Predicting orchid species richness in New Guinea
André Schuiteman, Research Leader in Identification and Naming, explains how we can predict the species richness of unexplored areas.
New Guinea: an orchid hotspot
Just south of the equator and north of Australia lies a huge tropical island, shaped like a legless bird: New Guinea. It is around 2,500 km long from west to east, with a central mountain range running almost throughout its length – its highest peak reaches nearly 4,900 m above sea level. To this day, most of the island is underexplored botanically, and large areas are even unexplored altogether.
The climate is predominantly ever-wet tropical, with most places receiving 2,500–4,500 mm of rain annually. Some areas however, especially in the southern lowlands, experience a pronounced dry season.
The flora of New Guinea is extremely rich, but just how rich is difficult to determine. Estimates range from 11,000 to 25,000 plant species.
As an orchid researcher I am perhaps biased when I say that the orchids take pride of place, but it is certainly by far the largest plant family in New Guinea, with an estimated 2,800 species. Of these, about 90% are endemic, that is, they are only found in New Guinea.
Just how limited our knowledge of these plants is becomes clear when we count the number of collecting records (herbarium collections) for each species. It turns out that more than half of the species are only known from one or two collections.
When we plot the known collecting localities of all orchids on a map, it is obvious that the dots are by no means evenly distributed across the island. In fact, they are strongly clustered in certain areas, and along roads and rivers.
We can also see that the eastern half of the island (Papua New Guinea) has far more dots than the western half (Indonesian New Guinea). The absence of a dot could mean the complete lack of orchids, but in reality most of the absences are simply there because nobody has looked for orchids in those places. These blank areas literally represent the gaps in our knowledge of orchid (and plant) distributions in New Guinea.
The situation is even worse than it looks at first glance. This is because each dot can represent anything from a single species record to more than a hundred different records collected from a given locality. Very few localities in New Guinea are so well sampled that we can be sure that all of the orchid species have already been found there.
In short, we are dealing with an enormous collectors' bias (a highly non-random sampling), which makes it difficult to say how rich in orchid species the various parts of New Guinea really are. It is evident to anyone who has been there (a privilege I have had four times so far) that some parts of New Guinea are far richer in orchids than others. Is it possible to produce a more useful map, which tells us how many species of orchid can be found at each locality?
Unfortunately, to do so we would need to possess full inventories of every locality or grid cell, depending on the desired resolution, throughout the island. It would take an army of botanists, years of fieldwork, and lots of money to carry out such an enormous project. In practice, there are only a handful of qualified botanists, who are chronically underfunded and short of time. A full island-wide inventory is not going to happen any time soon.
Niche modelling to the rescue
How can we possibly tell how many orchid species there are in an area where nobody has ever set foot, let alone collected orchids? Here is where the recently developed techniques of ecological niche modelling can help.
For any given species we need at least five collecting localities (but the more the better). We also need a list of environmental variables for which we know (or can estimate) the values at each locality. These variables could include: the amount of rainfall in the driest quarter, the annual temperature range, the mean temperature in the driest quarter, the pH of the soil, etc. Using these values we can try to predict (using complicated statistical methods) under which conditions the species might occur in nature. This set of conditions is an approximation (a model) of the ecological niche of the species.
Next, we need to check each locality in the study area (in this case New Guinea) to determine if the conditions are inside or outside the approximated ecological niche of the species. If they are inside, then we will assume that the species occurs there (perhaps with a certain probability only); if outside, we will assume that the species is absent. If we do this for not just one species, but for hundreds, we can easily produce a map which tells us exactly how many species (among those that we modelled) are predicted to occur at every single locality, even in unexplored areas, assuming that conditions in these areas are similar to those in better sampled ones.
In this way we can create a map of the estimated species richness for every locality (or grid cell) in the area of interest. Vollering et al. (2016) have done this for the orchids of New Guinea and obtained a map which does indeed show that some parts of the island are far richer in orchids than others. There is, for example, a rather unexpected and unexplained area of lower diversity in the middle of the central mountain range.
Because this is only a prediction, we still need botanists to go in and verify these results ('ground truthing'). Besides, there are undoubtedly hundreds of undescribed orchid species waiting to be discovered in underexplored areas. Further exploration is still urgently needed, before deforestation and climate change force species to go extinct that haven't even been described. By using maps based on ecological niche modelling we can target areas that are both under-collected and thought to be exceptionally rich. Who knows what discoveries are still awaiting us there?
Marshall, A.J. & Beehler, B.M. (2007). The Ecology of Papua (vol 1 & 2). Periplus, Hong Kong.
Vollering, J., Schuiteman, A., de Vogel, E., van Vugt, R. & Raes, N. (2016). Phytogeography of New Guinean orchids: patterns of species richness and turnover. Journal of Biogeography 43: 204-214. doi:10.1111/jbi.12612. Available online.