1. Solitary bees are central place foragers returning to their nests several times a day with pollen and nectar to provision their brood cells. They are especially susceptible to landscape changes that lead to an increased spatial separation of suitable nesting sites and flower rich host plant stands. While knowledge of bee foraging ranges is currently growing, quantitative data on the costs of foraging flights are very scarce, although such data are crucial to understand bee population dynamics.
2. In this study, the impact of increased foraging distance on the duration of foraging bouts and on the number of brood cells provisioned per time unit was experimentally quantified in the two pollen specialist solitary bee species Hoplitis adunca and Chelostoma rapunculi. Females nesting at different sites foraged under the same environmental conditions on a single large and movable flowering host plant patch in an otherwise host plant free landscape.
3. The number of brood cells provisioned per time unit by H. adunca was found to decrease by 23%, 31% and 26% with an increase in the foraging distance by 150, 200 and 300 m, respectively. The number of brood cells provisioned by C. rapunculi decreased by 46% and 36% with an increase in the foraging distance by 500 and 600 m, respectively.
4. Contrary to expectation, a widely scattered arrangement of host plants did not result in longer mean duration of a foraging bout in H. adunca compared to a highly aggregated arrangement, which might be due to a reduced flight directionality combined with a high rate of revisitation of already depleted flowers in the aggregated plant arrangement or by a stronger competition and disturbance by other flower visitors.
5. The results of this study clearly indicate that a close neighbourhood of suitable nesting and foraging habitats is crucial for population persistence and thus conservation of endangered solitary bee species.
The mere knowledge of maximum bee foraging distances, however, is not enough for the preservation of populations of endangered species. To understand bee population dynamics, we need detailed and quantified information about the costs of foraging flights, but such data are very scarce. Foraging flights may impose high costs on solitary bees. A large proportion of foraging females of Hylaeus punctulatissimus and Hoplitis adunca were found to discontinue foraging already at distances considerably shorter than the species’ maximum foraging distance (Zurbuchen et al. 2010a). Similarly, females of Chelostoma florisomne preferred host plant patches relatively close to their nesting sites (Zurbuchen et al. 2010b), and travel costs were assumed to render distant plant patches less profitable to foraging Osmia lignaria than closer plant patches (Williams & Tepedino 2003). Further, the number of progeny produced within a reproductive season was negatively affected by increasing foraging distance in Megachile rotundata (Peterson & Roitberg 2006b), and offspring production in Osmia lignaria was sufficient to ensure population persistence when nesting sites were surrounded by natural habitat offering suitable floral resources, but not when nesting at sites more distant from natural habitats (Williams & Kremen 2007).
The distribution of host flowers within the flight radius of a bee is expected to influence foraging bout duration and thus reproductive output. Indeed, pollinators were shown to adjust their foraging strategy to different plant distribution patterns (Klinkhamer & Dejong 1990; Cartar & Real 1997; Cresswell 1997; Kunin 1997). To the best of our knowledge, however, the quantitative influence of host plant distribution on the reproduction of solitary bees has not yet been addressed.
For this study, we selected the two differently sized and pollen specialist solitary bee species Hoplitis adunca (Panzer, 1798) (Megachilidae) and Chelostoma rapunculi (Lepeletier, 1841) (Megachilidae). Females of both species build several brood cells during their lifetime as adult insects, which lasts maximally 4–6 weeks (Westrich 1990). Each cell is provisioned with pollen and nectar before a single egg is laid. The hatched larva feeds on the pollen–nectar mixture and develops inside the cell within 1 year to the adult insect.
We compared the impact of (i) foraging distance and (ii) spatial arrangement of host plants on bee reproduction. First, we quantified the effect of long compared to short foraging distances on the duration of foraging bouts and the number of progeny reared by a female bee, i.e. the number of brood cells provisioned per time unit. Secondly, we quantified the effect of aggregated compared to scattered arrangements of host plants on the duration of foraging bouts. We then set the findings from the two experiments into the context of bee ecology and bee population dynamics.
Materials and methods
The two solitary bee species H. adunca and C. rapunculi, which have their reproductive period in summer (June–August), exclusively collect pollen from Echium (Boraginaceae) and Campanula (Campanulaceae), respectively (Westrich 1990; Sedivy et al. 2008). They naturally nest in beetle burrows in dead wood or hollow stems (Westrich 1990), allowing for artificial breeding in hollow bamboo sticks or pre-drilled burrows in wooden blocks, and transport pollen back to the nest in a hair brush (scopa) on the ventral side of the abdomen (Westrich 1990). Hoplitis adunca is a rather large species with an average dry body mass of 19·7 mg and a body length of 11–13 mm, C. rapunculi is a medium-sized species with an average dry body mass of 8·6 mg and a body length of 8–10 mm (Müller et al. 2006). Nests of the two species were collected at different locations in Switzerland from bamboo sticks that had been offered the preceding year as nesting sites. These nests were transferred to artificial nesting stands in the study area prior to adult emergence.
To yield movable host plant patches, Echium vulgare (for H. adunca) and Campanula rapunculus (for C. rapunculi) were planted in pots (diameter: 20 cm, volume: 3520 cm3). Both host plant species produce many flowers per individual plant. As these flowers open throughout the day, resource supply was not expected to vary substantially in the course of the day. To ensure a stable quantity and quality of pollen over the complete period of the experiments, flowering was phased by repeatedly trimming shoots before flowering and by cooling plants in a greenhouse chamber (15 ± 2 °C, 70 ± 5% RH, 16L : 8D light regime). The host plant stock was kept outdoors under an insect net to prevent flowers from being exploited by insects prior to the experiments.
Experimental agricultural landscape
The experimental area was an agricultural landscape intensively used for field crops, devoid of the specific bee host plants, in western Switzerland near Selzach, Solothurn (7°27′78″ E, 47°11′63″ N, 420 m a.s.l.). Both experiments were conducted along a straight and unpaved track leading in North–South direction. Wind was generally absent or only weak during the experiments (field weather station CR10 Measurement and Control Module; Campbell Scientific Ltd, Shepshed Leicestershire, UK). In case weak wind came up during the experiments, the tested females were unlikely faced by any direct headwind or downwind, as the prevailing wind direction was from the West, i.e. perpendicular to the North–South direction of the track.
Bee establishment and marking
Hollow bamboo sticks and wooden nesting blocks (150 × 150 × 400 mm) with pre-drilled burrows (120 mm in length, 4–6 mm in diameter for C. rapunculi and 6–9 mm for H. adunca) were prepared as artificial nests and placed in a covered shelf to protect them from rain. To support initiation of nesting activity by the newly emerged females, flowering host plants in pots (50 plants of E. vulgare and 100 plants of C. rapunculus) were placed at a distance of <1 m from the nesting stands. These plants were removed before the onset of the experiments.
Females that started to nest were caught, immobilized by placing them for 2–3 min in a cool box at 5 °C and marked individually with fast-drying enamel paint (Revell, Bünde, Germany) on the thorax (one or two positions) and the abdomen (one position), applying colour codes with eight different colours.
Impact of foraging distance and spatial host plant arrangement
Females of the two tested bee species nested at different distances from a single large and movable flowering host plant patch in an otherwise host plant free environment (Fig. 1). Specifically, H. adunca was offered a single host plant patch between the two nesting stands A and B in year 1 (2007) and between nesting stands C and E in year 2 (2008). The following distance pairs were tested simultaneously each: 100 vs. 300 m (year 1); 225 vs. 375 m and, after moving the plant patch, 450 vs. 150 m (year 2). To test a total of three distances at the same time, C. rapunculi was offered a single host plant patch North of the three nesting stands C, D and E in year 2, resulting in a distance triplet of 1000 vs. 500 vs. 400 m. A total of 80 and 70 nests were transferred to each of the nesting stands of H. adunca and C. rapunculi, respectively. The number of potted plants used to form a movable host plant patch amounted to 150 for E. vulgare, covering an area of 4 m2, and 250 for C. rapunculus, covering an area of 6 m2. Withered plants were replaced at regular intervals with fresh plants from the host plant stock. To ensure that no naturally occurring host plants were available, the area was checked prior to the start of the experiment and repeatedly during the experimental period. No flowering plants of the genus Echium were found within a radius of a minimum of 1200 m from nesting stand A, 800 m from nesting stand B, 1200 m from nesting stand C and 600 m from nesting stand E. During the experiments, a small patch of non-flowering E. vulgare was located and eliminated. No flowering plants of the genus Campanula were found within a radius of a minimum of 1200 m from nesting stand C, 700 m from nesting stand D and 600 m from nesting stand E.
Two spatial arrangements of host plant distribution were tested in H. adunca. In the aggregated arrangement, plants were concentrated in an area of 4 m2, at a density of 37·5 plants m−2. In the scattered arrangement, plants were placed perpendicular to the North–South axis of the experimental track in 2-m intervals and in groups of one to three plants along a line of 160 m length extending both to the West and East, yielding a density of 4·7 plants m−2.
To test the effect of the foraging distance and the spatial arrangement of host plants on the time needed by a bee for a single foraging bout, the duration of foraging bouts of marked females was recorded to the nearest second simultaneously by one observer each per nesting stand. The duration of a foraging bout was defined as the time an individual female bee needed from leaving the nest until return to the nest with pollen and nectar. Only foraging bouts of bees returning with pollen visible in the abdominal scopa were considered. The duration of two to six foraging bouts per individual female and per observation day was recorded. For every individual bee, the mean duration of a foraging bout per distance and per spatial host plant arrangement was calculated by pooling the data from all days of observation. At each day of observation, the presence of marked females on the host plant patch was recorded during 2 h by an additional observer. Foraging bouts of females that were observed at the nest but not on the host plant patch the very same day were excluded. All observations were carried out on sunny days with maximum daily temperatures ranging from 23·2 to 30·7 °C between 10.00 and 18.00 h when bee activity was high (field weather station CR10 Measurement and Control Module; Campbell Scientific Ltd).
Impact of foraging distance on pollen load
The experiments conducted in this study rely upon the crucial assumption that the number of pollen grains transported per foraging flight is independent of the foraging distance. To test this assumption, we compared the mean number of pollen grains transported back to the nest by females of H. adunca foraging at the experimental host plant patch at a distance of 450 and 150 m from their nests, respectively (Fig. 1). On 3 days in July 2008, a total of 33 pollen-loaded females returning to their nests were caught by one observer each per nesting stand during the same observation period. Pollen contained in the abdominal scopa was washed off with 1 mL of ethanol (70%) to remove the adhesive pollenkitt. After ultrasonic treatment for 2 min with a ultrasonic bar (Vibra Cell 72446; Bioblock, Illkirch, France) at 20 kHz to loosen the pollen grains from each other, 1 mL of the homogeneous ethanol–pollen mixture was centrifuged at 2500 r.p.m. for 5 min. To remove the ethanol, 0·8 mL of the supernatant was discarded and replaced by 0·8 mL H2O for a final volume of 1 mL. After thoroughly stirring to disperse evenly the pollen grains within the solution, one droplet of the pollen solution was transferred to the chamber of a haemocytometer (Neubauer improved; Brand, Wertheim, Germany). The pollen grains in each of the four corner squares characterized by an exactly defined volume (0·0001 mL) were counted under a microscope (Olympus BX 50; Tokyo, Japan) at a magnification of 100×. Each pollen sample was processed three times, resulting in a total of 12 squares counted per pollen load. To obtain an estimate of the total number of pollen grains per scopal load, the mean number of pollen grains counted per square was extrapolated in proportion to the initial volume of the pollen solution.
Impact of foraging distance on the number of brood cells provisioned
Experiments with H. adunca were conducted in July 2007 and in June and July 2008. The duration of foraging bouts for the three distance pairs of 100 vs. 300, 225 vs. 375 and 450 vs. 150 m was recorded for 2, 7 and 6 days, respectively. A total of 26, 35 and 43 females were tested for the first, second and third distance pair, respectively. Experiments with C. rapunculi were conducted in July 2008. A total of 23 females were tested for the distance triplet of 1000 vs. 500 vs. 400 m during 3 days.
To quantify the effect of different foraging distances on reproduction, the mean number of brood cells provisioned per time unit was estimated for each species and foraging distance tested. The quantity of pollen and nectar contained in the brood cells of solitary bees is known to vary considerably within species depending on the sex of the offspring, the size of the adult females or the quantity and availability of host plants (Bosch & Vicens 2006; Peterson & Roitberg 2006a, b; Bosch 2008). Therefore, the mean time required to provision an average-sized standard brood cell was determined for each species and foraging distance by multiplying the mean duration of a foraging bout by the average number of flights needed to provision a single brood cell. The average number of flights (FBC) needed to provision a brood cell was calculated for each species by dividing the mean number of pollen grains contained in a brood cell (PBC) by the mean number of pollen grains transported per foraging flight (PF). To estimate PBC and PF, freshly completed brood cells, in which the larvae had not yet hatched, as well as unmarked females upon their return from a foraging bout with a filled abdominal scopa were randomly collected at the nesting stands. The procedure of pollen counting was as described above except for: (i) the pollen content of each brood cell was initially dispersed in 10 mL of ethanol (70%) for H. adunca and 5 mL of ethanol for C. rapunculi; and (ii) the number of pollen grains contained in a brood cell was estimated based on two samples of 1 mL each, resulting in a total of 24 haemocytometer squares (volume 0·0001 mL) counted per brood cell.
The indirect method applied in this study to quantify the number of progeny reared by a female ruled out any bias that might originate from possible unequal conditions at the nesting sites, such as abundance of natural enemies.
Impact of spatial host plant arrangement on the duration of foraging bouts
To test the effect of different spatial arrangements of host plants on the duration of foraging bouts in H. adunca, a single plant patch composed of 150 potted E. vulgare was offered to the foraging female bees either in an aggregated arrangement with a density of 37·5 plants m−2 or in a scattered arrangement with a density of 4·7 plants m−2 (see Experimental design) at a distance of 450 and 150 m from the nesting stands C and E, respectively (Fig. 1). Duration of foraging bouts for both plant arrangements was recorded twice during two periods of 3 days each. Both periods were characterized by sunny and windless weather conditions with very similar maximum daily temperatures (July 15–17 and 24–26, 2008). During the first period, data for the aggregated plant arrangement were collected on 1 day, followed by data collection for the scattered plant arrangement on 2 days. During the second period, data for the aggregated plant arrangement were collected on 2 days, followed by data collection for the scattered plant arrangement on 1 day. A total of 21 individual females were tested for the distance of 150 m and 30 individual females for the distance of 450 m. Data for the aggregated plant arrangement are part of the data set collected for the distance pair 450 vs. 150 m in the experiment on foraging distances detailed above. The exact time and position of each marked female observed on the plants in the scattered plant arrangement was recorded by at least two observers to obtain information on the spatial use of the host plants.
Impact of foraging distance on pollen load
The number of pollen grains transported per foraging flight by females of H. adunca was not significantly different between foraging distances of 150 and 450 m (Fig. 2) (two-sample t-test, P = 0·511, n150 = 15, n450 = 18).
Impact of foraging distance on the number of brood cells provisioned
Foraging over longer compared to foraging over shorter distances required significantly more time for H. adunca females in all three distance pairs tested (Table 1, Fig. 3) [two-sample t-tests: 225 vs. 375 m, P < 0·01, n225 = 18, n375 = 17; 100 vs. 300 m, P < 0·01, n100 = 9, n300 = 17; 150 vs. 450 m, P < 0·001 (Bonferroni corrected), n150 = 18, n450 = 25]. Chelostoma rapunculi females needed significantly more time to forage at 1000 m compared to either 400 or 500 m, whereas no difference in the duration of foraging bouts was found between females foraging at 400 and 500 m (Table 1, Fig. 4) (anova: P < 0·05, Tukey’s HSD: P1000–400 < 0·05, P1000–500 < 0·05, P500–400 = 0·676, n400 = 11, n500 = 6, n1000 = 6).
Table 1. Mean duration of a foraging bout (tF), mean time to provision an average-sized standard brood cell (tBC), the proportion of a brood cell provisioned per hour (BC/h) and calculated decrease in the number of brood cells provisioned per time unit for different distance pairs/triplets and plant distribution arrangements in the two solitary bee species Hoplitis adunca and Chelostoma rapunculi. n, number of female bees tested. Different letters indicate a significant difference in the mean duration of a foraging bout. tBC = tF*FBC, where FBC is the mean number of foraging flights needed to provision a brood cell. NS, no significant difference. Experiments with H. adunca were conducted on 16–18 July 2007, 24 June–5 July 2008 and 10–29 July 2008 for the distance pairs 100 vs. 300, 225 vs. 375 and 150 vs. 450 m, respectively. Experiments with C. rapunculi were conducted on 22 June–9 July 2008
Plant distribution pattern
Distance pairs/triplets (m)
t-Test, P < 0·01
t-Test, P < 0·01
t-Test, P < 0·001
NS (400 vs. 500 m)
anova, P < 0·05 with Tukey’s HSD
36% (1000 vs. 400 m)
46% (1000 vs. 500 m)
Two-way anova, distance P < 0·001, distribution P = 0·721
An average number of 46 foraging flights were needed by females of H. adunca to provision a standard brood cell based on the mean number of pollen grains transported per foraging flight (mean ± SE = 501 700 ± 44 700, n = 56) and the mean number of pollen grains stored in a brood cell (mean ± SE = 23 083 000 ± 1 120 000, n = 30). An average number of 19 foraging flights were needed by females of C. rapunculi to provision a standard brood cell based on the mean number of pollen grains transported per foraging flight (mean ± SE = 187 500 ± 17 900, n = 33) and the mean number of pollen grains stored in a brood cell (mean ± SE = 3 596 300 ± 284 600, n = 30).
The time needed to provision a standard brood cell significantly increased with increasing foraging distance in both bee species (Table 1), except for C. rapunculi foraging at 400 vs. 500 m. Correspondingly, the proportion of brood cells provisioned per time unit decreased in H. adunca by 23% with an increase in the foraging distance by 150 m, by 31% with an increase in the foraging distance by 200 m and by 26% with an increase in the foraging distance by 300 m (Table 1). In C. rapunculi, the proportion of brood cells provisioned per time unit decreased by 46% with an increase in the foraging distance by 500 m and by 36% with an increase in the foraging distance by 600 m (Table 1).
Impact of spatial host plant arrangement on the duration of foraging bouts
No significant difference in the duration of foraging bouts was found between the aggregated and the scattered host plant arrangement in H. adunca (Table 1, Fig. 5), while the duration of foraging bouts was significantly longer for the distance of 450 m compared with the distance of 150 m [two-way anova: distance P < 0·001; distribution: P = 0·721; interaction between distance and distribution: P = 0·532 (Bonferroni corrected); naggr_150 = 17, nscatt_150 = 18, naggr_450 = 23, nscatt_450 = 21].
Whenever distances between nesting stands and a single experimental host plant patch increased by more than 100 m, the duration of foraging bouts consistently increased in both solitary bee species tested. This increase in the mean duration of a foraging bout cannot be attributed to a larger amount of pollen collected requiring more time, as the quantity of pollen transported back to the nest by the individual female bees was found to be independent of the foraging distance. Instead, it is likely due to longer flight durations needed to cover the increased distance between nesting stand and plant patch. In fact, a direct linear relationship between duration of foraging bout and foraging distance is apparent in C. rapunculi. Under the assumption of a direct and linear flight between nest and host plants with a constant flight velocity of 1·5 m s−1, which applies to slowly flying honeybees and orchid bees (Brodschneider, Riessberger-Galle & Crailsheim 2009; Combes & Dudley 2009), the extra time needed by the females of C. rapunculi to cover the additional distances of 2 × 500 and 2 × 600 m, respectively, is roughly equivalent to the increase in the mean duration of a foraging bout measured in the field. In contrast, assuming again a straight flight at the same velocity, substantially more time is needed by the females of H. adunca. While under these assumptions flights would last only 3 min 20 s (for 2 × 150 m), 4 min 27 s (for 2 × 200 m) and 6 min 40 s (for 2 × 300 m), respectively, females spent between 4 and 6 min longer for a single foraging bout. This deviation suggests that yet unknown factors might influence the duration of foraging bouts in H. adunca. Hardly anything is known about the directionality of a bee’s flight between foraging habitat and nesting site. While the assumption of a direct and linear non-stop flight applies to certain species, such as C. rapunculi, it might be too simple for other species, such as H. adunca. Longer foraging distances could require more frequent or more extended nectar uptakes to meet the higher energy expenditure or time consuming exploratory loops for orientation. Indeed, feeding on nectar is well known to increase the flight capacity of insects, such as parasitoid wasps (Wanner, Gu & Dorn 2006; Rousse et al. 2009).
By applying an indirect method to quantify the time required to provision a standard brood cell based on the duration of foraging bouts and the average number of foraging flights, the number of brood cells provisioned per time unit was found to decrease substantially with increasing foraging distance in both bee species tested. The reduction in the number of brood cells in H. adunca amounted to 23%, 31% and 26% with an increase in the foraging distance by 150, 200 and 300 m, respectively, and in C. rapunculi to 46% and 36% with an increase in the foraging distance by 500 and 600 m, respectively. The nonlinearity of the calculated percentage decreases with increasing foraging distance might reflect varying abiotic and biotic conditions during the experiments, including temperature and host plant quality. Temperature, humidity and light intensity strongly influence bee behaviour (Corbet et al. 1993; Stone 1994; Bosch & Kemp 2002; Klein, Steffan-Dewenter & Tscharntke 2004) and a low pollen and nectar supply adversely affects bee reproduction (Peterson & Roitberg 2006a). The number of offspring produced by solitary bees is very low compared to many other insect taxa (Westrich 1990). Foraging for pollen and nectar accounts for an average of 72% of the total time needed to build, provision and close an entire brood cell in three different species of Osmia (Maddocks & Paulus 1987; Müller 1994), the sister genus of Hoplitis, and 51% in Chelostoma florisomne (Herrmann 1999), a member of the same genus as C. rapunculi. By applying these percentages to the two bee species investigated and considering the weather conditions at the study site in 2008, we estimate that the females of H. adunca and C. rapunculi were able to construct 7 and 22 brood cells, respectively, for the longer foraging distances tested, and 15 and 32 brood cells, respectively, for the shorter distances. Given this low number of brood cells, already a moderate increase in foraging distance might lower offspring production below the threshold value needed to ensure persistence of a bee population.
The magnitude of the reduction in the number of brood cells provisioned per time unit with increasing foraging distance as found in this study is a conservative estimate as our experimental design did not consider the impact of longer foraging distances on bee senescence and life span. Increased flight activity is expected to further senescence and reduce life span. Indeed, life span was shown to be shortened by a high flight activity in honeybees (Neukirch 1982; Schmid-Hempel & Wolf 1988). Similarly, the wear of parts of the exoskeleton, e.g. wings or pollen collecting apparatus, as well as the physiological ageing of the flight muscles reduce flight and foraging capacity and increase mortality in adult bees (Torchio & Tepedino 1980). Longer foraging distances are not only expected to affect female reproductive capacity, they may influence offspring survival as well. With increasing costs of foraging flights, fewer resources are allocated to offspring (Kim & Thorp 2001; Peterson & Roitberg 2006a), resulting in a lowered survival rate of overwintering bee larvae (Bosch & Kemp 2004; Peterson & Roitberg 2006b; Bosch 2008). In addition, parasitization of the open brood cells might increase at longer foraging distances as females spend more time away from their nest allowing for more or longer attacks by natural enemies (Goodell 2003; Peterson & Roitberg 2006b; Seidelmann 2006). In conclusion, we argue that the decrease in the number of brood cells with increasing foraging distance as found in this study for H. adunca and C. rapunculi would have been even more pronounced when considering the effect of longer foraging distances on senescence, life span and larval survival. Indeed, in a study that included the effects of senescence and mortality, the mean number of brood cells built during one season was 74% lower in females of Megachile rotundata foraging in plant patches 150 m away from their nests compared to females foraging in plant patches directly adjacent to the nests (Peterson & Roitberg 2006b).
Surprisingly, no effect of host plant distribution on the mean duration of foraging bouts was found in H. adunca. This finding is in contrast to studies on bumblebees showing that flight durations between plants increased with decreasing plant density (Harder 1988; Ohashi & Yahara 2002). As flowers of E. vulgare contain only small amounts of pollen, which is successively shed by the five anthers, females of H. adunca have to visit a high number of flowers to fill their abdominal scopa with pollen. Thus, the considerably lowered host plant density in the scattered host plant arrangement actually forced the foraging female bees to fly substantially longer distances between individual host plants on the very same foraging bout compared to the aggregated arrangement. In fact, several marked females of H. adunca were observed to visit host plants spread over a distance of at least 140 m during a single foraging bout. We suppose that the lack of difference in foraging bout duration between the two arrangements of plant distribution was due to the very high density of flowers in the aggregated host plant patch. Hence, less efficient pollen collection may have blurred the advantage of the short inter-plant flight distances. This hypothesis is in line with studies on bumblebees, which showed that increasing plant density resulted in a decrease of flight directionality (Cartar & Real 1997; Cresswell 1997, 2000), that the frequency of revisitation of the same flowers was almost twice as high in patchy flower arrays than in uniform arrays (Cresswell 2000), that the search for unvisited inflorescences might be easier in small than in large host plant patches (Goulson 2000), and that a higher abundance of flower visitors restricts the foraging area due to interactions with other pollinators (Comba 1999; Makino & Sakai 2005).
In conclusion, this study clearly shows that increasing spatial separation of nest and host plants substantially reduces offspring production in two solitary bee species already at a small spatial scale. Thus, spatial separation of nesting and foraging habitats due to past and present landscape changes might be an important reason for the current decline in local species diversity and population size in solitary bees. For the conservation of endangered solitary bee species, a close neighbourhood of suitable nesting sites and flower rich foraging habitats within a maximal distance of few hundred metres appears to be crucial, even if the species-specific maximal foraging distance is considerably longer.
We thank all the farmers involved for their generous collaboration, the municipal administration of Selzach for their support and the authorities of the Canton Solothurn for allowing us to work in the agricultural conservation area. We thank L. Landert for help with field work, S. Scholl for watering our plants, and K. Tschudi-Rein, L. Lubitz and D. Mazzi for comments on earlier drafts of the manuscript. This study was supported by the Competence Centre Environment and Sustainability (CCES) and the Biedermann-Mantel foundation.