Effects of Wetland Presence and Upland Land Use on Wild Hymenopteran and Dipteran Pollinators in the Rainwater Basin of Nebraska, USA

Pollination is an ecosystem service that is in jeopardy globally. Decreasing trends in wild pollinator populations are primarily attributed to habitat loss and degradation. These concerns are especially apparent in the Rainwater Basin of Nebraska, USA where more than 90% of the land has been cultivated and 90% of the wetlands have been lost. We compared hymenopteran and dipteran pollinator abundance, richness, and diversity in 28 wetlands and their adjacent uplands within 3 dominant land-condition types: cropped, restored, and reference state. We used vane traps to capture local pollinator insects and sweep nets to collect insect pollinators on flowers. Vane-trap captures revealed a greater abundance, richness, and diversity of hymenopteran pollinators in uplands as compared to wetlands over the entire growing season in all land-condition types. However, net collections showed that hymenopterans foraged more in wetlands than in uplands, especially in restored wetlands. The exception was September when hymenopterans were captured in uplands more than wetlands while feeding on late season forbs. Dipteran vane-trap and sweep-net captures primarily consisted of hoverflies (Syrphidae). Hoverflies were more abundant in wetlands than uplands. All pollinators used both wetlands and uplands for resources. Thus, insect-pollinator conservation can be enhanced by promoting native forb communities and pollinator habitat through wetland restoration and the planting of buffer strips.


Introduction
Global pollinator declines can jeopardize agricultural production as well as wild-plant populations and their associated biota ). There are myriad issues that contribute to pollinator declines in wild pollinator populations. Interrelated factors include pesticides, pathogens, exotic species, climate change, and habitat loss and degradation (Biesmeijer et al. 2006;Winfree et al. 2009;Potts et al. 2010). In this paper, we focus on habitat loss and degradation, which is the primary threat to native bee populations (Winfree et al. 2009).
Bees are recognized as the primary pollinators (Willmer 2011), however, the importance of dipterans, especially non-Syrphid dipterans (Orford et al. 2015), is often overlooked (Larson et al. 2001). Many are major contributors to pollination services in agriculture and biodiversity, and are present in nearly all habitats (Ssymank et al. 2008). There is evidence that some populations of syrphids are in decline in North America, likely due to displacement by non-native species (Skevington et al. 2019). However, dipterans have different ecological requirements than wild bees, and could provide pollination service "insurance" in habitats where bees are threatened (Kearns 2001;Orford et al. 2015).
We conducted this study in one of the most productive cropland regions in the world, the Rainwater Basin (RWB) located in the Northern High Plains (Smith et al. 2011) (Fig. 1). One of the key features of this region, playa wetlands (hereafter referred to simply as "playas") are a crucial component of the Northern High Plains because they provide ecosystem services such as floodwater attenuation, recharging of the Ogallala aquifer, water-quality improvement, and biodiversity support (Haukos and Smith 2003;Smith et al. 2011). However, North American grasslands have declined by as much as 99%, largely due to cultivation, and less than 10% of the historically present playas are estimated to remain in the RWB (Samson and Knopf 1994;LaGrange 2005;RWBJV, 2018).
However, more than 90% of the RWB has been cultivated and 90% of the region's playas have been converted to other land-condition types (). Cultivating watersheds increases erosion and other runoff, and sacrifices floral diversity in uplands and downslope wetland communities (O'Connell et al. 2012), consequently influencing the abundance and diversity of plants for pollinating insects.
Wild pollinators need abundant and diverse floral resources (Kwaiser and Hendrix 2007;Fründ et al. 2010), and nesting habitat that includes areas of bare soil for ground-nesting bees and pithy material for stem nesters (Potts et al. 2005;Black et al. 2011). Proximity to natural or semi-natural areas has shown to increase abundance and species richness of wild pollinators (Garibaldi, 2011;Smith et al. 2013;Alomar, 2018). Native landscapes historically provided a diversity of floral resources that sustained diverse mutualistic pollinator-plant relationships . Although some generalists can survive in a changing habitat, many native species of pollinators with more specialized requirements cannot (Kwaiser and Hendrix, 2007;Aizen et al. 2012). If habitat quality degrades and homogenization of the landscape becomes significant, an ecosystem may suffer from a loss of pollinator species richness (Mogren et al., 2016) and consequently, pollination services. Most invertebrate pollinator species successfully reproduce if their habitat provides nest sites, building materials, and food plants in close proximity and therefore, these remaining pieces of suitable habitat may be essential for pollination in the region (James and Pitts-Singer 2008;Hoffman Black et al. 2011). Wetlands have been demonstrated to be a landscape feature associated with increased pollinator abundance and diversity (Evans et al. 2018). Therefore, our research focused on how the presence, restoration, and surrounding land use of playa systems in the RWB can influence pollinator populations in this highly altered landscape.
The purpose of our research was: & to compare abundance, richness, and diversity among the major land-condition types in the RWB; cropland, restored, and reference; & to compare abundance, richness, and diversity of hymenopteran and dipteran pollinators in playas and their adjacent uplands; & determine if playas serve as important habitat for hymenopteran and dipteran pollinators; & provide management recommendations based on our results.

Study Area
We studied playas and their adjacent upland watersheds within three land-condition types: cropped, restored, and reference. This design allowed us to evaluate the effect of restoration practices on native pollinators. Reference sites were the best representation of the original landscape that we could find. The RWB comprises approximately 15,907 km 2 in southcentral Nebraska (LaGrange 2005) (Fig. 1). Historically, the landscape was flat to gently rolling plains with abundant natural playas that held rain and runoff water . Grasslands in this region consisted of mixed grasses including bluestems (Andropogon spp.), wheatgrasses (Agropyron spp.), and needle grasses (Stipa spp.). Tallgrass plant communities that included bluestems along with switchgrass (Panicum virgatum), and Indian grass (Sorghastum nutans) occurred in the extreme eastern portion of the region (Kuchler 1964). Restored sites were enrolled in the United States Department of Agriculture's Wetlands Reserve Program (WRP), now known as the Agricultural Conservation Easement Program's Wetland Reserve Easements (WRE). While conservation programs that focus on upland restoration are rare in the RWB, wetland restoration programs, such as the WRP, are prominent (Smith et al. 2011). This program offers financial and technical support to landowners for enhancing, restoring, or protecting wetlands and the immediate watershed (NRCS 2008a). Conservation practices included planting a filter strip of perennial cover to reduce sedimentation and nutrient runoff, filling of reuse pits (an excavation constructed to capture irrigation water for reuse (Rainwater Basin Joint Venture 2019)), and removal of sediment to improve hydrology and plant communities (Gleason et al. 2011;Beas et al. 2013). Even with the great loss of playas, the RWB is a focal point of the Central Flyway that over 170 different species of birds rely on (NRCS 2008b). Consequently, RWB conservation primarily focuses on providing wetland habitat and critical food resources for midcontinent migrating waterfowl (Webb et al. 2010). State and federal wildlife agencies have conserved over 2500 ha of playas since the 1960s (Walker 2016). Over 75% of the conserved area is protected in the form of federal Waterfowl Production Areas, state Wildlife Management Areas, and conservation easements such as those obtained through the WRP (Grosse 2014). Therefore, remaining playa wetlands and their surrounding uplands in the RWB may be important as pollinator habitat in this highly altered landscape.
Currently, the RWB has been overwhelmingly converted to soybean (Glycine max) and corn (Zea mays) production (Smith 2003). We selected cropland sites based on presence of a playa within an active cropping operation and willingness of the landowner to cooperate.

Field Sampling
Insects were collected from the first week of April until mid-October in 2014 and 2015 in the RWB. Twenty-eight different sites were sampled 14 times (once every 2 weeks for 7 months) in 2014 and 2015. Each sampling site featured a playa that was embedded within one of three different land-condition types: cropped (n = 10), restored (n = 9), and reference (n = 9). At each site, we selected 6 random sampling points; 3 within the playa and 3 within the adjacent uplands immediate to the playa (Fig. 2). We collected insects with vane traps (Springstar™ LLC, Woodinville, WA, USA) and insect nets (Stephen and Rao 2005;Stephen and Rao 2007). Vane traps consisted of a bright yellow plastic jar fitted with a blue fabricated polypropylene screw cap funnel into which two, blue, cross vanes were inserted (Stephen and Rao 2007). A wooden stake was placed at each sampling point and a trap was attached with wire so it was visible in the surrounding vegetation. Upland points were established at least 25 m away from the playa edge so that they were located within the upland plant community. There was no maximum distance as it varied based on the vegetation characteristics of the playa watershed. Traps were set up between 0900 and 1100 CDT at different sites each day and then picked up the following morning during the same time-period, allowing for approximately 24-h exposure. We then replaced the funnel with a lid and the jars were placed in a freezer to kill captured specimens.
We used 30.5-cm-diameter insect nets for targeted netting. We established a 25-m-long, 1.0-m-wide transect belt at each vane trap location and positioned it so the wooden stake with the trap was at the midline of the belt (12 m up the transect with 0.5 m on either side). Technicians scanned the belt transect for 6 min, observing flowers and captures insects when they fed on or manipulated flowers within the transect belt. We paused the timer while the insect was placed in a killing jar of ethyl acetate. Once it was dead, we placed the captured insect in a plastic sample bag labeled with site, transect number, date, and species of the flower on which it was captured. Targeted sampling started at approximately 1100 CDT, after the vane traps for the day's sites had been set out.

Insect Identification
All specimens were kept in a freezer at −10°C until they were identified. Insect specimens were first identified to order by simple observation from a local entomologist and then to genera and species using taxonomic keys as needed. A complete list of taxonomic keys and identification methods used can be found in Park et al. (2017). Because Hymenoptera and Diptera are the two most important pollinator orders in North America, we focused on these two orders (Inouye et al. 2015;Orford, 2015).
Specimens of additional insect orders were collected, including Coleoptera and Lepidoptera, but we did not include those orders in this analysis (see Park et al. 2017 for complete list). We used Willmer (2011) to determine the families of Hymenoptera and Diptera recognized as pollinators. We used Triplehorn et al. (2005) to confirm dipteran life history and anthophilous behavior, recognizing that pollinators in one environment may only be anthophilous in another habitat, or potential pollinators.

Statistical Analyses
We measured abundance, richness, and diversity (Shannon-Weiner index) of hymenopteran and dipteran pollinators among land-condition types and landscape positions. We used a subset of data from Park et al. (2017). Insects collected in the 3 vane traps in the playas and the 3 vane traps in the uplands were pooled to create a playa and upland sampling unit for each site per sampling period. We performed an analysis of variance (ANOVA) (Minitab 17 Statistical Software, State College, PA) to compare the main effects and interactions of biweekly sampling period, land-condition type (cropped, restored, reference) and landscape position (playa, upland), for insect vane trap and net-sampling techniques. Land-condition type, land position, and sampling unit were classified as fixed effects and year was a random effect. When significant relationships occurred we used Tukey's pairwise comparisons to examine differences in means. We used an alpha of 0.1 because we wanted to detect large-scale biological effects if they exist.

Hymenoptera
Hymenopterans comprised almost 80% of all specimens caught. Over 86,500 bees from five different families were collected. Apidae and Halictidae made up 98% (62% and 36%, respectively) of the hymenopteran samples collected. Additional families included Andrenidae, Colletidae, and Megachilidae. Apidae was the most species rich family with 55 total species, including 11 Bombus and 16 Melissodes species. The 16 Melissodes species made up 76% of all specimens collected within Apidae. Less than 2% of.
Hymenopterans caught were not bees; wasps and ants each equaled 1% of total insects. Two bee families, Andrenidae and Megachilidae, and the wasp family Vespidae had greater representation in net collections than traps. However, Halictidae was greater in trap than net collections and Apidae and Colletidae frequencies were not different between the two sampling techniques.
With vane traps, reference uplands (x ̅ = 70.67, SE = 8.28) had higher hymenopteran abundance than crop uplands (x ̅ = 58.06, SE = 6.8, [F (2, 1420) = 4.06, p = 0.017] (Fig. 3a). Hymenoptera abundance was higher in early September (pe- google.com/web/). The WMA is rangeland with a playa located in the middle. Three traps were placed in the playa (represented by triangle points) and three traps in the upland (represented by square points). The upland traps were placed at a minimum of 25 m from the playa edge with goal of placing the trap in an upland plant community that differs from the playa plant community. Triangles represent the playa traps and squares represent the upland traps. The playa traps were pooled at each site to make a playa sampling unit and the upland traps were pooled to make an upland sampling unit (gh), 2 (x ̅ = 3.71, SE =1.55) (h), 1 (x ̅ = 1.3, SE =0.41), [F (13, 1420) = 29.35, p = <0.001) (h) (Fig. 3b) (Table 1). Tukey's post hoc results are indicated in parentheses when not included in the accompanying figures throughout this section.
Hymenoptera comprised 26% of insects collected with nets; 93% were bees. Three hundred seventy-five bees from five different families (Andrenidae, Apidae, Colletidae, Halictidae, and Megachilidae) were caught on flowers. Apidae had the highest abundance with Bombus and Melissodes being the most common genera. Twenty-nine wasps from six families were collected. Vespidae had the most individuals overall, comprising over half of the wasps caught.
Sampling period was the primary driver of dipteran richness variation, and there were no significant differences among landcondition types within a sampling period. Richness increased in early summer and again in October with the maximum richness observed in sampling period 14 (Fig. 5b) (Table 3). Sampling period was also the primary driver with diversity, and it was not significantly different among land-condition types within the same sampling period (Table 3).

Hymenoptera
When we examined vane trap data, there were no strong patterns of differences in Hymenoptera abundance, richness and diversity among land-condition types and among playas and upland plant communities. Reference uplands had a greater abundance of hymenopterans than cropped uplands, but hymenopteran abundance in reference uplands was not statistically different from their abundance in restored uplands or playas in all land-condition types. All playas, regardless of land use, had similar hymenopteran abundance, richness, and diversity to adjacent uplands in trap collections, suggesting that surrounding land-condition does not affect, or only minimally affects, use of a playas by bees. Similar studies have conflicting results. Evans et al. (2018) determined that wetland presence enhanced wild bee abundance and richness in agricultural landscapes. Mogren et al. (2016), however, found that if wetlands surrounding agricultural lands were in a degraded condition, wildpollinator abundance and richness did not differ from surrounding land-conditions. We also need to consider capture methods. Vane traps are powerful attractants (Stephen and Rao 2007) and could be drawing in bees that would otherwise move to other foraging areas. Our data did support that wild-pollinator richness at cropland sites was lower than at reference and restoration sites, which is consistent with findings that habitat loss is a major contributor to pollinator decline (Winfree et al. 2009) and distance from natural areas results in reduced pollinator richness (Ricketts et al. 2008). However, these results also showed that playas in all landcondition types were statistically similar to cropland. Grasslands are an important habitat for wild bees (Kimoto et al. 2012). Therefore, although wetlands in all landcondition types had similar abundances of bees, grassland watersheds may be providing additional resources (i.e., nesting habitat) (Westrich 1996;Hoffman Black et al. 2011) for a wider array of bee species that are not found in or utilizing wetlands. Although one of the goals of our Net captures tell a different story from trap samples, and this is consistent with results from Meyer et al. (2017). These data suggest that bees forage more in playas than uplands in July and August. However, in September, bees foraged more in adjacent uplands, suggesting bees shift foraging areas as food supplies change. For example, a greater abundance of bees captured in playas than uplands during mid-summer was because of foraging on wetland plants such as smartweed (Polygonum spp. and Persicaria spp.) and plains coreopsis (Coreopsis tinctoria) (Park 2017). Foraging later in the season shifted to surrounding uplands to take advantage of late-season forbs (e.g. goldenrods [Solidago spp. and sunflowers [Helianthus spp.]) blooming in the uplands during late summer (Park 2017). Vickruck et al. (2019) determined that wetlands were an important seminatural habitat feature in the Prairie Pothole Region, and supported wild bee populations that were distinct. The targeted netting data allowed us a more specific window into how pollinators are using uplands and playas differently, and that playas may be distinctly important for some bee populations.

Diptera
Although flies only comprised 12% of the trap data, they had the most individuals of any order in net collections (32%). The largest Dipteran family was Syrphidae, or hoverflies -a Hymenoptera mimicker and an efficient pollinator in temperate zones (Willmer 2011). Syrphid flies are recognized pollinators (Willmer 2011), and can use a variety of habitats (Kearns 2001), and playas can provide additional resources not as readily available in grasslands alone. Hoverflies from the tribes Eristalini and Sericomyiini deposit eggs in water and the larvae are aquatic (Vockeroth and Thompson 1987), Five genera from Eristalini were represented in our trap samples. Hoverfly larvae use rotting organic matter in wetlands that can be provided by playas (Gilbert 1986;Vockeroth and Thompson 1987). In trap collections, flies were more abundant in playas than in uplands. This was also observed in net collections, especially during mid-summer. This indicates that flies are utilizing habitat in proximity of the playas, enhancing the scope of pollinating insects in the landscape. Net and trap abundance, richness, and diversity metrics varied throughout the season. However, playas typically had higher abundance of flies than uplands. Abundance and richness peaked in May/early June, declined in summer, and then increased again in September. Syrphids annually produce two or three generations (Gilbert 1986). The peaks in abundance and richness may be due to generations of larvae hatching in late spring and fall when playas typically have more water present. The increase could also be due to different species emerging at various times throughout the season.
Although playas, even when surrounded by crop fields, can provide reproductive sites for hoverflies, pollen and nectar producing forbs are still required to complete their life cycle. Therefore, promoting forb production by removing sediment from playas and planting forb-rich buffer strips can assist resilient pollinators, such as hoverflies, to persist in an agriculturally dominated landscape.

Recommendations
Although our data did not show strong evidence of differences among land-condition types and between playas and uplands, it did indicate that reference and restored sites support greater richness of hymenopterans, and playas are likely providing additional habitat resources for dipterans and additional floral resources for both orders. This is consistent with research that has demonstrated the importance of natural areas for wild bees (Garibaldi et al. 2011;Smith et al. 2013;Alomar, 2018) and dipterans (Kearns 2001). Playas in the RWB are the primary remaining natural habitats available to provide necessary resources to pollinators to complete their life history. However, playas are a small fraction of the landscape-approximately 0.4% (Walker 2016). Conserving these wetland areas could be beneficial to both hymenopteran and dipteran pollinators.
The WRP has been critical in protecting and restoring the few remaining playas in the RWB. Restoration practices for playas such as sediment removal are important for restoring the natural hydroperiod, removing invasive species, and allowing the native seed bank to establish (Beas et al. 2013), thus providing more forage for pollinators. Furthermore, restoring adjacent uplands could be helpful to pollinators, especially for bees that need grasslands for nest sites, nest material, and overwintering habitat (Westrich 1996;Hoffman Black et al. 2011). This is also important in late season when many pollinators use the adjacent uplands.
In the agricultural matrix of the RWB, playas and adjacent uplands are providing important habitat (i.e. food, nesting resources, and overwintering sites) for bees. In order to maintain the ecosystem service of flowering-plant pollination, it is critical to promote programs such as the WRP (now WRE), which are restoring playas through watershed development, diverse forb communities, and enhancement of wetland seed banks through sediment removal (Beas et al. 2013). Playas void of buffer strips are beneficial for habitat generalists such as hoverflies, but lack the necessary resources to sustain bee populations. Continuing to protect and enhance the entire playa ecosystem, including the watershed, is essential to maintaining natural or semi-natural areas to protect the full diversity of pollinators in the RWB and support pollination services.
creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.