Grantee Research Project Results
Final Report: Demographic and genetic factors affecting population viability of Lupinus perennis, an indicator species of Oak Savanna.
EPA Grant Number: R826596Title: Demographic and genetic factors affecting population viability of Lupinus perennis, an indicator species of Oak Savanna.
Investigators: Michaels, Helen J. , Mitchell, Randall J.
Institution: Bowling Green State University - Main Campus , University of Akron
EPA Project Officer: Hahn, Intaek
Project Period: January 1, 1999 through December 31, 2002
Project Amount: $289,178
RFA: Ecological Indicators (1998) RFA Text | Recipients Lists
Research Category: Ecological Indicators/Assessment/Restoration , Aquatic Ecosystems
Objective:
The overall objective of this research project was to improve understanding of whether and how population decline develops for a model organism, Perennial Lupine (Lupinus perennis: Fabaceae). The specific objectives of this research project were to: (1) determine whether variation in population size and density explain patterns of variation in reproduction; (2) determine whether reduced reproduction in small populations is associated with habitat degradation, as indicated by differences in environmental factors that influence plant reproduction; (3) determine whether pollinator visitation or foraging patterns have changed in response to reductions in population size or density; and (4) quantify whether differences in levels of genetic variation, inbreeding depression, and mating system occur as a function of population size, and examine their contributions to reduced reproduction in impacted populations.
Plants in small and sparse populations often have low reproductive success, indicating that fragmentation of populations by human activities may prevent populations from being self-sustaining. L. perennis not only suffers from habitat loss and fragmentation, but also is the only host plant for three federally or state endangered butterfly species. Lupine also is an important indicator species for the imperiled Oak Openings savanna community of the Great Lakes ecosystem. Thus, understanding how demographic and genetic factors contribute to the decline of this plant species likely is to be useful in understanding this community, and will help in recovery efforts for imperiled butterfly species.
The possible causes of declines in plant reproductive success in small and sparse populations include decrease in pollinator services, loss of genetic diversity, inbreeding depression, and combinations of these factors. This project examined the importance of these factors, utilizing a blend of observational and experimental techniques, and merging precise data from novel DNA-based indicators of genetic diversity with classical ecological data on reproductive ecology. The observational studies focused on how existing variation in population size affects the factors listed above. Two multipopulation experiments, a reciprocal transplant of seedlings between large and small populations and an inbreeding depression study, tested the role of population size as the causal factor behind the observed patterns. These data have improved understanding of the fundamental biological principles at work in small populations at several levels-from gene, organism, population, to community levels-of biological organization. This research documents changes across multiple spatial scales (local, regional), as well as in functional processes between producers and mutualistic pollinators, and also provides guidance in designing management strategies.
Summary/Accomplishments (Outputs/Outcomes):
Major findings to date for this research include:
· Small populations have slightly lower genetic diversity.
· Small populations have reduced outcrossing rates.
· Smallest populations (N <700) have reduced pollination and, by inference, reduced pollinator visitation.
· Small populations do not differ in habitat quality (based on preliminary data).
· Inbreeding depression is strong, but does not significantly vary with population size, although small populations have higher seed abortion rates.
· A new technique developed for isolating microsatellite loci.
Figure 1. Plant Size in Large and Small Populations (Mean ± standard error [SE]; N = 5 Populations Per Point). Analysis of variance (ANOVA) reveals no significant effect of population size on plant size (F1,323 = 1.2, P>0.2), implying that habitat quality is similar for large and small populations.
We identified and monitored 10 Lupine populations throughout Northwestern Ohio and Southeastern Michigan. We chose pairs of "large" and "small" Lupine populations that past experience indicated should markedly differ in population size, but were as closely matched as possible in all factors (e.g., location, vegetation, openness, soils, management agency). Our goal was that "large" populations should have more than 900 individual flowering Lupine, and "small" populations should have 130-700 flowering individuals (some of our experiments and observations made it inappropriate to include smaller populations in our study). Population size estimates based on 1-m quadrants (26-355 per population) confirmed our size categorizations in all but one pairing, where recent management activities in populations caused the two populations in that pair to converge in population size.
In each population, we chose more than 20 random plants to monitor for reproduction and survivorship during the 3 years of our study. Plant size for small and large populations did not differ (see Figure 1), which is consistent with the hypothesis that habitat quality does not covary with population size.
Figure 2. Seedling Size After 5 months Growth in the Field. Mean ± SE. In January 2000, we planted 160 seeds from each site in the greenhouse, and by April 2000, more than 1,300 seedlings were vigorous and healthy. We then transplanted 800 of these seedlings into the field in a reciprocal transplant design (using five pairs of populations: one large and one small population in each pair). Plants established well, with less than 10 percent mortality by June 2000. "Source" refers to where the seeds were collected; "To" refers to the site in which seedlings were grown. ANOVA: Size of Source population: P = 0.07, Size of Growth population: P = 0.7.
To more directly test for differences in habitat quality, and in offspring vigor for large and small populations, we used reciprocal transplants to act as "phytometers." Analysis to date shows no indication that habitat quality differs as a function of population size, and no indication that offspring vigor is greater for larger populations (e.g., Figure 2). Because we requested a 1-year no-cost extension to complete the genetic studies, we chose to gather a third year of data on both the phytometers and the focal monitoring plants. We completed data collection in June 2002, and currently are analyzing the growth, reproduction, and demographic performance of the more than 300 plants in the monitoring study and the 800 individuals planted for the phytometer experiment over the time period 1999-2002. We also have gathered environmental data associated with each focal plant in each of the 10 populations. Data on soil characteristics (gravimetric water content, nutrients, pH, organic matter), light, and other habitat data now are being analyzed to directly test for environmental differences between large and small populations, and to evaluate their roles in determining reproduction of small populations.
To test the hypothesis that pollinator visitation varies with population size, we observed visitation in our paired populations (>70 hour total pollinator observation time). Pollinator visitation data for these 2 years (1999, 2000) revealed large differences among populations, but no difference in visitation rates as a function of population size (see Figure 3), although visitation rates to dense areas within a population were significantly greater than those for areas with low Lupine density. A Master's Thesis (Bernhardt, 2000) on this was successfully defended in the fall of 2000. We currently are compiling the data for publication.
Figure 3. Flower Visitation Rate for Observation Plots; 1999 and 2000 Data Combined (mean ± SE). Population size had no significant effect on visitation rate (F1,158 = 0.06, P > 0.8), but local density did (F1,158 = 6.5, P < 0.02). Interaction was not significant (F1,158 = 1.63, P > 0.2).
Although necessary, these direct observations of pollinators are very time-consuming, and thus limited the number of sites we could assess. Therefore, we customized, field-tested, refined, and employed a method for counting pollen tubes in flowers as a quick (but indirect) method of assessing successful pollination of Lupine. This method also allowed us to monitor some of the very small (<130 plant) populations that initially helped motivate our study, but which could not be used for most of our experiments because of the negative impact these experiments would have on small populations. This new approach proved to be both low impact, and an excellent indicator of the effects of reduced population size on pollinator service. These data revealed highly variable, and significantly reduced pollination success in the smallest populations (see Figure 4).
We examined the role of genetics in contributing to reduced reproduction for small populations in several ways. To investigate genetic variation as a function of population size, we isolated 15 microsatellite loci, of which 7 were polymorphic. We also developed a novel, chromosome walking/PCR-based approach. We now are preparing a manuscript (Michaels, Shi, and Mitchell, in preparation) describing this approach, which achieved considerable improvement in efficiency in isolation of these markers over the more commonly employed methods.
Figure 4. Number of Pollen Tubes Reaching the Ovary in 23 Populations of Lupinus perennis. Means for each population based on 2 flowers from each of more than 20 flowering stems on different genetic individuals.
Figure 5. Mean ± SE Number of Alleles/Locus for Paired Populations. Means are based on 6 loci, 18-21 individuals/population. Regression of population means reveals a significant increase with population size (P < 0.04). ANOVA also reveals a significant difference between large and small populations (P < 0.02).
To look for effects of population size on genetic variation in each of our 10 populations, we obtained genotypes at 6 loci for the 18 to 25 random focal plants. ANOVA revealed that number of alleles/locus was significantly lower for small populations (see Figure 5). Large populations of L. perennis also tend to have higher expected and observed heterozygosity, though this was not significant. There was significant genetic differentiation among the 10 populations. Analysis of Molecular Variance revealed that population size explained 6 percent of genetic variance, suggesting that there is considerable genetic differentiation between large and small populations. These results indicate that the small populations may be at an early phase of genetic erosion in this species. The genetic differences documented in these studies will be useful to multiple regional agencies involved in actively restoring Lupine populations for Karner Blue butterfly reintroduction efforts in Ohio (Northwest Ohio Office of The Nature Conservancy, Toledo Metroparks, Toledo Zoo, U.S. Fish and Wildlife, Ohio Division of Natural Areas and Preserves, Michigan Fish and Wildlife).
Using the same genetic markers described above, we estimated multilocus outcrossing rates (tm) for eight populations, and determined that Lupine predominately outcrosses. Multilocus outcrossing rates varied among populations (0.716 to 0.951). However, even though outcrossing rates are high, outcrossing rate increased with population size, indicating an increase in the frequency of selfing in smaller populations (see Figure 6a). There also was a trend for outcrossing to decline as inflorescence density increased (p = 0.06). The correlation of outcrossed paternity within progeny arrays (an indicator of the degree to which a pair of siblings share a father) presented intermediate values (rp = 0.255 -0.839), indicating that in Wild Lupine, random outcrossing occurs to a pool of about two to four neighbors.
Figure 6a. Mean Outcrossing Rate (tm) for the Study Populations. Multilocus outcrossing rate increased with population size (R2 = 0.68, p = 0.04).
Preliminary analysis indicates that the correlation of outcrossed paternity increased with population size (see Figure 6b), suggesting that seeds from the same maternal plants in small populations had a greater variety of fathers than in large populations. These data are consistent with the hypothesis that population size changed the foraging behavior of pollinators, but a complete understanding of the complexities of these data requires additional analyses that are in progress.
Figure 6b. The Effect of Population Size on the Correlation of Outcrossed Paternity (rp). The correlation of outcrossed paternity increased with population size (R2 = 0.61, p = 0.02). Outcrossing rates are based on surveys of genetic variation at 5 microsatellite loci in 8 populations for offspring arrays (n=11) of up to 20 maternal plants.
Figure 7a. The Effects of Population Size on Cumulative Fitness Across Four Life Stages: Seed Production in the Field, Followed by Subsequent Seed Emergence, Seedling Survival, and Seedling Biomass When Grown in the Greenhouse. Values for each bar are based on 4 population means, which are themselves based on values from 20 maternal plants. The experiment began in spring of 2000, when we identified 20 plants in each of 8 of our sites (it was not logistically feasible to perform this experiment in all 10 sites), and generated seeds for 3 treatments: hand self-pollination, hand cross-pollination, and open pollination (natural visitation by bees). To compare the relative fitness of these offspring, we germinated more than 1,300 seeds and grew them in a randomized block design in the greenhouse for 15 weeks.
We investigated the extent of inbreeding depression in a multipopulation experiment in the field and greenhouse. Seed abortion was significantly greater in small populations than in large populations (P < 0.03), and fruits produced by self-pollination had greater seed abortion than those from natural or hand-outcross pollinations (P < 0.03). Seedlings from small populations performed poorly compared to those from large populations (see Figure 7a). Inbreeding depression across four life-history stages in L. perennis is substantial (see Figure 7b), but does not vary with population size (see Figure 7c). These data clearly indicate that small populations produce offspring with lower fitness, and suggest that changes in pollinator foraging that lead to self-pollination significantly will reduce reproduction in Wild Lupine. Seed production likely is to be reduced through seed abortions as very early acting inbreeding expression is expressed. This most likely is to be a significant factor in small populations. Furthermore, seed that matured following self-pollinations are of lower quality, further compromising the viability of Lupine populations that have reduced outcrossing rates. We currently are compiling the data for publication.
Figure 7b. Mean ± SE Cumulative Fitness of Seeds Produced by Natural Pollination, Hand-Outcrossed, and Hand-Self Pollinations Size. Relative performance is an index of inbreeding depression ranging from +1 (selfs much worse than crosses) to -1 (the reverse).
Figure 7c. Mean ± SE Relative Performance of Selfed Offspring as a Function of Population
The genetic work constitutes the bulk of Xiujie Shi's Ph.D. Dissertation, and will be defended in 2003. We also are preparing these data for publication.
Conclusions:
Our results do not point to a single overarching factor as the cause of reduced reproductive success for small plant populations. Instead, we found that several different factors each have fairly subtle, but consistently detrimental effects in small populations. These effects include reduced pollinator visitation, lower genetic diversity, and lower outcrossing rates (which should increase the expression of inbreeding depression in small populations). Although separately these factors are not very strong, it is reasonable to suggest that they are mutually self-reinforcing (e.g., small populations may get fewer pollinators, which causes less outcrossing and consequent loss of genetic diversity). If this is true, populations that dip below a particular population size threshold may not be able to recover. Our work is consistent with the hypothesis that for L. perennis, this threshold may be in the vicinity of 200-700 flowering individuals, a much higher threshold than most other workers have considered. Further analysis of our current data, and future work on how changes in reproductive success with population size affect population demography, should help to elucidate the complex interactions of these factors.
This research adds to an understanding of the problems faced by fragmented or degraded natural plant populations and their pollinator communities. One of our most valuable discoveries about this pattern is that a relatively inexpensive and quick method (pollen tube surveys) provides high quality data under a wider variety of circumstances than does the more laborious direct method of watching pollinator visitation. Pollen tube surveys promise to be useful indicators of the integrity of ecological communities, and more importantly, of the extent to which the interactions between mutualists (such as plants and their pollinators) are functioning adequately.
Findings from this work will be utilized in ongoing recovery and restoration efforts for the Oak Openings region, Perennial Lupine, and the federally endangered Karner Blue butterfly. Agencies involved include the Ohio Department of Natural Areas and Preserves, The Nature Conservancy, Toledo Metroparks, Michigan Fish and Wildlife, Ohio Fish and Wildlife, U.S. Fish and Wildlife Service, and the Toledo Zoo. These results also should be of interest to those working on other plants facing habitat fragmentation and degradation, such as Kincaid’s Lupine, an endangered plant that is host for Fender's Blue butterfly.
Activities That Deviated From the Original Research Plan. There were a number of activities that deviated from the original research plan, including:
(1) In addition to pollinator observations, we added an extra index of pollinator visitation-the number of pollen tubes growing in field-collected pistils, as described above.
(2) We conducted the inbreeding depression experiment in only 8 of our 10 sites because there simply was not enough time to properly complete hand pollinations on an appropriate number of plants across 10 locations within a short space of time and comparable weather conditions. In any case, our original proposal was to only study 8 sites, but we have continued to study all 10 of our candidate sites for other aspects of the study, despite our realization that we would be unable to maintain such a broad project for all aspects of the work.
(3) We have developed Lupine-specific microsatellite loci for our analysis of genetic variation, because the soybean primers did not amplify reliably enough. The effort required us to start these microsatellites from scratch, which slowed our progress on assessing genetic variation and the mating system, but the end results are better than we hoped; 7 polymorphic loci out of 15 were screened. Along the way, we developed a novel method for isolating microsatellite DNA markers using a commercially available kit. This new approach greatly reduces the time and labor involved and may be applied to any organism for which microsatellite markers need to be developed.
(4) During seed collections, we noted an insect seed predator that was not known to any of our colleagues interested in Lupine biology. Megalotomus quinquespinosus is a true bug (Hemiptera; Alydidae), but the immatures bear a striking resemblance to ants, in both morphology and behavior. Because the endangered Karner Blue butterfly (and other endangered butterflies in the area) often are tended by ants, this superficial but seldom-detected mimic may lead land managers to believe that ants are abundant (and therefore, suitability for Karner's is high) when they are not. Moreover, although these insects do not appear to kill the Lupine seeds they prey on, they do form a hole in the seed coat that breaks seed dormancy. We finally have an explanation for why so many Lupine seeds were germinating at the very inappropriate time of mid-summer. We have continued to investigate this unexpected discovery, and are now working on a manuscript. Our preliminary surveys have not indicated any clear variation in Megalotomus abundance related to population attributes such as size or density.
Journal Articles on this Report : 1 Displayed | Download in RIS Format
Other project views: | All 30 publications | 1 publications in selected types | All 1 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Shi XJ, Michaels HJ, Mitchell RJ. Effects of self-pollination and maternal resources on reproduction and offspring performance in the wild lupine, Lupinus perennis (Fabaceae). Sexual Plant Reproduction 2005;18(2):55-64. |
R826596 (Final) |
Exit |
Supplemental Keywords:
ecological effects, sensitive populations, animal, population, genetic polymorphisms, scaling, ecosystem, indicators, restoration, terrestrial, habitat, biology, ecology, botany, entomology, great lakes, conservation, Midwest, Ohio, OH, EPA Region 5., RFA, Scientific Discipline, Geographic Area, Ecosystem Protection/Environmental Exposure & Risk, Midwest, Ecology, Ecosystem/Assessment/Indicators, Forestry, Monitoring/Modeling, Ecological Effects - Environmental Exposure & Risk, Environmental Monitoring, Great Lakes, Ecological Indicators, ecological exposure, risk assessment, anthropogenic stresses, ecological effects, habitat, demographic, Oak Savanna, biodiversity, butterfly, conservation, demographic factors, multiple spatial scales, ecosystem indicators, DNA, environmental stress, Lupinus perennis, defoliation, genetic differentiation, indicator species, reproductive healthProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.