RESEARCH STATEMENT

I am interested in a wide variety of questions, both theoretical and empirical, relating to biodiversity and conservation biology. Most of my research focuses on questions relating to species detectability and extinction, and orchid ecology; in particular the response of orchids to climate change, epiphyte community ecology and modelling epiphyte seed dispersal. In pursuit of these goals, I have travelled widely, focusing on study systems in the western Indian Ocean islands and Africa. Although I am primarily an "orchid" person, I have worked on a diverse range of taxa including, ants, the dodo, mammoths and the North Atlantic Right Whale.

  1. Extinction & Biodiversity

    Following the 2002 World Summit in Johannesburg, the Convention of Biological Diversity has called for a decrease in the rate of biodiversity loss by 2010 (www.biodiv.org/2010-target). However, a 2003 UK Royal Society report on "Measuring Biodiversity for Conservation" discussed the unavailability of satisfactory measures of biodiversity, and the difficulty of reporting accurately on the loss of biodiversity by 2010. Related to this problem is the growing demand for information such as those held in biological collections to be provided over the internet (e.g. www.gbif.org). It is estimated that there are around 2.5 billion specimens in biological collections. These records represent primary, verifiable observations on the distribution of taxa through time and space. However, it is the temporal element of such collections that is of particular importance.

    1. Extinction models & their application

      Estimating the probability that a species is extinct and the timing of extinctions is of fundamental importance in a range of biological fields including paleoecology, epidemiology, conservation biology and invasive species eradication. According to the IUCN Red List criteria, a species is considered Extinct "when there is no reasonable doubt that the last individual has died" (IUCN 2001). In these guidelines, the IUCN go on to say that "a taxon is presumed Extinct when exhaustive surveys in known and/or expected habitat... throughout its historic range have failed to record an individual." Even for species that attract considerable attention, such as those described here, this level of certainty is difficult to achieve because any statement of extinction is probabilistic and because extremely intensive survey efforts are required for a high confidence of extinction from an area. Conservation priorities are currently guided more by raw estimates of diversity than by any measure of how quickly that diversity is being lost. Estimating the probability that a species is extinct and the timing of those extinctions is of fundamental importance in setting priorities to manage biodiversity loss.

      Much of my work in this field is aimed at the development of statistically rigorous methods for understand extinction process. Several methods have been developed to provide a probabilistic basis for an extinction hypothesis based on a sighting record (see Publication and Downloads). This area of research has many applications from the study of endangered species to epidemiology. With my collaborators, we are testing these models through the development of large datasets to test the various models using L-moment diagrams and probability plot correlation coefficient hypothesis tests to evaluate their goodness-of-fit (Vogel et al. 2009). In addition there is a need to develop Bayesian methods for inference about extinction, as they are especially powerful when external information is available about extinction time or sighting rate. Expanding the application of these extinction models on a spatial dimension, I am starting to investigate species decline and extinction for the identification of extinction hotspots.

      In recent years a number of species have been rediscovered after being presumed extinct (Lazarus taxa). This has implications for conservation biology, (1) we can loose public confidence by 'crying wolf' and (2) there is the possibility of committing Romeo's Error. I have built a dataset of such taxa with the aim of understanding the process that lead to the assumption of extinction. This is based on how surprising or exceptional is a new record and whether it could have arisen from the same process which created the previous sighting record.

    2. Species detectability

      Any attempt to use biological collections to draw inferences about species needs to have an understanding of the collection process. For example, the relative paucity of specimens of some taxa may be related to their time since discovery and the time for specimens to accumulate in collections. When sampling effort is consistently lower for recently identified taxa, there will be a tendency to underestimate their status such as size ranges. This begs the question as to why taxa are discovered when they are and whether conservation and biodiversity prioritisation may reflect a level of conspicuousness and accumulation of knowledge? From this, perhaps even more importantly is whether what we are collecting is representative of biodiversity. This is particularly important given the time and money that is currently being spent on 'rapid biodiversity assessment'. I am interested in the following questions, (a) Why do we discover species when we do? (b) Do certain morphological, geographical and ecological characters (conspicuousness) increase the probability of a species being discovered or record? (c) How does this affect the species conservation status through the accumulation of data over time since data of discovery?

      I have recently developed a test for the preferential discovery of species that are morphologically dissimilar from those described previously (Novelty bias). Novelty bias has been found this to occur in certain orchid genera, however we have found only one case of preferential discovery of species that are morphologically similar to those described previously (Familiarity bias). The difference between discovery at random vs novelty may relate to the level of morphological complexity. This also relates to the persistence of certain taxa in the fossil record and our ability to discriminate between species (i.e. Elvis taxa).

  2. Orchid Evolutionary Ecology

    Three years after the publication of the On the origin of species..., Darwin (1862) published On the various contrivances by which British and foreign orchids are fertilised by insects... in support of his theory of evolution through natural selection. Since Darwin's initial observations, most orchid research has focussed on their pollination biology, predominately in temperate regions. However, the movement of pollinia is only one stage in the population dynamics of an orchid. Seed dispersal is an extremely important process, both in terms of speciation and conservation. Darwin (1862) speculated that "the great grandchildren of a single plant [of Orchis maculata] would nearly... clothe with one uniform green carpet the entire surface of the land throughout the globe". However, what puzzled him was "What checks this unlimited multiplication cannot be told. The minute seeds within their light coats are well fitted for wide dissemination...". This perception of their immense dispersability still continues with virtually no quantitative analysis or data. It has recently been suggested that orchids, with their small effective population size (Ne), potential evolved through initial genetic drift followed by subsequent diversification and speciation through Darwinian adaptation to the local pollinator population. However, we still have little knowledge of the dispersal distance of orchid seeds, and its consequence on their evolutionary ecology and disperability in fragmented landscapes.

    The aim of my research is to investigate the dispersability orchid species and subsequent recruitment, particularly in epiphytes. With Prof Gil Bohrer from Ohio State University, I am studying the effects of canopy heterogeneity, seed release height and inertia on wind dispersal kernels of epiphytic orchid seeds. This uses the recently developed Regional Atmospheric Modelling System (RAMS)-based Forest Large Eddy Simulation (RAFLES) by Gil. Further, simulations of seed dispersion of epiphytes, drag coefficient and wake-kinetic-energy generation of different tree species are essential parameters. These parameters are typically assigned arbitrarily due to the lack of observations. With Dr Bharathram Ganapathisubramani at the Dept. of Aeronautics, Imperial College, we are conducting detailed experimental observation of these two parameters in wind tunnels. Hot-wire measurements performed in the wake of cylinders with different bark surface roughness. This will provide average statistics such as drag and kinetic energy, and enable us to understand the unsteady flow effects that impact seed dispersion.