Saccharomyces cerevisiae and its close congener S. paradoxus are typically indistinguishable by the phenotypic criteria of classical yeast taxonomy, but they are evolutionarily distinct as indicated by hybrid spore inviability and genomic sequence divergence. Previous work has shown that these two species coexist in oak-associated microhabitats at natural woodland sites in North America. Here, we show that sympatric populations of S. cerevisiae and S. paradoxus from a single natural site are phenotypically differentiated in their growth rate responses to temperature. Our main finding is that the S. cerevisiae population exhibits a markedly higher growth rate at 37°C than the S. paradoxus population; we also find possible differences in growth rate between these populations at two lower temperatures. We discuss the implications of our results for the coexistence of these yeasts in natural environments, and we suggest that thermal growth response may be an evolutionarily labile feature of these organisms that could be analyzed using genomic approaches.
Saccharomyces cerevisiae has had a long association with humans as a fermenting agent and continues to be one of the most important industrial microorganisms . S. cerevisiae is also a prominent laboratory research organism; after many decades of work, biologists can access a wealth of S. cerevisiae cell biology, genetic information, and genomic resources, including the first fully sequenced and annotated eukaryote genome . Our understanding of the ecology and evolution of S. cerevisiae in nature remains very limited, but a growing number of studies demonstrate that the species is common in habitats that are undisturbed by human activity [3–6] and in vineyards [7–10]. These studies suggest that the ecology and evolution of this important organism can be investigated in natural populations (but see [11,12]).
Recent studies have shown that S. cerevisiae and its congener S. paradoxus occupy the same microhabitat type (oak exudates, oak bark, and oak-associated soils) in widely separated woodland sites in eastern North America [3,5]. S. cerevisiae and S. paradoxus are reproductively isolated by low hybrid spore viability  and exhibit substantial genomic sequence divergence [14,15]. However, these species are typically indistinguishable by the criteria of classical yeast taxonomy, which include phenotypic characters such as cell, spore, and ascus morphology, as well as features of obvious ecological importance such as profiles of assimilation and fermentation of organic compounds . This close phenotypic similarity raises the question of whether sympatric populations of these two species nonetheless differ in some ecologically significant way, as might be predicted by classical ecological theory [16,17]. Some previous reports [11,18] have suggested that S. cerevisiae and S. paradoxus have different thermal growth profiles, but those reports were based on allopatric isolates. Here, we compare the profiles of growth rate vs. temperature in S. cerevisiae and S. paradoxus populations sampled from a single natural site. Our results indicate that these two species do indeed have different thermal growth profiles in sympatry; however, the growth rate profiles for our populations differ somewhat from those suggested by earlier studies. We discuss the possible implications of our results for the coexistence of S. cerevisiae and S. paradoxus in nature, and we speculate that variability in thermal growth profiles might be observable among natural populations of these yeasts isolated from divergent thermal environments.
Materials and methods
Nine diploid homothallic isolates of both S. cerevisiae and S. paradoxus were studied. The S. cerevisiae isolates were strains YPS128, YPS129, YPS133, YPS134, YPS139, YPS141, YPS142, YPS143 and YPS154; the S. paradoxus isolates were strains YPS125, YPS126, YPS138, YPS145, YPS150, YPS151, YPS152, YPS155 and YPS158. Methods employed in the collection and identification of these isolates have been described previously . Briefly, all were collected during a 2-week interval in July 1999 at Tyler Arboretum, near Media, Pennsylvania in eastern North America. The arboretum includes a mature, second-growth, deciduous woodland of approximately 182 ha that is contiguous with 1052 ha of similar habitat at Ridley Creek State Park. Samples of exudate, bark, and soil taken from on or around multiple red and black oaks (Quercus spp.) were subjected to enrichment culturing in the laboratory. S. cerevisiae and S. paradoxus isolates were identified by test matings in the laboratory. All isolates are currently stored in the corresponding author's laboratory at the University of Pennsylvania.
Growth rate measurements
We estimated the growth rate of each isolate in laboratory culture in yeast extract–peptone–dextrose (YPD) broth  at five temperatures chosen to span most of the growth-permissive range beneath the forest canopy at our collection site: 10, 16, 23, 30, and 37°C. Prior to each growth rate assay, isolates were conditioned by inoculating them from freezer storage into 5 ml of YPD broth in loosely capped 50-ml Erlenmeyer flasks and growing them to stationary phase at 30°C with aeration. For the growth rate assays, three replicate cultures of each isolate for each temperature were started by inoculating 100-μl aliquots from the stationary phase cultures into tightly capped 15-ml screwcap glass tubes containing 5 ml of YPD. The tubes were positioned in a rack fastened to a rocking platform set at 100 rpm (VWR Scientific Model 100) and placed on the center shelf of a temperature-controlled incubator (Percival VL36, Boone, IA, USA). Culture turbidity during growth to stationary phase was measured as absorbance at 600 nm on a Spectronic 20+ Series unit (Spectronic, Rochester, NY, USA). The spectrophotometer was frequently recalibrated to zero absorbance during sets of readings using a control tube containing only YPD. Tubes were removed for spectrophotometric measurements of culture turbidity at hourly intervals for the 23, 30, and 37°C temperature levels; because of slow growth, longer time intervals were used between measurements for the 10 and 16°C levels. Cultures were vortexed thoroughly to suspend the cells before each measurement.
The response variable used in the analysis of growth rates at each temperature was the maximum change in absorbance observed between successive measurements during the growth of each culture to stationary phase, which is a valid proxy for growth rate because the inflection point of the growth curve is in all likelihood contained within that time interval. The data were analyzed as a two-way, mixed-model ANOVA in which nine isolates were nested within each of the two species across the five temperature levels. The three-fold replication of each isolate/temperature combination yielded 270 experiment-wide observations. Species, temperature, and their interaction were analyzed as fixed effects; isolate and its interaction with temperature were analyzed as random effects. All analyses were carried out using PROC GLM in SAS (SAS Institute, Cary, NC, USA).
Fig. 1 illustrates profiles of average growth rate vs. temperature observed in both species. These growth rate profiles were significantly different across the whole experiment (F1,16=15.43; P=0.0012), and the interaction between species and temperature was highly significant (F4,64=14.02; P<0.0001). Overall, no significant effect of isolates within species was detected (F16,64=0.88; P=0.596).
Growth rates of the two species appeared to differ at three temperatures: 16, 30, and 37°C. Before correction for an experiment-wide significance level of 0.05, S. cerevisiae exhibited a higher growth rate at 16°C (non-orthogonal unplanned comparison: F1,64=5.69; P=0.0201), S. paradoxus exhibited a higher growth rate at 30°C (F1,64=4.55; P=0.037), and S. cerevisiae exhibited a higher growth rate at 37°C (F1,64=59.28; P<0.0001). When the sequential Bonferroni correction  was applied, the differences observed at 16 and 30°C became marginally non-significant, whereas the difference at 37°C remained highly significant.
Unplanned comparisons indicated that all differences in growth rates between temperatures within each species were significant except the difference between S. paradoxus at 30°C and S. paradoxus at 37°C.
We have shown that S. cerevisiae and S. paradoxus populations sampled in sympatry from a Pennsylvania woodland have different thermal growth rate profiles. To our knowledge, our study is the first to compare ecologically important phenotypic characters in sympatric natural Saccharomyces populations, although previous work has considered thermal growth profile and other phenotypic differences between S. cerevisiae and S. bayanus strains isolated during wine making [21–23]. The possible influence of growth temperature relations in determining yeast species distribution has also recently been addressed by Lachance et al. , who conclude that maximum growth temperature may be a critical property of the fundamental niche of the Metschnikowia and Candida species associated with morning glory flowers in Hawai’i.
As mentioned previously, S. cerevisiae and S. paradoxus are typically indistinguishable by the standard phenotypic criteria of yeast taxonomy, which include fermentation and assimilation profiles for compounds likely to be available in natural habitats. It was therefore of interest to ask whether sympatric isolates of these species would nonetheless exhibit some consistent phenotypic difference, as might be predicted from the classical ecological theory that niche differentiation mediates species coexistence [16,17]. We focused upon the growth rate response to temperature because previous studies had suggested that S. cerevisiae and S. paradoxus have different optimal growth temperatures ; although the phenotypes of our study populations may also differ in other respects, we did not investigate this possibility.
Whether the thermal growth profile difference that we have observed between S. cerevisiae and S. paradoxus affects co-occurrence of these species at our study site is uncertain. Although S. cerevisiae grows markedly faster than S. paradoxus at the top end of the thermal range investigated (37°C), sustained temperatures as high as 37°C are probably uncommon at our study site. Our data also hint at differences in growth rate at lower temperatures, with S. cerevisiae growing faster at 16°C and S. paradoxus growing faster at 30°C. Assuming for the moment that these differences at lower temperatures are real, a speculative ecological hypothesis might be that growth rate advantage alternates across the fluctuating daily or seasonal temperature regime in our natural site in a manner that does not consistently favor one or the other species.
In general, whether niche differentiation is necessary for species coexistence has stimulated considerable debate among ecologists [25–31]. Certainly, several conditions must be met before niche differentiation needs to be invoked to explain the co-occurrence of closely similar species. One such condition is that the populations in question be stably coexisting in sympatry rather than merely coincident. A number of previous studies have suggested that oak exudates, bark, and associated soils are indeed a natural habitat or repository for S. paradoxus in temperate regions worldwide [3,5,13,32–34], and thus there is good reason to believe that S. paradoxus is a long-term resident of our study site. However, it is conceivable that S. cerevisiae has had only a short history at our study site; the woodland habitat might represent a sink for S. cerevisiae isolates of human origin, or S. cerevisiae may have recently colonized this habitat or site. Although further study is needed to address this point, two lines of evidence support long-term residence of S. cerevisiae in the woodland habitat: (1) the presence of S. cerevisiae isolates in sympatry with S. paradoxus at numerous woodland sites in eastern North America (H. Kuehne, unpublished data); and (2) the fact that an S. cerevisiae isolate from the population studied in this paper shows significantly higher freeze tolerance and lower copper tolerance than laboratory and vineyard isolates . (The latter findings are consistent with adaptation to the woodland habitat because freezing temperatures are common at our natural sites, and copper tolerance, which is probably an adaptation to the use of copper sulfate as a fungicide in vineyards, may well be costly to fitness under natural conditions.) A second condition that must be met before niche differentiation can be invoked to explain coexistence is that the species in question co-occur at the microgeographic scale required for direct interaction; to date, we have no data that would address this possibility for S. cerevisiae and S. paradoxus at our study site. Finally, the density of each species must be sufficient to affect growth of the other. It is possible that extrinsic disturbances such as seasonal mortality limit densities sufficiently to preclude competitive interaction between S. cerevisiae and S. paradoxus; again, data are lacking on this point.
In studies of animal and plant populations, the evolution of niche differentiation as a result of competition between closely related species may be supported by the observation that specific characters of ecological importance differ in sympatry but not in allopatry (e.g. [36,37]). Clearly, our study does not meet this criterion: we have only compared S. cerevisiae and S. paradoxus in sympatry. Obtaining sufficient data to conclude that a given site ever harbors only one or the other of these species may be very difficult, and this limits the prospect of testing the ecological importance of the observed thermal growth profile difference by comparing allopatric populations. On the other hand, because S. cerevisiae and S. paradoxus are readily culturable, it may be possible to examine the ecological implications of thermal growth profile differentiation experimentally in the laboratory; in principle, such an approach could test the hypothesis of fluctuating growth advantage mentioned above.
Whatever the precise ecological significance of the thermal growth profile difference between sympatric S. cerevisiae and S. paradoxus populations, our results have implications for future ecological and evolutionary work in Saccharomyces. Previous work has suggested that S. paradoxus exhibits its highest growth rate at 37°C or higher and that S. cerevisiae may exhibit its highest growth rate at lower temperatures . We find roughly the opposite pattern: the S. paradoxus isolates from our natural population apparently reach a growth rate plateau at around 30°C, whereas the S. cerevisiae isolates exhibit their highest growth rate at 37°C and may grow even faster at temperatures greater than 37°C. Our results thus suggest that thermal growth rate properties in Saccharomyces are evolutionarily labile and might vary with the prevailing environmental temperature regime (see also [38,39]). There is considerable interest in genomic characterization of ecologically and evolutionarily important variation in S. cerevisiae[35,40,41], and similar approaches are now possible in S. paradoxus with the recent publication of its genome sequence . In the future it may be possible to study evolved differences in thermal growth properties within and between S. cerevisiae and S. paradoxus using genomic approaches.
We thank Dr. Paul Schmidt for the use of his incubator. We are grateful to two anonymous reviewers and to E. Fingerman, C.A. Francis and H.A. Murphy for comments on the manuscript. This research was supported by a grant from the University of Pennsylvania Research Foundation to P.D.S.