swarm cluster date of 15 May fits the broad pattern of swarm timing across the United States; to the north in northern New York swarming peaks in early June (Fell et al. 1977), and to the south in Louisiana swarming peaks in late April (Villa 2004). Weather records from the Northeast Regional Climate Center indicate that for the 4 years studied here (2013–2016), the average daily temperatures in central New Jersey during the months of March, April, and May were 11.3, 10.8, 11.8, and 12.8°C, respectively. The mean for these 4 years, 11.7°C, is very similar to the 20-year mean of 11.9°C, suggesting that the 4 years studied here are representative of the climate of the region. Climate-Region Variation The calendar day of observed swarm clusters varied by climate region (P ≤ 0.001; Fig. 2a), suggesting adaptive decision making in swarm timing by honey bee colonies. The maximum magnitude of these climate-region differences, ~10 d, is relatively small compared to the documented differences across the United States. Pooling all 4 years of swarm dates together, New Jersey’s five climate regions appear to cluster into three distinct zones: 1) North and Coastal with mean date of ~18 May, 2) Central and Southwest with mean date of ~13 May, and 3) Pine Barrens with mean date of ~8 May. Simultaneous pair-wise comparisons cast doubt on the distinction between the Pine Barrens and the Central/Southwest cluster (Fig. 2a), but this might be an artifact of the relatively small sample size for the Pine Barrens region. Further investigation into the similarity between North and Coastal regions, two very different climates, may reveal important causal relationships between environmental factors and swarm timing. Climate regions differed in swarm-cluster degree-day (P = 0.001; Fig. 3a), suggesting that for the geographical scale examined in this study, local heat accumulation cannot fully account for differences in observed swarm cluster dates. Degree-day information did, however, eliminate the statistical differences between the North/Coastal regions, the Pine Barrens region, and the Central region, leaving the Southwest as an outlier with a late mean swarm date in terms of degree-days (pair-wise comparisons; Fig. 2a versus 3a). These results suggest that local heat accumulation might help account for regional differences in swarm-departure timing. Future studies using finer climate gradations (‘subzones’) or alternative divisions (e.g., inland vs. coastal) may be especially helpful in connecting geographical location to swarm-departure timing. Annual Variation The calendar day of observed swarm clusters varied by year (P ≤ 0.001; Fig. 2b), which was expected based upon previous research (e.g., Villa 2004). The statistical significance of this effect was, however, due mostly to 1 of the 4 years; the mean swarm cluster days for all years except 2016 did not significantly differ (Fig. 2b). The minimum difference among years was <1/10 of a day (2014 vs. 2015) and the maximum difference was 9 d (2013 vs. 2016). This was less annual variation than expected and is small enough to be explained by short-term weather patterns, such as periods of rain during which colonies avoid swarm departure (Henneken et al. 2012). If weather is sufficient to explain variability in swarm-departure timing, this calls into question the notion that honey bee colonies adaptively tune the timing of the fission process on a seasonal scale to track annual variation in floral phenology. An alternative adaptive explanation for the observed low variability in swarm departure dates among years is that honeybee colonies optimize fitness by fissioning as early as possible to enable the daughter colony to take maximum advantage of peak spring nectar abundance regardless of seasonal fluctuation. Two competing explanations for the unexpectedly low inter-annual variation we believe can be eliminated. First, it could be that the observed swarm clusters might have come predominantly from managed colonies sourced from bee breeders outside of the Northeast region, thus representing bees not well-tuned to local relationships between the time of year and the floral bloom times. This explanation is unlikely because the mechanism of seasonal tuning is presumably a flexible colony decision-making process rather than local genetic evolution. Second, it could be that the 4 years examined in this study were abnormally similar in weather resulting in uncharacteristically low variability in swarm-departure timing. This explanation appears unlikely because weather data indicate fluctuations in New Jersey during 2013–2016 consistent with the 20-year temperature trend (records from Northeast Regional Climate Center). Swarm-cluster degree-day differed significantly by year (P ≤ 0.001; Fig. 3b), suggesting that local heat accumulation does not effectively predict observed swarm cluster dates each year. This result may be considered unsurprising given that the growing degree-day concept assumes that 1) the organism to which the concept is applied cannot regulate its own temperature, and 2) the temperatures used for calculating growing degree-day are the same as those experienced by the organism (Higley et al. 1986). Experimental evidence supports direct associations between temperature and honey bee activity such as the first appearance of honey bees outside the nest (Gordo and Sanz 2006, Gordo et al. 2010) and first spring cleansing flights (Sparks et al. 2010, Langowska et al. 2017), but these assumptions do not hold when considering the growth and development of a homeothermic honey bee colony. Yet an indirect relationship between heat accumulation and swarm-departure timing doubtless exists, as immediate preparations for swarm departure (such as queen rearing) depend on colony cues (such as worker density; Grozinger et al. 2014) that, in turn, depend on access to early pollen-source plants. Identification of specific early pollen-source plants and documentation of their phenology alongside honey bee phenology should be considered a research priority for understanding how environmental factors affect honey bee colony phenology. Acknowledgments We thank Charles Ilsley for discussions that initiated this study and contagious enthusiasm, Kevin Inglin for creating the New Jersey Swarm Report website and providing the compiled data, and Dr. David Robinson and the Office of the New Jersey State Climatologist at Rutgers University for communication of information regarding state climate zones. We gratefully acknowledge the efforts of the New Jersey beekeeping community and the New Jersey Beekeepers Association leadership for its support. This study could not have been completed without the records and tools provided by the Northeast Regional Climate Center and the National Oceanic and Atmospheric Administration’s National Centers for Environmental Information. D.C.G. was supported by William Paterson University of New Jersey’s Assigned Release Time Program and Sabbatical Leave Program. References Cited Burgett , D. M. , and R. A. Morse . 1974 . The time of natural swarming in honey bees . Ann. Entomol. Soc. Am . 67 : 719 – 720 . Google Scholar CrossRef Search ADS Fell , R. D. , J. T. Ambrose , D. M. Burgett , D. De Jong , R. A. Morse , and T. D. Seeley . 1977 . The seasonal cycle of swarming in honey bees . J. Apic. Res . 16 : 170 – 173 . Google Scholar CrossRef Search ADS Gilley , D. C. , and D. R. Tarpy . 2005 . Three mechanisms of queen elimination in swarming honey bee colonies . Apidologie 36 : 461 – 474 . Google Scholar CrossRef Search ADS Gordo , O. , and J. J. Sanz . 2006 . Temporal trends in phenology of the honey bee Apis mellifera (L.) and the small white Pieris rapae (L.) in the Iberian Peninsula (1952–2004) . Ecol. Entomol . 31 : 261 – 268 . Google Scholar CrossRef Search ADS Gordo , O. , J. J. Sanz , and J. M. Lobo . 2010 . Determining the environmental factors underlying the spatial variability of insect appearance phenology for the honey bee, Apis mellifera, and the small white, Pieris rapae . J. Insect Sci . 10 : 1 – 21 . Google Scholar CrossRef Search ADS PubMed Grozinger , C. , J. Richards , and H. Mattila . 2014 . From molecules to societies: mechanisms regulating swarming behavior in honey bees (Apis spp.) . Apidologie 45 : 327 – 346 . Google Scholar CrossRef Search ADS Henneken , R. , S. Helm , and A. Menzel . 2012 . Meteorological influences on swarm emergence in honey bees (Hymenoptera: Apidae) as detected by crowdsourcing . Environ. Entomol . 41 : 1462 – 1465 . Google Scholar CrossRef Search ADS PubMed Hepburn , H. R. , and S. E. Radloff . 1995 . First approximation to a phenology of the honey bees (Apis mellifera) and flora of Africa . Oecologia 101 : 265 – 273 . Google Scholar CrossRef Search ADS PubMed Herms , D. A . 2004 . Using degree-days and plant phenology to predict pest activity , 49 – 59 . In V. Krischik and J. Davidson (eds.), IPM (Integrated Pest Management) of Midwest Landscapes: Minnesota Agricultural Experiment Station Publication SB-07645 . Minnesota Agricultural Experiment Station, St. Paul , Minnesota . Higley , L. G. , L. P. Pedigo , and K. R. Ostlie . 1986 . DEGDAY: a program for calculating degree-days, and assumptions behind the degree-day approach . Environ. Entomol . 15 : 999 – 1016 . Google Scholar CrossRef Search ADS Hodgson , J. A. , C. D. Thomas , T. H. Oliver , B. J. Anderson , T. M. Brereton , and E. E. Crone . 2011 . Predicting insect phenology across space and time . Global Change Biol . 17 : 1289 – 1300 . Google Scholar CrossRef Search ADS Langowska , A. , M. Zawilak , T. H. Sparks , A. Glazaczow , P. W. Tomkins , and P. Tryjanowski . 2017 . Long-term effect of temperature on honey yield and honey bee phenology . Int. J. Biometeorol . 61 : 1125 – 1132 . Google Scholar CrossRef Search ADS PubMed Ludlum , D. M . 1983 . The New Jersey Weather Book . Rutgers University Press , New Brunswick, NJ . Mitchener , A. V . 1948 . The swarming season for honey bees in Manitoba . J. Econ. Entomol . 41 : 646 – 646 . Google Scholar CrossRef Search ADS Mussey , G. J. , and D. A. Potter . 1997 . Phenological correlations between flowering plants and activity of urban landscape pests in Kentucky . J. Econ. Entomol . 90 : 1615 – 1627 . Google Scholar CrossRef Search ADS Neuheimer , A. B. , and C. T. Taggart . 2007 . The growing degree-day and fish size-at-age: the overlooked metric . Can. J. Fisheries Aquatic Sci . 64 : 375 – 385 . Google Scholar CrossRef Search ADS Page , R. E . 1981 . The seasonal occurrence of honey bee swarms in North-Central California . Am. Bee J . 121 : 266 . Renato , L. , Jr. , C. L. Oliveira , E. S. Ferreira , E. Cruiff , C. Tavares , A. C. Santos , C.R. Franke , and M. E. Bavia . 2009 . Spatial analysis of migrating Apis mellifera colonies in Salvador, Bahia, Brazil . Geospatial Health 4 : 129 – 134 . Google Scholar CrossRef Search ADS PubMed Seeley , T. D. , and P. K. Visscher . 1985 . Survival of honey bees in cold climates: the critical timing of colony growth and reproduction . Ecol. Entomol . 10 81 – 88 . Google Scholar CrossRef Search ADS Simpson , J . 1959 . Variation in the incidence of swarming among colonies of Apis mellifera throughout the summer . Insectes Sociaux 6 : 85 – 99 . Google Scholar CrossRef Search ADS Simpson , J. , and E. Moxley . 1971 . The swarming behaviour of honeybee colonies kept in small hives and allowed to outgrow them . J. Apic. Res . 10 : 109 – 113 . Google Scholar CrossRef Search ADS Sokal , R. R. , and F. J. Rohlf . 1995 . Biometry: the principles and practice of statistics in biological research . Freeman & Company , New York, NY . Sparks , T. H. , A. Langowska , A. Głazaczow , Z. Wilkaniec , M. Bieńkowska , and P. Tryjanowski . 2010 . Advances in the timing of spring cleaning by the honey bee Apis mellifera in Poland . Ecol. Entomol . 35 : 788 – 791 . Google Scholar CrossRef Search ADS Stenseth , N. C. , and A. Mysterud . 2002 . Climate, changing phenology, and other life history traits: nonlinearity and match–mismatch to the environment . Proc. Nat. Acad. Sci . 99 : 13379 – 13381 . Google Scholar CrossRef Search ADS Tarpy , D. R. , and D. C. Gilley . 2004 . Group decision making during queen production in colonies of highly eusocial bees . Apidologie 35 : 207 – 216 . Google Scholar CrossRef Search ADS Villa , J. D . 2004 . Swarming behavior of honey bees (Hymenoptera: Apidae) in southeastern Louisiana . Ann. Entomol. Soc. Am . 97 : 111 – 116 . Google Scholar CrossRef Search ADS Winston , M. L . 1987 . The biology of the honey bee . Harvard University Press , Cambridge, MA . Wright , W . 2002a . Articles-apply survival traits of honey bees for swarm prevention and increased honey production-part I . Am. Bee J . 142 : 123 – 124 . Wright , W . 2002b . Articles-apply survival traits of honey bees for swarm prevention and increased honey production-part II . Am. Bee J . 142 : 201 – 202 . © The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Environmental Entomology Oxford University Press

Phenology of Honey Bee Swarm Departure in New Jersey, United States

Loading next page...
 
/lp/ou_press/phenology-of-honey-bee-swarm-departure-in-new-jersey-united-states-yGOuXdJoiy
Publisher
Entomological Society of America
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0046-225X
eISSN
1938-2936
D.O.I.
10.1093/ee/nvy039
Publisher site
See Article on Publisher Site

Abstract

Abstract Departure of swarms from honey bee (Apis mellifera Linnaeus (Hymenoptera: Apidae)) nests is an important reproductive event for wild honey bee colonies and economically costly in managed bee colonies. The seasonal timing of swarm departure varies regionally and annually, creating challenges for honey bee management and emphasizing the potential for swarming behavior to be affected by plant-pollinator phenological mismatch. In this study, we first document variability in the timing of swarm departure across the large and heterogeneous geographical area of New Jersey over 4 years using 689 swarm-cluster observations. Second, hypothesizing that honey bee colonies adaptively tune the timing of swarm departure to match floral food-resource availability, we predicted that growing degree-days could be used to account for regional and annual variability. To test this idea, we used local weather records to determine the growing degree-day on which each swarm cluster was observed and tested for differences among climate regions and years. The state-wide mean swarm cluster date was May 15 (± 0.6 d), with moderate but significant differences among the state’s five climate regions and between years. Use of degree-day information suggests that local heat accumulation can account for some climate-region differences in swarm-departure timing. Annual variation existed on a scale of only several days and was not accounted for by growing degree-days, suggesting little adaptive tuning of swarm-departure timing with respect to local heat accumulation. phenology, swarm, growing degree-day, climate, temperature Swarm departure in honey bees (Apis mellifera L.) is the culmination of a complex colony fission process involving the buildup of worker population, rearing of replacement queens, preparation of the mother queen for flight, swarm departure, and monogyny restoration (summarized in Winston 1987, reviewed by Tarpy and Gilley 2004, Gilley and Tarpy 2005, Grozinger et al. 2014). Swarm departure results in significant economic loss for beekeepers by reduction of colony worker population, interruption of the brood cycle, and potential failure to re-queen. The timing of swarm departure is an adaptive decision by a honey bee colony (Seeley and Visscher 1985) that takes place on two timescales. In the short term, the day and hour of swarm departure is chosen to avoid poor weather (Henneken et al. 2012) and prevent fighting between the mother queen and daughter queen (Gilley and Tarpy 2005). In the long term, on a seasonal time scale, the timing of swarm departure is chosen to minimize the likelihood of colony death from late frost and maximize the likelihood of matching colony growth to bloom times of important forage plants (Seeley and Visscher 1985). We focus here on the timing of swarm departure on this seasonal scale because it is most directly relevant to swarm management practices and is most likely to be affected by mismatch between plant and pollinator phenologies caused by climate change (‘phenological mismatch,’ see Stenseth and Mysterud 2002). The seasonal timing of swarm departure has been documented for a number of locations around the world, including many cities and locales within North America (Mitchener 1948, Burgett and Morse 1974, Fell et al. 1977, Page 1981, Seeley and Visscher 1985, Villa 2004). Informal comparison of these studies suggests substantial variation in swarm-departure timing on a national scale (e.g., among central California, upstate New York, and southeastern Louisiana); formal meta-analysis is complicated by differences in sampling-area size and sampling approach. Annual variation also appears substantial (e.g., Villa 2004) but has not often been fully reported. Overall, current documentation of swarming phenology does not seem adequate to detect potential future phenological mismatch resulting from spatial and temporal temperature changes, which are likely to affect swarm-departure timing due to the dependence of this process on floral bloom timing as described below. More information is needed across large regions for longer periods of time to enable us to detect phenological mismatch and to overcome the challenges regional and annual variability creates for honey bee management. Effective swarm prevention typically involves identifying a causal factor leading to swarm departure, developing a procedure to disrupt that factor, and determining when to apply the disruption procedure (e.g., Wright 2002a,b). Optimal timing can be determined from knowledge of the typical earliest swarm departure date together with an estimate of the number of days that elapse before the causal factor results in swarm departure. Precise estimates of swarm timing are important for success because early intervention can be as unsuccessful as late intervention. Regional variation compounds these challenges by making specific management solutions difficult to apply in other locations. A likely driver of the seasonal timing of swarm departure is the bloom timing of key forage plants, which are a necessary prerequisite for the growth and reproduction of a honey bee colony. A strong relationship between swarming and the flowering of major honey bee plant genera was found in a meta-analysis of 23 regions of Africa (Hepburn and Radloff 1995). The bloom timing for a particular plant species can be predicted with some accuracy by measuring accumulated thermal energy as the cellular physiology that drives plant growth is temperature dependent (Higley et al. 1986). Accumulation of thermal energy can be measured using degree-days (sometimes referred to as growing degree-days [GDD]), which is calculated most simply for each day by subtracting a chosen base temperature (the standard used for many applications is 10°C [50°F]; Herms 2004) from the daily average temperature. Degree-days summed during the course of the growing season provide a metric of cumulative thermal energy that has been applied to pest insect development for decades (e.g., Mussey and Potter 1997), is important in studies linking climate change to plant-pollinator phenology (e.g., Hodgson et al. 2011), and has even been extended to growth in fish (Neuheimer and Taggart 2007). Given the dependence of honey bee colony growth and development on plant bloom timing, it is reasonable to hypothesize that there is a relationship between the timing of honey bee swarm departure and degree-day accumulation. The goals of this study are 1) to document the timing and variability of swarm departure across a large and geographically heterogeneous area and 2) to investigate the relationship between growing degree-days and the timing of swarm departure. Traditional approaches to documenting swarm departure are to closely monitor a small number of hives for swarming activity (e.g., Simpson 1959, Simpson and Moxley 1971, Seeley and Visscher 1985) or to rely on observations contributed by local beekeepers (e.g., Mitchener 1948, Burgett and Morse 1974, Fell et al. 1977, Villa 2004). Modern communications and database tools allow expansion of the latter approach to address phenological (e.g., Henneken et al. 2012) as well as public safety issues (e.g., Renato et al. 2009) related to honey bee swarming behavior. Here, we seek to address the first goal using data from a targeted effort to catalog swarm-cluster observations across the state of New Jersey between 2013 and 2016. To address the second goal, we use local weather records to determine the degree-day of each swarm departure, predicting that this measure of local heat accumulation will help account for variation in swarm-departure timing. Methods Data Collection Observations of swarm clusters were reported by observers across the state of New Jersey. These observations were submitted to the ‘New Jersey Swarm Report,’ a long-term project created and managed by members of the New Jersey Beekeepers Association. Since the project’s establishment in 2013, beekeepers have been urged by New Jersey Beekeepers Association leadership to make a report any time they encountered a swarm. Participants were able to submit their observations via the Swarm Report website or via email to the project’s coordinators. The information requested for each swarm was as follows: date captured or observed, swarm location address (including ZIP code and county), size of swarm (small/medium/large), and height of swarm (less than 6 feet/between 6 feet and 10 feet/higher than 10 feet) (where 6 feet = 1.8 m and 10 feet = 3.0 m). Between 1 January 2013 and 1 January 2017, 755 swarm clusters were reported. The raw data from the Swarm Report, which is not available via the website and has not been previously published, was communicated to us by the Swarm Report project coordinators. The analysis presented here uses all swarm observations between 1 January and 15 July of each year. Swarms reported after 15 July likely represent the beginning of a second minor cycle of swarming in the late summer, which we consider a distinct phenomenon from both biological and apicultural perspectives. A small number of incomplete swarm reports, which did not contain sufficient information for this analysis, were also excluded from the dataset, as were any obvious duplicate reports. The total number of swarm observations in the final dataset was 689. We assume that each reported cluster represents a unique swarming event, following a swarm departure from either a feral nest or a managed hive. False-positive identifications may exist in these data but are unlikely as the majority of observers are beekeepers and/or associated with the New Jersey Beekeepers Association, plus the details requested for a report are likely to dissuade false or inaccurate reports (e.g., mistaking a wasp nest for a honey bee swarm). Analysis Variables For each of these 689 swarms we determined the following variables. Calendar Day The calendar day number (from 1 to 365) was determined for each swarm. The year 2016 was a leap year, which was taken into account when converting reported date to calendar day. Degree-day The local cumulative degree-day on which each swarm cluster was observed was calculated using public information tools available from Northeast Regional Climate Center, based on data from weather stations across New Jersey. We used the NRCC CLIMOD 2 web tool to determine, based on reported ZIP code, the identity of the weather station closest to each swarm. If the nearest weather station had more than five missing daily records during the swarming period (a rare occurrence), then the next closest weather station to the swarm cluster was used. Degree-days were calculated for each weather station by the CLIMOD 2 web tool using the average-temperature method (growing degree days) with a base temperature of 10°C (50°F), and accumulation beginning on January 1 of each year. The degree-day for each swarm was defined as the cumulative degree-day on the reported date at the closest weather station to the swarm cluster. Climate Region The State of New Jersey, considered by the United States Geological Survey as a part of the Mid-Atlantic Region, consists of 22,600 km2 of land, ranging from sea-level to 550 m, includes a large coastal plain (~50% of area), significant forest area (~40% of area), and some agricultural land (~20% of area). Five climate zones are defined by the Office of the New Jersey State Climatologist (following Ludlum 1983): North, Central, Southwest, Pine Barrens, and Coastal. These zones we considered to be the most likely boundaries influencing the timing of reproductive swarming. In addition, the size of these regions allowed for sufficient sample sizes for comparative analysis. Full characterization and maps of the boundaries of these climate zones is available from the Office of the New Jersey State Climatologist (Rutgers University) and Ludlum (1983). Because the boundaries may be useful for readers of this study, approximate boundaries of these zones are given in Table 1. Each swarm was assigned to one of these five climate regions based on the location of the weather station nearest to the observed swarm cluster. Table 1. Boundaries of the climate regions used in this study Region name Description of boundaries Counties included North The inland portion of the northern half of the state Sussex, Warren, Morris, Hunterdon; western halves of Passaic and Somerset Central The coastal portion of the northern half of the state and the central inland portion of the state Bergen, Essex, Hudson, Union, Middlesex, Mercer; eastern halves of Passaic and Somerset Coastal The coastal border, ~20 km wide, of the southern half of the state Cape May; eastern halves of Monmouth, Ocean, and Atlantic Southwest The western portion of the south half of the state, bordering the Delaware Bay and Delaware River Western halves of Burlington, Camden, Glouchester, Salem; southwestern half of Cumberland Pine Barrens The interior of the southern half of the state Central Burlington; eastern halves of Camden, Glouchester, Salem; northeastern Cumberland; western halves of Monmouth, Ocean, and Atlantic Region name Description of boundaries Counties included North The inland portion of the northern half of the state Sussex, Warren, Morris, Hunterdon; western halves of Passaic and Somerset Central The coastal portion of the northern half of the state and the central inland portion of the state Bergen, Essex, Hudson, Union, Middlesex, Mercer; eastern halves of Passaic and Somerset Coastal The coastal border, ~20 km wide, of the southern half of the state Cape May; eastern halves of Monmouth, Ocean, and Atlantic Southwest The western portion of the south half of the state, bordering the Delaware Bay and Delaware River Western halves of Burlington, Camden, Glouchester, Salem; southwestern half of Cumberland Pine Barrens The interior of the southern half of the state Central Burlington; eastern halves of Camden, Glouchester, Salem; northeastern Cumberland; western halves of Monmouth, Ocean, and Atlantic View Large Table 1. Boundaries of the climate regions used in this study Region name Description of boundaries Counties included North The inland portion of the northern half of the state Sussex, Warren, Morris, Hunterdon; western halves of Passaic and Somerset Central The coastal portion of the northern half of the state and the central inland portion of the state Bergen, Essex, Hudson, Union, Middlesex, Mercer; eastern halves of Passaic and Somerset Coastal The coastal border, ~20 km wide, of the southern half of the state Cape May; eastern halves of Monmouth, Ocean, and Atlantic Southwest The western portion of the south half of the state, bordering the Delaware Bay and Delaware River Western halves of Burlington, Camden, Glouchester, Salem; southwestern half of Cumberland Pine Barrens The interior of the southern half of the state Central Burlington; eastern halves of Camden, Glouchester, Salem; northeastern Cumberland; western halves of Monmouth, Ocean, and Atlantic Region name Description of boundaries Counties included North The inland portion of the northern half of the state Sussex, Warren, Morris, Hunterdon; western halves of Passaic and Somerset Central The coastal portion of the northern half of the state and the central inland portion of the state Bergen, Essex, Hudson, Union, Middlesex, Mercer; eastern halves of Passaic and Somerset Coastal The coastal border, ~20 km wide, of the southern half of the state Cape May; eastern halves of Monmouth, Ocean, and Atlantic Southwest The western portion of the south half of the state, bordering the Delaware Bay and Delaware River Western halves of Burlington, Camden, Glouchester, Salem; southwestern half of Cumberland Pine Barrens The interior of the southern half of the state Central Burlington; eastern halves of Camden, Glouchester, Salem; northeastern Cumberland; western halves of Monmouth, Ocean, and Atlantic View Large Statistical Analysis To investigate the influence of climate region and year on the calendar day of swarm clustering, we used the General Linear Model with climate region (a = 5) and year (a = 4) as fixed factors. Swarm cluster calendar day deviated significantly from a normal distribution (Anderson–Darling Normality test, A2 = 2.980, P < 0.001), thus was first transformed with λ = −0.5, the suggested optimal transformation from the Box–Cox procedure (Sokal and Rohlf 1995), to yield approximately normal distributions with equal variances among both climate region and year. Tukey’s 95% confidence intervals with a group-wise error rate were used to evaluate all pair-wise comparisons within climate regions and within years. Identical analysis, but with a λ = 0.0 transformation, was conducted using the degree-day of swarm clustering as the response variable. Excluded from these analyses were observations from the Pine Barrens climate region in 2015 because an insufficient number of swarms were reported in that region in that year to give reliable parameter estimates (n = 4; for all other region-year combinations n ≥ 17). This exclusion resulted in a final sample size of 685 swarm cluster observations. Statistical analyses were performed using Minitab 17 (Minitab, Inc., State College, PA). Results Swarm Cluster Calendar Day A histogram of the calendar day of swarm clusters for all 4 years across all five regions is shown in Fig. 1. The overall mean calendar day was 135 (15 May) with a SD of 15.8, SE of 0.6, median of 133, minimum of 92 (2 April), and a maximum of 187 (6 July). The mean swarm cluster calendar day for each region and each year is reported in Table 2. Table 2. Mean swarm cluster calendar day (± SE) for each climate region and year Year Region 2013 2014 2015 2016 North 143 ± 2.6 (n = 34) 146 ± 3.0 (n = 27) 145 ± 2.2 (n = 23) 133 ± 2.2 (n = 125) Coastal 140 ± 2.9 (n = 17) 141 ± 2.1 (n = 33) 140 ± 2.7 (n = 18) 134 ± 2.6 (n = 32) Central 133 ± 3.6 (n = 26) 137 ± 1.9 (n = 57) 139 ± 3.4 (n = 18) 131 ± 2.2 (n = 66) Southwest 135 ± 4.4 (n = 25) 141 ± 2.4 (n = 31) 133 ± 2.8 (n = 17) 129 ± 1.9 (n = 75) Pine Barrens 129 ± 2.2 (n = 24) 135 ± 2.5 (n = 18) a (n = 4) 122 ± 3.0 (n = 19) Year Region 2013 2014 2015 2016 North 143 ± 2.6 (n = 34) 146 ± 3.0 (n = 27) 145 ± 2.2 (n = 23) 133 ± 2.2 (n = 125) Coastal 140 ± 2.9 (n = 17) 141 ± 2.1 (n = 33) 140 ± 2.7 (n = 18) 134 ± 2.6 (n = 32) Central 133 ± 3.6 (n = 26) 137 ± 1.9 (n = 57) 139 ± 3.4 (n = 18) 131 ± 2.2 (n = 66) Southwest 135 ± 4.4 (n = 25) 141 ± 2.4 (n = 31) 133 ± 2.8 (n = 17) 129 ± 1.9 (n = 75) Pine Barrens 129 ± 2.2 (n = 24) 135 ± 2.5 (n = 18) a (n = 4) 122 ± 3.0 (n = 19) aToo few observations were made to determine a reliable estimate of the mean. View Large Table 2. Mean swarm cluster calendar day (± SE) for each climate region and year Year Region 2013 2014 2015 2016 North 143 ± 2.6 (n = 34) 146 ± 3.0 (n = 27) 145 ± 2.2 (n = 23) 133 ± 2.2 (n = 125) Coastal 140 ± 2.9 (n = 17) 141 ± 2.1 (n = 33) 140 ± 2.7 (n = 18) 134 ± 2.6 (n = 32) Central 133 ± 3.6 (n = 26) 137 ± 1.9 (n = 57) 139 ± 3.4 (n = 18) 131 ± 2.2 (n = 66) Southwest 135 ± 4.4 (n = 25) 141 ± 2.4 (n = 31) 133 ± 2.8 (n = 17) 129 ± 1.9 (n = 75) Pine Barrens 129 ± 2.2 (n = 24) 135 ± 2.5 (n = 18) a (n = 4) 122 ± 3.0 (n = 19) Year Region 2013 2014 2015 2016 North 143 ± 2.6 (n = 34) 146 ± 3.0 (n = 27) 145 ± 2.2 (n = 23) 133 ± 2.2 (n = 125) Coastal 140 ± 2.9 (n = 17) 141 ± 2.1 (n = 33) 140 ± 2.7 (n = 18) 134 ± 2.6 (n = 32) Central 133 ± 3.6 (n = 26) 137 ± 1.9 (n = 57) 139 ± 3.4 (n = 18) 131 ± 2.2 (n = 66) Southwest 135 ± 4.4 (n = 25) 141 ± 2.4 (n = 31) 133 ± 2.8 (n = 17) 129 ± 1.9 (n = 75) Pine Barrens 129 ± 2.2 (n = 24) 135 ± 2.5 (n = 18) a (n = 4) 122 ± 3.0 (n = 19) aToo few observations were made to determine a reliable estimate of the mean. View Large Fig. 1. View largeDownload slide Histogram of observation day for 685 spring swarm clusters reported across the state of New Jersey from 2013 to 2016. The median cluster date was 13 May (Day 133). Fig. 1. View largeDownload slide Histogram of observation day for 685 spring swarm clusters reported across the state of New Jersey from 2013 to 2016. The median cluster date was 13 May (Day 133). Both climate region and year explained a highly significant amount of variance in swarm cluster calendar day (climate region: F = 10.37; df = 4, 677; P ≤ 0.001; year: F = 21.00, df = 3, 677; P ≤ 0.001). Analysis of potential interaction effects between climate region and year was possible only for four of the five climate regions due to missing data for the Pine Barrens in 2015; no significant interaction effects between climate region and year were observed (F = 0.88; df = 9, 608; P = 0.545). Mean swarm cluster calendar day is shown in Fig. 2 for each climate region (panel a) and each year (panel b). These mean days translate into the following average dates for each region: North = 18 May (SD = 15.5 d), Coastal = 18 May (SD = 13.0 d), Central = 14 May (SD = 16.5 days), Southwest = 13 May (SD = 17.1 days), and Pine Barrens = 8 May (SD = 12.3 days). Tukey’s Simultaneous Tests among all levels of year revealed significant differences only between 2016 and all other years (P ≤ 0.001 for all three comparisons); all other pair-wise comparisons were not significant. Tukey’s Simultaneous Tests among all levels of climate region revealed significant differences between North and Central, North and Southwest, North and Pine Barrens, and Coastal and Pine Barrens (P ≤ 0.001); all other pair-wise comparisons were not significant. Fig. 2. View largeDownload slide Mean swarm cluster day (a) for each climate region and (b) for each year, with error bars representing the SEM. Both region and year were significant predictors of swarm calendar day (GLM; df = 4, 3, 677; both P ≤ 0.001). Letters indicate pair-wise comparisons (Tukey’s simultaneous tests), where levels of each factor that share a letter were not significantly different from each other at a group alpha of 0.05. Fig. 2. View largeDownload slide Mean swarm cluster day (a) for each climate region and (b) for each year, with error bars representing the SEM. Both region and year were significant predictors of swarm calendar day (GLM; df = 4, 3, 677; both P ≤ 0.001). Letters indicate pair-wise comparisons (Tukey’s simultaneous tests), where levels of each factor that share a letter were not significantly different from each other at a group alpha of 0.05. Swarm Cluster Degree-Day The overall mean swarm degree-day was 179.4°C·d (323°F d), with a SD of 118.3°C d (213°F d), SE of 4.5°C d (8.1°F d), median of 142.8°C d (257°F d), minimum of 16.6°C d (30°F d), and a maximum of 778.9°C d (1,402°F d). The largest difference between years was 82.2°C d (2015 vs. 2016) and smallest difference was 19.4°C d (2013 vs. 2014). Both climate region and year explained a highly significant amount of variance in swarm cluster degree-day (climate region: F = 5.02; df = 4, 677; P = 0.001; year: F = 13.71; df = 3, 677; P ≤ 0.001). Mean swarm cluster degree-day is shown in Fig. 3 for each climate region (panel a) and each year (panel b). Tukey’s Simultaneous Tests among all levels of year revealed significant differences between 2015 and 2013 (P ≤ 0.001), 2015 and 2014 (P = 0.034), 2015 and 2016 (P ≤ 0.001), and 2014 and 2016 (P ≤ 0.001); all other pair-wise comparisons were not significant. Tukey’s simultaneous tests among all levels of climate region revealed significant differences between Southwest and Central (P = 0.024), Southwest and North (P = 0.002), and Southwest and Pine Barrens (P = 0.002); all other pair-wise comparisons were not significant. Fig. 3. View largeDownload slide Mean swarm cluster growing degree-days (GDD) (a) for each climate region and (b) for each year, with error bars representing the SEM. Both region and year were significant predictors of swarm GDD (GLM; df = 4, 3, 677; both P ≤ 0.001). Letters indicate pair-wise comparisons (Tukey’s simultaneous tests), where levels of each factor that share a letter were not significantly different from each other at a group alpha of 0.05. Fig. 3. View largeDownload slide Mean swarm cluster growing degree-days (GDD) (a) for each climate region and (b) for each year, with error bars representing the SEM. Both region and year were significant predictors of swarm GDD (GLM; df = 4, 3, 677; both P ≤ 0.001). Letters indicate pair-wise comparisons (Tukey’s simultaneous tests), where levels of each factor that share a letter were not significantly different from each other at a group alpha of 0.05. Discussion The overall mean swarm cluster date of 15 May fits the broad pattern of swarm timing across the United States; to the north in northern New York swarming peaks in early June (Fell et al. 1977), and to the south in Louisiana swarming peaks in late April (Villa 2004). Weather records from the Northeast Regional Climate Center indicate that for the 4 years studied here (2013–2016), the average daily temperatures in central New Jersey during the months of March, April, and May were 11.3, 10.8, 11.8, and 12.8°C, respectively. The mean for these 4 years, 11.7°C, is very similar to the 20-year mean of 11.9°C, suggesting that the 4 years studied here are representative of the climate of the region. Climate-Region Variation The calendar day of observed swarm clusters varied by climate region (P ≤ 0.001; Fig. 2a), suggesting adaptive decision making in swarm timing by honey bee colonies. The maximum magnitude of these climate-region differences, ~10 d, is relatively small compared to the documented differences across the United States. Pooling all 4 years of swarm dates together, New Jersey’s five climate regions appear to cluster into three distinct zones: 1) North and Coastal with mean date of ~18 May, 2) Central and Southwest with mean date of ~13 May, and 3) Pine Barrens with mean date of ~8 May. Simultaneous pair-wise comparisons cast doubt on the distinction between the Pine Barrens and the Central/Southwest cluster (Fig. 2a), but this might be an artifact of the relatively small sample size for the Pine Barrens region. Further investigation into the similarity between North and Coastal regions, two very different climates, may reveal important causal relationships between environmental factors and swarm timing. Climate regions differed in swarm-cluster degree-day (P = 0.001; Fig. 3a), suggesting that for the geographical scale examined in this study, local heat accumulation cannot fully account for differences in observed swarm cluster dates. Degree-day information did, however, eliminate the statistical differences between the North/Coastal regions, the Pine Barrens region, and the Central region, leaving the Southwest as an outlier with a late mean swarm date in terms of degree-days (pair-wise comparisons; Fig. 2a versus 3a). These results suggest that local heat accumulation might help account for regional differences in swarm-departure timing. Future studies using finer climate gradations (‘subzones’) or alternative divisions (e.g., inland vs. coastal) may be especially helpful in connecting geographical location to swarm-departure timing. Annual Variation The calendar day of observed swarm clusters varied by year (P ≤ 0.001; Fig. 2b), which was expected based upon previous research (e.g., Villa 2004). The statistical significance of this effect was, however, due mostly to 1 of the 4 years; the mean swarm cluster days for all years except 2016 did not significantly differ (Fig. 2b). The minimum difference among years was <1/10 of a day (2014 vs. 2015) and the maximum difference was 9 d (2013 vs. 2016). This was less annual variation than expected and is small enough to be explained by short-term weather patterns, such as periods of rain during which colonies avoid swarm departure (Henneken et al. 2012). If weather is sufficient to explain variability in swarm-departure timing, this calls into question the notion that honey bee colonies adaptively tune the timing of the fission process on a seasonal scale to track annual variation in floral phenology. An alternative adaptive explanation for the observed low variability in swarm departure dates among years is that honeybee colonies optimize fitness by fissioning as early as possible to enable the daughter colony to take maximum advantage of peak spring nectar abundance regardless of seasonal fluctuation. Two competing explanations for the unexpectedly low inter-annual variation we believe can be eliminated. First, it could be that the observed swarm clusters might have come predominantly from managed colonies sourced from bee breeders outside of the Northeast region, thus representing bees not well-tuned to local relationships between the time of year and the floral bloom times. This explanation is unlikely because the mechanism of seasonal tuning is presumably a flexible colony decision-making process rather than local genetic evolution. Second, it could be that the 4 years examined in this study were abnormally similar in weather resulting in uncharacteristically low variability in swarm-departure timing. This explanation appears unlikely because weather data indicate fluctuations in New Jersey during 2013–2016 consistent with the 20-year temperature trend (records from Northeast Regional Climate Center). Swarm-cluster degree-day differed significantly by year (P ≤ 0.001; Fig. 3b), suggesting that local heat accumulation does not effectively predict observed swarm cluster dates each year. This result may be considered unsurprising given that the growing degree-day concept assumes that 1) the organism to which the concept is applied cannot regulate its own temperature, and 2) the temperatures used for calculating growing degree-day are the same as those experienced by the organism (Higley et al. 1986). Experimental evidence supports direct associations between temperature and honey bee activity such as the first appearance of honey bees outside the nest (Gordo and Sanz 2006, Gordo et al. 2010) and first spring cleansing flights (Sparks et al. 2010, Langowska et al. 2017), but these assumptions do not hold when considering the growth and development of a homeothermic honey bee colony. Yet an indirect relationship between heat accumulation and swarm-departure timing doubtless exists, as immediate preparations for swarm departure (such as queen rearing) depend on colony cues (such as worker density; Grozinger et al. 2014) that, in turn, depend on access to early pollen-source plants. Identification of specific early pollen-source plants and documentation of their phenology alongside honey bee phenology should be considered a research priority for understanding how environmental factors affect honey bee colony phenology. Acknowledgments We thank Charles Ilsley for discussions that initiated this study and contagious enthusiasm, Kevin Inglin for creating the New Jersey Swarm Report website and providing the compiled data, and Dr. David Robinson and the Office of the New Jersey State Climatologist at Rutgers University for communication of information regarding state climate zones. We gratefully acknowledge the efforts of the New Jersey beekeeping community and the New Jersey Beekeepers Association leadership for its support. This study could not have been completed without the records and tools provided by the Northeast Regional Climate Center and the National Oceanic and Atmospheric Administration’s National Centers for Environmental Information. D.C.G. was supported by William Paterson University of New Jersey’s Assigned Release Time Program and Sabbatical Leave Program. References Cited Burgett , D. M. , and R. A. Morse . 1974 . The time of natural swarming in honey bees . Ann. Entomol. Soc. Am . 67 : 719 – 720 . Google Scholar CrossRef Search ADS Fell , R. D. , J. T. Ambrose , D. M. Burgett , D. De Jong , R. A. Morse , and T. D. Seeley . 1977 . The seasonal cycle of swarming in honey bees . J. Apic. Res . 16 : 170 – 173 . Google Scholar CrossRef Search ADS Gilley , D. C. , and D. R. Tarpy . 2005 . Three mechanisms of queen elimination in swarming honey bee colonies . Apidologie 36 : 461 – 474 . Google Scholar CrossRef Search ADS Gordo , O. , and J. J. Sanz . 2006 . Temporal trends in phenology of the honey bee Apis mellifera (L.) and the small white Pieris rapae (L.) in the Iberian Peninsula (1952–2004) . Ecol. Entomol . 31 : 261 – 268 . Google Scholar CrossRef Search ADS Gordo , O. , J. J. Sanz , and J. M. Lobo . 2010 . Determining the environmental factors underlying the spatial variability of insect appearance phenology for the honey bee, Apis mellifera, and the small white, Pieris rapae . J. Insect Sci . 10 : 1 – 21 . Google Scholar CrossRef Search ADS PubMed Grozinger , C. , J. Richards , and H. Mattila . 2014 . From molecules to societies: mechanisms regulating swarming behavior in honey bees (Apis spp.) . Apidologie 45 : 327 – 346 . Google Scholar CrossRef Search ADS Henneken , R. , S. Helm , and A. Menzel . 2012 . Meteorological influences on swarm emergence in honey bees (Hymenoptera: Apidae) as detected by crowdsourcing . Environ. Entomol . 41 : 1462 – 1465 . Google Scholar CrossRef Search ADS PubMed Hepburn , H. R. , and S. E. Radloff . 1995 . First approximation to a phenology of the honey bees (Apis mellifera) and flora of Africa . Oecologia 101 : 265 – 273 . Google Scholar CrossRef Search ADS PubMed Herms , D. A . 2004 . Using degree-days and plant phenology to predict pest activity , 49 – 59 . In V. Krischik and J. Davidson (eds.), IPM (Integrated Pest Management) of Midwest Landscapes: Minnesota Agricultural Experiment Station Publication SB-07645 . Minnesota Agricultural Experiment Station, St. Paul , Minnesota . Higley , L. G. , L. P. Pedigo , and K. R. Ostlie . 1986 . DEGDAY: a program for calculating degree-days, and assumptions behind the degree-day approach . Environ. Entomol . 15 : 999 – 1016 . Google Scholar CrossRef Search ADS Hodgson , J. A. , C. D. Thomas , T. H. Oliver , B. J. Anderson , T. M. Brereton , and E. E. Crone . 2011 . Predicting insect phenology across space and time . Global Change Biol . 17 : 1289 – 1300 . Google Scholar CrossRef Search ADS Langowska , A. , M. Zawilak , T. H. Sparks , A. Glazaczow , P. W. Tomkins , and P. Tryjanowski . 2017 . Long-term effect of temperature on honey yield and honey bee phenology . Int. J. Biometeorol . 61 : 1125 – 1132 . Google Scholar CrossRef Search ADS PubMed Ludlum , D. M . 1983 . The New Jersey Weather Book . Rutgers University Press , New Brunswick, NJ . Mitchener , A. V . 1948 . The swarming season for honey bees in Manitoba . J. Econ. Entomol . 41 : 646 – 646 . Google Scholar CrossRef Search ADS Mussey , G. J. , and D. A. Potter . 1997 . Phenological correlations between flowering plants and activity of urban landscape pests in Kentucky . J. Econ. Entomol . 90 : 1615 – 1627 . Google Scholar CrossRef Search ADS Neuheimer , A. B. , and C. T. Taggart . 2007 . The growing degree-day and fish size-at-age: the overlooked metric . Can. J. Fisheries Aquatic Sci . 64 : 375 – 385 . Google Scholar CrossRef Search ADS Page , R. E . 1981 . The seasonal occurrence of honey bee swarms in North-Central California . Am. Bee J . 121 : 266 . Renato , L. , Jr. , C. L. Oliveira , E. S. Ferreira , E. Cruiff , C. Tavares , A. C. Santos , C.R. Franke , and M. E. Bavia . 2009 . Spatial analysis of migrating Apis mellifera colonies in Salvador, Bahia, Brazil . Geospatial Health 4 : 129 – 134 . Google Scholar CrossRef Search ADS PubMed Seeley , T. D. , and P. K. Visscher . 1985 . Survival of honey bees in cold climates: the critical timing of colony growth and reproduction . Ecol. Entomol . 10 81 – 88 . Google Scholar CrossRef Search ADS Simpson , J . 1959 . Variation in the incidence of swarming among colonies of Apis mellifera throughout the summer . Insectes Sociaux 6 : 85 – 99 . Google Scholar CrossRef Search ADS Simpson , J. , and E. Moxley . 1971 . The swarming behaviour of honeybee colonies kept in small hives and allowed to outgrow them . J. Apic. Res . 10 : 109 – 113 . Google Scholar CrossRef Search ADS Sokal , R. R. , and F. J. Rohlf . 1995 . Biometry: the principles and practice of statistics in biological research . Freeman & Company , New York, NY . Sparks , T. H. , A. Langowska , A. Głazaczow , Z. Wilkaniec , M. Bieńkowska , and P. Tryjanowski . 2010 . Advances in the timing of spring cleaning by the honey bee Apis mellifera in Poland . Ecol. Entomol . 35 : 788 – 791 . Google Scholar CrossRef Search ADS Stenseth , N. C. , and A. Mysterud . 2002 . Climate, changing phenology, and other life history traits: nonlinearity and match–mismatch to the environment . Proc. Nat. Acad. Sci . 99 : 13379 – 13381 . Google Scholar CrossRef Search ADS Tarpy , D. R. , and D. C. Gilley . 2004 . Group decision making during queen production in colonies of highly eusocial bees . Apidologie 35 : 207 – 216 . Google Scholar CrossRef Search ADS Villa , J. D . 2004 . Swarming behavior of honey bees (Hymenoptera: Apidae) in southeastern Louisiana . Ann. Entomol. Soc. Am . 97 : 111 – 116 . Google Scholar CrossRef Search ADS Winston , M. L . 1987 . The biology of the honey bee . Harvard University Press , Cambridge, MA . Wright , W . 2002a . Articles-apply survival traits of honey bees for swarm prevention and increased honey production-part I . Am. Bee J . 142 : 123 – 124 . Wright , W . 2002b . Articles-apply survival traits of honey bees for swarm prevention and increased honey production-part II . Am. Bee J . 142 : 201 – 202 . © The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Environmental EntomologyOxford University Press

Published: Mar 31, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off