Three decades of native bee biodiversity surveys at Pinnacles National Park highlight the importance of monitoring natural areas over time

Thousands of species of bees are in global decline, yet research addressing the ecology and status of these wild pollinators lags far behind work being done to address similar impacts on the managed honey bee. This knowledge gap is especially glaring in natural areas, despite knowledge that protected habitats harbor and export diverse bee communities into nearby croplands where their pollination services have been valued at over $3 billion per year. Surrounded by ranch and farmlands, Pinnacles National Park in the Inner South Coast Range of California contains intact Mediterranean chaparral shrubland. This habitat type is among the most valuable for bee biodiversity worldwide, as well as one of the most vulnerable to agricultural conversion, urbanization and climate change. Pinnacles National Park is also one of a very few locations where extensive native bee inventory efforts have been repeated over time. This park thus presents a valuable and rare opportunity to monitor long-term trends of native bees in natural habitats. Fifteen years after a species inventory marked Pinnacles as a biodiversity hotspot for native bees, we resurveyed these native bee communities over two flowering seasons using a systematic, plot-based design. Combining results, we report a total of 450 bee species within this 109km2 natural area of California, including 48 new species records as of 2012 and 95 species not seen since 1999. As far as we are aware, this species richness marks Pinnacles National Park as one of the most densely diverse places known for native bees. We explore patterns of bee diversity across this protected landscape, compare results to other surveyed natural areas, and highlight the need for additional repeated inventories of bee biodiversity in protected areas over time amid widespread concerns of bee declines.

most valuable for bee biodiversity worldwide, as well as one of the most vulnerable to 23 agricultural conversion, urbanization and climate change. Pinnacles National Park is also one of 24 a very few locations where extensive native bee inventory efforts have been repeated over time. 25 This park thus presents a valuable and rare opportunity to monitor long-term trends of native 26 bees in natural habitats. Fifteen years after a species inventory marked Pinnacles as a 27 biodiversity hotspot for native bees, we resurveyed these native bee communities over two 28 flowering seasons using a systematic, plot-based design. Combining results, we report a total of 29 450 bee species within this 109km 2 natural area of California, including 48 new species records 30 as of 2012 and 95 species not seen since 1999. As far as we are aware, this species richness 31 marks Pinnacles National Park as one of the most densely diverse places known for native bees. 32 The importance of bees as critical ecosystem service providers can scarcely be 37 exaggerated. Twenty thousand species of bees worldwide provide the pollination services 38 required for reproduction in 85% of wild and cultivated plants [1,2]. In the United States, the 39 economic importance of bees to agriculture has been valued at up to $14.6 billion annually [3], 40 with $3.08 billion and up to 30% of the U.S. diet specifically credited to the four thousand North 41 American species of native, non-honey bees [4]. Diverse assemblages of native bees have been 42 found capable of enhancing fruit set and yield in the presence of imported honey bees, and of 43 providing adequate pollination for a majority of crops in their absence [5][6][7]. In natural areas, 44 without the manpower of imported, managed honey bee hives, native bees play a key role in 45 maintaining plant communities that provide soil structure, shelter other invertebrate ecosystem 46 service providers, and establish the base of the food chain [8,9]. 47 Although native bees are often observed pollinating agricultural fields, they seldom nest 48 there. Instead, they rely on nearby remnant patches of semi-natural habitat, a resource that is 49 rapidly disappearing with increasing agricultural intensification, habitat fragmentation, and urban 50 development [10][11][12]. Despite recognition of natural areas as valuable reservoirs of pollinators 51 [13,14], research on native bee ecology remains concentrated in urban or agricultural settings 52 where baselines may already reflect impacts of degraded ecosystems. Compared to massive 53 honey bee research efforts, progress towards a holistic understanding of how to protect wild bee 54 communities or the habitats they require has not matched their value as pollinators or the known 55 risks they face [15][16][17]. 56 The relative paucity of research on native bees is due, in part, to the complexity of their 57 biology and behaviors, particularly in wild landscapes. Efforts to monitor wild bees must 58 In 2005, Pinnacles National Monument acquired an additional 15km 2 of privately-owned 115 land that expanded the park boundary primarily to the east, but also incorporated some relatively 116 inaccessible lands to the north and south. In 2010, TLG initiated a follow-up biodiversity survey 117 of the bees at Pinnacles, including the new lands to the east. In order to better track temporal 118 trajectories in native bee biodiversity and phenology, we adopted a more systematic park-wide 119 sampling protocol and established long-term monitoring plots where timed, regular collecting 120 events using both nets and pan traps were conducted by JMM across the 2011 and 2012 121 flowering seasons. The following methods and results are focused on this most recent systematic 122 survey, since a summary of the 1996-1999 inventory has previously been published [37]. 123

124
For the 2011-2012 re-inventory effort, we established ten 1-hectare long-term plots 125 across a diversity of habitat types and reasonably-accesssible areas of the park. We placed three 126 plots on the western side of the rocky spine divide: two in grasslands and one in a Blue Oak 127 woodland. On the larger, lower-elevation eastern side, we set up three plots in alluvial habitats, 128 two in Live Oak woodlands, and one in a Blue Oak woodland. We also established one plot in a 129 Blue Oak woodland along the high rock spine bisecting the park. One-hectare rectangular plots 130 were roughly 200m by 50m, which fit the constraints of the narrow canyon landscapes. In 131 addition to sampling within plots, we visited areas sampled during the original inventory as well 132 as newly-aquired lands to conduct opportunistic aerial net collecting and set out pan traps at the 133 same locations that were sampled using pan traps in 2002 (Fig 1). 134 Spatially, our collecting extended beyond previous efforts to capture bee biodiversity in 135 three main ways: by traveling off the trail network (along which most collecting was conducted 136 in the 1990s, except for one extensive burned area) for plot and opportunistic sampling, by 137 explicitly establishing repeatedly-sampled plots in a diversity of habitat types across the park, 138 and by venturing into the 15km 2 of new lands acquired by Pinnacles National Monument in 2005 139 for both opportunistic and systematic sampling (Fig 1). Temporally, whereas sampling in the 140 1990s was somewhat irregular, in 2011-12 we sought to capture the full bee community 141 phenology by sampling plots fortnightly throughout the entire flowering season, beginning in 142 February before bee activity began and continuing through late June after most bloom had faded 143 [46]. 144 We sampled all ten plots, typically two per day, every fortnight on days that were mostly 145 sunny, without high winds, and over 15C. We conducted additional opportunistic net collecting 146 along the trail network or in new off-trail areas in between plot efforts. Immediately before each 147 collecting event, we recorded the ambient temperature, wind speed, humidity, barometric 148 pressure, and a categorical cloud cover value. During plot sampling, two collectors used aerial 149 nets to perform thirty-minute timed collections of all bees visually or auditorily detected in plots 150 at consistent times in both the morning and afternoon. In order to sample the community as 151 evenly and systematically as possible, we walked a steady pace through plots rather than 152 focusing on activity at flowers. We placed all netted bees in vials according to their floral host 153 and collected a voucher plant when the floral host was unknown. At the end of sampling days, 154 we pinned and labeled all specimens and froze them for 48 hours to prevent beetle infestation. 155 In addition to net collecting, we also set out thirty colored pan traps, a common passive 156 collection method, between 9am and 4pm in each plot on the day we net collected there. Pan 157 traps were made prior to going into the field by spraying 2-oz Solo cups with one of three colors 158 of paint: fluorescent blue, fluorescent yellow, and white, as indicated by the protocol set up for 159 native bee monitoring by Lebuhn et al. [47]. Traps were placed directly on the ground across 160 rectangular plots approximately 10m apart in an "X" pattern in alternating colors and were filled 161 3/4 full of mildly soapy water to break the surface tension and cause bees visiting the bowls to 162 sink to the bottom. At 4pm, we strained insects from the water and immersed them in 75% 163 ethanol until they could be rinsed, pinned and labelled. Data for each pan-trapped specimen 164 includes the color of the bowl from which it was collected. 165   (Table 1a). There was a 79% overlap in species and an 94% overlap in genera between the two 210 years (Table 1b). The preservation and curation of older specimens enabled us to update species 211 determinations from previous inventories based on more recent taxonomic changes to compare 212 and combine biodiversity records across inventory efforts (Table 2). 213

235
The combined results from all three inventories document a total of 450 species of bees 236 across 53 genera and all six North American bee families within the modest 109km 2 of Pinnacles 237 National Park ( Table 2). The most recent survey documented 48 new species records for the 238 Pinnacles National Park area and did not recapture 95 species that had been collected in earlier 239 studies (S1 Table). Of the 48 species recorded for the first time in 2011 and 2012, 47 were rare 240 (here defined as represented by fewer than ten specimens), and 20 were singletons (represented 241 by a single specimen) (S1 Table). Thirty of the 48 new species were captured in areas previously 242 sampled, while 18 were only captured in new lands added to the park since previous inventories 243 (Table 2). Overall, 51 of the 450 species were singletons (Table 2), and 95 were present in only 244 one year of sampling, with the majority of these temporally rare species being from the families 245 Apidae and Andrenidae (Fig 2a). The family Megachilidae had the most species present in all 246 seven years of sampling (N=38 out of 68 total) (Fig 2a). Overlap in species lists between years 247 ranged from 49% to 81% and overlap in genera ranged from 85-99% between any two years 248 (Table 1b).  (Table 1a). Between 2 and 26 species were unique to a particular year and not 255 recorded within the park during any of the other six years of surveys. The genus Ancylandrea 256 (family Andrenidae) was present only in the 1996 collection and 2012 was the only year that 257 three genera from the family Apidae (Neopasites, Peponapis, and Brachynomada) were 258 documented (S1 Table). For five out of six bee families, new species were added to the park list 259 nearly every year. Melittidae is represented by only two common species, both of which were 260 collected in the original year of sampling, and in every year thereafter (Fig 2b). 261 collected in plots across 150 plot samples shows that our efforts captured a majority of the 276 estimated true bee biodiversity within these areas (Fig 3). The leveling off of the curve at the far 277 right indicates that additional plot sampling would be unlikely to yield many more species to the 278 collection, especially for organisms like insects for which observed richness rarely reaches an 279 asymptote [50]. Expanding collecting efforts into areas not extensively surveyed in the more 280 remote and homogenous chaparall habitats in the northern and southern ranges of the park may 281 yield additional bee biodiversity (Fig 1). However, the number of singleton or doubleton species 282 recorded across many genera illustrates the frequency of rare bee species at Pinnacles National 283 Park, which additional sampling efforts may or may not detect (S1 Table). The blue curve and 284 vertical confidence interval lines indicate the estimated rate of species accumulation for a 285 random community with the same number of species and samples (Fig 3). That the observed 286 curve has an initially steeper slope than expected is indicative of Pinnacles' rich biodiversity 287 resulting in rapid early accumulation of common species. 288   (Table 3a). 313 The most bee-popular plants also varied between years. In 2011, more bees visited 314  (Table 3b). 322 Table 3  States that matched our criteria for comparison (N>100 species, extensive inventory-style 333 sampling in a natural area) (S3 Table). It is worth visualizing that, while efforts to survey native 334 bees have increased in recent years, these published inventories still only cover a small 335 proportion of natural areas and habitat types across the United States, and thus offer only a small 336 window into the status of native bees across the country (Fig 4).  bees detected over the past three decades across just 109km 2 , Pinnacles National Park again 364 ranked fourth in the measure of most densely-diverse areas for native bees. 365

Bee Family Genus Species
Because species density is known to not scale linearly across increasing area, we then 366 incorporated the knowledge of increasing diversity over space modeled by the species-area 367 curve. We plotted a power-law species-area relationship based on the reported species richness 368 and area covered by known bee inventories (S3 Table) to calculate which of the 23 listed studies 369 found lower-than-expected bee richness based on their size and which studies were likely true 370 hotspots of native bee biodiversity (Fig 5).  Table.  382 383 Based on this difference between observed and expected species richness per area (the 384 positive or negative distance of the point to the trend line in Fig 5) (Fig 5, S3 Table). Many more studies will be necessary to fill in the map of bee 395 biodiversity in natural areas (Fig 4) and interpret how the bee species-area relationship relates to 396 ecosystem, climate, or habitat stage (Fig 5). Until that time when more comparable data are available, our review of exhaustive bee 427 surveys conducted in the continental United States identified only three natural areas with more 428 recorded bee species than Pinnacles, all conducted across areas larger than Pinnacles by at least 429 an order of magnitude (S3 Table). When plotted along a species-area curve, Pinnacles National 430 Park also has the greatest deviation above the curve compared to any of these other studies, 431 suggesting it contains far more species per area sampled than would be expected in other 432 locations of similar size (Fig 6). Bee diversity in GSENM in Utah, with 1.5 time the number of 433 bees as Pinnacles over 70 times the area, also exceeds the diversity predicted by the species-area 434 curve, though not by the same margin. These results suggest that, while GSENM and several 435 other protected regions sampled are remarkably diverse for the area covered, Pinnacles National 436 Park is home to the most exceptional bee species density of any known place. 437 We attribute the extraordinarily rich bee fauna of Pinnacles National Park to its 438 Mediterranean climate, steep environmental gradients, and high habitat heterogeneity, the last of 439 which has been found in other research to be a stronger predictor of species richness than the 440 species-area relationship [60,61]. Habitat heterogeneity can occur over both space and time. 441 Mediterranean habitats, like those at Pinnacles, are known for rich 'flash-bloom' cycles during 442 spring months, followed by hot, dry summers and mild, wet winters, an environment that tends 443 to support a high biodiversity of many taxa by creating many temporal habitat niches [9,62]. 444 Among bees, the rapid turnover of floral resources in these areas may favor solitary species, 445 whose shorter flight periods and more specialized foraging behaviors may allow many species to 446 coexist in a single area, as each occupies a narrower temporal and foraging niche space than 447 longer-lived social or generalist species, which are more common in temperate areas [19,23]. to long-term residence [11,13]. Agricultural habitats fail to support diverse native bees due to 470 impacts of pesticides, nutritional deficits resulting from monocultures offering only one type of 471 bloom, and the practice of tilling and turning over the soil where many native bee species 472 overwinter [5,30,67]. The native bees known to pollinate crops persist not within the fields but in 473 nearby patches of natural, uncultivated land. California has increased efforts to restore habitat for 474 wild bees in agricultural lands. But less attention has been paid to bee source populations in 475 adjacent natural areas, even though source-sink dynamics have recently been determined to 476 influence bee population sensitivity to decline [68]. 477 If Pinnacles National Park is indeed a biological refuge for native bee populations within 478 a highly-altered landscape, it will be even more important to track trends in its bee biodiversity 479 over time. Our establishment of ten 1-hectare plots and repeatable methodology will facilitate 480 ongoing monitoring activities and better comparisons of bee biodiversity and population stability The need for multi-year, temporally replicated bee surveys to better quantify trends and 492 declines in native bees over time is further highlighted by the recent increase in the use of 493 chronosequences, which substitute space as a proxy for time in restored habitats to model 494 changes in native bee dynamics [69,70]. This is a clever approach but increasing efforts to repeat 495 surveys using the same methodology in the same natural areas over actual timespans would be 496 better. Spatial coverage of published bee inventory studies is sparse (Fig 5), and temporal 497 coverage is worse. Expanding long-term bee biodiversity monitoring to additional habitats and 498 supporting the museum work and collection maintenance that enable these important temporal 499 comparisons will bolster our chances of protecting native bee biodiversity and agricultural 500 stability. 501 502

503
Here we reported details of the third extensive bee inventory effort at Pinnacles Naitonal 504 Park in California over three decades in order to share ongoing findings from a native bee 505 biodiversity hotspot and to highlight the need for more studies like this to evaluate trends among 506 our most important polliantors. Our study is the first to compile and compare information on 507 native bee biodiversity from published inventory studies in natural areas across the United States. 508 With 450 species of native bees, we found that Pinnacles National Park houses a higher density 509 of species than any other natural area studied, and compared to seven California urban areas 510 surveyed (species N = 60-80) [54]. This result indicates that America's newest national park may 511 be a substantial exporter of free, native pollinators into economically valuable agricultural lands. 512 However, our discovery that Pinnacles is the only area to have been extensively and repeatedly 513 surveyed for bee biodiversity over multiple decades underscores our call for increased repeated 514 monitoring efforts to facilitate research on bee population decline and variability at its source. 515 516 Acknowledgements 517 We are grateful to Therese Lamperty for dedication in the field, and to Harold Ikerd and 518 Skyler Burrows for assistance in the lab. This work would not have been possible without 519 generous guidance from USU co-P.I. Edward W. Evans and from Pinnacles wildlife biologist 520  collection as separate from original 1996-9 inventory, but still prior to recent study      Table. Floral taxa visited by bees at Pinnacles National Park (unique groups, identified 751 to lowest possible level), and their relative popularity by year. Plants are marked with "R" for 752 rare if bee visits were fewer than 10 in that year, with "U" for uncommon if bee visits ranged 753 between 10-100, and "C" for common when over 100 bees were collected on that plant.