Species Richness of Yeast Communities in Floral Nectar of Southern Spanish Plants
|
|  Yüklə 0.72 Mb.
|
a Cryptococcus spp. includes Cryptococcus aerius, Cryptococcus diffluens, Cryptococcus uzbekistanensis, and Cryptococcus victoriae b Rhodotorula spp. includes Rhodotorula colostri and Rhodotorula mucilaginosa 
species were recovered from each nectar drop. The two dominant species, M. reukaufii and M. gruessii, were recorded in 73.4% and 29.7% of nectar drops, respectively (Fig. 1). The rest of species were much less frequent, and included A. pullulans (found in 7% of nectar samples), Rhodotorula colostri (3.9%), Rhodotorula mucilaginosa (1.6%), and Sporobolomyces roseus (1.6%). Species that were recorded only once (<1% of nectar samples) included Cryptococcus aerius, Cryptococcus diffluens, Cryptococcus victoriae, Cryptococcus uzbekistanensis, L. hoffmannii, and the unidentified Metschnikowia sp. DOTUR analysis of the DNA sequence data for these isolates provided a slightly higher estimate of total species richness, as a total of 18 OTUs were identified at the 3% DNA dissimilarity cut-off. The possibility that additional undescribed species were present in our samples cannot therefore be ruled out. When plant species rather than nectar drops were considered as the samples for analyses, the pattern of yeast community species richness was similar: M. reukaufii and M. gruessii were isolated from 20 and ten, respectively, of the 24 plant species surveyed. The rest of species occurred in a much reduced subset of plant species. A. pullulans and S. roseus were found in four and two plant species, respectively, while the other species occurred in a single plant species each. The most species-rich yeast community occurred in the nectar of Digitalis obscura (six species, Table 2), followed by Erinacea anthyllis and Atropa baetica, with four yeast species each. Antirrhinum australe and Lonicera implexa harbored three species each, and two species were recovered from the nectar of Gladiolus illyricus, Prunella grandiflora, Anthyllis vulneraria, and Marrubium supinum. Only one yeast species occurred in the nectar of each of the remaining species (Table 1).
Figure 1 Yeast and yeast-like growing species frequency in nectar samples: drop-based (upper panel) and plant species-based (lower panel) approaches
Species relative frequency in nectar drops   Rhodotorula mucilaginosa Rhodotorula colostri Metschnikowia sp. Lecythophora hoffmanii Cryptococcus victoriae Cryptococcus uzbekistanensis Cryptococcus diffluens Cryptococcus aerius Sporobolomyces roseus Aureobasidium pullulans Metschnikowia gruessii
Metschnikowia reukaufii 0 20 40 60 80 100 Species relative frequency by plant species   
        examined (Fig. 2a). This finding reveals that, although additional rare species are expected to arise by further increasing the total sampling effort on the set of 24 plant species surveyed, results of this survey can be considered as providing a reliable basis for estimating overall yeast species richness in the floral nectar of this set of host species. Species richness estimates obtained from all nectar samples combined were 25.7 (ICE estimator) and 21 species (Chao2 estimator; Fig. 2b), which denotes that our sampling recovered around
50% of the total number of species occurring in the nectar of sampled plant species in the study area.       Rarefaction analyses were also conducted using plant species rather than individual nectar samples as the units for the analyses. Using this approach, the yeast species accumulation curve is not linear and does not reach a distinct plateau for the N = 24 plant species sampled (Fig. 2c). The two nonparametric richness estimators, ICE and Chao2, gradually drifted apart from the Mao Tau function (observed diversity) with increasing number of plant species (Fig. 2d). The steadily increasing curves shown by these two estimators denote that our sampling of the plant community was able to detect only one third of the total estimated yeast species richness occurring in floral nectar in the regional plant community as a whole.
Yeast Community Composition and Osmotolerance       Variation among yeast species in their frequency of occurrence in nectar drops or plant species (Fig. 1) was related to interspecific differences in physiological traits, with osmophilic species occurring most frequently. The magnitude of the effect differed slightly depending on whether species frequencies were computed on the basis of their occurrence in nectar drops or plant species. When frequencies of occurrence were computed in respect to nectar drops, the set of osmophilic yeast species occurred, on average, in 27.7% of samples, while non-osmophilic yeasts occurred in 1.5% of samples, the difference being statistically significant (P= 0.048; Wilcoxon rank-sum test). Frequency of occurrence across plant species was also significantly greater for the osmophilic yeasts, which on average occurred in 35.5% of plant species, as compared to non-osmophilic ones which occurred in only 4.8% of the plant species surveyed (P= 0.024).
Yeast Diversity and Plant Pollinators   Table 2 summarizes data for the subset of 12 plant species with information simultaneously available on yeast incidence and pollinator diversity and composition. Interspecific varia- tion in yeast diversity and density was not significantly related to quantitative differences in pollinator composition (rs =0.12, P= 0.71 in both cases). Pollinator composition, as described
  40 a
30 40 c  30
 20 20
  10 10
0 0   40 b
30 40 d        30
     20 20
10 10 0 0 0 20 40 60 80 100 120 Number of nectar drops sampled 0 5 10 15 20 25 Number of plant species sampled Figure 2 Species richness vs. species density. a Individual-based rarefaction curve (black solid line, Mao Tau function) and 95% confidence intervals (long dash lines) for Cazorla 2008 spring nectar drops dataset. b Performance of nonparametric estimators of species richness for Cazorla 2008 spring nectar drops dataset: Singletons (dash-dotted line), Doubletons (medium dash), ICE (long dash), and Chao2 (dotted line), in comparison with rarefaction curve (Mao Tau function, solid line). c Individual-based rarefaction curve (black solid line, Mao Tau function) and 95% confidence intervals (long dash lines) for Cazorla 2008 spring dataset considering plants species surveyed as sampling units. d Performance of nonparametric estimators of species richness for Cazorla 2008 dataset by plant species surveyed: Singletons (dash-dotted line), Doubletons (medium dash), ICE (long dash), and Chao2 (dotted line), in comparison with rarefaction curve (Mao Tau function, solid line). Yeast species appearance was based on incidence data by scores on the first principal component, correlated weakly with yeast frequency, but the relationship did not reach statistical significance (rs = −0.448, P=0.14). No correlations were found between pollinator diversity, as computed with the Shannon index, and yeast variables (frequency, density, and diversity; rs =0.01–0.18). Discussion As habitat for yeast growth, floral nectar is characterized by high cell densities [12, 18] and low species richness [4, 17], in contrast to other microbial habitats such as aquatic environments that are characterized by high species rich- nesses [31] and low cell densities [35]. This paper describes a particularly species-poor nectar yeast community, largely composed of only two species (M. reukaufii and M. gruessii). The rest of species identified were rare in nectar samples, although some of them are common inhabitants in other environments like plant surfaces, soil, air, and water [3, 40]. Species accumulation curves showed that further rare species would appear if more nectar samples from the 24 plant species surveyed were sampled, or if additional plant species were sampled, but the main conclusion would most likely remain unchanged that nectar yeast communi- ties in our study region are characteristically species-poor. A number of mechanisms may contribute to the low yeast species richness found in nectar communities [see also 4]. Such factors may include the fact that we have used culture-dependent methods to obtain yeast isolates, irre- spective of how the unknown pool of yeast species growing in nectars tolerates the growth medium's conditions. Nevertheless, culture-dependent methods are widely used for the characterization of yeast communities for clinical or ecological purposes, and recent studies on yeast diversity in natural conditions have revealed a better performance of culture-dependent methods over purely molecular techni- ques like temperature gradient gel electrophoresis or restriction fragment length polymorphism [15, 19]. The existence of undescribed species could be another source of bias in our GenBank-querying species identification meth- od, but results of DOTUR-based analyses tend to rule out this possibility as a major biasing factor. A total of 18 OTUS were identified at the 3% dissimilarity cut-off used, which is not dramatically different from the 12 species identified using BLAST queries. The low yeast species
richness found in this study could also been explained by the fact that we focused exclusively on nectar samples rather than on flowers as a whole. Other surveys of flower- dwelling yeasts yielding longer species lists do not provide detailed information on which floral parts were sampled [27, 29]. In fact, a higher species richness is expected to occur at the level of whole flowers, since flowers harbor contrasting microhabitats for microbial growth such as pollen, perianth surfaces, and nectaries. Soil or plant surfaces could represent macroscales of sampling in relation to microbial body size [14], which might partly account for the high species richnesses revealed by surveys of these microbial habitats. The predominance of ascomycetous over basidiomyce- tous yeasts found in this study probably reflects that, as a microhabitat, floral nectar favors fermentative, osmotoler- ant, copiotrophic species. This interpretation is supported by our analyses on the relationship between species occurrence and ability to grow in 50% glucose, and also agrees with some earlier predictions [26]. When interpret- ing the results of these analyses, however, it must be kept in mind that individual yeast species might be internally heterogeneous in respect to their responses to osmotic stress, and that focusing at the multispecific level as done here misses such source of variation. Additional work on different yeast strains from the different species involved would be needed to explore the potential influence of that effect on our conclusions. Two groups of yeast species could be distinguished, namely those few present in nearly all nectar samples and the majority that occurred in only a few plant species. The first group includes two specialist species in the Metschni- kowia clade that have been frequently isolated from flowers, nectars, and pollinators [29]. The second group included generalist species that are found frequently in other microenvironments like leaf surfaces, soil, freshwater, and air. These include species found in decomposing organic matter, phylloplane, and soils (Aureobasidium) along with ubiquitous species of Cryptococcus and Rhodo- torula. Additional work is needed to dissect the different ecological origins of yeast species found in floral nectar, discerning which species originate from pollinators' glossae, pollinators body surfaces, corolla inner surface or pollen. Although rarefaction curves do not provide an estimation of asymptotic species richness [34, 46], they are often used to evaluate sampling adequacy by assessing whether the cumulative number of species reaches a plateau, and to compare observed species richness with figures obtained the values predicted by the ICE and Chao 2 estimators, thus corroborating that observed species accumulation curves will frequently underestimate actual species richness [10]. Nonparametric estimators incorporate information on the distribution of rare species in the dataset (i.e., those represented once, twice, or only a few times), so the fact that the majority of species recovered from our samples were rare account for the marked dissimilarity found here between observed and predicted species richness. Rarefaction curves did not reach a plateau, thus revealing that additional sampling effort would be needed to assess the true diversity of nectar yeasts in the study area. The total species diversity in a given area can be broken down into within-habitat or alpha-diversity, and between- habitat or beta-diversity [32]. Since alpha-diversity refers to a group of organisms interacting and competing for the same resources or sharing the same environment, results obtained from our drop-based analyses would roughly correspond to this diversity component. In contrast, plant species-based approach would fall conceptually closer to beta-diversity, allowing comparisons between different ecosystems or along environmental gradients [32]. Because of practical limitations, total sampling effort was restricted to 150 nectar drops spread among plant species. If nectar yeast communities were characterized by high beta- diversity, which would imply a high species turnover between plant species, then maximizing the spread of nectar drops across plant species would lead to an optimal sampling design. The use of individual plant species as units of analysis achieved lower accuracy at assessing true diversity at the plant community level than the use of pooled nectar drops, which is not surprising given that the number of samples involved in the nectar drops approach was considerably larger than in the plant species approach. Considering also that each plant species comprises a large array of elemental nectar “habitats,” using species as sampling units may provide an excessively coarse scale
with nonparametric estimators [10, 11, 47]. Nonparametric richness estimators, infrequently used in microbial diversity estimates [22, and references therein], may be used for inferring true species richness [49]. As shown in this study, observed species richness can be substantially lower than 0 5 10 15 20 25 Number of plant species sampled
for estimating species richness of nectar yeast communities at the habitat level unless a much larger number of nectar samples is collected from each plant species. Although a number of microbiological surveys of nectar yeasts have been carried out previously [e.g., 4, 13, 19, 23, 41], direct comparisons between our survey and previous work are problematical, since species identifications in the early literature on nectar yeasts was mostly based on traditional microbiological tests [e.g., 6, 41, 48]. The list of genera recorded in our study is closely similar to those reported for Asclepias syriaca nectar in North America and a broad survey of nectar yeasts in central European plants [4, 13]. Among recent work using molecular identification methods, the study of Herzberg [19] in Germany remains the most thorough investigation to date on nectar yeasts where the majority of yeast identifications was based on DNA sequences. Results reported in this study and those of Brysch-Herzberg for Germany are similar in that M. reukaufii was the commonest species at both sites (65% and 47% of nectar samples in southern Spain and Germany, respectively) followed by M. gruessii (23% and 17%). Species of Cryptococcus were isolated more frequently in Germany (18% of samples, compared to only 2% in southern Spain). The published data for Germany are amenable to rarefaction analysis, and a comparison of sample-based rarefaction curves for the two regions is shown in Fig. 3. The two species accumulation curves bear strong resemblance and, although similarly low at both regions, observed species richness was slightly higher in Germany than in southern Spain. Although further studies are obviously needed, the provisional conclusion can be drawn from these two studies [see also 6] that European nectar yeast communities are characterized by generally low species richness and a marked numerical dominance by very few species, represented here by M. reukaufii and M. gruessii. The ecological mechanisms accounting for these patterns remain to be elucidated, but factors related to variable, species-specific colonization opportunities, inter- specific antagonistic relationships, and growth limitations imposed by osmotic stress, low nitrogen content, and presence of secondary compounds in the nectar [17] could all play some role in determining the composition and diversity of nectar yeast communities.
supported by a predoctoral grant from the Spanish Ministerio de Educación y Ciencia. M. I. P is grateful to A. P. López, C. Rosell, and M. Alonso for field help and support and would like to dedicate this article to the memory of Miguel B., for the papers he will never be able to write. |
