Abstract Evolutionary biologists remain puzzled by the often dramatic variation of mating strategies within single species. Of particular interest is the extent to which environmental conditions shape patterns of variation of mating system components within mixed mating species, and how widespread anthropogenic manipulations may influence these associations. Here, we address this question in the common morning glory (Ipomoea purpurea) by combining a dataset of floral traits, estimates of the mating system, and relevant environmental factors compiled for 22 populations of this species distributed along a wide range of environments from the Southeastern and Midwestern United States. We identify a disparate set of environmental factors to influence population-level variation in selfing, inbreeding, and flower morphology. Although floral traits are primarily associated with climatic variation, the outcrossing rate and inbreeding coefficient are primarily influenced by the level of herbicide resistance. Furthermore, we find that populations with higher levels of herbicide resistance exhibit a stronger correlation between mating system-floral traits and mating system estimates (outcrossing rate and inbreeding coefficient). Altogether, these results demonstrate the dominant role that herbicide application plays in the determination of I. purpurea’s mating system, and more generally uncover the complex and unforeseen evolutionary consequences of anthropogenic manipulations in natural systems. inbreeding, Ipomoea purpurea, mating system, morning glory, selfing The diverse number of mating system types among flowering plants have both interested and puzzled evolutionary biologists for centuries. In particular, the ability of many species to produce progeny through both selfing and outcrossing (i.e., mixed mating) and the often dramatic variation of selfing and outcrossing rates across populations of a species have challenged simple evolutionary models (Goodwillie et al. 2005; Karron et al. 2012). Several theoretical explanations have been put forward to reconcile these patterns, and these explanations generally focus on the negative fitness consequences of selfing, such as inbreeding depression and pollen discounting (i.e., reduction in the opportunities for pollen to contribute to the outcrossing pollen pool, Kalisz 1989; Goodwillie et al. 2005; Glémin et al. 2006). A major remaining question is the extent to which environmental forces condition the relative rate of outcrossing and self-fertilization within a species (Karron et al. 2012). Natural processes such as interactions between conspecifics, pollinators, and abiotic conditions (Routley et al. 1999; Kalisz et al. 2004; Lankinen et al. 2016) as well as human manipulations such as deforestation, changing climatic conditions, and the application of herbicides (Dubois et al. 2003; Baucom et al. 2008; Eckert et al. 2010; Ivey and Carr 2012) all influence mating system dynamics. How do these natural and human-mediated forces together affect the relative rate of outcrossing and selfing, and do they explain the wide variation identified among populations in mating system estimates? The combined effects of natural and human-mediated impacts on the mating system likely depend on how they influence the trade-off between reproductive assurance, the avoidance of inbreeding depression (Lande and Schemske 1985; Goodwillie et al. 2005; Barrett 2014), and plasticity in the mating system within a species (Levin 2010; Peterson and Kay 2015). On one hand, local environmental conditions and/or human manipulations may limit opportunities for outcrossing (e.g., by reducing the abundance of pollinators or conspecifics; Sagarin et al. 2006; Eckert et al. 2010; Scheper et al. 2013; Cusser et al. 2016). Under these circumstances, plants should benefit from the reproductive assurance associated with self-fertilization (Goodwillie et al. 2005), assuming that this benefit is not counterbalanced by other evolutionary forces such as inbreeding depression (Fisher 1941; Stone et al. 2014). In this scenario, reproductive assurance, along with the automatic transmission advantage of selfing (i.e., the proportionately higher representation of selfed genes among offspring; Goodwillie et al. 2005), is expected to favor increased selfing rates. On the other hand, if local environmental conditions do not limit outcrossing opportunities, selfing could be detrimental as it increases the chances of inbreeding depression and pollen discounting (Chang and Rausher 1998; Harder and Wilson 1998; Fishman 2000). Together, these environmentally dependent interactions should ultimately determine the mating system of individual populations (Barrett and Eckert 1990). Despite these simple expectations, however, there are few studies that examine how complex and multifaceted environmental factors—both natural and anthropogenic—affect mixed mating in nature (Lankinen et al. 2016). Thus, there are significant gaps in our knowledge regarding the relative influence of human-mediated environmental regimes on mating system dynamics compared to natural environmental variables such as temperature, rainfall, and soil quality. Ipomoea purpurea, a hermaphroditic, mixed mating species, is particularly suitable for examining the relative influence of natural and anthropogenic influences on the mating system. Because of its wide distribution, populations of this species experience a range of environmental conditions—i.e., differences in elevation, precipitation, and soil constitution—and these environmental differences likely impact mating system dynamics. Further, populations are typically found in agricultural fields and as such frequently experience herbicide application. As a result, some populations have evolved resistance to glyphosate (Kuester et al. 2015), which is the active ingredient in the commonly used herbicide RoundUp (Giesy et al. 2000). We have recently documented variation in estimates of the mating system across populations (Kuester et al. 2017) and have shown that this variation is strongly correlated with the level of glyphosate resistance among populations, with the most resistant populations exhibiting a lower outcrossing rate compared to susceptible populations (Kuester et al. 2017). We, likewise, found that the anther–stigma distance (ASD; the distance between the tallest stamen and the stigma) correlates to the level of herbicide resistance across populations (Kuester et al. 2017). This indicates that floral traits associated with the mating system may, likewise, evolve when under selection from human-mediated influences, as ASD is a floral trait that positively correlates with the outcrossing rate (i.e., plants with greater ASD exhibit higher rates of outcrossing; Chang and Rausher 1998). Thus, although we have documented the influence of human-mediated selection regimes on components of I. purpurea’s mating system (Kuester et al. 2017), we know relatively little about the effect of other and more natural environmental influences, such as temperature, humidity, or rainfall, on the mating system and how human-mediated influences may interact with natural environments to influence mating system components. Natural environmental factors are likely to affect mating in this species, as work in other mixed mating species finds that populations in areas with lower precipitation or at greater elevations show lower outcrossing rates due to a reduction in pollinators or by influencing the availability of moisture and nutrients to plants (Waller 1980; Olivieri et al. 1983; Schoen and Brown 1991; Levin 2010). Here, we use mating system estimates and measures of floral phenotypes from populations of I. purpurea distributed across a wide portion of the species range to examine the environmental factors that influence mating system variation in this species. We specifically address the following questions: first, is variation in the mating system and associated floral traits geographically structured? If natural or anthropogenic environmental processes such as broad climatic patterns or patterns of herbicide use similarly influence the mating system of populations in close proximity, we would expect to see evidence of spatial structuring in mating system estimates across the landscape. If, however, local processes influence populations differently, we would expect to find little spatial structure in trait values across the landscape. Second, do anthropogenic factors, such as the level of herbicide resistance, explain more of the variation in mating system components compared to natural factors? If human activities are a main driver of mating system variation in I. purpurea (as reported in Kuester et al. 2017), then anthropogenic effects should outweigh natural influences in explaining mating system variation in this species. Finally, are floral traits other than ASD strongly correlated with mating system parameters? Additionally, is the strength of mating system-floral trait correlation influenced by natural or anthropogenic factors? If certain natural or anthropogenic environmental factors simultaneously affect the outcrossing rate and associated floral traits, the strength of the correlations between these components should depend on the magnitude of environmental factors (e.g., stronger mating system-floral trait correlation at greater levels of human environmental manipulations). By addressing these questions, we offer insights into the proximate determinants of mating system variation and provide compelling evidence of the dominant role that human activities can play in shaping mating system in natural populations. Methods Study System Ipomoea purpurea, a weed of major agricultural concern worldwide (Baucom and Mauricio 2004), is a climbing annual vine with a wide distribution across both Central and North America (Ennos 1981; Defelice 2001). It grows mostly in disturbed habitats such as agricultural fields and roadsides and begins flowering in mid-May (Rausher and Fry 1993; Defelice 2001). This species is primarily pollinated by bumblebees, and will set mature seeds within 4 weeks of cross- or self-fertilization. The mixed mating strategy of I. purpurea is unlikely to be maintained by inbreeding depression (Chang and Rausher 1999; Mason et al. 2015), but it has been suggested that pollen discounting may prevent the fixation of selfing alleles (Rausher and Chang 1999). Neither of these processes by themselves, however, are considered to be sufficient to maintain the variability observed in outcrossing rates and inbreeding coefficients in this species (Rausher and Chang 1999). Additionally, these processes do not explain the heritable variation in ASD (Chang and Rausher 1998) or in other floral traits associated with selfing rates such as floral size (Sicard and Lenhard 2011; Duncan and Rausher 2013). Previous work has identified modest but significant genetic differentiation among I. purpurea populations located across the Southeastern and Midwestern United States (FST = 015; Kuester et al. 2015). Further, estimates of population connectivity in this species based on over 8000 SNPs identified low levels of recent migration between populations and found that migration is primarily conditioned by climate and human population density (Alvarado-Serrano et al. 2017). This indicates that, for the most part, I. purpurea populations are locally stable and under the effect of local (i.e., population-specific) environmental conditions. Data Compilation We compiled previously reported morphological and genetic data on 22 populations of I. purpurea sampled from the Southeastern and Midwestern United States in 2012 (Kuester et al. 2017) as well as environmental data compiled from a variety of sources (listed in Supplementary Table S1). We used a wide range of abiotic environmental factors that could (directly or indirectly) influence mating system variation in this species (Supplementary Table S1). Given the difficulty of measuring all these environmental factors in situ, we chose to use remote sensing and census data. Although this decision carries an intrinsic spatial resolution limitation, GIS data at moderate to coarse resolutions have been shown to reasonably capture biologically relevant population-level processes (Kerr and Ostrovsky 2003; Kozak et al. 2008). In addition, we included population-level glyphosate resistance estimates—measured as the proportion of surviving individuals after the application of manufacturer’s recommended doses of glyphosate (Kuester et al. 2015), along with county-level estimates of the cumulative amount of glyphosate applied to these populations over the last 2 decades (years 1992–2012). Our complete environmental dataset comprised a total of 31 predictor variables (Supplementary Table S1) with several that were highly correlated with each other. Therefore, we performed a hierarchical agglomerative clustering in R3.3.3 (R Core Team 2017), using package ClustOfVar (Chavent et al. 2013) to select a non-redundant set of environmental predictors. This analysis clusters variables into statistically homogeneous sets to identify groups of variables that contribute similar information (Chavent et al. 2012), and is a variable reduction approach appropriate for our environmental data because it simplifies interpretation when multiple clusters of highly correlated variables exist. In contrast to a principal components analysis, this method has the advantage of retaining original variables instead of producing new synthetic variables, which could be difficult to interpret (Dormann et al. 2013). Using this analysis we retained a set of 8 environmental variables that were not highly correlated with one another (average absolute Pearson’s coefficient = 0.36). The mating system estimates used in this work were obtained from genotyping 4584 individuals from 22 populations (median number of individuals per population: 207 [29–417]) using 15 microsatellite loci (previously published in Kuester et al. 2017). Mating system estimates were determined using default parameters in the programs BORICE (Koelling et al. 2012) and MLTR (Ritland and Jain 1981; Ritland 2002). Because BORICE and MLTR estimates were correlated with each other and BORICE is known to outperform MLTR when maternal genotypes are unavailable (Koelling et al. 2012), we kept only BORICE’s per population estimates of the outcrossing rate (t) and inbreeding coefficient (F) for all subsequent analyses (for further details see Kuester et al. 2017). We examined both t and F because each represents different aspects of the mating system; although outcrossing rate indirectly reflects the amount of offspring produced by self-pollination, the inbreeding coefficient reflects the history of selfing and/or close relative mating (Koelling et al. 2012). We used estimates of the following 4 floral traits: the length of the tallest stamen to the top of the anther (TAL), height of the pistil to the top of the stigma (SL; data from Kuester et al. 2017), and length and width of the corolla (CL and CW, respectively; previously unpublished data from the experiment described in Kuester et al 2017; Figure 1). TAL and SL were used to determine the anther–stigma distance (ASD, which is the distance between the tallest anther and the stigma) of each flower and TAL and SL together with CL and CW were used to produce an estimate of floral size, which is of interest given the commonly reported association between flower size and selfing (Sicard and Lenhard 2011; Duncan and Rausher 2013). These measurements were taken over multiple dates in the fall of 2014 from a total of 445 individuals collected as seeds from all 22 populations and grown at the Matthaei Botanical Gardens at the University of Michigan (Ann Arbor, MI; further details in Kuester et al. 2017). All 5 floral traits were individually averaged for each population across flowers, dates, and individuals so that we have a single estimate per population in order to be comparable to the population-level estimates of the mating system (t and F). Given the high correlation between 4 of our floral measurements (TAL, SL, CL, and CW; Supplementary Figure S1), we condensed them into a single estimate of flower size by running principal component analyses on their covariance matrix after standardizing all of them. The retained first principal component, which accounted for 73.77% of the total variance, was equitably negatively associated with all 4 floral measurements (Supplementary Figure S1), and thus primarily summarized overall flower size (with lower scores corresponding to bigger flowers). For ease of interpretation, however, populations’ scores on this axis were multiplied by −1 so that flower size increased as PC scores increased. We used this component (PC1) in all subsequent analyses together with ASD as our floral traits. Figure 1. View largeDownload slide Samples distribution of I. purpurea populations with floral measurements and mating system estimates (a). The four floral measurements included in the analysis are shown in upper-right inset (b). See online version for full colors. Figure 1. View largeDownload slide Samples distribution of I. purpurea populations with floral measurements and mating system estimates (a). The four floral measurements included in the analysis are shown in upper-right inset (b). See online version for full colors. Data Analyses Using this combined dataset, we performed three sets of analyses. First, we determined if the populations exhibited spatial structuring in the mating system and trait values, which is a prerequisite to our hypothesis of local or population-specific and not regional environmental processes (whether anthropogenic or natural) primarily driving mating system variation in this species. Second, we assessed whether natural or anthropogenic environmental predictors best explained individual variation in mating system components and investigated whether common responses to environmental factors were present among mating system components. Finally, we assessed whether natural or anthropogenic environmental variation influences the strength of the correlation between floral traits and mating system estimates, which is expected if some environmental factors simultaneously influence both mating system parameters and associated floral traits. Spatial Structure We investigated whether floral traits and mating system estimates are spatially clustered, which would be expected if regional environmental processes condition mating system variation in I. purpurea. Such analysis is especially important in light of the clustered geographic distribution of our samples (Figure 1), which reflects the distribution of crop fields with which this species is tightly associated (Alvarado-Serrano et al. 2017). To do this, we assessed the degree of global and local spatial autocorrelation on our data by calculating global and local Moran’s I (Moran 1950; Anselin 1995) in R3.3.3 (R Core Team 2017) using packages ape (Paradis 2011) and spdep (Bivand and Piras 2015), respectively. For the local analysis, we adjusted P values for multiple comparisons using a false discovery rate method (Benjamini and Hochberg 1995). In addition, we assessed the association between each individual mating system component and geographic location by running independent multivariate linear regressions against longitude and latitude. Trait Variation To identify whether natural and/or anthropogenic environmental predictors best explained individual variation in mating system components, we examined the relationship between the 8 selected environmental factors and our 4 mating system components (ASD, flower size, t, and F). Specifically, we ran multivariate linear regressions for each mating system trait and performed backward stepwise variable selection using a resampling model calibration strategy with 500 bootstrap replicates. This strategy allows for bias-correction of error estimates based on nonparametric smoothers and hence avoids possible overfitting given our sample size (Harrell 2015). All these analyses were run on standardized environmental variables using the package rms (Harrell 2017) in R3.3.3 (R Core Team 2017). In addition, we estimated the relative contribution of each predictor retained by comparing their associated regression coefficients (i.e., the expected change in the independent variable per unit change of a predictor when all other predictors in the model remain constant). Given our sample size, however, no interactions were included in any model. In addition, because several assumptions of linear regression might potentially be violated by our dataset we verified the robustness of our results by running homologous model-free regressions using machine-learning tools (Random Forest Regressions). For simplicity, we present these models in the Supplemental Material. Trait Correlation Because humans’ overall effect on the mating system may depend on the intensity of the environmental manipulation (e.g., greater herbicide use potentially leads to a more pronounced effect on multiple components of the mating system), we assessed whether the magnitude of anthropogenic impact has an effect on the strength of the correlation between mating system parameters and associated floral traits. We, likewise, performed the same analysis using natural environmental variables for comparison. To do this, we employed a novel analysis that effectively runs separate correlation tests on groups of environmentally similar samples. Analogous to Geographically Weighted Regression (Brunsdon et al. 1998), this analysis investigates whether subdividing the sample into relatively homogeneous groups (in our case, according to their environmental characteristics) significantly affects the analysis’ estimates. To run this analysis, we first calculated pairwise Pearson’s simple correlation coefficients between ASD and outcrossing rate (t) and inbreeding coefficient (F) as well as between flower size and t and F using our entire set of populations (hereafter called global coefficients). Then, we re-calculated Pearson’s correlation coefficients on subsamples of populations grouped according to their environmental variable values (using variable quantiles to define groups; hereafter called subsample coefficients). Finally, we compared the subsample coefficients against the corresponding global coefficients. Because of our small sample size after subsampling the data (range: 4–5 populations per subsample), we did not use the correlation P-value between traits within each subsample to assess significance. Instead, we assessed significance by comparing the correlation coefficient obtained for each group against similarly obtained coefficients from a set of 100 randomly split data sets that share the same number of observations as the environmentally grouped data (bootstrap coefficients hereafter). Significant differences between our bootstrap and subsample coefficients suggest that the environmental variable used for binning has an effect on the degree of association between mating system components, and thus, that, depending on its magnitude, this environmental factor simultaneously affects more than one mating system component. Results Spatial Structure We did not uncover significant geographic structure across I. purpurea populations for any of our floral traits (ASD and flower size) or any of our mating system estimates (t and F), suggesting that local (i.e., population-specific) environmental processes primarily drive mating system variation in this species. No mating system component was significantly associated with either latitude or longitude (Supplementary Figure S2), nor did we find evidence of geographically proximate populations showing similar trait values—as measured by either global or local Moran’s I—in any of our individual variables (Table 1; Figure 2). We further verified the finding of little spatial structure by comparing the results of nonspatial analyses with 2 spatially explicit analyses. First, we compared linear regressions results against those of Geographically Weighted Regressions (GWR), which instead of fitting a global model using all observations, fit a set of linear regressions exclusively using geographically proximate observations that are iteratively selected through a moving window (Fotheringham et al. 2002). Second, we compared simple correlation analyses against partial correlation analyses that used latitude and longitude as covariates. Neither comparison identified significant differences between nonspatial and spatially explicit analysis (GWR: BFC99 (Brunsdon et al. 1999): 0.67–1.21, P-value: 0.37–0.79; partial correlation analysis: Mann–Whitney U = 0.75, P-value: 0.69). Therefore, we present the results only of the nonspatial analyses. Table 1. Summary of Global and Local Moran’s I analyses on the extent of local spatial correlation among mating system estimates and associated floral traits Variable Global Moran’s I (estimate / P value) Local Moran’s I mean [range] Standard deviate Local Moran’s I [range] Local Moran’s I P-value range ASD 0.045 (0.360) −0.132 [−1.160–0.52] −0.250 [−3.598–1.278] 0.503–1.000 Flower size −0.056 (0.930) −0.197 [−1.078–0.813] −0.397 [−3.139–1.917] 0.138–1.000 t −0.062 (0.889) −0.048 [−0.572–0.921] −0.022 [−1.534–1.848] 0.129–1.000 F 0.001 (0.997) 0.113 [−0.512–0.888] 0.394 [−1.234–2.170] 0.075–1.000 Variable Global Moran’s I (estimate / P value) Local Moran’s I mean [range] Standard deviate Local Moran’s I [range] Local Moran’s I P-value range ASD 0.045 (0.360) −0.132 [−1.160–0.52] −0.250 [−3.598–1.278] 0.503–1.000 Flower size −0.056 (0.930) −0.197 [−1.078–0.813] −0.397 [−3.139–1.917] 0.138–1.000 t −0.062 (0.889) −0.048 [−0.572–0.921] −0.022 [−1.534–1.848] 0.129–1.000 F 0.001 (0.997) 0.113 [−0.512–0.888] 0.394 [−1.234–2.170] 0.075–1.000 View Large Figure 2. View largeDownload slide Geographic variability in mating system estimates and associated floral traits: (a) ASD, (b) flower size, (c) outcrossing rate (t), and (d) inbreeding coefficient (F). See online version for full colors. Figure 2. View largeDownload slide Geographic variability in mating system estimates and associated floral traits: (a) ASD, (b) flower size, (c) outcrossing rate (t), and (d) inbreeding coefficient (F). See online version for full colors. Individual Trait Variation Our linear regressions identified unique sets of environmental predictors for the different components of I. purpurea’s mating system. Not only were there different best predictors for each mating system component analyzed (Table 2), but also the relative magnitude—measured as the net effect that change in the predictor causes in the response trait—and explanatory power of the associations differed among components (Supplementary Figure S3). Specifically, ASD decreased as elevation, relative humidity, and mean annual temperature increased (Table 2, Supplementary Figure S3a). Flower size on the other hand was only marginally associated with mean annual temperature (Table 2, Supplementary Figure S3b). In contrast, as previously observed (Kuester et al. 2017), outcrossing rate (t) was primarily affected by herbicide resistance, being the lowest at higher resistance values (Table 2, Supplementary Figure S3c). Similarly, the inbreeding coefficient was strongly associated with herbicide resistance and proportionally increased as resistance increased, as previously described (Kuester et al. 2017). The inbreeding coefficient (F) was also positively associated with mean temperature, and inversely associated with annual precipitation and annual temperature range (Table 2, Supplementary Figure S3d). Thus, though we found significant associations between all components of the mating system and environmental factors, the relationships between these components and environment factors was not similar between t and F. After other environmental variables were taken into account, floral traits were exclusively affected by natural environmental factors, whereas anthropogenic factors strongly impacted outcrossing rate (t) and inbreeding coefficient (F), supporting our hypothesis that human activities are a main driver of mating system variation in I. purpurea. These results were consistent with our Random Forest (RF) regressions (Supplementary Material A) despite the different statistical approaches of linear and RF regressions (Breiman 2001). Table 2. Summary of Ordinary Least Square linear regressions of mating system components on environmental variables Dependent LR X2 a P-value R2 Retained Predictorsb Beta coefficientsc ASD 9.11 (3, 18) 0.028 0.34 Elevation Relative humidity Mean temperature −0.064 (±0.026) −0.052 (±0.018) −0.049 (±0.022) Flower size 2.97 (1, 20) 0.10 0.11 Mean temperature 0.105 (±0.395) t 9.05 (1, 20) 0.002 0.34 Herbicide resistance −0.078 (±0.024) F 18.44 (4, 17) 0.001 0.57 Herbicide resistance Annual precipitation Temperature range Mean Temperature 0.059 (±0.014) −0.039 (±0.018) −0.038 (±0.018) 0.037 (±0.019) Dependent LR X2 a P-value R2 Retained Predictorsb Beta coefficientsc ASD 9.11 (3, 18) 0.028 0.34 Elevation Relative humidity Mean temperature −0.064 (±0.026) −0.052 (±0.018) −0.049 (±0.022) Flower size 2.97 (1, 20) 0.10 0.11 Mean temperature 0.105 (±0.395) t 9.05 (1, 20) 0.002 0.34 Herbicide resistance −0.078 (±0.024) F 18.44 (4, 17) 0.001 0.57 Herbicide resistance Annual precipitation Temperature range Mean Temperature 0.059 (±0.014) −0.039 (±0.018) −0.038 (±0.018) 0.037 (±0.019) Results obtained after 500-bootstrapped backward selection are presented. Corresponding Random Forest regression results are given in Supplemental Material. aModel likelihood ratio chi-square statistic. Degrees of freedom are below in parentheses. bPredictors are ordered by absolute beta coefficient magnitude. Predictors also identified in Random Forest regressions are in bold. cStandard errors are in parentheses underneath. View Large Trait Correlation As previously reported (Chang and Rausher 1998), we found that ASD was significantly correlated with the outcrossing rate (t; Figure 3), which increases as the distance between the anther and stigma becomes greater. We found no correlation between ASD and the inbreeding coefficient of maternal individuals (F), nor did we find a correlation between floral size and either t or F (Figure 3). We assessed whether the strength of these correlations varied across environmental gradients and found that no natural environmental factor significantly affected the correlation between ASD and t or F or between floral size and t or F (results not shown). In contrast, herbicide resistance and herbicide use significantly affected the correlation between ASD and outcrossing rate (t) and ASD and inbreeding coefficient (F), respectively. Specifically, the ASD-t correlation was stronger at the highest values of herbicide resistance (Figure 4a), whereas the ASD-F correlation became significant and negative only at moderately high herbicide use (Figure 4b). These results suggest that under high herbicide use, humans simultaneously affect floral and genetic components of the mating system. No significant effect of either herbicide use or herbicide resistance was recovered on the correlation between floral size and outcrossing rate (t; Figure 4c) or inbreeding coefficient (F; Figure 4d). Figure 3. View largeDownload slide Correlation between floral traits and mating system parameters. (a) ASD and outcrossing rate (t), (b) ASD and inbreeding coefficient (F), (c) flower size and outcrossing rate (t), and (d) flower size and inbreeding coefficient (F). Figure 3. View largeDownload slide Correlation between floral traits and mating system parameters. (a) ASD and outcrossing rate (t), (b) ASD and inbreeding coefficient (F), (c) flower size and outcrossing rate (t), and (d) flower size and inbreeding coefficient (F). Figure 4. View largeDownload slide Correlation of floral traits and mating system parameters across anthropogenic selection regimes. Estimates of Pearson’s correlation coefficient across environmentally binned samples are shown by circles, whereas a dashed horizontal line indicates the global estimate (with shaded box indicating its confidence intervals). Grey vertical lines indicate 95% confidence estimates based on random resampling. Significant estimates are highlighted by brighter symbols. Figure 4. View largeDownload slide Correlation of floral traits and mating system parameters across anthropogenic selection regimes. Estimates of Pearson’s correlation coefficient across environmentally binned samples are shown by circles, whereas a dashed horizontal line indicates the global estimate (with shaded box indicating its confidence intervals). Grey vertical lines indicate 95% confidence estimates based on random resampling. Significant estimates are highlighted by brighter symbols. Discussion In this study, we examined the potential that populations exhibit spatial structure in mating system and floral traits, that particular environmental variables, both natural and human-mediated, influence mating system components, and that the strength of correlations between mating system traits may be influenced by the environment. We found little evidence for geographic structure in the mating system of I. purpurea, supporting the hypothesis that natural and anthropogenic environmental processes at the level of the population primarily drive mating system variation in this species. Further, we identified a set of disparate environmental factors to influence individual mating system components. Specifically, we find that mating system estimates (t and F) are influenced by the level of herbicide resistance, whereas associated floral traits (ASD and flower size) are primarily influenced by climatic factors. Accordingly, floral traits and mating system estimates are not strongly correlated with each other, with the exception of ASD and outcrossing rate. Finally, the strength of the correlation between ASD and outcrossing rate (t) and inbreeding coefficient (F) seems to be influenced by the selection pressure imposed by high herbicide use, which suggests that intense anthropogenic manipulation has the potential to simultaneously affect both mating system estimates and associated floral traits. Interestingly, no other environmental factor influenced the correlation between floral traits and genetic estimates of the mating system of I. purpurea. Taken together, these results support our hypothesis that human manipulations are a dominant driver of mating system variation in I. purpurea (as reported in Kuester et al. 2017) even after taking into consideration other plausible natural drivers. Furthermore, these results highlight the complex influence of anthropogenic and natural environmental factors on the mating system of this agricultural weed and highlight the unique responses of individual mating system components. Below we discuss our results in light of the disparate environmental responses among mating system components and present them in context with the dominant impact of herbicide application. Disparate Environmental Responses Compelling empirical evidence supports an association between individual mating system components and environmental conditions. For example, outcrossing rates have been found to covary in a variety of plant systems with elevation (Neale and Adams 1985), humidity (Brown et al. 1978; Shea 1987), and temperature (Holtsford and Ellstrand 1992). Similarly, ASD has been found to strongly respond to environmental factors, including humidity (Elle and Hare 2002; Van Etten and Brunet 2013), water and nutrient availability (Vallejo-Marín and Barrett 2009), light regime (Brock and Weinig 2007), and temperature (Lankinen et al. 2016). Nevertheless, a relatively small number of studies have simultaneously explored variation in patterns of multiple mating system components in natural populations across environmental gradients. Those that have tend to find that separate mating system components are primarily associated with different environmental variables (e.g., Lankinen et al. 2016). In agreement with these findings, our study also identifies a combination of multiple different environmental factors acting on different components of the I. purpurea mating system. Specifically, we find that ASD is primarily associated with elevation, humidity and temperature, flower size is marginally associated with temperature, and outcrossing rate and inbreeding coefficient are primarily associated with herbicide resistance. Together these findings support the relative independence of mating system components under natural conditions (Johnston and Schoen 1996; Dudley et al. 2007), which raises the question of how these separate responses are integrated into concerted mating strategies, and, whether the strength of their correlation varies across environmental gradients. Considering the dramatic evolutionary consequences that reproductive strategies carry (Kalisz 1989; Glémin et al. 2006), individuals with reproductive strategies ill-matched to their complex environment are expected to experience strong detrimental fitness consequences. Thus, plants may fine-tune their overall mating strategies through plasticity and/or adaptation to local environmental conditions. For this reason, it might be potentially beneficial to maintain flexibility in environmental responses among mating system components. In particular, given the multiplicity of conflicting selective pressures, plants experience in natural environments (Holtsford and Ellstrand 1992; Sagarin et al. 2006), such flexibility—and presumably weak underlying genetic correlations between traits (Dudley et al. 2007; Lankinen et al. 2007)—may offer a solution to a disparate set of environmental pressures by allowing mating strategies to “match” local environmental conditions (Wolters and Jürgens 2009; Fournier-Level et al. 2011). However, the correlation between different components of the mating system may become tighter in novel stressful conditions, under which selective forces favoring selfing and a selfing-prone phenotype often prevail (Antonovics 1968; Levin 2010; Kuester et al. 2017). Given extreme conditions, selection to reproduce in low-density populations and simultaneously maintain genotypes adapted to the extreme environment is expected to outweigh other concurrent selection for pollinator attraction (Uyenoyama and Antonovics 1987; Parachnowitsch and Kessler 2010; Opedal et al. 2017). This expectation fully lines up with our finding that current agricultural practices, and in particular herbicide application, seems to overcome other disparate environmental factors and elicit a concerted response in I. purpurea among mating system components. Although this latter finding needs to be further validated using a larger sample size, it points towards an overall shift towards selfing caused by the development of herbicide resistance. Below we discuss this possibility in more detail. Dominant Effect of Anthropogenic Manipulations The impact of human activities on world’s ecosystems is undeniable (Vitousek et al. 1997; Haberl et al. 2007). Humans influence evolutionary processes at all levels, from those governing community structure to those governing individuals’ life strategies (Hendry et al. 2016), including the mating patterns of organisms. Indeed, compelling evidence supports an indirect effect of human-mediated processes on plant mixed mating systems, whether through habitat modifications (e.g., disturbance; Eckert et al. 2010; Levin 2012) or selection for increased selfing (e.g., in response to metal pollution; Antonovics 1968; Barrett and Bush 1991). In contrast, the influence of human-driven processes in other systems such as agricultural weeds is less well known. Our previous work on the impact of herbicide resistance on mating system variation in I. purpurea supports the hypothesis of human-mediated selection leading to increased selfing (Kuester et al. 2017). However, it was unknown if the association between the mating system and herbicide resistance was potentially mediated by other environmental factors—an important consideration, as herbicide resistance is itself correlated with precipitation and soil variables (Alvarado-Serrano and Baucom, unpublished data). Further, the influence of herbicide use on mating system variation relative to natural factors remained unexamined, and we had yet to consider the potential that resistance evolution favored an overall concerted response among mating system components. Here, by simultaneously exploring the association of mating system estimates and floral traits with multiple environmental variables, both natural and anthropogenic, we provide further support for the dominant role of herbicide use on mating system variation across populations in I. purpurea. Herbicide resistance is not only by far the strongest predictor of outcrossing rate (t) and inbreeding coefficient (F) in I. purpurea, but together with herbicide use also has a significant impact on the association of these estimates and ASD (although this conclusion needs further confirmation given the sample size limitation for this analysis, see above). These findings highlight the dominance of this anthropogenic influence over that of natural environmental factors. Although the ultimate mechanisms underlying these findings are beyond the scope of our study, especially considering the difficulty of assessing direct causality from statistical associations (Mac Nally 2000), our results point again to strong anthropogenic selective pressures driving mating system variation in this species (Kuester et al. 2017). Under consistent herbicide use, it is likely that other (arguably weaker) selective forces that act on the different mating system components might be superseded by the remarkably strong selection imposed by herbicide use (Culpepper et al. 2001). If this is indeed the case, the dominant influence of herbicide application could diminish natural mating system variation in this species by favoring selfing over outcrossing across herbicide-exposed populations, and potentially lead to increased selfing rates across populations. In line with this expectation, we predicted that a shift to selfing in the highly resistant populations may influence flower size, as highly selfing species often exhibit smaller, less visible flowers (Sicard and Lenhard 2011; Button et al. 2012). Our data did not support this idea; however, as no other floral trait beyond ASD appeared to be influenced by herbicide resistance or herbicide use. It is important to note that there is likely a dynamic and intricate evolutionary response behind the associations observed in this study. On one hand, although the observed mating system responses parallel other anthropogenic induced changes (Antonovics 1968; González-Varo et al. 2010), the underlying, causal mechanisms may not be the same (i.e., compare agricultural practices with deforestation; Eckert et al. 2010). Further, the evolutionary dynamics of the response to herbicide application would presumably vary over time as the continued use of herbicide likely increases resistance levels across populations. Under these circumstances, more individuals should be able to survive its application (reducing the limitation of conspecific mates) while non–herbicide-adapted individuals should be purged out of the population (reducing the risks of maladaptive gene flow). Further, this evolutionary dynamic may also be affected by the expected increased frequency of more extreme natural environmental variation (Rahmstorf and Coumou 2011) and its effect on relevant floral traits (Hargreaves and Eckert 2013). Therefore, further studies of this and other similar anthropogenically driven interactions, including long-term analyses, are necessary to better understand the maintenance of mixed mating systems and the general evolutionary impacts of human activities. Future Directions and Conclusions Although years of research on mixed mating systems have identified many possible mechanisms underlying their maintenance (Barrett 2014), there has been seldom empirical evidence to assess their relative contribution in natural populations, let alone the effects of the ever increasing influence of human activities. Our examination of a dataset that combines floral measurements and mating system estimates over a wide portion of a species’ range provides further support for the dominant influence of human activities on species’ mating system (Antonovics 1968; Levin 2010; Kuester et al. 2017). Not only does this anthropogenic manipulation outweigh the effect of natural environmental factors, but it also appears to influence the correlation strength between mating system components by favoring concerted shifts towards a selfing strategy. Although it remains to be seen how prevalent these effects are in less extreme selective regimes, our findings reveal the high evolutionary lability of mating system components and hence their potential sensitivity to anthropogenic impacts (Eckert et al. 2010; Opedal et al. 2017). Given the relatively short time scale over which resistance has developed in I. purpurea (Baucom and Mauricio 2004; Kuester et al. 2016), our results call attention to the need for considering the potential major impact of human-driven selection on such a fundamental life history trait in management and conservation efforts of this and other species. The results from the current study thus offer practical information to forecast plausible consequential responses of species to anthropogenic manipulations, a topic of major importance given the ever increasing reach of human activities. Supplementary Material Supplementary material can be found at https://academic.oup.com/jhered/. Funding United States Department of Agriculture (07191, 04180) to R.S.B. Acknowledgments We are grateful to Adam Kuester for collecting the data used in this study, and to Megan Van Etten and others in the Baucom lab for helpful discussions and comments on previous drafts. We also thank the reviewers and associate editor for their insightful suggestions and John Megahan for his illustration of I. purpurea flower. Data Availability All original data available at Dryad: http://dx.doi.org/doi:10.5061/dryad.dn6vr. References Alvarado-Serrano DF, Van Etten M, Chang S-M, Baucom RS. 2017. Landscape connectivity of a noxious invasive weed: human-aided or natural dispersal? bioRxiv. doi: http://dx.doi.org/10.1101/054122. Anselin L. 1995. Local indicators of spatial association—LISA. Geogr Anal . 27: 93– 115. Google Scholar CrossRef Search ADS Antonovics J. 1968. Evolution in closely adjacent plant populations V. Evolution of self-fertility. Heredity . 23: 219– 238. Google Scholar CrossRef Search ADS Barrett SCH. 2014. Evolution of mating systems: outcrossing versus selfing. In: Losos JB, Baum DA, Futuyma DJ, Hoekstra HE, Lenski RE, Moore AJet al. , editors. The Princeton guide to evolution . Princeton University Press: Princeton. p. 356– 362. Google Scholar CrossRef Search ADS Barrett SCH, Bush EJ. 1991. Population processes in plants and the evolution of resistance to gaseous air pollutants. In: Ecological Genetics and Air Pollution , Springer, pp 137– 165. Google Scholar CrossRef Search ADS Barrett SCH, Eckert CG. 1990. Variation and evolution of mating systems in seed plants. In: Kawano S, editor. Biological approaches and evolutionary trends in plants . Burlington: Academic Press. p 229– 254. Google Scholar CrossRef Search ADS Baucom RS, Mauricio R. 2004. Fitness costs and benefits of novel herbicide tolerance in a noxious weed. Proc Natl Acad Sci USA . 101: 13386– 13390. Google Scholar CrossRef Search ADS PubMed Baucom RS, Mauricio R, Chang S-M. 2008. Glyphosate induces transient male sterility in Ipomoea purpurea. Botany . 86: 587– 594. Google Scholar CrossRef Search ADS Benjamini Y, Hochberg Y. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Statist Soc B . 57: 289– 300. Bivand R, Piras G. 2015. Comparing implementations of estimation methods for spatial econometrics. J Stat Softw . 63: 1– 36. Breiman L. 2001. Statistical modeling: the two cultures. Stat Sci . 16: 199– 231. Google Scholar CrossRef Search ADS Brock MT, Weinig C. 2007. Plasticity and environment-specific covariances: an investigation of floral-vegetative and within flower correlations. Evolution . 61: 2913– 2924. Google Scholar CrossRef Search ADS PubMed Brown A, Zohary D, Nevo E. 1978. Outcrossing rates and heterozygosity in natural populations of Hordeum spontaneum Koch in Israel. Heredity . 41: 49– 62. Google Scholar CrossRef Search ADS Brunsdon C, Fotheringham S, Charlton M. 1998. Geographically weighted regression. J R Stat Soc D . 47: 431– 443. Google Scholar CrossRef Search ADS Brunsdon C, Stewart Fotheringham A, Charlton M. 1999. Some notes on parametric significance tests for geographically weighted regression. J Reg Sci . 39: 497– 524. Google Scholar CrossRef Search ADS Button L, Villalobos AL, Dart SR, Eckert CG. 2012. Reduced petal size and color associated with transitions from outcrossing to selfing in Camissoniopsis cheiranthifolia (Onagraceae). Int J Plant Sci . 173: 251– 260. Google Scholar CrossRef Search ADS Chang S-M, Rausher MD. 1998. Frequency-dependent pollen discounting contributes to maintenance of a mixed mating system in the common morning glory Ipomoea purpurea. Am Nat . 152: 671– 683. Google Scholar PubMed Chang S-M, Rausher MD. 1999. The role of inbreeding depression in maintaining the mixed mating system of the common morning glory, Ipomoea purpurea. Evolution . 53: 1366– 1376. Google Scholar CrossRef Search ADS PubMed Chavent M, Kuentz-Simonet V, Liquet B, Saracco L. 2012. ClustOfVar: An R package for the clustering of variables. J Stat Softw . 50: 1– 16. Google Scholar CrossRef Search ADS PubMed Chavent M, Kuentz-Simonet V, Liquet B, Saracco J. 2013. ClustOfVar: Clustering of variables R package . Version 0.8; [accessed 2017 September 15]. Available from: https://CRAN.R-project.org/package=ClustOfVar. Culpepper SA, Gimenez AE, York AC, Batts Roger B Wilcut John. 2001. Morningglory (Ipomoea spp.) and large crabgrass (Digitaria sanguinalis) control with glyphosate and 2, 4-DB mixtures in glyphosate-resistant soybean (Glycine max). Weed Technol . 15: 56– 61. Google Scholar CrossRef Search ADS Cusser S, Neff JL, Jha S. 2016. Natural land cover drives pollinator abundance and richness, leading to reductions in pollen limitation in cotton agroecosystems. Agric Ecosyst Environ . 226: 33– 42. Google Scholar CrossRef Search ADS Defelice MS. 2001. Tall morningglory, Ipomoea purpurea (L.) roth—flower or foe? Weed Technol . 15: 601– 606. Google Scholar CrossRef Search ADS Dormann CF, Elith J, Bacher S, Buchmann C, Carl G, Carré Get al. 2013. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography . 36: 27– 46. Google Scholar CrossRef Search ADS Dubois S, Cheptou P-O, Petit C, Meerts P, Poncelet M, Vekemans Xet al. 2003. Genetic structure and mating systems of metallicolous and nonmetallicolous populations of Thlaspi caerulescens. New Phytol . 157: 633– 641. Google Scholar CrossRef Search ADS Dudley LS, Mazer SJ, Galusky P. 2007. The joint evolution of mating system, floral traits and life history in Clarkia (Onagraceae): genetic constraints vs. independent evolution. J Evol Biol . 20: 2200– 2218. Google Scholar CrossRef Search ADS PubMed Duncan TM, Rausher MD. 2013. Evolution of the selfing syndrome in Ipomoea. Front Plant Sci . 4: 301. Google Scholar CrossRef Search ADS PubMed Eckert CG, Kalisz S, Geber MA, Sargent R, Elle E, Cheptou PO, Goodwillie C, Johnston MO, Kelly JK, Moeller DAet al. 2010. Plant mating systems in a changing world. Trends Ecol Evol . 25: 35– 43. Google Scholar CrossRef Search ADS PubMed Elle E, Hare JD. 2002. Environmentally induced variation in floral traits affects the mating system in Datura wrightii. Funct Ecol . 16: 79– 88. Google Scholar CrossRef Search ADS Ennos RA. 1981. Quantitative studies of the mating system in two sympatric species of Ipomoea (Convolvulaceae). Genetica . 57: 93– 98. Google Scholar CrossRef Search ADS Fisher RA. 1941. Average excess and average effect of a gene substitution. Ann Eugen . 11: 53– 63. Google Scholar CrossRef Search ADS Fishman L. 2000. Pollen discounting and the evolution of selfing in Arenaria uniflora (caryophyllaceae). Evolution . 54: 1558– 1565. Google Scholar CrossRef Search ADS PubMed Fotheringham SA, Brunsdon C, Charlton M. 2002. Geographically weighted regression: the analysis of spatially varying relationships . Chichester: Wiley. Fournier-Level A, Korte A, Cooper MD, Nordborg M, Schmitt J, Wilczek AM. 2011. A map of local adaptation in Arabidopsis thaliana. Science . 334: 86– 89. Google Scholar CrossRef Search ADS PubMed Giesy JP, Dobson S, Solomon KR. 2000. Ecotoxicological risk assessment for Roundup® herbicide. In: Ware GW, editor. Reviews of Environmental Contamination and Toxicology, Volume 167 , New York: Springer, pp 35– 120. Google Scholar CrossRef Search ADS Glémin S, Bazin E, Charlesworth D. 2006. Impact of mating systems on patterns of sequence polymorphism in flowering plants. Proc Biol Sci . 273: 3011– 3019. Google Scholar CrossRef Search ADS PubMed González-Varo JP, Albaladejo RG, Aparicio A, Arroyo J. 2010. Linking genetic diversity, mating patterns and progeny performance in fragmented populations of a Mediterranean shrub. J Appl Ecol . 47: 1242– 1252. Google Scholar CrossRef Search ADS Goodwillie C, Kalisz S, Eckert CG. 2005. The evolutionary enigma of mixed mating systems in plants: occurrence, theoretical explanations, and empirical evidence. Annu Rev Ecol Evol Syst . 36: 47– 79. Google Scholar CrossRef Search ADS Haberl H, Erb KH, Krausmann F, Gaube V, Bondeau A, Plutzar C, Gingrich S, Lucht W, Fischer-Kowalski M. 2007. Quantifying and mapping the human appropriation of net primary production in earth’s terrestrial ecosystems. Proc Natl Acad Sci U S A . 104: 12942– 12947. Google Scholar CrossRef Search ADS PubMed Harder LD, Wilson WG. 1998. A clarification of pollen discounting and its joint effects with inbreeding depression on mating system evolution. Am Nat . 152: 684– 695. Google Scholar CrossRef Search ADS PubMed Hargreaves AL, Eckert CG. 2013. Evolution of dispersal and mating systems along geographic gradients: implications for shifting ranges. Funct Ecol . 28: 5– 21. Google Scholar CrossRef Search ADS Harrell FEJr. 2015. Regression modeling strategies: with applications to linear models, logistic and ordinal regression, and survival analysis . 2nd ed. Springer: New York. Harrell FEJr. 2017. rms: Regression Modeling Strategies R package. Version 5.1; [accessed 2017 September 15]. Available from: https://CRAN.R-project.org/package=rms. Hendry AP, Gotanda KM, Svensson EI. 2016. Human influences on evolution, and the ecological and societal consequences. Philos Trans R Soc Lond B Biol Sci . 372: 1– 13. Holtsford TP, Ellstrand NC. 1992. Genetic and environmental variation in floral traits affecting outcrossing rate in clarkia tembloriensis (onagraceae). Evolution . 46: 216– 225. Google Scholar CrossRef Search ADS PubMed Ivey CT, Carr DE. 2012. Tests for the joint evolution of mating system and drought escape in Mimulus. Ann Bot . 109: 583– 598. Google Scholar CrossRef Search ADS PubMed Johnston MO, Schoen DJ. 1996. Correlated evolution of self-fertilization and inbreeding depression: an experimental study of nine populations of amsinckia (boraginaceae). Evolution . 50: 1478– 1491. Google Scholar CrossRef Search ADS PubMed Kalisz S. 1989. Fitness consequences of mating system, seed weight, and emergence date in a winter annual, collinsia verna. Evolution . 43: 1263– 1272. Google Scholar CrossRef Search ADS PubMed Kalisz S, Vogler DW, Hanley KM. 2004. Context-dependent autonomous self-fertilization yields reproductive assurance and mixed mating. Nature . 430: 884– 887. Google Scholar CrossRef Search ADS PubMed Karron JD, Ivey CT, Mitchell RJ, Whitehead MR, Peakall R, Case AL. 2012. New perspectives on the evolution of plant mating systems. Ann Bot . 109: 493– 503. Google Scholar CrossRef Search ADS PubMed Kerr JT, Ostrovsky M. 2003. From space to species: ecological applications for remote sensing. TREE . 18: 299– 305. Koelling VA, Monnahan PJ, Kelly JK. 2012. A Bayesian method for the joint estimation of outcrossing rate and inbreeding depression. Heredity (Edinb) . 109: 393– 400. Google Scholar CrossRef Search ADS PubMed Kozak KH, Graham CH, Wiens JJ. 2008. Integrating GIS-based environmental data into evolutionary biology. Trends Ecol Evol . 23: 141– 148. Google Scholar CrossRef Search ADS PubMed Kuester A, Chang SM, Baucom RS. 2015. The geographic mosaic of herbicide resistance evolution in the common morning glory, Ipomoea purpurea: Evidence for resistance hotspots and low genetic differentiation across the landscape. Evol Appl . 8: 821– 833. Google Scholar CrossRef Search ADS PubMed Kuester A, Fall E, Chang SM, Baucom RS. 2017. Shifts in outcrossing rates and changes to floral traits are associated with the evolution of herbicide resistance in the common morning glory. Ecol Lett . 20: 41– 49. Google Scholar CrossRef Search ADS PubMed Kuester A, Wilson A, Chang SM, Baucom RS. 2016. A resurrection experiment finds evidence of both reduced genetic diversity and potential adaptive evolution in the agricultural weed Ipomoea purpurea. Mol Ecol . 25: 4508– 4520. Google Scholar CrossRef Search ADS PubMed Lande R, Schemske DW. 1985. The evolution of self-fertilization and inbreeding depression in plants. I. genetic models. Evolution . 39: 24– 40. Google Scholar CrossRef Search ADS PubMed Lankinen A, Armbruster WS, Antonsen L. 2007. Delayed stigma receptivity in Collinsia heterophylla (Plantaginaceae): genetic variation and adaptive significance in relation to pollen competition, delayed self-pollination, and mating-system evolution. Am J Bot . 94: 1183– 1192. Google Scholar CrossRef Search ADS PubMed Lankinen Å, Madjidian JA, Andersson S. 2016. Geographic variation in floral traits is associated with environmental and genetic differences among populations of the mixed mating species Collinsia heterophylla (Plantaginaceae). Botany . 95: 121– 138. Google Scholar CrossRef Search ADS Levin DA. 2010. Environment-enhanced self-fertilization: implications for niche shifts in adjacent populations. J Ecol . 98: 1276– 1283. Google Scholar CrossRef Search ADS Levin DA. 2012. Mating system shifts on the trailing edge. Ann Bot . 109: 613– 620. Google Scholar CrossRef Search ADS PubMed Mac Nally R. 2000. Regression and model-building in conservation biology, biogeography and ecology: the distinction between – and reconciliation of – ‘predictive’ and ‘explanatory’ models. Biodivers Conserv . 9: 655– 671. Google Scholar CrossRef Search ADS Mason CM, Christopher DA, Rea AM, Eserman LA, Pilote AJ, Batora NLet al. 2015. Low inbreeding depression and high plasticity under abiotic stress in the tall morningglory (Ipomoea purpurea). Weed Sci . 63: 864– 876. Google Scholar CrossRef Search ADS MORAN PA. 1950. Notes on continuous stochastic phenomena. Biometrika . 37: 17– 23. Google Scholar CrossRef Search ADS PubMed Neale DB, Adams WT. 1985. Allozyme and mating-system variation in balsam fir (Abies balsamea) across a continuous elevational transect. Can J Bot . 63: 2448– 2453. Google Scholar CrossRef Search ADS Olivieri I, Swan M, Gouyon PH. 1983. Reproductive system and colonizing strategy of two species of Carduus (compositae). Oecologia . 60: 114– 117. Google Scholar CrossRef Search ADS PubMed Opedal ØH, Bolstad GH, Hansen TF, Armbruster WS, Pélabon C. 2017. The evolvability of herkogamy: Quantifying the evolutionary potential of a composite trait. Evolution . 71: 1572– 1586. Google Scholar CrossRef Search ADS PubMed Parachnowitsch AL, Kessler A. 2010. Pollinators exert natural selection on flower size and floral display in Penstemon digitalis. New Phytol . 188: 393– 402. Google Scholar CrossRef Search ADS PubMed Paradis E. 2011. Analysis of phylogenetics and evolution with R , 2nd edn. Springer: New York. Peterson ML, Kay KM. 2015. Mating system plasticity promotes persistence and adaptation of colonizing populations of hermaphroditic angiosperms. Am Nat . 185: 28– 43. Google Scholar CrossRef Search ADS PubMed Rahmstorf S, Coumou D. 2011. Increase of extreme events in a warming world. Proc Natl Acad Sci U S A . 108: 17905– 17909. Google Scholar CrossRef Search ADS PubMed Rausher MD, Chang S. 1999. Stabilization of mixed-mating systems by differences in the magnitude of inbreeding depression for male and female fitness components. Am Nat . 154: 242– 248. Rausher MD, Fry JD. 1993. Effects of a locus affecting floral pigmentation in Ipomoea purpurea on female fitness components. Genetics . 134: 1237– 1247. Google Scholar PubMed R Core Team. 2017. R: A language and environment for statistical computing . Vienna: R Foundation for Statistical Computing. Version 3.3.3; [accessed 2017 September 15]. Available from: https://www.R-project.org/. Ritland K. 2002. Extensions of models for the estimation of mating systems using n independent loci. Heredity (Edinb) . 88: 221– 228. Google Scholar CrossRef Search ADS PubMed Ritland K, Jain S. 1981. A model for the estimation of outcrossing rate and gene frequencies using n independent loci. Heredity . 47: 35– 52. Google Scholar CrossRef Search ADS Routley MB, Mavraganis K, Eckert CG. 1999. Effect of population size on the mating system in a self-compatible, autogamous plant, aquilegia canadensis (Ranunculaceae). Heredity (Edinb) . 82 (Pt 5): 518– 528. Google Scholar CrossRef Search ADS PubMed Sagarin RD, Gaines SD, Gaylord B. 2006. Moving beyond assumptions to understand abundance distributions across the ranges of species. Trends Ecol Evol . 21: 524– 530. Google Scholar CrossRef Search ADS PubMed Scheper J, Holzschuh A, Kuussaari M, Potts SG, Rundlöf M, Smith HG, Kleijn D. 2013. Environmental factors driving the effectiveness of European agri-environmental measures in mitigating pollinator loss–a meta-analysis. Ecol Lett . 16: 912– 920. Google Scholar CrossRef Search ADS PubMed Schoen DJ, Brown AHD. 1991. Whole- and part-flower self-pollination in glycine clandestina and G. argyrea and the evolution of autogamy. Evolution . 45: 1651– 1664. Google Scholar CrossRef Search ADS PubMed Shea KL. 1987. Effects of population structure and cone production on outcrossing rates in engelmann spruce and subalpine fir. Evolution . 41: 124– 136. Google Scholar CrossRef Search ADS PubMed Sicard A, Lenhard M. 2011. The selfing syndrome: a model for studying the genetic and evolutionary basis of morphological adaptation in plants. Ann Bot . 107: 1433– 1443. Google Scholar CrossRef Search ADS PubMed Stone JL, VanWyk EJ, Hale JR. 2014. Transmission advantage favors selfing allele in experimental populations of self-incompatible Witheringia solanacea (solanaceae). Evolution . 68: 1845– 1855. Google Scholar CrossRef Search ADS PubMed Uyenoyama MK, Antonovics J. 1987. The evolutionary dynamics of mixed mating systems: on the adaptive value of selfing and biparental inbreeding. In: Bateson P, editor. Perspectives in Ethology . Springer, p 125– 152. Google Scholar CrossRef Search ADS Vallejo-Marín M, Barrett SC. 2009. Modification of flower architecture during early stages in the evolution of self-fertilization. Ann Bot . 103: 951– 962. Google Scholar CrossRef Search ADS PubMed Van Etten ML, Brunet J. 2013. The impact of global warming on floral traits that affect the selfing rate in a high-altitude plant. Int J Plant Sci . 174: 1099– 1108. Google Scholar CrossRef Search ADS Vitousek PM, Mooney HA, Lubchenco J, Melillo JM. 1997. Human domination of earth’s ecosystems. Science . 277: 494– 499. Google Scholar CrossRef Search ADS Waller DM. 1980. Environmental determinants of outcrossing in impatiens capensis (balsaminaceae). Evolution . 34: 747– 761. Google Scholar CrossRef Search ADS PubMed Wolters H, Jürgens G. 2009. Survival of the flexible: hormonal growth control and adaptation in plant development. Nat Rev Genet . 10: 305– 317. Google Scholar CrossRef Search ADS PubMed © The American Genetic Association 2017. All rights reserved. For permissions, please e-mail: firstname.lastname@example.org
Journal of Heredity – Oxford University Press
Published: Mar 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera