Katherine A. Holt

Pollen Image Classification Using the Classifynder System: Algorithm Comparison and a Case Study on New Zealand Honey

We describe an investigation into how Massey Universitys Pollen Classifynder can accelerate the understanding of pollen and its role in nature. The Classifynder is an imaging microscopy system that can locate, image and classify slide based pollen samples. Given the laboriousness of purely manual image acquisition and identification it is vital to exploit assistive technologies like the Classifynder to enable acquisition and analysis of pollen samples. It is also vital that we understand the strengths and limitations of automated systems so that they can be used (and improved) to compliment the strengths and weaknesses of human analysts to the greatest extent possible. This article reviews some of our experiences with the Classifynder system and our exploration of alternative classifier models to enhance both accuracy and interpretability. Our experiments in the pollen analysis problem domain have been based on samples from the Australian National Universitys pollen reference collection (2,890 grains, 15 species) and images bundled with the Classifynder system (400 grains, 4 species). These samples have been represented using the Classifynder image feature set. We additionally work through a real world case study where we assess the ability of the system to determine the pollen make-up of samples of New Zealand honey. In addition to the Classifynders native neural network classifier, we have evaluated linear discriminant, support vector machine, decision tree and random forest classifiers on these data with encouraging results. Our hope is that our findings will help enhance the performance of future releases of the Classifynder and other systems for accelerating the acquisition and analysis of pollen samples.

Separating morphologically similar pollen types using basic shape features from digital images: A preliminary study(1.).

Premise of the study: One of the many advantages offered by automated palynology systems is the ability to vastly increase the number of observations made on a particular sample or samples. This is of particular benefit when attempting to fully quantify the degree of variation within or between closely related pollen types. Methods: An automated palynology system (Classifynder) has been used to further investigate the variation in pollen morphology between two New Zealand species of Myrtaceae (Leptospermum scoparium and Kunzea ericoides) that are of significance in the New Zealand honey industry. Seven geometric features extracted from automatically gathered digital images were used to characterize the range of shape and size of the two taxa, and to examine the extent of previously reported overlap in these variables. Results: Our results indicate a degree of overlap in all cases. The narrowest overlap was in measurements of maximum Feret diameter (MFD) in grains oriented in polar view. Multivariate statistical analysis using all seven factors provided the most robust discrimination between the two types. Discussion: Further work is required before this approach could be routinely applied to separating the two pollen types used in this study, most notably the development of comprehensive reference distributions for the types in question.

Modern pollen distribution and its relationship to vegetation from the south-western part of the Ganges-Brahmaputra Delta, India

ABSTRACT Thirty surface samples collected from the Pakhiralaya area, south-western Ganges-Brahmaputra Delta, India, have been analysed for pollen in order to investigate modern pollen-vegetation relationships. Data on non-pollen palynomorphs (fungal spores, dinoflagellate cysts and algal cysts) were also obtained. There is currently a paucity of modern pollen studies from this region, which places limits on the interpretation of pollen data from other types of studies, for example studies aimed at understanding the Holocene vegetation history of this globally renowned tropical forest region. The local vegetation of the area is characterised by mangrove, marsh, herb and open land. The results of our study indicate that the pollen from locally growing taxa are the major components of modern pollen assemblages. Thus, pollen spectra show close linkages between modern pollen and local vegetation. Within the samples collected from the mangrove forest, the dominant mangrove taxa (Rhizophora mucronata, Bruguiera gymnorrhiza, Excoecaria agallocha, Sonneratia and Avicennia marina) are also the most frequently encountered pollen types, along with the mangrove associate taxa Thespesia sp., Pongamia pinnata and Phoenix paludosa. Thus, undisturbed mangrove forests are characterised mainly by a dominance of true mangrove and mangrove associate species in the pollen spectra. In contrast, pollen spectra from samples from ‘open land’ areas contained significant proportions of pollen from midland taxa. Anthropogenic impact on this area is captured in the samples through pollen of introduced plants such as Eucalyptus and Casuarina equisetifolia. The present study provides a basis for useful interpretation of Late Quaternary pollen sequences from the Ganges-Brahmaputra Delta, since modern pollen studies are still meagre in the poorly investigated delta region.

Stable isotope (δD–δ18O) relationships of ice facies and glaciological structures within the mid-latitude maritime Fox Glacier, New Zealand

ABSTRACT Relationships between stable isotopes (δD–δ18O), ice facies and glacier structures have hitherto gone untested in the mid-latitude maritime glaciers of the Southern Hemisphere. Here, we present δD–δ18O values as part of a broader study of the structural glaciology of Fox Glacier, New Zealand. We analyzed 94 samples of δD–δ18O from a range of ice facies to investigate whether isotopes have potential for structural glaciological studies of a rapidly deforming glacier. The δD–δ18O measurements were aided by structural mapping and imagery from terminus time-lapse cameras. The current retreat phase was preceded by an advance of 1 km between 1984 and 2009, with the isotopic sampling and analysis undertaken at the end of that advance (2010/11). Stable isotopes from debris-bearing shear planes near the terminus, interpreted as thrust faults, are isotopically enriched compared with the surrounding ice. When plotted on co-isotopic diagrams (δD–δ18O), ice sampled from the shear planes appears to show a subtle, but distinctive isotopic signal compared with the surrounding clean ice on the lower glacier. Hence, stable isotopes (δD–δ18O) have potential within the structural glaciology field, but larger sample numbers than reported here may be required to establish isotopic contrasts between a broad range of ice facies and glacier structures.