Herbarium Biases

Humans carry countless biases with them wherever they go. Even the logical mind of a scientist is no stranger to prejudice. Identifying such biases in the way we do science is key to improving the discipline and, as computing power and access to big data increases, we are gaining a better understanding of just how prevalent our biases really are. A recent study that looked at herbarium collections around the world aims to do just that.

With herbaria closing shop around the globe, the need to digitize collections has never been more urgent. Although more and more collections are finding their way into digital libraries, a vast majority of herbarium collections risk being lost forever. This alone represents a major bias. Such organismal science has sadly been scoffed at in recent decades. Still, enough collections have been entered into databases that interesting patterns are starting to emerge. A team of researchers recently took a closer look at roughly 5 million digitized floras representing the most complete digital floras from Australia, South Africa, and New England.

In doing so, the team was able to find some startling biases in these collections. They broke them down into a handful of categories with the hope that botanists and ecologists can start to improve on these gaps over the coming decades. Although the floras they examined by no means represent anything close to a complete picture of our floristic understanding of the world, they nonetheless mirror issues that are sure to crop up no matter where collections have been made.

The first major category is that of spatial or geographic bias. This occurs whenever specimens are collected at a higher frequency in one place over another. There are likely many reasons for this - ease of access, proximity to research institutions, just to name a few. The team found that herbarium collections tended to occur in the same areas through time. What's more, they tended to occur more often near roads with a surprising 50% of specimens collected within 2 km of a roadside. This can result in a highly skewed perspective of the kind of taxa represented in a region. Roadside vegetation is comprised of species capable of dealing with runoff, soil compaction, and pollution, and is likely depauperate of taxa less able to handle such conditions. They also found a elevational bias, with a majority of specimens having been collected below 500 meters. 

Maps demonstration spatial biases in herbarium collections. Those in red have more collections and those in blue have fewer collections.

Maps demonstration spatial biases in herbarium collections. Those in red have more collections and those in blue have fewer collections.

The second major category is that of temporal bias. This occurs whenever specimens are collected more frequently during certain parts of the year over others. The team found that collections disproportionately occurred during spring and summer months. As anyone who hikes can tell you, there is a lot of variation among plant communities from season to season and any good collection should sample a location multiple times a year. In addition to seasonal biases, the team also found extreme biases in terms of history. Collections in South Africa and Australia started to rise shortly after World War II and peaked in the 1980's and 1990's respectively. Compare this to New England where peak collections occurred nearly 100 years prior. If we are to track long term trends and changes in the flora of various regions, collections need to occur far more regularly. Obviously institutions have shied away from such investigations in recent decades. Only public interest and funding can reverse such trends, hopefully not before it is too late.

The third major bias they found is that of trait bias. This occurs whenever a collector specifically aims for species with a certain life history characteristic (annual vs. perennial, woody vs. herbacious) as well as species of conservation concern. Indeed, the team found that perennial species were over-represented in most herbarium collections. Also, gramminoids dominated herbarium collections in Australia and South Africa whereas herbs and trees were over-represented in New England. Another interesting pattern that emerged is that short plants had higher representation in harbaria than taller species. Obviously this has a lot to do with ease of collection.

Another pattern that emerged which is of conservation concern is that threatened or endangered species are severely under-represented in herbarium collections. Although care must be taken to not over-collect species whose numbers are dwindling, their lack of representation in herbarium collections can seriously hinder conservation efforts. Such under-represenation can lead to erroneous estimations of species abundances and distributions. It can also hinder our understanding of plant community dynamics.

The fourth major bias is that of phylogenetic bias. Certain clades are more sought after than others. This leads to a disproportionate amount of showy or valuable species turning up in herbaria around the globe. It also leads to an over-representation of potentially "useful" plant species in terms of things like medicines or dyes. This leaves a large portion of regional floras under-sampled. This in turn exacerbates issues relating to our understanding of plant community dynamics and the change in plant abundance and distribution through time.

Finally, the fifth major bias is that of collector bias. This pattern stems from the fact that in all the regions sampled for this study, a majority of the collections were made by only a handful of individuals. This means that all of these collections are the products of the habits and preferences of these collectors. Some collectors may favor sampling the entire flora of a region whereas others may favor certain clades over others. Similarly, some collectors may favor plants with interesting physiologies whereas other may favor plants with peculiar life-histories such as carnivores or succulents.

The use and importance of herbaria has changed a lot over the last two centuries. Whereas they largely started out as a tool for taxonomists, the utility of herbarium collections has since expanded into areas that were never thought possible. With the advent of new technologies, who knows what the future holds. Of course, this means nothing if interest and support for herbarium collections continues to decline. Their utility in the context of research and conservation cannot be understated. We need herbaria now more than ever. Understanding biases is a great step towards improving the discipline. We must aim to improve collections in these so-called cold spots and to avoid as many biases as possible in doing so.

Photo Credits: Wikimedia Commons

Further Reading: [1]