Ecology of yeasts in plant–bumblebee mutualism in Central Europe


  • Michael Brysch-Herzberg

    Corresponding author
    1. Department of Biology, Philipps-University Marburg, Karl-von-Frisch-Straße, 35043 Marburg, Germany
      E-mail address:
    Search for more papers by this author

E-mail address:


Yeast community involved in plant–bumblebee mutualism was investigated in three successive years. Yeasts were isolated from floral nectar, bumblebee queens after hibernation, bumblebee workers, and the honey provisions in nests. From the distribution of yeast species in the various microhabitats in the course of the year their ecology was assessed. Nectar of numerous plant species belonging to various plant families was analyzed in order to uncover possible impacts on the yeasts present in the nectar. Only ascomycetous yeasts were autochthonous members of the communities in the plant–bumblebee mutualism. Species in the Metschnikowia clade, the Starmarella clade, and the genera Debaryomyces and Zygosaccharomyces were associated with the mutualism. Some species appeared highly specialized, whereas others had a broader distribution. While physical and chemical properties of nectar had only limited influence on the abundance of nectar yeasts, the attractiveness of plants to the flower-visiting insects appears to have had a greater impact on the abundance and frequency of yeasts in the nectar of different plant species.


Yeast communities associated with mutualism of nectar-producing plants and flower-visiting insects were first examined by Boutroux [1] who reported that yeasts regularly occur in floral nectaries and not just by chance. He concluded that insects, in particular bees (Subfamily: Apoidae), are the vectors distributing yeasts among flowers. Later, Grüß[2] described morphological adaptation of Metschnikowia gruessii to its dispersal by bumblebees (Bombus sp.) and honeybees (Apis mellifera).

Researchers interested in yeasts in the plant–bee mutualism had long wondered how nectar yeasts survive winter in temperate zones where flowers and vector insects are absent during winter. The yearly cycle of yeasts was first investigated by Hansen [3,4]. From his studies on Hanseniaspora uvarum he concluded that all yeasts over-winter in soil and that they are distributed to flowers and fruits by wind in early spring. In contrast to these findings, Rommier [5] and Berlese [6] proposed that H. uvarum is distributed by insects and that the yeast survives through the winter with the insects. This theory was later adopted by others [2,7–9] who suspected that M. gruessii over-winters inside bumblebees and is inoculated into floral nectars in early spring.

The chemical and physical properties of nectar in different plant species were believed to impact the diversity of yeast species present [10,11]. Grüß[2] was convinced that the differences in nectar properties among various plant species led to the development of different morphological types of M. gruessii. Contrarily, other authors could not find any relation between different nectar types and the occurrence of specific yeast species or morphotypes [12–14].

Capriotti [15] and Lund [13] were the first to make a reliable investigation on the species spectrum of yeasts in floral nectar. With the description of M. gruessii and the resolution of a long-standing confusion in nomenclature [16], it became possible to correctly identify the most abundant yeast species in nectars.

Few authors [1–3,17] have investigated the ecology of yeasts associated with floral nectar and bees in Central Europe, and none of their studies explored the whole spectrum of known autochthonous species and their yearly activity. The aim of the current study was to gain further knowledge on the ecology of yeast species associated with nectar-producing plants and bumblebees in Central Europe. The spectrum of yeasts in the floral nectaries of Helleborus foetidus was investigated from winter to spring in order to detect seasonal changes in the composition of the yeast community. Since those changes may be induced by bumblebee queens that begin to forage nectar after hibernation, these animals were included in the study. H. foetidus produces large amounts of highly concentrated nectar [18] and has the advantage of flowering over the whole period of the investigation, i.e., from February to the beginning of May. Bagging experiments performed after the appearance of the first bumblebees were aimed at differentiating between seasonal changes induced by environmental factors (e.g., temperature) and those caused by bumblebees. The nectar concentration of H. foetidus was measured in spot checks to characterize the microhabitats occupied by the yeasts.

It has been hypothesized that factors other than the presence or absence of bumblebees affect the composition of the yeast communities. This study thus investigated the distribution of yeast species in plant–bumblebee mutualism in the course of an entire year. Different physical and chemical properties of nectar in plant species from various plant families were analysed. Another environmental factor considered and also recorded, was the temperature. By means of statistical analysis the impact of the environmental factors on the frequency and abundance of yeasts has been investigated.

2Materials and methods

All environmental samples were collected in the New Botanical Garden in Marburg, Germany. Bumblebees were caught from colonies that were reared for this investigation.

2.1Isolation media

Sugar agar (0.2% w/w MgSO4, 0.3% KH2PO4, 0.5% peptone, 13.3% sucrose, 13.3% glucose, 13.3% fructose and 2% agar) [8] was used for the isolation of yeasts from environmental samples in 2001. As some yeast species do not grow well with peptone as the sole nitrogen source, YM agar [19] supplemented with 100 mg/L chloramphenicol was used for the isolation in 2002.

2.2Isolation of yeasts from nectar, other plant materials, and honey

Nectar samples were taken exclusively from H. foetidus (Ranunculaceae) every week from February to the end of April 2001. For a better comparability only flowers of the same age were investigated. A subset of H. foetidus plants was covered with gauze to protect them from flower visiting insects after the appearance of the first bumblebee queens. All nectar samples were taken with sterile scaled capillary pipettes (1–5 μl) and directly plated on the isolation medium. Different parts of H. foetidus plants and any kind of animal (Table 2) other then bumblebees found on the plants were collected. The plant material was gently pressed on the isolation medium and the animals were allowed to walk on agar plates for 1 h.

Table 2.  Number of samples from different substrates, which harboured the various yeast taxa, N = number of samples
Isolation substrateNN with yeastsM. reukaufiiM. gruessiiM. pulcherrimaC. rancensisC. kunwiensisD. maramusD. hanseniiZ. rouxiC. bombiBasidiomycetes
Plant surfaces
Green and decaying leaves1210         10
Calyx leaves65         5
Nectar February to April 2001
Before the appearance of bumblebees26311         11
After the appearance of bumblebees            
Flowers protected from insects633         3
Flowers not protected from insects5085932       721
Nectar 2002
April74331661      13
May to September1201067752312 11  30
Queens March to April 2001
Bombus spec. Proboscis1752       21
Bombus spec. Surface2016    7   122
Queens from hibernation 2003
Surface, proboscis, digestive tract66    31  42
Workers 2002
Bombus spec. surface2921145 1 4   11
Bombus spec. Proboscis34271816    1  3
Bombus spec. honey bulb875 1      2
Bombus spec. digestive tract21211845161 1 7
Bumblebee honey
Honey pots9942   6148 
Other animals March to April 2001
F. auricularia33         23
M. aenneus31         1
Snails76         6
Other animals May to October 2002
M. aenneus 200299913      7
Other insects333        2

In 2002, yeasts were isolated from 194 nectar samples. These samples came from 140 plant species, which belonged to 27 families (Table 1). Isolation started with the appearance of the first bumblebee at the beginning of April and was continued until the end of September. The sample volume was 4 μl. If a single flower did not yield 4 μl, nectar samples were taken from more than one flower to reach the complete volume. After 5 days of incubation at 20 °C the total number of colonies was estimated as a relative measure of the abundance of yeasts in the nectar.

Table 1.  Plant families, number of plant species, number of nectar samples for the isolation of yeasts, number of samples containing different yeast taxa, nectar pH, nectar sugar molarity, sucrose/hexose ratio of nectar sugars
Plant familyPlant species, NSamples, NBasidiomycetes, NAscomycetes, NM. reukaufii, NM. gruessii, NC. rancensis, NNectar pHNectar sugar molarity [M/L]Nectar sucrose/hexose ratio
        Mean (SD)RangeMean (SD)RangeMean (SD)Range
  1. N, number of cases; SD, standard deviation.

Asteraceae564322 5.3 (±0.9)4.3–6.50.63 (±0.21)0.5–0.91.03 (±0.11)0.89–1.14
Berberidaceae33 322 5.3 (±0.4)5.0–5.70.54 (±0.16)0.4–0.71.02 (±0.36)0.65–1.37
Boraginaceae9134661 4.9 (±0.2)4.7–5.20.86 (±0.34)0.5–1.70.79 (±0.08)0.70–0.94
Brassicaceae44 321 5.8 (±0.7)4.8–6.30.31 (±0.04)0.2–0.30.77 (±0.03)0.74–0.80
Campanulaceae553433 6.4 (±0.4)6.0–6.70.76 (±0.58)0.2–1.50.85 (±0.11)0.69–0.95
Caryophyllaceae443321 3.9 (±0.8)3.0–4.30.87 (±0.16)0.8–1.10.70 (±0.19)0.47–0.94
Dipsacaceae332332 5,20.76 (±0.28)0.6–1.00.70 (±0.07)0.65–0.75
Ericaceae911 422 7.6 (±1.6)5.7–100.80 (±0.21)0.5–1.10.79 (±0.24)0.48–1.11
Fabaceae591645 5.8 (±0.7)4.8–6.32.06 (±1.30)0.9–3.80.78 (±0.34)0.31–1.18
Fumariaceae33     5.5 (±0.9)4.5–6.31.33 (±0.96)0.7–2.41.01 (±0.14)0.88–1.16
Grossulariaceae44 11  7.7 (±0.7)6.7–8.30.58 (±0.14)0.5–0.81.03 (±0.21)0.85–1.34
Hydrophyllaceae142111 4.5 0.4 1.03 
Lamiaceae2138331241984.7 (±0.8)3.5–6.51.00 (±0.39)0.6–2.21.00 (±0.20)0.72–1.46
Liliaceae662122 5.8 (±0.6)5–6.30.98 (±0.54)0.5–1.80.60 (±0.13)0.37–0.75
Lythraceae122    5.8 0.90 0.56 
Oleaceae22 11  4.84.7–4.80.750.3–1.20.620.46–0.78
Primulaceae57332  6.1 (±0.4)5.7–6.51.10 (±0.50)0.6–1.80.670.61–0.72
Ranunculaceae1725312112 7.6 (±1.0)6.2–9.70.89 (±0.56)0.3–2.01.12 (±0.58)0.52–2.95
Salicaceae221    5.64.2–7.00.90 0.610.59–0.62
Rosaceae12143754 8.3 (±1.1)6.5–9.80.57 (±0.28)0.3–1.20.68 (±0.12)0.46–0.86
Saxifragaceae22222  6.86.3––0.40.470.06–0.87
Scrophulariaceae7111185 4.9 (±0.3)4.5–5.31.31 (±0.46)0.7–2.10.92 (±0.21)0.60–1.16
Solanaceae221111 6.35.8––1.10.790.78–0.8
Thymelaeaceae11     4.8 0.800,840.71 
Violaceae22 0.730.49–0.97
Total143189421159157126.1 (±1.5)3.5–100.89 (±0.55)0.1–3.80.87 (±0.31)0.06–2.95

At the end of September 2002, after the young bumblebee queens had left their nests, two nests of Bombus terrestris and two nests of Bombus pascuorum were taken to the laboratory for isolation of yeasts from the honey provisions. Samples were taken with sterile loops and plated directly on the isolation medium.

2.3Isolation of yeasts from bumblebees

Bumblebees of the species B. terrestris, Bombus lucorum, Bombus cryptarum, Bombus lapidarius, B. pascuorum, Bombus pratorum, and Bombus hortorum were captured in the field and anesthetized with CO2. The animals' glossa and their proboscis were gently stroked over the surface of the isolation medium. In addition, the rest of the bumblebees' body was gently pressed on the agar. After this procedure the animals were fed with honey water and set free. Some animals from the colonies reared for this investigation were dissected and the content of their honey bulb and the rest of their digestive tract was plated separately on isolation medium. In March 2003 young bumblebee queens from the reared colonies were caught as soon as they left their hibernation locations. The yeast communities on and in the animals were investigated as described above.

2.4Purification of culture material

Agar plates were incubated at 20 °C for 6 days. At least two representatives of each colony type were transferred to YM agar. After two weeks at 20 °C the cultures were examined with a Leitz Laborlux microscope. For every environmental sample, one culture type was separated by macroscopic and/or microscopic characteristics and purified by producing single-cell cultures according to the method of Vobis [20] in 2001 and by the standard method of Yarrow [19] in 2002.

2.5Phenotypic characterization

All yeast strains that were isolated in 2001 were characterized nutritionally by replica plating [21] and by standard methods [19]. Additionally, tolerance to high sugar concentrations was tested as described by Yarrow [19]. Selected yeast strains isolated in this study were tested for their ability to grow on 2% and the tolerance to 50%, 60%, 70%, 2 M and 3 M of the main nectar sugars sucrose, glucose and fructose. Phenetic similarities with other described yeasts were examined with the computer program YEASTCOMPARE [22]. For at least two strains of each ascomycetous species identified on the basis of physiological tests, the D1/D2 domain of the large subunit rDNA was sequenced (see below) to confirm identity. All strains isolated in 2002 were subjected to the diazonium blue B test as described by Hagler and Ahearn [23]. Strains identified as basidiomycetous were not further considered.

2.6Molecular genetic characterization

All ascomycetous strains isolated in 2002 were identified by means of molecular genetic methods. It was known from earlier investigations [24–26] that Metschnikowia reukaufii, M. gruessii, and Metschnikowia kunwiensis were likely to be the most frequent species in the current study. Therefore, species-specific amplification primers could be used to identify strains of those species.

Species-specific binding sites for amplification primers were identified in the D1/D2 domains of the large subunit rRNA gene. Specific primers were used in combination with universal primers NL1 (5′-GCA TAT CAA TAA GCG GAG GAA AAG-3′) and NL4 (5′-GGT CCG TGT TTC AAG GAC GG-3′) [27]. Species-specific primers: M. reukaufii: Mr (5′-CGAGTATTATAGCCTTTTTCTC-3′) with NL4; M. gruessii: Mg (5′-GGAAGATATG TATTGTGTTG AATCAG-3′) with NL1; M. kunwiensis: Primer Mk (5′-GGAGGGAGGA ACCATCTAGC CGGG-3′) with NL4. To avoid time-consuming DNA isolation, amplification was performed with whole yeast cells of 24 h old cultures [28].

The amplification was conducted as recommended by the manufacturer of Taq polymerase (Q-Biogene, Heidelberg, Germany) in 10 μl glass capillaries (Idaho Technologies, USA) in a Rapid-Cycler (Idaho Technologies, USA). Temperature program: 95° for 5 min, 35 cycles of 95 °C for 15 s, annealing temperature for 10 s, 72 °C for 20 s, and finally 72 °C for 1 min. Annealing temperatures for the different primer pairs were: Mr, NL4: 58 °C; Mg, NL1: 56 °C; Mk, NL4: 62 °C.

The specificity of each pair of primers was tested under the conditions described above with the type strains of the following species from the Metschnikowia clade: Metschnikowia pulcherrima (CBS 5833T), Metschnikowia fructicola (CBS 8853T), Metschnikowia koreensis (CBS 8854T), Metschnikowia hibisci (CBS 8433T, CBS 8434T), Metschnikowia continentalis (CBS 8429T, CBS 8430T), Metschnikowia borealis (CBS 8431T, CBS 8432T), C. ipomeae (CBS 8466T), C. kofuensis (CBS 8058T), M. gruessii (CBS 7657T) and M. reukaufii (CBS 1903T), M. kunwiensis (CBS 9676T).

Strains that did not yield a band with any of the primers were tested again and if no signal was obtained, the D1/D2 domain of the large subunit rRNA gene was sequenced following the instructions by Lachance et al. [28]. Blast search was performed with the obtained sequences in GenBank. According to Kurtzman and Robnet [29], strains that exhibited more than 98% sequence similarity to the closest related type strain were considered to belong to the corresponding species. Two strains of each species were deposited at the CBS (Centraalbureau voor Schimmelcultures, Utrecht, The Netherlands) and sequences were deposited in GenBank.

2.7Temperature and nectar properties

Temperatures were recorded automatically by the New Botanical Garden. In the statistical analysis, the mean temperature three days prior to the day of sampling was used.

In 2001, total nectar sugar concentration of H. foetidus was measured with a laboratory bench refractometer (Zeiss Abbe refractometer). Sugar concentrations (w/w) from 23 blossoms of different plants were measured. In 2002, the total nectar sugar molarity and composition of 143 plant species was measured by HPLC. An HPLC-System (Sykam, Gilching, Gemany) with a refractive index detector (Schammbeck, Germany), an autosampler (Jasco, Japan), an injection valve with a 20 μl loop, and a CARBOSep CHO-682 Pb column (length: 30 cm; diameter: 0,78 cm) (TransgenomicsTM) was used (oven temperature 80 °C; rate 0.4 ml min−1; mobile phase: bidistilled water). The system was calibrated with a wide range of standards. Chromatograms were analyzed via the computer program PEAK SIMPLE (SRI Instruments, Earl St. Torrance, USA). Results were expressed as mol/L. The sugar composition was expressed by a ratio as proposed by Baker and Baker [30]. The ratio (R) of sucrose (S) and the hexoses fructose (F) and glucose (G) was calculated as follows: R= (S+1)/(G+F+1). The value of 1 was added to the numerator and to the denominator of the quotient to avoid invalid cases (if S= 0) or outliers (if the denominator would be < 1). The pH was measured with tri-test indicator paper (Macherey and Nagel, Düren, Germany). For every plant species nectar properties were measured three times from independent samples.

2.8Rearing of bumblebees

Bumblebee queens of the species B. terrestris, B. pascuorum, B. lapidarius, B. pratorum and B. hortorum were captured in the New Botanical Garden during late March and early April. The queens were confined in the greenhouse for colony initiation [31,32]. To prevent the contamination of the colonies with foreign yeasts the bumblebees were not supplied with honey water, which contains numerous yeast and bacteria [33], but with a sterile sugar solution (10% fructose w/w, 10% glucose, 10% sucrose). The bees were fed with pollen from the field. After the colonies had reached the size ofabout 20 animals they were placed in the New Botanical Garden, where the bees found a broad spectrum of food plants. In this way, a natural yeast community could develop in the nests during summer.

2.9Microscopic studies

In June 2002, nectar samples from Digitalis purpurea were taken early in the morning before the bumblebees had visited the flowers. Yeast cells were counted under the microscope using a Neubauer particle counting chamber. In September 2002, the proboscis of B. terrestris and B. pascuorum individuals from the reared colonies were investigated under the microscope for the presence of yeast cells. Different parts of the proboscis were put in a drop of lacto cotton blue without phenol to stain the plasma of the yeast cells blue and get a better contrast with the brown hair of the glossa.

2.10Statistical analysis

Due to sample size constraints, only data for the year 2002 were statistically analyzed. The tests were performed with the SPSS 11 statistic package (SPSS Inc., Chicago). Distribution of variables was analyzed with the Kolmogorov–Smirnov test. Total molar sugar concentration, pH, and temperature deviated significantly from a normal distribution. Accordingly the concentration was transformed to its natural logarithm and the pH to its reciprocal. The distribution of the temperature could not be improved by transformation, and so the raw variable was analyzed. For the entire study period, mean sucrose/hexose ratio, mean sugar molarity, and mean pH of the plant species were analyzed by one-way analysis of variance (ANOVA) with plant families as the explanatory factor. For the period from May 01, 2002 to October 01, 2002 the number of colonies recovered from nectar samples was analyzed by one-way ANOVA, with the plant family as the explanatory factor. Last, the number of colonies recovered from nectar sample was analyzed by one-way ANOVAs with the presence of basidiomycetes and ascomycetes as explanatory factors. For the periods April 1, 2002 to May 1, 2002 and April 1, 2002 to October 1, 2002 the number of colonies recovered from nectar samples were analyzed with two-way ANOVAs using the mean temperature of the three days preceding the day of sampling as a cofactor and the plant families as a factor. When plant families served as explanatory factors, only those families were considered for which four or more plant species were included in the analysis.

3Results and discussion

3.1Colony counts, cell counts and isolation media

The number of colonies recovered from nectar samples were used to estimate the abundance of yeasts in the nectar. In D. purpurea, up to 16,000 yeast cells per microliter nectar were counted under the microscope. Such a high number of colonies has never been observed on isolation medium, indicating that the number of colonies recovered from a sample can only be used as a relative measure for the abundance of yeasts and not as a measure for the total number of yeast cells. For simplicity the term “abundance” is used as a synonym of “number of colonies per nectar sample” (sample volume 4 μl) throughout the discussion.

All yeast species isolated on YM agar in the second year appeared to grow well on Hautmans sugar agar used in the first year and vice versa. Therefore, the results of the isolations in the different years where different isolation media were used could be compared without limitation.

3.2Factors influencing the abundance and frequency of yeasts in nectar

Two hundred and twelve yeast strains (thus not counting the repeated isolation of the same yeast species from a sample) were isolated from 195 nectar samples in 2002. Seventy two percent of the samples contained yeasts. In the first half of April, 28% of the samples hosted yeasts and only 0–100 colonies were recovered per nectar sample. After the appearance of the first ascomycetes in the nectar the average number of colonies per sample increased within 3–4 weeks to an average level of 100–1000 colonies at the beginning of May and remained at this level until the end of September (Fig. 1). During this period 88% of the samples contained yeasts. This is in good agreement with the study of Jimbo [34], who found yeasts in 30% of the flowers in spring and in 68% of the flowers in summer. Similar values were reported by Hilkenbach [35], who detected yeasts in 62% of the flowers collected in summer and Lund [13], who isolated yeasts from 68% and 78% of the flowers in July and August, respectively. Sandhu and Wairaich [11] demonstrated the presence of yeasts in the nectar of a similar portion of flowers (68%). In agreement with several other researchers (e.g., [1,2,10,17,36–38]), it can be conclude that yeasts occur in the nectar regularly, and that they are typical members of the microbial community associated with flowers.

Figure 1.

Mean number of colonies recovered from nectar samples over the whole growing season in 2002. N, number of samples. Bars indicate the standard error of mean values.

3.3Nectar properties

For the time of the maximum abundance and frequency of yeasts in nectaries, from May 1, 2002 to October 1, 2002, the one-way ANOVAs (Table 4) demonstrated that the mean number of colonies recovered from nectar samples of various plant families differed significantly. Also, the plant families differed significantly in mean nectar pH, nectar sugar molarity, and sucrose/hexose ratio of the nectar sugars (Tables 1 and 4). This is in agreement with the findings of other authors who reported the sucrose/hexose ratio [39,40], the total sugar concentration and the pH [41] to be characteristic of each plant family. However, sugar concentration may vary considerably within a species. For example, in H. foetidus the average total sugar concentration in nectar varied from 34% (w/w) to 57% (w/w) (mean 49%, SD ±6.6). Many factors influence nectar concentration in a single plant including relative humidity and how long the nectar is exposed to air in the flowers [42], the water supply of the plant [43], the temperature [44], the age of the flowers [45] and the time of day [46]. Moreover, nectar properties also depend on the plant family. These factors cannot be treated as independent explanatory variables. Based on the data of this study it could not be determine whether nectar properties or some other characteristics of the various plant families have greater influence on the abundance of yeasts in the nectar. Many factors are known to affect the behavior of flower-visiting insects. Flower morphology, sugar production of flowers and sugar concentration in the nectar all affect the frequency of flower visits. The more often a flower is visited, the greater is the probability that the nectar will be colonized by yeasts immediately after the corolla has opened. The sooner the nectar of a certain flower type is inoculated the higher the probability of recovering a high number of yeast cells from it.

Table 4.  One-way ANOVAs for the nectar properties in different plant families; the number of colonies recovered from nectar samples in the presence or absence of ascomycetes and basidiomycetes; the number of colonies recovered from nectar samples in different plant families
  1. NS =p>0.05.

  2. **p < 0.01.

  3. ***p < 0.001.

Sucrose/hexose ratio
 Plant families142.9960.2142.589**
Total sugar molarity
 Plant families1412.5630.8974.015***
 Plant families12171.14514.26217.860***
Number of colonies
 Presence of ascomycetes1278.5278.5320.4***
Number of colonies
 Presence of basidiomycetes10.1470.150.063 NS
Number of colonies
 Plant families833.5034.1882.888**

Even though it could not be distinguish among the respective influences of nectar properties and other plant characteristics on the abundance of yeasts in nectar, conclusions can be drawn from the results of the physiological tests and the nectar analyses. The mean total sugar molarity for all plant species was 0.89 M (Table 1). Only in some species of Fabaceae did the mean sugar molarity exceeded 2 M. The two dominant nectar yeasts, M. reukaufii and M. gruessi (see below), grow well in the presence of 2 M glucose (Table 3). Therefore, the growth of those yeasts will be reduced only in a very few plant taxa under dry and hot conditions that lead to extreme nectar sugar concentrations.

Table 3.  Growth response of different yeast species to various glucose concentrations
 N2%50%60%70%2 M3 M
  1. N, number of strains that were included in the tests; +, strong growth; s, slow; w, weak; −, negative.

C. bombi10++s++
D. hansenii2++s+s/w
D. maramus5++s+s
Z. rouxi5w++w++
M. gruessii5++w+s/w
M. reukaufii7++s++
C. rancensis14+ss/w
M. kunwiensis6++s+s

Only few plant species exhibited extreme pH values in their nectar. A minimum pH below 4 was found in two and a maximum pH of 8 or higher was measured in four families. The average pH was 6.1 (Table 1). Most yeast species grow well in the entire pH range [19]. Since the two main nectar yeasts grow well on sucrose, fructose, and glucose, it is not likely that the sucrose/hexose ratio has any direct impact on the abundance of yeasts in nectar.

We can thus conclude that chemical and physical nectar properties most likely do not have a significant impact on the abundance of nectar yeasts. Differences in the abundance of yeasts in nectar are more likely to be affected by the attractiveness of flowers to various insects. Differences in community composition were not observed. The theories of Reukauf [10] and Sandhu and Waraich [11], claiming that different nectar properties lead to different yeast communities, can therefore be rejected, at least in the cases of this study.


In contrast to nectar properties, the temperature had a significant impact on the abundance of yeasts in nectar. The two-way ANOVAs (Table 5) revealed such an influence for the time period between April 1 and May 1, 2002, whereas later, during the same year, the abundance of yeasts was influenced solely by plant family. In the first 3–4 weeks of the investigation, temperature may have influenced yeast frequency and abundance in two ways: First, the yeasts depend on bumblebees for their dispersal and bumblebees are more active at higher temperatures, visiting flowers more frequently. In addition, the bumblebees become more numerous as more queens leave their hibernation places. Second, yeasts themselves need higher temperatures for a higher growth rate, which results in higher numbers of different yeast populations.

Table 5.  Two-way ANOVAs for the number of colonies recovered from nectar samples in the various plant families and under changing temperatures
  1. ***p < 0.001, NS=p>0.05.

  2. **p < 0.01.

Constant term16.326.323.56 NS
Plant families620.213.371.9 NS
Constant term117.9417.9412.24**
Temperature10.190.190.13 NS
Plant families833.14.132.82**

3.5Distribution of different yeast taxa

The sequences of both strains (if available) of all yeast species that were identified using the analysis of D1/D2 domain of the 26S rRNA gene were deposited in EMBL. The respective strain cultures were deposited at the Centraalbureau voor Schimmelcultures in Utrecht, The Netherlands. Accession numbers and strain numbers are given in Table 6.

Table 6.  Accession numbers of the D1/D2 sequences of the 26S rRNA genes for strains that were deposited at the Centraalbureau voor Schimmelcultures in Utrecht, The Netherlands
SpeciesStrain numberEMBL Accession No.
M. koreensisCBS 9443AJ716120
 CBS 9444AJ716121
M. kunwiensisCBS 9679AJ716106
 CBS 9680AJ716107
M. gruessiCBS 9697AJ716111
 CBS 9698AJ716112
M. reukaufiiCBS 9706AJ717114
 CBS 9707AJ716113
C. rancensisCBS 9436AJ716123
 CBS 9438AJ716122
C. cf. fragiCBS 9669AJ716104
C. friedrichiiCBS 9670AJ716105
S. bombicolaCBS 9710AJ716115
 CBS 9711AJ716117
C. bombiCBS 9659AJ699426
 CBS 9663AJ699427
D. hanseniiCBS 9683AJ716108
 CBS 9685AJ716109
D. maramusCBS 9692AJ716110
 CBS 9695AJ716116
Z. rouxiiCBS 9714AJ716118
 CBS 9719AJ716119
C. bombiphilaCBS 9712AJ620185
 CBS 9713AJ620186


The most frequent basidiomycetes in nectar were Cryptococcus victoriae, Cystofilobasidium capitatum, Cryptococcus albidus and Cryptococcus spec. Basiodmycetous yeasts were isolated during the entire period of the investigation (Table 2). They were detected in the same low frequency in flowers that were protected from insects as in non-protected flowers. As basidiomycetes were much more frequent on plant surfaces (Table 2) it became apparent that they belong to phyloplane and not to nectar. The association of basidiomycetous yeasts with the phyloplane has been reported before [47]. In the present investigation, basidiomycetous yeasts were detected on the inside of the calyx leaves of H. foetidus. In view of the short distance between some plant surfaces and the nectaries it is evident that basidiomycetes can occasionally be isolated from nectar.

The results of the one-way ANOVAs for the abundance of yeasts in nectar with the presence of ascomycetes or basidiomycetes as explanatory factors (Table 4) can be interpreted in a similar manner as the data about frequency of basidiomycetous yeasts in nectar. In this analysis, only the presence of the ascomycetes significantly affected the abundance of yeasts, suggesting that only ascomycetous yeast cells multiplied well in nectar, whereas basidiomycetous yeast cells were present but not able to reproduce well. This finding corresponds with the results of the physiological tests: None of the basidiomycetes grew on 50% glucose. This value is exceeded quite often in many plant species [41]. Sugar concentrations between 30% and 50% considerably lower the growth rate of most basidiomycetous yeasts. Only those yeasts that are able to grow fast in a high osmotic environment are able to reach new flowers before their host flower wilts. In agreement with Lachance et al. [38], basidiomycetes are evidently autochthonous members of the yeast communities in flowers.


3.7.1Metschnikowia reukaufii and Metschnikowia gruessii

From April 1 to October 1, 2002 the species with the highest frequencies, M. reukaufii and M. gruessii, were present in 48% and 30% of the nectar samples, respectively. At the time of the maximum frequency and abundance of yeast in nectar, from May 1 to October 1, 2002, M. reukaufii was present in 64% and M. gruessii in 43% of the samples. It is difficult to compare the current results with the literature before 1992 when M. gruessii was described [16]. Prior to that time, M. reukaufii and M. gruessii were assumed to be the same species. The descriptions, drawings, and micrographs of the characteristic airplane cell configurations [16] in earlier studies strongly suggest that M. gruessii was isolated more frequently. The species was detected, but not necessarily identified correctly in flowers in Britain [48], Portugal [16], France [1], Switzerland [17], all over Germany [2,7,10,14,26,35,36,49,50], the Czech Republic [8,37,51], California (USA) [52] and in Ontario (Canada) [53]. For several decades, the cross forms of M. gruessii were thought to be a characteristic of M. reukaufii[54]. Therefore, it remains unclear, which of the two species was actually found in the various studies [13,52,55–59]. After the description of M. gruessii, M. reukaufii was detected in Japan, Canada, the state of Washington [60] and Virginia (USA) [61], Portugal [16], Russia (Yurkov, personal communication) and Germany [26]. Both species may be present in various habitats, but the repeated isolation of the yeasts by various authors over a period of almost 100 years in various locations in the northern hemisphere strongly suggests that flowers are their main habitat.

In spite of the assumption of other authors [2,8,9] that M. gruessii over-winters inside bumblebees or on their proboscis, neither M. gruessii nor M. reukaufii were present in or on the bumblebee queens that were captured directly after they had left their hibernation localities in 2003 and on the body surface of queens examined in early spring 2001 (Table 2). M. gruessii was detected before in the soil under flowering plants [8], but it was shown not to be an autochthonous member of the yeast community in soil [13,57]. Therefore, the hibernation of the species in the soil, as proposed for H. uvarum[3,4], is most unlikely. In the current study, M. gruessii and M. reukaufii were absent on plant materials other than the nectar. M. reukaufii was present in nectar only two weeks after the appearance of the first bumblebee queens in spring 2001 and it was only detected on the proboscis of the animals two days after being isolated from the nectar for the first time. It is clear that the bumblebees did not inoculate the initial cells into the nectar. These yeasts were not found on other insects, such as nitidulid beetles Meligethes aenneus or Forficula auricularia, collected from the flowers in early spring. The question of where they survive through the winter remains open. In agreement with Babjeva and Goring [9], it can be assumed that the over-wintering sites are other insects.

Although the over-wintering sites of M. reukaufii and M. gruessii are unknown, it is clear that only a few cells of those species in the nectar are enough to infect the honey pots to which they are brought by foraging queens. Later, workers receive the yeasts from the honey pots and distribute them to a wide spectrum of nectar-producing plant species. Both yeast species depend on bumblebees or other flower-visiting insects for their dispersal during the summer. The yeasts were absent in nectar of protected flowers or flowers that had opened in the climate chamber. Similar observations were made by Schoellhorn [17], who found M. gruessii only in flowers that were visited by insects. Like other authors [35,36,56], he could not find any yeasts in closed flower buds. M. gruessii and M. reukaufii were frequently isolated from the proboscis of all bumblebee species in 2002 (Table 2). Typical airplane cell configurations (Figs. 2(a) and (b)) of M. gruessii were observed under the microscope. They were attached to the glossa like an anchor (Figs. 2(c) and (d)). Grueß[2] compared the number of airplane cell configurations with single cells of other shapes and found those cell configurations stick more easily and in greater numbers to the fine hairs of the glossa than other yeast cells. This may be the reason why M. gruessii is more successful in plant–bumblebee interaction than other yeasts. M. reukaufii and M. gruessii are presumably highly specialized yeasts that have a narrow realized niche in the interaction of nectar-producing-plants and flower-visiting insects.

Figure 2.

Typical cross-form cell configurations of M. gruessii. (a, b) Cells in the nectar of Epilobium angustifolium; (c, d) Cells attached to the hair of the glossa of B. pascuorum (arrow). Bars indicate 20 μm.

3.7.2Candida rancensis

Like the latter species, C. rancensis is another yeast that was present only in the nectar, on the proboscis and in the digestive tract of bumblebees (Table 2) in the current study. The species was described in 1984 by Ramirez and Gonzales [62] and later suspected to be synonymous with M. reukaufii[60]. Only in a very recent study [63]C. rancensis was shown to be a separate species. Most likely, this species was isolated in other investigations and identified as M. reukaufii. In the current study, C. rancensis was present in nectar from mid-July to the end of September. It was detected only in the nectar of Epilobium angustifolium, Epilobium hirsutum (Onagraceae) and from various Lamiaceae. The narrow spectrum of plant species and the short period in which C. rancensis was present may be due to its vector. Probably, the yeast was carried into the nectars by an insect that becomes active in July and that is flower-constant. The reason why this yeast was not distributed by other insects to other nectar-producing plants may be that its growth was slowed down by the sugar contents of the nectar. C. rancensis was the only ascomycetous species in the study that exhibited delayed growth on 50% glucose (Table 3).

3.7.3Metschnikowia pulcherrima

In contrast to M. gruessii and M. reukaufii, M. pulcherrima was present only in very few nectar samples. The presence of M. pulcherrima in nectar has been recorded before [13,55,58]. In the current investigation, the species was most common in the digestive tract of the bumblebees and on the nitidulid beetle M. aenneus (Table 2). M. pulcherrima is known to have a wider distribution in nature than M. gruessii and M. reukaufii. It has been found on various animals, such as: A. mellifera[13,55] wild bees [55], Helix spec. [57] and Drosophila spec. [64]. It was also isolated from a wide spectrum of fruits [13,50,65–69], wounded root crops [13], exudates of trees [13,66], and different parts of flowers outside the nectaries [15]. Hence M. pulcherrima is present in the plant–bumblebee mutualism accidentally and this is not its main habitat.

3.7.4Debaryomyces hansenii

Debaryomyces hansenii was rare in all habitats (Table 2). Other authors [9,13,50,70] detected the species in nectaries, but it was isolated also from a wide spectrum of habitats, like cacti [71,72], agave rots [73], fruits [13,74], wounded root crops [13], honey [58,75] and Apoidae [13,58]. Because the species is widespread and abundant in habitats with high sugar concentrations, it is likely one of the most abundant species at least in the honey pots. It remains unclear why it was so rare in this study.

3.7.5Metschnikowia kunwiensis

Metschnikowia kunwiensis was present in the digestive tract of worker bumblebees and bumblebee queens in all seasons. It was also present on the body surface of the queens after hibernation. Most likely, it attaches itself to the body surface with the faeces in winter. M. kunwiensis was not detected in the honey, nectar or the honey bulb. In contrast to the current study, Hong et al. [76] isolated one strain from a flower. Most likely, the species happened to be in the flower by chance. The results of this study indicate that this yeast species is restricted to the digestive tract of bumblebees and that it permanently depends on the animals as a habitat.

3.7.6Candida bombi

Like M. kunwiensis, C. bombi over-wintered in the digestive tract of the bumblebee queens and was detected on the queen's body surface in spring (Table 2). In the summer, it was the most abundant yeast in the honey pots to which it is brought by the queen in spring. Most likely, the young queens become infected from the honey while they still work in the nests. The species has been shown earlier to be associated with bumblebees [77].

3.7.7Zygosaccharomyces rouxi and Debaryomyces maramus

Beside M. reukaufii and C. bombi, Z. rouxi and D. maramus were the most frequent yeasts in the honey provisions in nests (Table 2). Since honey is more concentrated than nectar, osmotic pressure may be the most important environmental factor in this community. All species mentioned above were osmotolerant. Z. rouxi exhibited very weak growth even on 70% glucose. Growth with 60% glucose was stronger than with 2% glucose, suggesting that Z. rouxi is not only osmotolerant but also osmophilic. Its presence in the honey of A. mellifera has been reported earlier [78,79] and it was isolated from the nests of different wild bees [55]. Batra et al. found the species in floral nectar and in the honey stomach of various bees. Z. rouxi has been isolated from soil, and it has been suggested that yeasts in soil may be a source for the infection of A. mellifera honey [80]. Since most bumblebee species build their nests in the ground, this may apply even more to bumblebee honey. Similarly, D. maramus has been isolated from a broader spectrum of substrates [81]. Spencer [82] investigated the yeast community in Canadian bumblebee honey. He reported Z. rouxii and M. reukaufii and three other species to be the most abundant species. Z. rouxi and D. maramus are typical members of the yeast communities in bumblebee honey, but they are not restricted to it. C. friedrichii, Starmarella bombicola, M. koreensis and C. bombiphila were only isolated ones or twice. Therefore, no conclusions could be drawn about their ecology.


Some yeast taxa, including all basidiomycetes, show a ubiquitous distribution. Other species such as Z. rouxi or M. pulcherrima exhibit a more limited distribution, but they are not restricted to the plant–bumblebee mutualism. M. kunwiensis and C. bombi are highly specialized. They depend on the bumblebees as a host for hibernation and live in close association with the insects the year round. M. reukaufii and M. gruessii reproduce in the flowers very well. They are closely associated with plant–bumblebee mutualism as well, but it remains unclear where they survive through the winter. Nevertheless, for those yeasts and even for C. rancensis, which was present in the nectar for only 2 or 3 months, the flowers may be of great importance. Flowers are traffic junctions for all kinds of flower-visiting insects. Yeasts may use flowers for mass reproduction and as a platform from which they will meet their next host insect for hibernation. The various nectar properties in different plant species appeared to have a limited impact on the abundance of yeasts in nectar. Hence, the attractiveness of particular flowers to nectar foraging insects is of greater importance.


This work was funded by the Deutsche Bundesstiftung Umwelt. Appreciation goes to Marc-André Lachance for his help in physiological tests and for numerous discussion, to R. Fischer for the materials for molecular work; A. Titze for the initial idea for the study and helpful discussions; V. Roberts from CBS for the type strains; A. Yurkov for help with the articles in Russian; V. Brysch for help with articles in French; A. Esposito for help with articles in Italian; B. Nüsslein for his assistance with HPLC; the working group of Peter Frenzel for access to HPLC and the Microscope; R. Peilstöcker for help with the rearing of the bumblebee colonies; H. Berg and D. Matthies for statistical analysis and B. LaMar for critical reading of the manuscript.