A Data-base for manual Pollen Recognition (Developed at Massey University)

Pollen Data-Base

This data base is designed to aid the identification of pollen by user selection of identifying features and and showing images of possible candidates for selection of the correct one. The data for that pollen type is displayed.

Palynology - The Study of Pollen

Pollen is technically termed “the multinucleate gametophyte generation of flowering plants” . It carries the male gametes, or sex cells, for fertilization of plants. Wind and insects are the main carriers of pollen with the grains themselves having characteristics that promote, usually one or the other of the two methods of transport. Birds and bats help out the insects in that task. Spores, included in studies under palynology, are “asexual reproductive bodies of lower vascular plants” and “algae, fungi, and mosses” . They are asexual in that they are themselves able to grow into an organism. Wind is an inefficient method of transport so, for wind dispersal, vast numbers of pollen or spores are produced: roughly one million per cone for pine and up to 500 million for a single marijuana plant shoot.

Palynology has numerous applications. Fossil pollen analysis (palaeopalynology) is used to identify the plant taxa, from which, can be deduced :

•  vegetation variations with time

•  climate and its temporal variation

•  evidence of human activities including

•  land clearing

•  burning

•  atmospheric pollution (also natural, e.g. volcanic)

•  salinity

•  soil degradation and changes

•  archaeological information

•  dating of sediment levels and ages of artefacts found

•  what people may have eaten

•  what was buried with them – flowers etc.

•  oil deposit locations

Honey type, and location of origin, can be indicated by the pollen found in the honey (melissopalynology). Inhalant allergy sufferers can be advised of high pollen counts in the air (Aeropalynology) . Forensic investigations can be aided by determining if an object has been in a certain general location by identifying the pollen grains attached.

The layers making up a pollen grain wall are labelled in Figure 2 - 1 .

Figure 2 - 1 : A section of pollen wall showing structure and some features.

The taxa, or type, of pollen can be determined by pollen morphology, or structure. Large data bases of images and diagrams are being built so pollen can be identified by referring to them. Pores, (holes), culpi, (furrows), and the numbers of them are clues to pollen taxa.

The exine ( Figure 2 - 1 ) is made of sporopollenin. Sporopollenin is a substance composed of oxidative copolymers of carotoid and carotenoid esters. It is an extremely durable substance and can be found in anaerobic sediments dating back hundreds of millions of years. A pore is shown in the diagram but these may be elongated to a furrow and are called colpi. The number of pores or culpi, and their arrangement on the sphere surface, is a strong indication of type. Surface features are used for identification using mathematical feature extraction in AutoStage. Some of the surface features can be seen in the scanning electron microscope images of pollen in Figure 2 - 2 .

Figure 2 - 2 : Golden Rod (echinate), Oak pollen(colpi) and Birch pollen (pores). Scanning Electron Microscope (SEM) images of pollen from “ National Pollen and Aerobiology Research Unit” web site .

Modern “pollen rain”, or the spatial distribution of pollen, is modelled to determine the vegetation taxa and climatic conditions that would have caused the pollen distribution found in sediment core samples. Sample cores are extracted from lake beds, swamps etc., sliced horizontally to divide into samples of sequential depths, and analysed to make stratigraphic maps of the region. Carbon dating and other techniques are used to compare depth relationships to dates of deposits. The thickness of each core slice taken, determines the temporal resolution. Fine resolution, down to one year representing perhaps a millimetre of core depth, is often required but requires the preparation of hundreds of samples and many slides of pollen to prepare and count. A greater number of cores, taken in close proximity, will increase the reliability of the data and add information regarding localised variation. Regionalised layer variation is obtained by multiple core sampling at larger spacings. The counting of pollen from core samples requires experts to identify the pollen taxa as they count pollen grains on a slide. This is very time consuming and laborious work for trained people whose time could well be spent on less mundane tasks. This is one reason that AutoStage is being developed.

The AutoStage will also aid pollen counting by aeropalynologists who advise of pollen counts in the air for benefit of inhalant allergy and asthma sufferers. At present there are a number of pollen counting stations, mainly in Europe and North America . Many have volunteer based counting , where volunteers spend 2-3 hours per day, 3 days a week counting pollen captured in pollen traps.

A PhD student who uses Palynology as a prime research tool may spend up to 30 months of microscope work counting pollen. Any study using palynology would be advanced considerably if the counting were performed automatically and the time taken to collect research data reduced consequentially.

Preparation of samples is also time consuming and important to the outcome of the counting process. Fossil samples are treated to remove as much foreign materials as possible by sieving and chemical treatments. Density gradient centrifugation may have a part to play in the classification of pollen but at least it appears that it will be helpful in separating pollen from organic matter not removed by chemicals and sieving. The problem of organic matter remaining is applicable to automatic pollen counting as the organic matter, as seen in the images in Figure 6 - 2 , reduces segmentation effectiveness. Ultrasonic filtration, is an advance that may prove useful. Preparation is a separate part of the process but important to the aim of this project in two ways:

•  preparation involves laborious, time consuming work

•  preparation quality can influence the AutoStage processes and the reliability of the results.