New Insights from Old Herbarium Specimens
Contributed by Richard B. Primack (Boston University; primack@bu.edu) and Charles G. Willis (Harvard University; charleswillis@fas.harvard.edu)
“Live in each season as it passes - breathe the air, drink the drink, taste the fruit & resign yourself to the influence of each.” Thoreau, in his Journal. 1835
Millions of herbarium specimens are being digitized every year and made available online. In the USA, this effort is being spear-headed by the iDigBio program, and involves hundreds of institutions around the country. Similar initiatives have also been implemented in Europe, China, Australia, and elsewhere. As a result, scientists now have instant access to datasets of plant distributions, morphology, and phenological behavior that were once unimaginable in terms of their scale. These massive digital datasets allow scientists to carry out ecological, evolutionary, and climate change research far more quickly and easily than ever before. Furthermore, the sheer scale of these data make it possible for scientists to pursue new and exciting research projects that could fundamentally reshape our understanding of the pattern—as well as the future—of plant and animal diversity on Earth.
In three recent papers, many scientists, including ourselves, have shown some of the opportunities and limitations of research using herbarium specimens.
Willis, C.G., E.R. Ellwood, R.B. Primack, C.C. Davis, K.D. Pearson, A.S. Gallinat, J.M. Yost, G. Nelson, S.J. Mazer, N.L. Rossington, T.H. Sparks, P.S. Soltis. 2017. Old plants, new tricks: Phenological research using herbarium specimens. Trends in Ecology and Evolution 32: 531-546.
Botanists have always known that herbarium specimens can be a valuable source of information on when plants flower and mature their fruits. For instance, information included in a regional flora on what months you might find a certain species in flower is often synthesized from herbarium specimens. These specimens, however, are taking on new importance in terms of providing a detailed record of how climate change is altering plant phenology. In their review article, Willis et al. (2017) summarize the current state of this nascent, but important field. They detail the range of climate change projects that have used herbarium specimens to study phenology. They also describe many of the peculiarities of herbarium specimen collecting that might create biases in the data, such as the tendency for many collectors to collect specimens in peak flower rather at the start or end of the flowering period. They also highlight the advantages of using herbarium specimens to study phenology and climate change in contrast with long-term field studies. In two case studies, they demonstrate that herbarium specimens typically cover a broader geographic area and sample from a wider variety of climatic situations than field studies, and that herbarium specimens can be used to detect the effects of climate change on a wider range of phenological events, like the leafing out times of trees. The review also underscores the promise of integrating herbarium specimen data of flowering dates with other historical datasets of flowering times, such as field observations and dated photographs, to best characterize the impacts of climate change on plant phenology.
Figure 1. Phenology studies using herbarium specimens (blue-gray dots) are increasing in number and their geographical distribution, and this process is being facilitated in regions of the world with a high density of digitized herbarium specimens (yellow-to-red dots) (From Willis et al. 2017; Figure 1).
Pearse, W.D., C.C. Davis, D. Inouye, R.B. Primack, T.J. Davies. 2017. Measuring the limits of phenology: estimating synchrony and variation in contemporary, historic, and citizen-science datasets. Nature Ecology and Evolution 1, 1876–1882
Investigations of the impacts of climate change often depend on comparing past phenological observations with modern ones. However, field observations of flowering times are often based on when the first open flower of a species is observed, whereas herbarium specimens are often collected when plants are in peak flower. Until recently, there was no good method for combining these two different types of phenological data, that is first flowering dates and peak flowering dates. In this recent paper, Pearse et al. (2017) demonstrate a new, robust statistical method that makes it possible to estimate historical first flowering dates from peak flowering dates taken from herbarium specimens. The approach proposed in the paper was, surprisingly, based on an approach to estimate extinction dates in rare animals, like the dodo bird. To test the validity of this approach, they estimated historical first flowering dates from herbarium specimens collected from around Massachusetts and then compared these estimated first flowering dates to actual observations from Concord, MA made by the famous environmental philosopher Henry David Thoreau and the botanist Alfred Hosmer in Concord in the 19th century. Strikingly, they found that estimated first flowering dates from herbarium specimens were comparable to the first flowering dates observed by Thoreau and Hosmer. Pearse and colleagues also demonstrate that their approach can be applied to large citizen science datasets like those gathered by the National Phenology Network and the extensive monitoring of flowering times records over decades at the Rocky Mountain Biological Lab. This new statistical method opens the way for an expanded role for herbarium specimens in climate change research and is an example of a creative way to combine diverse types of phenological data.
Figure 2. For each of seven orchid species in Massachusetts, the estimated date of first flowering (red dot) was calculated using the dates of herbarium specimens collected in full flower (black tick marks) and compared to the earliest first flowering date observed in the field by Thoreau and Hosmer (blue dot). (From Pearse et al. 2017; Figure 3, Credit: images of plant species Pogonia ophioglossoides, Platanthera lacera, Cypripedium acaule, Corallorhiza maculata and Arethusa bulbosa, Steven J. Baxter; Platanthera psycodes, Rob Routledge/Sault College; Platanthera grandifolia, Arnold T. Drooz/USDA Forest Service).
Daru, B.H., D.S. Park, R.B. Primack, C.G. Willis, D.S. Barrington, T.J.S. Whitfeld, T.G. Seidler, P.W. Sweeney, D.R. Foster, A.M. Ellison, C.C. Davis. 2017. Widespread sampling biases in herbaria revealed from large-scale digitization. New Phytologist, doi: 10.1111/nph.14855
Herbarium collections often reflect the preferences and habits of their collectors. This is well known to botanists that work in museums, but is less appreciated by others. If ecologists, climate change biologists, and other researchers do not appreciate the biases of collections, this could lead to mistaken conclusions when using herbarium specimens for research. In this study, Daru and his colleagues examined over 5 million digitized herbarium specimens from South Africa, Australia, and New England and determined that there were significant and consistent geographical, seasonal, and phylogenetic biases to the collections. For example, specimens were more frequently collected closer to roads and herbaria than other places, threatened species were collected less than other species, and collection efforts were disproportionately focused in some closely related groups (i.e., clades) more than others. Also, a high percentage of specimens were collected by a small number of very active collectors, likely biasing collections toward the groups of plants these collectors preferred. Future studies using herbarium specimens for novel research projects, especially focused on diversity, distribution, and comparative evolution, will need to take these biases into account in order to avoid erroneous conclusions.
Figure 3. There is a strong geographical bias in collecting in Australia, South Africa and New England, with a greater density of collections nearer to herbaria, roads, and urban centers, and with smaller, redder triangles expressing greater sampling effort and larger, blue triangles representing less sampling effort. (From Daru et al. 2017; Figure 1a, b, c.).
As these three papers demonstrate, there are many exciting new opportunities for ecological, evolutionary, and climate change studies using herbarium specimens, and the potential of these studies will greatly expand in coming years as millions more digitized herbarium specimens become available online.