MS data was obtained using the positive scan mode for all of the extracts considered in the study. Compared to the underivatised UV-VIS and DPPH‚ chromatograms, the positive scan mode MS chromatograms show very different profiles . The cinnamon myrtle chromatograms are dominated by a peak that elutes with a retention time of 11 min. This peak does not appear in either the underivatised UV-VIS or DPPH‚ chromatograms. A small number of secondary peaks occur in both chromatograms, although they too do not appear to correspond to any major peak in either the underivatised UV-VIS or DPPH‚ chromatograms. Unlike the cinnamon myrtle, the negative scan MS chromatograms of the lemon myrtle are not dominated by a single peak, but show a number of peaks with similar intensity. Additionally, some of these peaks match in retention time to peaks that appear in the underivatised UV-VIS and DPPH‚ chromatograms. In particular, peaks with retention times of 3 and 4.5 min appear in a similar area of the chromatogram to peaks that respond to both underivatised UV-VIS and DPPH‚ . Additionally, the large peaks at around 12 min in the underivatised UV-VIS chromatograms appear as small peaks in the MS negative scan. As the number of peaks identified in the MS in positive scan mode was smaller than expected, growing raspberries in container the multiplexing experiment was repeated for the water extracts using both negative and positive MS scan modes. The water extracts were chosen as they showed the greatest number and intensity of antioxidant peaks in the DPPH‚ chromatograms. Both cinnamon myrtle and lemon myrtle showed a greater abundance of peaks in negative scan mode compared to positive scan mode.

Additionally, both chromatograms showed a large number of peaks eluting within the first 5 min of the chromatogram where the majority of the compounds that gave a response to DPPH‚ eluted. Table 1 shows the peaks that were detected in the DPPH‚ chromatograms along with the masses of those peaks as determined in the MS scans and possible identification based on the MS data. Due to the non-specificity of the MS scan, a number of the peaks that were identified in the MS data were due to more than one major m/z value indicating the presence of two co-eluting species being detected. Furthermore, a number of peaks that were detected in the DPPH‚ chromatograms did not show peaks in the MS data, indicating that these species did not ionise in the MS conditions used in the method. Additionally, it can be seen that a number of peaks that eluted had a m/z ratio that is either lower than typically observed in antioxidants, such as the peak at 0.5 min in the cinnamon myrtle water extract, or higher than that of typical antioxidants, such as the peaks at 3.3 and 4.6 min in the lemon myrtle water extract . This indicates that two different peaks may be eluting at these retention times, one of which is observed in the DPPH‚ chromatogram and the other that is observed in the MS chromatogram. Finally, it can be seen that where MS peaks were evident in both positive and negative scan modes, most of the peaks observed had very different m/z ratios in each mode, indicating the presence of co-eluting species. From the MS data it is possible to perform some investigation into the identification of the antioxidants that were present in the extracts. The m/z ratio of each of the peaks identified in the MS chromatograms was compared to the molecular masses of known antioxidants. If a match between the m/z ratio of the peak and the molecular mass of one or more antioxidants was found, this was considered a possible identification for that peak. For example, both lemon myrtle extracts show peaks with a m/z ratio of 139 Da indicating that the peak may be due to hydroxybenzoic acids.

However, positive identification is impossible without additional information such as MS/MS data and/or the comparison of the peaks with standard solutions. Due to the non-specific nature of the MS data that was collected in the study, only preliminary identification of the peaks could be performed. Thus a number of peaks could either not be identified or were identified as one of a number of possible antioxidants.In recent decades, advances in the final frost dates of winter or early spring have been observed throughout North America while advances in the timing of flowering have been documented in many angiosperm taxa . In response to recent climate warming, the flowering times of many species have changed, which may alter the risk of reproductive structures being exposed to spring frosts . Exposure of reproductive tissues to frost is hazardous for many plant species, as floral tissues are often the most vulnerable to frost damage, and the exposure of floral tissues to frost or freeze events can reduce pollen and seed production or result in reproductive failure . Over multiple generations, reductions in reproductive success due to increases in frost exposure may lead to progressive declines in local abundance, potentially resulting in local extirpation . Accordingly, the ability to initiate and to complete flowering and fruiting without exposure to frost or freeze events plays a major role in determining the geographic range of many species . Previous studies have predicted that progressive warming could increase the risk of frost damage to floral tissues for many species if,in response to warming, flowering times advance more rapidly than the date of last frost, defined as the date that marks the beginning of that portion of each year during which daily minimum temperatures remain above 0°C . This pattern has been particularly well documented among shrub and forb species whose flowering time is primarily driven by snowmelt , resulting in reductions to annual flower and seed production . Conversely, warming climates may advance the date of last frost more rapidly than plant species advance their flowering times, thereby reducing their risk of frost exposure ; this pattern has been detected among 14 European angiosperm species .

Warming conditions may also delay bud break and flowering of those taxa that require an extended period of winter chilling to break dormancy, protecting them from flowering prior to the onset of the frost-free period . While the phenological responses of flowering time to climate warming have been measured in thousands of species , and broad-scale temporal reductions in frost risk to developing leaves have been detected among North American trees , no large-scale examinations of shifts in frost risk have yet been conducted on a sufficient array of taxa to detect or to characterize general trends in a continental flora. As a result, the general effects of recent climate change on the risk of frost exposure to floral tissues remain largely unknown. Additionally, flowering phenology has previously been documented to be evolutionarily conserved among co-occurring taxa that are closely related . Given that exposure to frost depends on a species’ phenology at a given location, it is also possible that frost risk is phylogenetically conserved. However, no systematic examination of the degree to which frost risk is phylogenetically conserved among closely related taxa has yet been conducted. To address these gaps, we conducted the first continent-wide assessment of frost risk by evaluating the flowering times of 1,653 species collected in flower from 1920 to 2015 and represented by 475,694 digital records of herbarium specimens collected throughout North America, with specimens primarily concentrated in the Western and Eastern United States. By comparing rates of temporal changes in dates of last frost experienced by each species among the sites where it was sampled to rates of temporal changes in flowering date from 1920 to 2015, we determined that, for most species, the advancement of the last frost date has outpaced the advancement of flowering date, resulting in a reduction in the risk of floral exposure to frost. Furthermore, this pattern persisted across regions that historically experienced both early and late dates of last frost. We also conducted a phylogenetically informed analysis to determine whether,as has been found for flowering time itself , the risk of exposure to frost exhibits a phylogenetic signal. Finally, we compared the degree of frost risk experienced by native versus exotic species, and evaluated whether the relatively low risk exhibited by the latter is due to differences in the mean climate conditions they occupy or to differences between natives and exotics in the degree of phenological change that they exhibited.Phenological data pertaining to flowering times in this study consisted of 475,694 specimen records of angiosperm species collected in flower. These data were derived through filtering of a larger dataset consisting of 894,392 specimen records accessed from the digital archives of 72 herbaria , and cleaned using several criteria described below.

Estimates of mean flowering date from herbarium specimens have been reported to provide accurate estimates of species’ flowering times and have yielded estimates of phenological change similar to those derived from in situ observations of living plants across both temporal and spatial climate gradients . To ensure the quality of the data used in this study, large plastic pots for plants specimens were included in the dataset analyzed here only if, at the time of digitization, herbarium personnel had: verified that the specimens were collected when in flower; recorded GPS coordinates of the location from which the specimen was collected; and provided the precise date of collection . Only those specimens that were explicitly recorded as being in flower within either the DarwinCore “reproductivecondition” or “lifestage” fields of their source’s database were included in this study. Specimens that were listed only as “buds present” or “fruiting” were not considered to be in flower for purposes of this analysis, as some perennial species collected during the winter may be described as “buds present” when buds are completely dormant, or may retain aborted or unripe fruits that cannot be distinguished from recently matured fruits preserved on herbarium specimens. The taxonomic nomenclature used to identify all specimens, which sometimes changed over time or differed among collectors, was standardized according to The Plant List and TROPICOS using the Taxonomic Name Resolution Service iPlant Collaborative, Version 4.0 and subsequently filtered to eliminate all taxa not identified to species level within the megaphylogeny used by the PhyloMaker package in R , which similarly used a standardized taxonomy derived from TPL and TROPICOS . To avoid pseudoreplication, duplicate specimens were also removed. The resulting dataset included 475,694 specimens representing 1,653 species distributed throughout North America . In this study, we calculated the frost risk of each sampled species using annual estimates of the date of last frost at each collection site obtained from ClimateNA version 5.5.1. Frost risk of each species was defined as the proportion of its specimens collected in flower before the date of last frost in the years and locations in which they were collected. Frost risk in this context does not invariably predict the risk of reproductive damage, which depends not only on species- and population-specific cold tolerances, which are undocumented for most taxa , but also on microclimate conditions that cannot be easily incorporated into continental-scale datasets, such as humidity, wind speed, and recent precipitation . Nevertheless, temperatures of 0°C have been documented to damage floral tissues ofa wide variety of species , as radiative cooling often results in damage to floral tissues and emerging leaves under nighttime temperatures of 0°C even in species that otherwise remain hardy to subzero temperatures . Thus, frost risk is used here as a standardized metric indicating the likelihood of exposure of floral tissues to frost or freeze events.To estimate historical frost risk for each species, we calculated the proportion of specimens of each species collected from 1920 to 1979 that were collected prior to the date of last frost at the site and year of their collection. To estimate recent frost risk, we similarly calculated the proportion of specimens of each species collected from 1980 to 2015 that were collected prior to the date of last frost at the site and year of their collection . To ensure that a sufficient number of observations of each species were available to produce meaningful estimates of frost risk within both periods, we eliminated all species that were not represented by at least 50 specimens both prior to the year 1980 and after the year 1979.